Study sites

Samples of surface microlayer water and larval samples were collected from two sites in Southern Ghana: Anyakpor (5°46′51.96″N, 0°35′12.84″E) in the Ada East District and Dodowa (5° 52’ 58.3212’‘ N 0° 5’ 52.9548’‘ W) . The vegetation in Anyakpor is similar to that of the coastal savannah, and a majority of the population grows vegetables on raised beds with irrigated fields with irrigation systems that could serve as Anopheles mosquito breeding grounds.

However, Dodowa has a secondary forest type vegetation with little original virgin forest left as a result of deforestation that has created potential breeding habitats for mosquitoes. A total number of 59 mosquito larvae breeding habitats were identified within the study sites, with 18 in Dodowa (the savannah-forest transition area) and 42 in Ada (the coastal savannah area). These habitats were mapped using a Global Positioning System (GPS) and monitored over the 24 weeks (from October, 2020 to May, 2021) to record the stability and productivity. The habitats were surveyed once every two weeks for the presence of aquatic stages of Anopheline and to generate stage-specific estimates of larval densities as described previously [31].

Mosquito larvae density estimation

At the time of collection of environmental samples, the density of Anopheles larvae was measured. Larvae were collected by standard dipping technique, using three dips with a 300 mL dipper of 9 cm diameter. The number of larvae was counted, and representative larvae were brought to the laboratory. In this study, we define a productive and non-productive habitat as a habitat with persistent presence and absence of anopheline larvae, and semi-productive habitat as habitats that switched between productive and non-productive habitats in no order at the time of habitat visitation and monitoring [31].

Collection of larvae and water samples

From each productivity habitat type (productive, semi productive, and non-productive), water samples were collected directly into sterile-capped disposable centrifuge tubes. Water was vigorously mixed prior to sampling to ensure that biofilms were included. In addition, representative sample of 1^st, 2^nd, 3^rd, and 4^th instar larvae (or all if less than ten) was collected from each mosquito-positive breeding habitat and transferred to sterile 15 ml Eppendorf tubes. All the samples were kept on ice and brought to the Medical Microbiology laboratory at the University of Ghana within 4 h of collection.

Processing of larval and water samples

In the laboratory, each water sample was filtered through a cellulose nitrate membrane filter (0.22 μm pore size, 47 mm dia., Sartorius Stedium®) using aseptic vacuum filter units (Millipore®). Larval samples were surface sterilized; first rinsed in 70% ethanol, then suspended in 90% ethanol and agitated with a vortex mixer for about 10 seconds, and finally rinsed with sterile, DNA-free water. Immediately after filtration and larval surface sterilization, the samples were stored at -80^◦C.

Genomic DNA extraction

Samples were thawed at room temperature before DNA extraction, and each membrane filter was cut into small pieces and placed into sterile 1.5 ml Eppendorf tubes for DNA extraction. Bacterial genomic DNA was extracted from sterilized Anopheles larvae and bacterial cells retained on each filter membrane using the ZR Fungal/Bacterial/Algal DNA MicroPrep™ kit following the manufacturer’s instructions. Bacterial DNA extracted from the larvae and water samples were quantified with Nanodrop One™ (Thermo Fisher Scientific, USA) and stored at -20^oC for subsequent PCR.

PCR amplification

The full length of the 16S rRNA gene was amplified using the universal primers 27F 5’- AGAGTTTGATCCTGGCTCAG-3’ and 1492R 5’-GGTTACCTTGTTACGACTT-3’ [32,33], which yielded PCR fragments of approximately 1,500 bp in size. The PCR was set up with the NEB OneTaq 2X master mix (New England Biolabs, USA), 0.2 μM of each primer and 3 μl DNA template. Reaction conditions were as follows; 94˚C for 10 mins followed by 35 cycles of 94˚C for 30 secs., 58˚C for 1 min., 68˚C for 1min 30 sec. and a final extension at 68˚C for 5 mins. The amplification was done in triplicate and pooled for each sample. PCR products for each sample were examined on an agarose gel and cleaned using the QIAquick® PCR Purification Kit (QIAGEN GmbH, GERMANY). Purified PCR products were checked for purity using the NanoDropOne spectrophotometer (ThermoFisher Scientific, USA) and quantified with the Qubit dsDNA HS Assay Kit and read on the Qubit 4 Fluorometer (ThermoFisher Scientific, USA).

