Background.Behavioral resilience in mosquitoes poses a significant challenge to mosquito control. Although behavior changes in anopheline vectors have been reported over the last decade, there are no empirical data to suggest they compromise the efficacy of vector control in reducing malaria transmission.Methods.In this study, we quantified human exposure to both bites and infective bites of a major malaria vector in Papua New Guinea over the course of 4 years surrounding nationwide bednet distribution. We also quantified malaria infection prevalence in the human population during the same time period.Results.We observed a shift in mosquito biting to earlier hours of the evening, before individuals are indoors and protected by bednets, followed by a return to preintervention biting rates. As a result, net users and non–net users experienced higher levels of transmission than before the intervention. The personal protection provided by a bednet decreased over the study period and was lowest in the adult population, who may be an important reservoir for transmission. Malaria prevalence decreased in only 1 of 3 study villages after the distribution.Discussion.This study highlights the necessity of validating and deploying vector control measures targeting outdoor exposure to control and eliminate malaria.

BackgroundThe major malaria vectors of Papua New Guinea exhibit heterogeneities in distribution, biting behaviour and malaria infection levels. Long-lasting, insecticide-treated nets (LLINs), distributed as part of the National Malaria Control Programme, are the primary intervention targeting malaria transmission. This study evaluated the impact of LLINs on anopheline density, species composition, feeding behaviour, and malaria transmission.MethodsMosquitoes were collected by human landing catch in 11 villages from East Sepik Province and Madang Province. Mosquitoes were collected for 3 years (1 year before distribution and 2 years after), and assayed to determine mosquito species and Plasmodium spp. infection prevalence. The influence of weather conditions and the presence of people and animals on biting density was determined. Determinants of biting density and sporozoite prevalence were analysed by generalized estimating equations (GEE).ResultsMosquito biting rates and entomological inoculation rates decreased significantly after the distribution. Plasmodium falciparum and P. vivax sporozoite prevalence decreased in year 2, but increased in year 3, suggesting the likelihood of resurgence in transmission if low biting rates are not maintained. An earlier shift in the median biting time of Anopheles punctulatus and An. farauti s.s. was observed. However, this was not accompanied by an increase in the proportion of infective bites occurring before 2200 hours. A change in species composition was observed, which resulted in dominance of An. punctulatus in Dreikikir region, but a decrease in An. punctulatus in the Madang region. When controlling for village and study year, An. farauti s.s., An. koliensis and An. punctulatus were equally likely to carry P. vivax sporozoites. However, An. punctulatus was significantly more likely than An. farauti s.s. (OR 0.14; p = 0.007) or An. koliensis (OR 0.27; p < 0.001) to carry P. falciparum sporozoites.ConclusionsLLINs had a significant impact on malaria transmission, despite exophagic and crepuscular feeding behaviours of dominant vectors. Changes in species composition and feeding behaviour were observed, but their epidemiological significance will depend on their durability over time.Electronic supplementary materialThe online version of this article (doi:10.1186/s12936-015-1067-7) contains supplementary material, which is available to authorized users.

Understanding mosquito host choice is important for assessing vector competence or identifying disease reservoirs. Unfortunately, the availability of an unbiased method for comprehensively evaluating the composition of insect blood meals is very limited, as most current molecular assays only test for the presence of a few pre-selected species. These approaches also have limited ability to identify the presence of multiple mammalian hosts in a single blood meal. Here, we describe a novel high-throughput sequencing method that enables analysis of 96 mosquitoes simultaneously and provides a comprehensive and quantitative perspective on the composition of each blood meal. We validated in silico that universal primers targeting the mammalian mitochondrial 16S ribosomal RNA genes (16S rRNA) should amplify more than 95% of the mammalian 16S rRNA sequences present in the NCBI nucleotide database. We applied this method to 442 female Anopheles punctulatus s. l. mosquitoes collected in Papua New Guinea (PNG). While human (52.9%), dog (15.8%) and pig (29.2%) were the most common hosts identified in our study, we also detected DNA from mice, one marsupial species and two bat species. Our analyses also revealed that 16.3% of the mosquitoes fed on more than one host. Analysis of the human mitochondrial hypervariable region I in 102 human blood meals showed that 5 (4.9%) of the mosquitoes unambiguously fed on more than one person. Overall, analysis of PNG mosquitoes illustrates the potential of this approach to identify unsuspected hosts and characterize mixed blood meals, and shows how this approach can be adapted to evaluate inter-individual variations among human blood meals. Furthermore, this approach can be applied to any disease-transmitting arthropod and can be easily customized to investigate non-mammalian host sources.