Abstract
BACKGROUND: Assessing adult mosquito populations is an important component of disease surveillance programs and ecosystem health assessments. Inference from adult trapping datasets involves comparing populations across space and time, but comparisons based on different trapping methods may be biased if traps have different efficiencies or sample different subsets of the mosquito community. METHODS: We compared four widely-used trapping methods for adult mosquito data collection in Kruger National Park (KNP), South Africa: Centers for Disease Control miniature light trap (CDC), Biogents Sentinel trap (BG), Biogents gravid Aedes trap (GAT) and a net trap. We quantified how trap choice and sampling effort influence inferences on the regional distribution of mosquito abundance, richness and community composition. RESULTS: The CDC and net traps together collected 96% (47% and 49% individually) of the 955 female mosquitoes sampled and 100% (85% and 78% individually) of the 40 species or species complexes identified. The CDC and net trap also identified similar regional patterns of community composition. However, inference on the regional patterns of abundance differed between these traps because mosquito abundance in the net trap was influenced by variation in weather conditions. The BG and GAT traps collected significantly fewer mosquitoes, limiting regional comparisons of abundance and community composition. CONCLUSIONS: This study represents the first systematic assessment of trapping methods in natural savanna ecosystems in southern Africa. We recommend the CDC trap or the net trap for future monitoring and surveillance programs.