Sequencing library preparation and sequencing

The sequencing library was prepared using the Native Barcoding Expansion 96 (EXP-NBD196) with the Ligation Sequencing Kit (SQK-LSK109) (Oxford Nanopore Technologies, UK) following the manufacturer’s instructions. Prepared libraries were pooled in equimolar amounts into a single library. The pooled library was sequenced using the MinION Mk1C device (Oxford Nanopore Technologies, UK) and a MinION R9.4.1 SpotON flow cell device according to the manufacturer’s instructions. Each run was set up for 72 hours.

Sequence data processing

Sequencing reads (.fast5) were basecalled, demultiplexed, and trimmed of adapters and barcodes to.fastq using Guppy (Oxford Nanopore Technologies, UK). Quality control of the sequenced data was carried out with FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) through the Galaxy server. A quality score of 7 was given for each barcoded read. The minimum length filter was set at 200, and the maximum length filter was set at 1500. Minimum coverage was increased from the default level of 30 to 50. Other parameters that were set included a minimum identity percentage of 80% and a maximum target sequence of 3.

Data analysis

The taxonomy classification was performed g in Galaxy using Kraken2 against the PlusPF database (version: 2022-06-07) with a confidence threshold of 0.7 and a minimum hit group of 2. Output was clustered into operational taxonomic units (ASVs) at kingdom, phylum, class, order, family, and genus levels. The generated Kraken2 report file was used with sample metadata to generate the biological observation matrix (BIOM) file. All downstream analyses were performed in R (RStudio 203.12.0 build 369), using the phyloseq package. Visualization of bacterial composition abundances among samples, sampling sites, and larvae stages was done using bar charts showing the distribution of bacterial classes. Bar charts were created using the “ggplot2” package, and where required, a Venn diagram was generated using the “ggvenn” package in R. Venn diagrams were used to visualize shared ASVs when comparing between samples, sampling locations, habitats, and larvae stages. Alpha diversity was estimated using the observed species richness, Shannon diversity index (H) and chao richness. At various levels we compared the alpha diversity indices between groups with a one-way analysis of variance (ANOVA) followed by the Kruskal-Wallis test (followed by Dunn’s post hoc test if a significance difference were detected). Alpha diversity plots and statistical analysis were performed in R. Beta diversity between samples was estimated using the Bray-Curtis distance matrix method. Visualization for overall differences between bacterial communities’ structure between habitats types (Non-productive, Productive, and Semi-productive) and larvae stages (stage 1, stage 2, stage 3 and stage 4) was performed using nonmetric multidimensional scaling (NMDS). Permutational multivariate analysis of variance (PERMANOVA) and an analysis of similarity (ANOSIM) were used to determine significant differences between subgroups.

Bacterial profiles and diversity across different sample types

To compare samples across habitats reads from samples collected from water or larvae from specific habitats were merged into 5 specific sample types (Productive MF [PRO_MF], Productive Larvae [PRO_LAE], Semi-Productive water [SEMIPRO_MF], Semi-Productive larvae [SEMIPRO_LAE], and Non-Productive water [NONPRO_MF]). Using a 0.5% cut-off of relative abundance, a total of 15 taxa were identified at the class level across all habitats and larvae. The most abundant classes across the habitats were Gammaproteobacteria, Betaproteobacteria and Epsilonproteobacteria. For the non-productive habitat, the dominant class was Epsilonproteobacteria (58.1%), followed by Cytophagia (22.4%), the least occurring classes were Bacilli (0.65%) and Clostridia (0.63%) (Fig 1A). In the productive larvae, the predominant taxa were Gammaproteobacteria (33.2%) and Betaproteobacteria (30.5%) while the lowest taxa were Chitinophagia (0.81%) and Actinomycetia (0.64%). In the productive water habitat Betaproteobacteria dominated with 52.4% and was followed by Cytophagia with 18.6% (Fig 1A). The least occurring taxa in this habitat was the Oligoflexia (0.57%) and Baccili (0.55%). In the semi-productive larvae, the predominant taxa were Gammaproteobacteria (60.3%) and Betaproteobacteria (12.75%), while the lowest taxa were Alphaproteobacteria (1.48%) and Tissierella (0.95%). In the semi-productive water habitat, the predominant taxa were Betaproteobacteria (72.7%) and Gammaproteobacteria (18.0%) while the lowest taxa were Flavobacteria (0.86%) and Alphaproteobacteria (0.51%) (Fig 1A).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Bacterial profiles and diversity across different sample types.

