Field study
We conducted field studies in La Lopé, Gabon in Central Africa from November to December 2016, and in Rabai, Kenya in East Africa from April to May 2017. The period of field study overlapped with the rainy season in each locality during which mosquitoes had a large population size. La Lopé has an extensive continuous tropical rainforest surrounding La Lopé village (Figure 1a). The forest in Rabai, on the other hand, is more fragmented, with several villages scattered around the forest patch (Figure 1b). In each location, we searched for water-holding containers as potential mosquito larval breeding sites in both the forests and nearby villages. A potential larval breeding site was defined as a container holding at least one mosquito larvae (not necessarily Ae. aegypti ) at the time of sampling, which suggested that the site had been present long enough for a mosquito to lay eggs. We categorized larval breeding sites into three habitat groups: forest, peridomestic (outdoor containers in a village area), and domestic (indoor containers) (Table 1), according to their locations. We separated indoor and outdoor containers because classical studies from the 1970s reported that, at least in Rabai, Kenya, Ae. aegypti living indoor and outdoor showed distinct behavioral and genetic differences (Leahy et al., 1978; McBride et al., 2014; Petersen, 1977; Tabachnick et al., 1979; Trpis & Hausermann, 1975). Genetic analysis showed that these indoor mosquitoes in Rabai were likely descendent of non-African Aaa (Brown et al., 2011; Gloria‐Soria et al., 2016). However, this previously describedAaa -like indoor form was no longer found during our field sampling (Rose et al., 2020; Xia et al., 2020).
In La Lopé, we visited 60 larval breeding sites in seven forest locations and 38 sites in six village locations (Figure 1a, one village location, Kazamabika village, is further away from the other village locations). The sampling locations separate by 5-17 km. Forest larval breeding sites were predominantly rock pools (n=49) around streams and tree holes accumulating rainwater (n=11). Previous studies have considered tree holes and rock pools as distinct mosquito larval habitat groups (Soghigian et al., 2017). However, we only found a few tree holes with complete data (n < 6 in all analyses), and comparing between tree holes and rock pools is beyond the scope of this study. Therefore, we grouped them as ‘natural containers.’ In the village, mosquito larvae were found in a variety of artificial containers, including construction bricks, tires, metal cans, and plastic containers. Because residents in the village rarely store water indoors, all village larval breeding sites were ‘peridomestic.’ In Rabai, Kenya, we sampled 31 larval breeding sites consisting of mainly plastic buckets, earthenware pots, and metal barrels in four villages. They were mostly indoor (i.e., ‘domestic’) containers. The 37 larval breeding sites in the Rabai forest were all tree holes holding rainwater (Figure 1b). We recorded the GPS coordinates of each sampling location (consist of multiple larval breeding sites) in La Lopé, and of each larval breeding site in Rabai, Kenya (Figure 1).
Upon identifying a potential larval breeding site in any habitat, we measured 11-16 physical variables and collected water samples to analyze bacterial and volatile profiles. Sample sizes for each category of environmental variables were summarized in Table 1. Method details are described in the following sections and the Appendix. We also collected all mosquito larvae and pupae using pipets and reared them to adults in field stations, keeping collections from different larval breeding sites separate. Upon eclosion, adults were identified to genus and species based on taxonomic keys using a dissection microscope in the field (Rueda, 2004). We kept Ae. aegypti adults alive to establish lab colonies for later behavioral experiments.
We categorized each larval breeding site (i.e., container) as ‘Ae. aegypti present’ or ‘Ae. aegypti absent’ based on whether it held any Ae. aegypti larvae or pupae (Table 1). It is worth noting that the absence of Ae. aegypti did not necessarily suggest an avoidance. Some sites may be suitable for oviposition and larval development but not yet colonized by Ae. aegypti , and we also could not observe unhatched eggs. Bearing this potential caveat, we combined the three habitat categories and the two Ae. aegyptipresence status to generate six ‘larval breeding site groups’ (Table 1). We focused on three comparisons for the analysis of environmental conditions: 1) across larval breeding site groups, 2) across habitat categories regardless of Ae. aegypti presence status, and 3) between Ae. aegypti present and absent sites regardless of habitats. In Rabai, almost all peridomestic and domestic larval breeding sites sampled were present with Ae. aegypti . The only peridomestic Ae. aegypti absent site was excluded from analyses comparing between larval breeding site groups, but retained in comparisons between habitats or between Ae. aegypti present vs. absent sites.
The fieldwork in La Lopé was approved by the CENAREST with the authorization AR0013/16/MESRS/CENAREST/CG/CST/CSAR, and by the La Lopé National Parks with the authorization AE16008/PR/ANPN/SE/CS/AEPN. The fieldwork in Rabai was approved by the Kenya Medical Research Institute Scientific and Ethical Review Unit with the authorization KEMRI/SERU/3433.