Discussion
In both La Lopé and Rabai, our study found that mosquito larval breeding sites in the forest and villages (including peridomestic and domestic sites) had different physical and biological characteristics, though this between-habitat contrast varies among variables. Notably, bacterial community composition showed clear and consistent difference between habitats in both localities. Despite this environmental difference, behavioral investigations suggested that Ae. aegypti in the forest readily accepted artificial containers as oviposition and larval breeding sites. Aedes aegypti colonies derived from the forest and villages also showed similar weak oviposition preferences in the lab. These results are consistent with the hypothesis that Ae. aegypti are generalists in larval breeding site choice. This hypothesis was also supported by the indistinguishable conditions between Ae. aegypti present and absent larval breeding sites within each habitat, suggesting that the mosquitoes were likely not selective and can accept a wide range of larval habitats. Lastly, oviposition choices in the laboratory were highly heterogeneous, consistent with a lack of strong preference.
Being versatile in larval habitat allows Ae. aegypti to take advantage of novel artificial containers when natural breeding sites are scarce. This has been proposed as a key driver for this mosquito to move into domestic habitats in the first place (Brown et al., 2014; Powell et al., 2018; Rose et al., 2020). Consistent with this hypothesis, movement between habitats was suggested by genetic studies showing little genetic differentiation between forest and village Ae. aegyptipopulations in La Lopé and Rabai (Kotsakiozi et al., 2018; Paupy et al., 2014; Rose et al., 2020; Xia et al., 2020). It should be noted that this genetic similarity in Rabai, as well as the lack of behavioral difference observed in this study and Rose et al. 2020, contrasts to studies in the 1970s and 2009 where Rabai forest and village Ae. aegypti showed significant genetic differences, feeding preference difference, and ovipositional difference (Brown et al., 2011; McBride et al., 2014; Petersen, 1977; Tabachnick et al., 1979). These strong differences found before 2017 resulted from the introduction of non-African Aaa to Rabai villages (Brown et al., 2011; Gloria‐Soria et al., 2016). The exotic Aaa population was no longer found during our fieldwork in 2017, and the village mosquito population was likely originated from the local forest (i.e.,Aaf ).
Once the mosquito established themselves in the novel habitat (likely moving from forest to domestic habitat), ecological divergence could take place. However, this study was unable to detect evidence of consistent ovipositional divergence by the laboratory oviposition experiments. One possibility is that the habitat shifts in La Lopé and Rabai happened recently and that the extensive connectivity in the local scale between habitats may hinder phenotypic divergence. Between more distant localities when gene flow is less frequent, there may be more differences between mosquitoes from different habitats or places with different human population density, as found for host preference (Rose et al., 2020). A third possibility is that ecological divergence may happen in the immature stages, e.g., egg, larvae, and pupae (Saul et al., 1980). This study did not examine larval performance, but future investigations comparing eggs hatching and larval development in different water conditions that mimic either forest or village larval breeding sites could be insightful in this regard. For example, microbial density was significantly lower in Rabai domestic containers, which might pose selection pressure on larval starvation resistance (Barrera & Medialdea, 1996; Souza et al., 2019), leading to higher resistance in the domestic population.
In addition to these plausible ecological and evolutionary considerations, we cannot rule out the possibility that our laboratory oviposition experiments lacked the power to detect oviposition preference or differences between colonies, although the two-choice or multi-choice assays have been used widely to investigate Ae. aegypti oviposition preference (but see Singer (2004) for more discussion on measuring preference). Colonies may have also lost distinctive traits due to adaptation to laboratory conditions (Hoffmann & Ross, 2018). Moreover, the design of using five females per cage instead of one female might introduce some unknown complexity, for instance, interference between individuals (Allan & Kline, 1998). Lastly, the contrast of oviposition choices might not be of a magnitude detectable by female Ae. aegypti . However, the choices used in this study were informed by characteristics of natural oviposition sites, and therefore should be ecologically relevant for the mosquitoes. A recent study using the same Ae. aegypti colonies did find between-habitat ovipositional difference towards more extreme but unnatural conditions (Xia, 2021). The complexities regarding Ae. aegypti oviposition and experimental design warrant future studies to examine more environmental conditions or combinations of multiple variables, applying multiple preference measurements, and use younger colonies in a more natural setting (e.g. conducting choices assays in the field with mosquitoes collected from larval breeding sites).
Besides adding to our understanding of the domestication history ofAe. aegypti , this study also provided the first detailed physical and biological characterization of Ae. aegypti larval breeding sites, at least in Africa. Dickson et al. (2017) described the bacterial community composition in larval breeding sites in La Lopé and found a strong difference between habitats, echoed in our study. Yee et al. (2012) found consistent differences between tree holes and tires in the U.S. although Ae. aegypti were not present in most containers. While our work provides useful baseline information for future studies onAe. aegypti ecology and behavior, we acknowledge that some caveats still exist in our field sampling, so the results should be interpreted with caution. For example, the absence of Ae. aegyptiin a larval breeding site did not necessarily reflect avoidance by the female nor that it is inhospitable for the larvae, especially as we could not inspect the existence of unhatched eggs. However, this does not affect our speculation that Ae. aegypti are not selective about larval breeding sites, as the Ae. aegypti present and absent sites had similar range of variation. We also only characterized larval breeding sites in a narrow temporal window during the rainy season. Future studies examining larval breeding sites throughout the year would be particularly relevant to the recent work suggesting the importance of seasonality in driving the domestication of Ae. aegypti (Rose et al., 2020). Furthermore, the sample sizes in our field study were relatively small. Field studies with larger sample sizes and more balanced sampling between different habitats and larval breeding site groups could further validate this study’s results. We also grouped tree holes and rock pools as “natural” containers due to the limitation of sample sizes, yet previous studies have implied that they could be two distinct larval habitats (Soghigian et al., 2017). However, our preliminary analysis suggested that grouping or separating them did not affect the main findings from the field data. We also need to acknowledge that the chemical profiles of larval breeding sites in Rabai reported in this study were probably not complete and therefore calls for future studies with improved sample collection and analysis techniques. Lastly, in addition to the condition of each individual larval breeding sites, the local context could also be important, e.g., vegetation around the sites (Rey & O’Connell, 2014).
In summary, this study suggested that Ae. aegypti in Africa were likely generalists in their larval habitat choice, which allowed them to readily accept artificial containers as larval breeding sites and potentially facilitated their introduction into domestic habitats. Being flexible in oviposition and larval breeding site choices could benefitAe. aegypti by spreading the risk during reproduction and reduce larval competition. This is consistent with the observations that this mosquito has a bet-hedging ‘skip oviposition’ behavior (i.e., lay small batches of eggs in multiple containers) (Colton et al., 2003; Starrfelt & Kokko, 2012). However, outside of Africa, Ae. aegypti are closely associated with human communities and use almost exclusively artificial containers for larval breeding sites (Day, 2016; Swan et al., 2018; Vezzani, 2007; Yee, 2008), raising the interesting question of when and how this specialization on artificial containers evolved. A few recent studies suggested that human specialization may happen somewhere in West Africa, such as Sahel or Angola (Crawford et al., 2017; Powell et al., 2018; Rose et al., 2020). On the other hand, the human-specialized non-African Aaa could also move back to ancestral breeding sites, for instance, in the Caribbean (Chadee et al., 1998). It would be interesting to examine such processes and test whether the mosquitoes resumed generalist in larval breeding site choice during this process.