Bacterial community composition in An. gambiae larvae and water collected from breeding sites (Ada Foah and Dodowa) classified based on productivity of the habitats (Productive, semi-productive and non-productive) at the class taxonomy level. Only classes making up > 0.5% in any group of samples were included. Other classes present are clustered as “Class 0.5 abend” together with unknown classes (B) Mean alpha diversity metrics calculated within the various sample groups by (i) Observed species richness, (ii) Chao1, (iii) Shannon diversity index, (C) Four-way Venn diagram depicting ASVs numbers that overlap and do not overlap between various breeding habitat groups, (D) NMDS ordination plot based on Bray-Curtis distances and (E) NMDS ordination plot based on ANOSIM test. Each sample type is indicated by a different colour.

https://doi.org/10.1371/journal.pone.0323464.g001

Each habitat type was characterised by unique bacterial composition. The most dominant class in the non-productive habitat was the Epsilonproteobacteria but was present in lower amounts in the productive and semi-productive habitats (Fig 1A). Interestingly, the bacterial composition in the water habitat mirrored what occurred in the Anopheles larvae. In the productive water and larvae, the Betaproteobacteria was a dominant taxon while in the semi-productive water and habitat Gammaproteobacteria was the dominant taxa. Cytophagia, Clostridia, and Bacilli were the classes identified between the productive and semi-productive habitats (Fig 1A). The Actinomtcetia was identified exclusively in the productive larvae, while Tissierellia was present in both productive and semi-productive larvae (Fig 1A). Furthermore, the most diverse bacteria was found in the productive samples, whereas fewer bacteria were found in the non-productive samples (Fig 1A, S1-S2 Figs).

Alpha diversity was estimated to determine the richness and evenness within the different habitat types (Fig 1B). Using the observed taxa (richness) measure at the class level, a statistically significant difference in taxa was observed (p = 0.0491, S2 Table). Water samples from the non-productive habitat had the highest mean observed diversity (959), followed by the water (919) and larvae (950) samples from the productive habitat. The least observed class taxa were the water sample (353) from the semi-productive habitat (Fig 1B [Observed]). Shannon and Chao1 alpha diversity indexes were not statistically significant (p = 0.6083 and p = 0.1202 respectively) (Fig 1B). Using the Shannon’s diversity index as measure for both richness and evenness, the water from the productive habitat had the highest mean index of 3.28, followed by the larvae samples from the productive habitat (3.15) (Fig 1B) [Shannon]).

Using a four-way Venn diagram, we compared the number of amplicon sequence variants (ASVs) that overlap across the habitats. The analyses revealed that 23 ASVs (53.1%) were shared between all the habitats (Fig 1C). Among these common taxa are Actimomycetia, Alphaproteobacteria, Gammaproteobacteria, Clostridia and Deinococci. The productive water habitat had the highest number of unique ASVs (7 [14.3%]), and 6 ASVs (12.2%) were shared exclusively between the productive water and larvae samples (Fig 1C, S3 Table). The non-productive water and the semi-productive larvae samples did not have any exclusive taxa. Four-way Venn diagram depicting ASVs that overlap and do not overlap between various breeding habitats at the genus and species taxonomic level is presented in S3 Table.

The composition and structure of bacterial communities (Beta diversity) between the various habitats and in the various larvae were evaluated with Bray–Curtis’s dissimilarity matrix and visualized with a nonmetric multidimensional (NMDS) ordination plot (Figs 2D and 2E). Samples did not show any clear clustering pattern either on type (non-productive, semi-productive, or productive) or on description (larval stage 1, stage 2, stage 3 or stage 4) (Fig 1E). A PERMANOVA analysis showed significant differences between bacterial profiles of the samples (Description: R²= 0.246, P = 0.014 and Type: R²= 0.072, P = 0.324). The ANOSIM test showed that there were statistical differences between mosquito larvae stages (R = 0.28, P = 0.005) but not for different habitat types (R = -0.08, P = 0.853).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Bacterial community composition in An. gambiae breeding habitats and larvae in Ada Foah and Dodowa.

