In this study, four major aspects of blood feeding by
An. farauti were examined: reproductive fitness associated with host feeding (characterized by fecundity), host attractiveness (defined by host choice experiments and calculations of feeding indices), human host-feeding success (measured by the HBI) and fidelity of host selection (characterized by mark-release-recapture experiments). A potential factor driving host blood meal choice in mosquitoes is reproductive fitness as expressed by fecundity [
6]. In this study in the Solomon Islands,
An. farauti was slightly more attracted to humans than pigs in both the animal-baited tent trap study and the feeding index calculation. However, any fitness advantage for
An. farauti associated with feeding on humans in Haleta village is uncertain as the advantage of increased fecundity after feeding on human blood was offset by a longer time to oviposition. If larger blood meals are taken from human hosts, they may require a longer time to digest; as reported previously blood meal size is linked with increased fecundity [
36‐
38].
The success of feeding on humans, the HBI, for
An. farauti varies widely among the geographic areas where it has been measured. In Papua New Guinea, the HBI of females collected indoors was 0.65–1.00 [
5,
39‐
41], when pooled for both indoors and outdoors was 0.68–0.88 [
42,
43], and when collected outdoors was 0.07–0.85 [
2,
5,
39‐
41,
44]. In the Solomon Islands, the HBI of females collected outdoors on Guadalcanal was 0.43 [
45]. In this study in Central Province, the HBI (0.93) for
An. farauti was higher than previous reports for outdoor resting collections. In villages where the outdoor HBI was low, the density of domestic animals was high and conversely in villages with a high outdoor HBI the density of alternative hosts was low [
5,
39,
41,
44], as was the situation in Haleta where the numbers of domestic animals were scarce compared to humans. The influence of host availability in determining anthropophagy in
An. farauti populations is particularly evident when the Haleta village population in the Solomon Islands is contrasted with the
An. farauti population in Maraga village, Madang Province, Papua New Guinea, where pigs were abundant compared to humans and the HBI of
An. farauti was 0.07. Consistent with the low HBI, the feeding index in Maraga Village, Papua New Guinea, indicated that
An. farauti preferred pigs to humans (feeding index ranged from 3.37–6.80 across villages) [
5]. Across the region where it is found, populations of
An. farauti exist in isolation with restricted gene flow [
46] and thus it is plausible that these populations are under different selective pressures for host selection. The observed differential feeding indices for
An. farauti in the Solomon Islands and Papua New Guinea are consistent with the existence of isolated populations.
Fidelity of host selection for humans can have significant repercussions on the effectiveness of interventions that target human host feeding. If populations of mosquitoes are composed of subpopulations associated with different behaviours (such as host species preference, time or location of blood feeding [indoor/outdoor]), the impact of interventions to control malaria will differ than if the vector population is a single population. Host preferences of mosquitoes have a genetic basis [
22,
25‐
27], and heterogeneous subpopulations were identified from mark-release-recapture experiments in some species (i.e.
An. balabacensis in Malaysia [
17‐
19],
An. minimus in Thailand [
20],
An. vestitipennis in Mexico [
21] and
An. gambiae in Tanzania [
22]) but not in others (e.g., anthropophagic and zoophagic subpopulations were not found for
An. maculatus in Malaysia [
47] and
An. culicifacies in Sri Lanka [
48]). If subpopulations of anthropophagic and zoophagic feeding mosquitoes do not exist, then even where the HBI is low (because of the presence of a large number of domestic animals, for example), vector control strategies targeting human blood feeding (such as insecticide treated nets or indoor residual spraying) can still be effective, as a significant proportion of the population will eventually seek a blood meal on a human during at least one feeding cycle during the period of the extrinsic incubation period.