There appeared to be two clusters of Ae. albopictus between which there is restricted gene flow. Despite the presence of these two genetic clusters in all the locations sampled, Ae. albopictus populations were structured in the eastern regions. In these regions, genetic clusters were significantly linked to habitat (gully or urban) with a mixed population that is present in both areas regardless of the type of larval development site or the season of sampling.
Significantly younger (nulliparous) and virgin (unmated) Ae. albopictus
females were found in gullies compared to urban areas regardless of season and the side of the island sampled. This result suggests that the availability of human hosts and stable containers serving as larval development sites support large are very important factors for Ae. albopictus
population dynamics. We hypothesize that gullies provide a larval development sites for mosquito populations, but that migration in search of hosts is likely to occur from gullies to urban areas. This is corroborated by the fact that the number of filled spermathecal capsules is higher overall in urban areas than in gullies. Similar results were found in experiments conducted on laboratory populations of Ae. albopictus
under optimal conditions, where only 8% of the females had three filled spermathecal capsules 
. The exchange of genes is supported by the study of the spermathecae, the parity and age, which showed a population movement from the gullies into urban areas.
The behaviour of the local zoophilic and anthropophilic Ae. albopictus
, could explain the relative importance of the larval development sites compared to the availability of hosts, which are always present in gullies and urban areas. It is likely that given the high number of Aedes
and the large number of hosts and their availability, vector control has caused a division between gully and urban clusters. These results differ from results for other Aedes
species, such as Ae. aegypti
populations in Cambodia, where patterns of differentiation between sympatric collections were associated with different container types 
. Ae. albopictus
was more abundant urban areas increasing the risk of virus transmission. The productivity of artificial immature production sites was much higher in urban areas than in gullies. This could be explained by the long-term availability of the former (anthropic immature production sites maintained by human activity, such as plates under flowerpots or containers for water storage) compared to natural sites. The existence of torrents in gullies, subjected to periodic flooding and drying, could explain why fewer Aedes
were observed in those areas (but still with a high number observed, ). In addition, there were fewer natural immature production sites and these are subject to frequent drainage because of the high rainfall. Nonetheless, natural environments in Réunion Island, such as gullies, should be considered as a potential risk for human health and as a nuisance, given the large Ae. albopictus
population observed. The natural areas may not act as a barrier but could constitute a reservoir, particularly because they are available all year round (after anti-vectorial control, for example, which is largely targeted in urban areas). No restriction of gene flow was observed in the western region, while gene flow present in the eastern region it was restricted.
The differences between the eastern and western regions may be associated with climate; the eastern part of the island is very humid, with an average annual rainfall of 3,563 mm compared to the west, with an average of 1,030 mm rain/year (Météo France, 2005). Therefore, in the east there are more suitable larval habitats (). In contrast, on the west side of the island where suitable larval development sites are scarce, Ae. albopictus
are more likely to migrate in search of suitable oviposition sites. This leads to an increase in the gene flow and is advantageous for population panmixia. Thus, dispersal of Ae. albopictus
appears, in part to be driven by the availability of oviposition sites. This is demonstrated by the ability of Ae. albopictus
to re-colonise neighbourhoods rapidly after environmental sanitation operations 
. Containers that are near to other larval habitats are more likely to be productive and have a higher number of pupae than areas where larval habitats are scarce, as has been demonstrated for Aedes aegypti
. The isolation of potential oviposition sites reduced the likelihood that they would contain pupae and reduced the average number of pupae per container 
. Furthermore, skip oviposition, where the females prefer laying eggs in multiple water collection 
, has been observed in Aedes
, thus enhancing population migrations when immature production sites are scarce.
Production of immature Aedes albopictus
were correlated with the abundance of mosquito-positive containers. In most cases, the population density of the species is associated with the number of discarded containers in the habitat 
. In Cambodia, similar results were obtained for habitat segregation (linked to levels of urbanisation), where authors found a habitat that genetically structured Ae. aegypti
. In Peru, Ae. aegypti
were spatially clustered indicating limited dispersal between households 
. This has also been shown between species of mosquitoes. In Florida, for example, habitat segregation has been observed according to habitat variables associated with urbanisation and rural characteristics (Ae. albopictus, Ae. aegypti, Culex quinquefasciatus, C. nigripalpus
. In Mayotte, this was observed between Ae. albopictus
and Ae. aegypti
; both species were capable of re-colonising the same larval development sites 
. The fact that no temporal or container type clusters were observed in Ae. albopictus
populations, suggests that there is no genetic adaptation to a particular type of larval habitat in this species, consistent with the observation that Ae. albopictus
is thought to have broad ecological plasticity 
No differentiation between vectorial competence for CHIKV was observed in populations from different localities in Réunion Island (not even between the eastern populations in Saint Benoît and the western populations in Saint Pierre) 
. However, differences in human CHIKV infection rates were observed 
which might probably due to the density of vectors.
We have shown that urban areas are preferred by Ae. albopictus for mating and oviposition. This is likely due to host availability and the existence stable and abundant artificial containers that serve as larval development sites facilitating large mosquito densities. Gullies and other natural environments however, are potential reservoirs for Ae. albopictus on Réunion Island, for re-colonising the urban areas after a population reduction (for example, following vector control). Nevertheless, when available suitable larval development sites are abundant, low production of mosquitos and population structuring is observed. This suggests that females have a preference for certain habitats and reproductive isolation depending on the habitat. An important consequence of the existence of highly clustered, local spatial patterns is that if some houses are missed during vector control operations, it is possible that the remaining intact mosquito clusters could subsequently repopulate the area. These results underline the need to use new control methods as an alternative to chemical control, such as the sterile insect technique.