Bacterial composition in An. gambiae water and larval samples collected from different breeding sites (Ada Foah and Dodowa) classified based on productive (Productive, semi-productive and non-productive) at class level (B) Dominant bacterial community composition at family level for Gammaproteobacteria. Only family making up > 0.1% in any sample are named, other family present are clustered as “Other” together with unknown the family (C) Dominant bacterial community composition at family level for Betaproteobacteria.

https://doi.org/10.1371/journal.pone.0323464.g002

Comparison of bacterial profiles and diversity between sites.

To determine the similarities and differences between the two sampling sites, samples were compared in different locations (Ada Foah or Dodowa), and the specimen types (water or larvae). This generated 4 sample types: Ada Productive Water, Ada Productive Larvae, Dodowa Productive Water and Dodowa Productive Larvae samples. Using a 0.5% cut off relative abundance, a total of 15 taxa were identified at the class level across all habitats. The most abundant class ASV across all the productive habitats was the Betaproteobacteria (Fig 2A), with an average of 51.2%. The common ASVs classes in all habitat groups were Bacilli, Alphaproteobacteria, Betaproteobacteria, Cytophagia, Flavobacteria, Gammaproteobacteria and Mammalia (Fig 2A). The Betaproteobacteria (62.2% and 54.8%) and Cytophagia (8.89% and 12.5%) were the dominant taxa in the water samples from the two sites (Ada Foah and Dodowa), while the Betaproteobacteria (59.9% and 35.9%) and Gammaprobacteria (17.9% and 25.4%) were the dominant classes in the larvae samples (Fig 2A). Actinomycetia were only found exclusively in the water samples from both sites, while Oligoflexia, Opitutae, and Saprospiria were only found exclusively in water samples collected from Ada Foah (Fig 2A). Collectively, the bacterial communities were much similar between water samples and larval samples.

The two dominant classes (Betaproteobacteria and Gammaproteobacteria) for the predominant bacteria were presented by filtering orders with greater than 5% abundance (Figs 3B and 3C). For the Gammaproteobacteria, the order Enterobacterales was the dominant order, and this occurred across all sample’s groups, and in the Betaproteobacteria, the order Burkholderiales was the dominant order.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Bacterial profiles and diversity across habitat types and larval stages (L1, L2, L3 and L4) in Ada Foah and Dodowa.

Bacterial community composition in An. gambiae larval stages collected from (A) Ada Foah site classified at class level and (B) Dodowa site classified at class level. Mean alpha diversity metrics calculated within the various sample groups by (i) Observed species richness ii) Chaol (iii) Shannon diversity index in (C) Ada Foah site (D) Dodowa site. Four-way Venn diagram depicting ASVs numbers that overlap and do not overlap between different larval stages (L1, L2, L3 and L4) in (E) Ada Foah (F) Dodowa.

https://doi.org/10.1371/journal.pone.0323464.g003

Comparison of bacterial profiles and diversity between mosquito larval stages

To determine the similarities and differences between the different larval stages, we compared the bacterial abundances, and diversities in the different larvae stages (L1, L2, L3 and L4) from the two study sites. In general, the bacterial taxa identified in larvae stages closely mirrored bacterial taxa in the water from which they were collected at the different sites (Figs 3A and 3B). For the different larval stages in samples from Ada Foah (Fig 3A), a total of 15 taxa were identified at the class level. Betaproteobacteria and Gammaproteobacteria were the predominant class across all the larvae stages in the habitats. Several bacterial classes were identified in all larval stages, particularly Cytophagia, Bacilli, Flavobacteria, and Epsilonproteobacteria. These bacterial taxa occurred across all larval stages but differed in abundance between larvae stages. For the different larval stages in samples from Dodowa (Fig 3B), a total of 14 taxa were identified at the class level. The Betaproteobacteria, Epsilonproteobacteria and Gammaproteobacteria were the predominant classes across all the larvae stages except for the L4 stage. In the L4 stage larvae from the Dodowa habitat the predominant bacterial class was the Episilonproteobacteria, but Gammaproteobacteria were not identified. Bacterial classes such as Bacilli, Alphaproteobacteria, Flavobacteria, and Cytophagia were also common across different larvae stages in different abundance (Fig 3B). In comparing the larval stages in the different locations, a distinct observation was the presence of Actinomycetia in the water samples in both locations but absent in the various larvae stages. The class Verrucomicrobiae was present in the L4 stage in Ada Foah samples but absent across all larvae stages in the Dodowa samples. In a similar manner, Nitrospra was identified in the L3 stage in Dodowa samples but absent from the Ada Foah samples (Figs 3A and 3B).

