The variability of Ae. aegypti-associated bacteria may trigger a differential influence on adult mosquito phenotypic traits such as survival and vector competence [19]. Hence, its characterization and the identification of habitat features influencing microbiome structuration is of major interest for risk assessments and vector control. In this study, we found that i) the physicochemical properties and bacterial communities of Ae. aegypti breeding sites on Guadeloupe differ substantially from those in French Guiana, (ii) the container type strongly influences the physicochemical parameters but not the microbiome at breeding sites, (iii) the microbiome associated with Ae. aegypti larvae, even if influenced by that of the breeding sites, was highly conserved in both territories, and (iv) dissolved oxygen, conductivity and metal content are strongly associated with the composition of Ae. aegypti microbiome. The characterization was based on comprehensive sampling of 161 breeding sites conducted during the dry season in both territories to ensure that containers were not rinsed by rainwater and to better observe permanent breeding sites. We found that drums were among the most common Ae. aegypti breeding sites on Guadeloupe and in French Guiana, as in neighbouring regions, such as Martinique and Suriname [44, 45, 46]. Drums are used extensively to store water because of the irregular water supply, which provides an excellent habitat for Ae. aegypti.
An appropriate aquatic habitat for Ae. aegypti is regulated by biotic and abiotic factors and their interactions. As expected, the temperature and pH were generally around 29 °C and 7.5, respectively, regardless of the type of container and the territory, as they have been shown to be favourable for larval development and survival [47, 48, 49]. Dissolved oxygen, electrical conductivity, salinity, COD, turbidity and the concentrations of Ca, Mg, Fe, Zn and Cu depended on the territory and/or the container type, the highest values being associated with tyres and plant containers. The high content of mineral and organic compounds may be due either to the small volume of containers (i.e. tyres), the presence of plants or the container material (i.e. metal), which could result in a higher ion content, turbidity, COD and less dissolved oxygen [50, 51, 52]. Our results confirm that Ae. aegypti does not breed only in clear water, as it can breed in containers with high turbidity (> 133 FNU), high DCO (> 242 mg/L) and low dissolved oxygen, nearly reaching anoxia, like Culex quinquefasciatus mosquitoes [53]. These findings are consistent with reports of successful Ae. aegypti development in septic tanks and drums, indicating tolerance of high organic pollution [54], which, in the absence of a preferred habitat, allows Ae. aegypti to extend its niche and breed in marginal habitats to maintain its population, especially during unfavourable seasons.
The differences found in the physicochemical parameters of breeding sites raise questions about their possible impact on the structure and composition of Ae. aegypti communities. Such interactions remain largely unexplored but should be considered, as habitat disturbance can alter the microbiome and thus affect adult vector phenotypic traits [55]. In agreement with other studies, we found that the predominant microbiota at Ae. aegypti breeding sites were Proteobacteria [64, 65]. Significant differences in the relative abundance of bacteria such as Pseudomonas were recorded between Guadeloupe and French Guiana. Our regression analysis revealed positive correlations with Mg, Ca and conductivity, which is consistent with the higher abundance of these bacteria on Guadeloupe as they are known to be more abundant in environments with high electrical conductivity [66]. Roseococcus, Pseudomonas and Brevundimonas are strict aerobic bacteria [66, 67, 68], which could explain the correlations obtained in our study between their abundance and higher dissolved oxygen, as seen on Guadeloupe. Curvibacter was particularly abundant in French Guiana breeding sites. These bacteria are frequently detected in iron-rich environments dominated by chemolithoautotrophic species or contaminated with toxic metals [69, 70]. In our study, Curvibacter were positively correlated with Fe, Zn and Cu and negatively with dissolved oxygen, conditions that were characteristic of French Guiana breeding sites. This finding may be due to the types of containers sampled in this locality – abandoned, rusty freezers, which may have increased metal concentrations in stagnant water, as seen elsewhere [71]. Conversely, the container type did not affect the structure of the bacterial communities or the water or larvae at breeding sites.
By contrast to water samples, no major differences in species composition were observed in the microbiota associated with Ae. aegypti larvae between Guadeloupe and French Guiana (Fig. 4A). This is presumably due to host selection for microbial communities that can colonize the larval gut environment, which is consistent with the lower alpha diversity recorded in larvae when compared to water samples. The shared microbiome of Ae. aegypti larvae consisted of abundant genera such as Herbiconiux, Bosea, Bacillus and Kaitsia, which were more prevalent in the midgut than in the breeding sites. Herbiconiux can degrade the cellulose and xylan found in the gut of some insects, which may explain the abundance of this ubiquitous genus in larval samples [72, 73]. The three other genera (Bosea, Bacillus and Kaitsia) may be part of the core microbiome of mosquitoes, as they have also been identified in the midgut of other Culex and Aedes species [74, 75]. The role of these bacteria in gut vector biology may have led to their evolutionary conservation; however, little information is available on the functional roles of these bacteria in mosquito hosts.
Differences in the microbiota associated with Ae. aegypti larvae and those in breeding sites are also due to the relative abundance of bacteria the development of which is constrained by the larval environment. In our study, Polynucleobacter and Emticicia bacteria were significantly more abundant in water samples from Guadeloupe than from French Guiana, while their relative abundance was significantly lower in larval samples from both territories. As suggested by Coon and colleagues [76], the gut hypoxia, which is probably a cue for growth and moulting of larval mosquitoes, may constrain the growth of such aerobic bacteria. Nevertheless, how insects “select” and acquire from breeding sites the microorganisms that become part of their semi-stable microbiota is still not well understood.
Finally, we observed territory-specific differences in the relative abundance of certain bacteria genera in Ae. aegypti larvae (i.e. Methylobacterium, Xanthobacter, Roseoccocus, Microbacterium, Microvirga, Pseudomonas) that were consistent with the differences between breeding sites on Guadeloupe and in French Guiana. These findings confirm that the larval microbiota are significantly affected by the microbial communities in water at breeding sites [18, 19]. The influence of breeding sites on the mosquito microbiome raises questions about the transmission of pathogens. We found that Pseudomonas was more abundant at breeding sites and larvae on Guadeloupe than in French Guiana, whereas Serratia was more abundant in water and larvae from French Guiana. It has been found that Serratia marcescens increases the susceptibility of Ae. aegypti females for dengue virus, while Pseudomonas rhodesiae can inhibit La Crosse virus replication in Ae. albopictus cells [77, 78]. In those studies, the mosquitoes were reared in laboratory conditions, as is the case in most estimates of vector competence, and it is unknown whether the influence of Serratia and Pseudomonas spp. on pathogen transmission by mosquitoes is maintained in natural conditions. Laboratory studies have also shown that Ae. aegypti from Guadeloupe and French Guiana have similar vector competence for arboviruses such as Zika and chikungunya [79, 80]; however, as larval exposure to bacteria can alter the vectorial capacity of this species [19], the differences we observed between Guadeloupe and French Guiana in terms of microbiota from breeding sites and larvae may result in differences in the vectorial capacity of the corresponding Ae. aegypti populations.