In this study, a notable difference was found in the distribution of molestus and pipiens forms according to collection methods. While the pipiens form was sampled by all methods, molestus individuals were caught only in IR collections. This result suggests differences between forms in biting and resting behaviours. When placed outdoors, CDC light traps are appropriate for sampling both host seeking mosquitoes and recently blood-fed mosquitoes searching for a suitable resting site . These traps have been successfully used as an alternative to outdoor resting collections in feeding pattern studies of Cx. pipiens s.l. conducted in the USA [17, 50]. The absence of the molestus form from outdoor CDC light trap collections may suggest a more endophagic and endophilic behaviour of this form. A tendency of the molestus form to bite indoors was further highlighted by its absence from outdoor landing catches. These results point to a predominantly indoor and synanthropic behaviour of the molestus form, as described for populations of this form at northern latitudes where inter-form hybridization is rare [6, 10, 16]. Therefore, it appears that in spite of the high hybridization levels and in addition to autogeny and stenogamy, the molestus population of Comporta maintains behavioural phenotypes typical of this form. This observation is consistent with a pure molestus genetic background found in the region, which contrasted with a more introgressed pipiens background .
Also compatible with a pattern of asymmetric hy-bridization, with more molestus genes introgressing the pipiens form, is an apparently more plastic resting behaviour of the pipiens form, suggested by the fact that blood-fed females of this form were collected both indoors and outdoors. However, the number of blood-fed Cx. pipiens s.s. females collected in outdoor CDC light traps (9 out of 1,718) was much lower than those in IR collections (174 out of 235). Furthermore, the apparent behavioural differences observed between pipiens and molestus forms should be considered with caution given the sampling design used in this study, which did not include paired collections with the same method. Additional surveys involving paired indoor/outdoor landing catches (to directly evaluate endo/exophagy) and indoor/outdoor resting collections would be required to confirm these observations.
The approach used for the selection of microsatellites to differentiate molestus and pipiens forms allowed reduction of the number of loci to be genotyped from 13 to 6 whilst maintaining high accuracy and power. The efficiency of multilocus analyses tends to increase with the number of microsatellite . However, the use of a more limited number of loci can benefit their application in surveillance studies by minimising genotyping costs and thus allowing genotyping of larger sampling sizes. Given the importance of accurately determining the intra-specific composition of Cx. pipiens s.s. it is recommended that similar microsatellite-based approaches are used in epidemiological surveys to complement the information based on a single marker (CQ11FL) that has limitations in areas of continued introgression [12, 33].
As in the survey conducted in 2005-2006 , sympatric molestus and pipiens populations displaying high hybridization levels were identified above ground in the region of Comporta. However, a higher proportion of the molestus form was found in a 2005-2006 survey (66.2%; Gomes et al., 2009), whereas in the present study the pipiens form prevailed (70.1%). This difference most likely reflects the outdoor sampling carried out in this study and which was not carried out in the previous survey. In addition, the survey of 2005/2006 was mainly concentrated in the locality of Pego (ca. 77% of females), where 79% of molestus individuals were collected in the present survey. The reasons for a higher frequency of the molestus form in Pego remain unclear. Although the proportion of chicken coops sampled in this locality was lower than in those localities where the pipiens form predominated, this may not explain the differences since the vast majority (>90%) of Cx. pipiens s.s. was sampled inside chicken coops in the three localities where the molestus form was detected. Other factors may affect the apparently heterogeneous distribution of the forms, such as the type of animal shelters (e.g. construction materials) or differences in breeding site availability and exploitation. More detailed ecological studies are thus needed to further clarify the presence and determinants of spatial heterogeneities between forms in this region.
Blood meal analysis revealed that the great majority of Cx. pipiens s.s. females fed on avian hosts. The pipiens form showed a slightly higher proportion of avian blood feeds when compared with the molestus form. However, this difference was non-significant and the proportion of avian blood meals was above 90% in both forms, suggesting an ornithophilic tendency for Cx. pipiens s.s. in the region. Ornithophilic tendency was also observed in a study analysing Cx. pipiens s.l. from urban and countryside areas of Portugal, in which over 70% of the females fed on birds , and from south-west countryside areas of Spain, where over 80% of the females fed on birds . The pipiens form has been described as ornithophilic, whereas molestus populations were recognised as being mammophilic [8, 9]. However, the feeding patterns of Cx. pipiens s.s. populations depend not only on their genetic background but also on the availability of vertebrate hosts and on host defensive mechanisms [15, 21]. Consequently, exceptions to the general feeding pattern have been reported for both forms in the USA and in the Mediterranean region [15, 51, 52]. Furthermore, hybridization between the two forms may also promote a more opportunistic feeding behaviour in Cx. pipiens s.s.. Such a catholic behaviour would thus increase the relative importance of host availability and host defensive mechanisms in the feeding pattern of the mosquito population. Therefore, continued hybridization between forms coupled with a greater availability of avian hosts in the study area may explain the greater proportion of avian feeds taken by the molestus population, which is otherwise considered as being mammophilic.
While molestus and pipiens appear to be mainly ornithophilic in the Comporta region, this may reflect host availability in the region rather than an intrinsic host preference. A lower availability of mammals (including humans) is suggested by a higher proportion of chicken coops (46.4%) when compared to mammalian shelters without domestic birds (25.0%), and by the well-built and protected human dwellings with door and window screens that prevent mosquito entry . On the other hand, pipiens form mosquitoes were caught biting humans outdoors in HLC collections which play in favour of a more opportunistic feeding pattern promoted by hybridization. Altogether, these findings suggest a closer association of both molestus and pipiens forms with avian hosts and that this ornithophilic tendency, albeit possibly genetically conditioned, may be mainly determined by host availability in the region. In this scenario, molestus females, with a preference for biting mammalian hosts, may feed more readily on the available bird hosts, which may increase the odds for alternate feeding on birds and mammals. This feeding behaviour may increase the risk of WNV transmission from birds, which are natural amplification hosts, to accidental hosts such as humans and domestic mammals.
It is worth noting, however, that the two mosquito collections carried out in this study took place in mid-summer. This did not allow inference on the possible seasonal variations in the feeding patterns of the forms, which may vary through time due to factors such as temperature and bird migration . Further studies, involving longitudinal sampling, will be required to further clarify the intrinsic host preference of Cx. pipiens forms in the region and if host availability is the main factor modulating the feeding patterns over time, in a similar manner to what has been observed in the USA .
Blood meal host identification based on mtDNA sequencing identified bird species from Passeriformes and Strigiformes orders. Birds from these orders were identified with anti-WNV antibodies in Portugal indicating the circulation of WNV in these populations . The Passeriformes are a well-known WNV reservoir [55, 56]. Species of this order, such as Passer domesticus, displayed the highest WNV prevalence in USA [36, 57].