Date Published: December 19, 2018
Publisher: Public Library of Science
Author(s): Melissa J. Bell, Luigi Sedda, Mikel A. Gonzalez, Cristian F. de Souza, Erin Dilger, Reginaldo P. Brazil, Orin Courtenay, James G. C. Hamilton, Mary Ann McDowell. http://doi.org/10.1371/journal.pntd.0007007
Abstract: In South America, the Protist parasite that causes visceral leishmaniasis, a potentially fatal human disease, is transmitted by blood-feeding female Lutzomyia longipalpis sand flies. A synthetic copy of the male produced sex-aggregation pheromone offers new opportunities for vector control applications. We have previously shown that the pheromone placed in plastic sachets (lures) can attract both females and males to insecticide treated sites for up to 3 months. To use the pheromone lure in a control program we need to understand how the application of lures in the field can be optimised. In this study we investigated the effect of increasing the number of lures and their proximity to each other on their ability to attract Lu. longipalpis. Also for the first time we applied a Bayesian log-linear model rather than a classic simple (deterministic) log-linear model to fully exploit the field-collected data. We found that sand fly response to pheromone is significantly related to the quantity of pheromone and is not influenced by the proximity of other pheromone sources. Thus sand flies are attracted to the pheromone source at a non-linear rate determined by the amount of pheromone being released. This rate is independent of the proximity of other pheromone releasing traps and indicates the role of the pheromone in aggregation formation. These results have important implications for optimisation of the pheromone as a vector control tool and indicate that multiple lures placed in relatively close proximity to each other (5 m apart) are unlikely to interfere with one another.
Partial Text: The sand fly Lutzomyia longipalpis (Diptera: Psychododae) is the major vector of Leishmania infantum, a Protist parasite and the causative agent of Zoonotic Visceral Leishmaniasis (ZVL) in Latin America. Approximately 90% of the cases of ZVL that occur in the Americas are recorded in Brazil where the greatest number of cases were found in the North East of the country . The range of the vector has been gradually expanding and consequently human and canine cases of the disease are now found throughout the central and southern states where it was previously absent [1, 2].
The results of the sand fly captures are presented in contingency tables (Tables 2 and 3) for each experiment. The raw data is presented in S1 and S2 Tables respectively.
This is the first time that a Bayesian log-linear model has been employed to quantify exogenous effects on sand fly catches. The model used allowed analysis of the contingency table obtained from multiple concurrent experiments containing categorical variables only, to identify the most important factors affecting the number of sand fly catches and to include model and data uncertainty in the model inference. Classically, analyses using simple (deterministic) log-linear models of sand fly count data are applied (e.g. [29, 30]), this can lead to limited interpretation of the β coefficients, and therefore of the effect of each factor, since a measure of uncertainty is missing. In addition, simple (deterministic) log-linear models do not allow for a comparison between the distributions (the values) of the β coefficients, which allows us to obtain credible interval data from differences between two β coefficients (Fig 2).