The alpha diversity estimates for the larval stages revealed the richness index showed a statistically significant difference (P = 0.0364, S4 Table), with the highest mean index being the larval stage 2 (336) and the lowest being larval stage 1 (240) (Fig 3C, S4 Table). The Chao1 and Shannon indexes were not significant (p = 0.188 and p = 0.101). For Chao1 the pattern was consistent with the Observed index, with the highest mean index in the larval stage 2 (542) and the larval stage 3 being the lowest (465) (Fig 3C [Chao1]). For the Shannon index, the highest mean was larval stage 3 (3.14) and the lowest was larval stage 1 (2.15) (Fig 3D [Shannon]). All alpha diversity estimates for the larval stages from Dodowa were not statistically significant (Observed: P = 0.216, Chao1: P = 0.243, and Shannon: P = 0.215) (Fig 3D). For the Observed diversity, the highest mean index was in larvae stage 3 (506) and the lowest in larval stage 4 (274) (Fig 3D[Observed]). For Chao1, the pattern was consistent with the Observed index, with the highest mean index in the larvae stage 3 (723) and the lowest mean index in larvae stage 4 (471) (Fig 3D [Chao1]). Similarly, the highest mean Shannon index was in larvae stage 3 (3.11), and the lowest was in larval stage 4 (2.23) (Fig 3D [Shannon]).

A four-way Venn diagram was generated to compare the number of ASVs that overlap between the larval stages (Fig 3E and 3F). For larval stages from Ada Foah, there were 18 ASVs (41.9%) overlap across all stages, with the largest similarity occurring between the L3 and L4 stages (29 ASVs (67.5%) (Fig 3E, S5 Table). Except for larvae stage 1, all other stages had distinctly associated ASVs. For both larvae stages 2 and stage 3, the only unique taxa identified were Thermoleophilia and Coriobacteriia respectively. Larvae stage 4 had the highest number of unique ASVs (7) these were Blastocatellia Candidatus, Brocadiia Candidatus, Saccharimonia Candidatus, Thermofonsia, Chlorobia, Thermodesulfovibrionia, and Vicinamibacteria (Fig 3E). For larvae from Dodowa, there were 21 ASVs (51.2%) that overlapped across all stages; as in the Dodowa samples, the largest similarity occurred between the L3 and L4 stages (33 ASVs = 80.5%) (Fig 3F). There were no unique ASVs associated with larvae stages 1 and 2. Larvae stage 3 had 3 unique ASVs (7.3%) (Acidimicrobiia, Chlorobia, and Chrysiogenetes) while larval stage 4 had 4 unique ASVs (9.8%) and these were Blastocatellia, Candidatus, Saccharimonia, Pisoniviricetes, and Thermoleophilia (Fig 3F, S5 Table). The Four-way venn diagram depicting ASVs that overlap and do not overlap between various larvae in the different sites at the genus and species taxonomic level is presented in S5 Table.

