High Spatial Resolution Ensemble Species Distribution Modeling of Rhodnius prolixus, Vector of Chagas Disease, in Western Venezuela

Abstract Rhodnius prolixus is the most common and abundant kissing bug found in Royal and other native palms from western Venezuela. R. prolixus is a dominant vector of Trypanosoma cruzi, the parasite causing Chagas disease. Here we use species distribution models (SDMs) to estimate habitat suitability for R. prolixus. Based on habitat suitability we estimate the population at risk of Chagas disease transmission. We fitted an ensemble SDM with 250 m spatial resolution using remote sensing covariates, processed for the same time when kissing bugs were sampled, for modeling R. prolixus habitat suitability, based on 67 samples (from 41 locations) collected between 2004 and 2012. The ensemble SDM included prediction using six different machine learning algorithms, which include: generalized linear model, multiple adaptive regression splines, regression trees, random forests, generalized boosted regression trees, and extreme gradient boosting. The final SDM included 9 out of 13 variables selected using variable importance. The final ensemble SDM had an average receiver operating curve (±SD) of 0.833 ± 0.114 for the best model. The model suggested a high habitat suitability for R. prolixus along the eastern slope of the Venezuelan Andean Cordillera spread along the states of Merida, Barinas, Tachira, Trujillo, and Portuguesa. Validation with an independent data set, collected between 2012 and 2018, showed that higher suitability predicted occurrence of R. prolixus (p < 0.025). The results show how ensemble SDMs can provide high spatial resolution distribution information for R. prolixus, which can be used to accurately estimate Chagas disease transmission risk.