Abundance of bacterial in productive habitats

Information about bacteria that either enhance or hamper the survival of mosquito larvae has the potential to be useful in different aspects of mosquito vector control. In this study, the bacterial community composition of Anopheles mosquitoes breeding waters in Ghana was evaluated using 16S rRNA gene amplicon sequencing. In this study, productive habitats and their larvae from all the sites had similar bacterial composition with Gammaproteobacteria and Betaproteobacteria dominating. The two dominant classes in the productive habitats are consistent with previous studies by Nilsson et al., [28] in Latin America with Anopheles darlingi breeding habitats and in Cameroon with An. coluzzii and An. gambiae breeding waters by Buck et. al., [34]. Similarly, Dada et al., [7] have reported the dominance of Gammaproteobacteria, and Betaproteobacteria in domestic water storage containers containing with Aedes aegypti in Thailand and Laos. However, in semi-natural Anopheles gambiae breeding sites in Kenya, Wang et al., [12] reported Alphaproteobacteria and Cyanobacteria were abundant in the domestic water-storage containers with Anopheles and Aedes, and in India, Nilsson et al., [27] detected Alphaproteobacteria and Betaproteobacteria were abundant in potential Aedes and Anopheles breeding sites. In eastern Puerto Rico, Caragata, et al., [35] detected the abundance of Alphaproteobacteria, Deltaproteobacteria, and Epsilonproteobacteria in Aedes aegypti associated breeding habitats water, and in the USA, Njoroge et al., [36] detected Gammaproteobacteria (Enterobacteriaceae) and Epsilonproteobacteria (Arcobacter) in water samples with Culex mosquito larvae, suggesting that there are some class-specific differences in bacteria that are associated with different species of mosquitoes positive breeding habitats.

Prevalence of bacterial in non-productive habitats

The selection of larval breeding habitats is a critical stage in the life history of gravid mosquitoes because it can ultimately influence the fitness of progeny, distribution of larvae, and larval population dynamics [37]. Selection of breeding habitat sites by gravid mosquitoes can be stimulated by the presences of mixture(s) of bacteria [9,10] and odours emitted from the living microorganisms in soil and water in the habitats [11]. In this study, Epsilonproteobacteria and Actinomycetia were abundant in habitats where mosquito larvae were absent (non-productive). In contrast to this study, Nilsson et al., [27] found that in the absence of mosquito larvae, Acidimicrobiia, Sphingobacteria and Bacilli were significantly abundant, and the family Flavobacterium with Rhodobacter was detected in mosquito-negative waters by Dada et al., [7]. Epsilonproteobacteria are currently considered to be exclusively hydrogen-oxidizing bacteria [38] and Actinobacteria is a class of bacteria that contains nonfilamentous forms of bacteria that produce a large number of biologically active secondary metabolites [39]. Metabolites from Actinobacteria have been shown to have strong larvicidal activity against larvae of Anopheles, Aedes [39,40], and Culex [41] mosquitoes. The finding of Epsilonproteobacteria and Actinomycetia being in waters devoid of mosquitoes could suggest that there are potential candidates that could be explored for repelling the mosquitoes from ovisiting in potential breeding habitats. These groups of bacteria, and the environmental conditions they reflect, should be further investigated for vector control purposes.

Bacterial diversity and abundance

There were slight differences in bacterial richness within the water and larval samples studied at the different sites. The Observed (richness) showed that the microbiota of breeding habitats were richer compared to the larvae they contain (Fig 1B). This suggests that the sampling site environment is a key determinant of the bacterial profiles in mosquitoes’ larvae and habitats. Compared to Dodowa, Ada Foah breeding sites were located in farming communities, and the exposure of the bacterial communities in the habitats to pesticides can have some influence on bacterial richness in the aquatic environments [42], partly due to the pesticides suppressing the bacteria in habitats, thus favouring the growth and enrichment of more bacterial species that can degrade the pesticides in the larvae and enable their survival. The importance of the environmental impact of pesticides in shaping the bacterial communities of juvenile mosquitoes and their environment has been documented by Juma et al., [43]. Based on findings from this study, there is the need for future studies to evaluate mosquito larvae microbial responses to exposures to pesticides and their possible implications for mosquito ecology and vector control.

Location-driven variability in bacterial composition may arise in various breeding habitats; however, studies have shown that the presence of specific bacterial groups correlates with the presence of mosquito larvae [7,27,44]. Altogether, our investigation showed different bacterial profiles for the different larval stages; however, a fundamental group of bacterial classes predominates at the different stages, with changing abundances. In line with Mosquera et al., [45] study, Gammaproteobacteria and Betaproteobacteria were the main ASVs linked to Anopheles larvae. However, their variation in proportion during the development of the larvae likely reflects that there might be some selection induced by the alkaline environment in the larval gut due to the presence of the microorganisms [46].