Anopheles albimanus (Diptera: Culicidae) Ensemble Distribution Modeling: Applications for Malaria Elimination

Abstract

In the absence of entomological information, tools for predicting Anopheles spp. presence can help evaluate the entomological risk of malaria transmission. Here, we illustrate how species distribution models (SDM) could quantify potential dominant vector species presence in malaria elimination settings. We fitted a 250 m resolution ensemble SDM for Anopheles albimanus Wiedemann. The ensemble SDM included predictions based on seven different algorithms, 110 occurrence records and 70 model projections. SDM covariates included nine environmental variables that were selected based on their importance from an original set of 28 layers that included remotely and spatially interpolated locally measured variables for the land surface of Costa Rica. Goodness of fit for the ensemble SDM was very high, with a minimum AUC of 0.79. We used the resulting ensemble SDM to evaluate differences in habitat suitability (HS) between commercial plantations and surrounding landscapes, finding a higher HS in pineapple and oil palm plantations, suggestive of An. albimanus presence, than in surrounding landscapes. The ensemble SDM suggested a low HS for An. albimanus at the presumed epicenter of malaria transmission during 2018–2019 in Costa Rica, yet this vector was likely present at the two main towns also affected by the epidemic. Our results illustrate how ensemble SDMs in malaria elimination settings can provide information that could help to improve vector surveillance and control.

Mapping potential malaria vector larval habitats for larval source management: Introduction to multi-model ensembling approaches

“…Data mining scientists recommend using diverse and independent models for ensemble modeling [29,37]. Most medical and biological application studies have not mentioned how individual models were selected for ensemble modeling [20,21,[24][25][26]32,60]. However, model selection may affect the eventual predictions of the ensembled model.…”

“…The copyright holder for this preprint this version posted October 7, 2022. ; https://doi.org/10.1101/2022.10.06.511086 doi: bioRxiv preprint included both models in the ensembled model and used accuracy or AUC as the weight [24,25],…”

“…Indeed, stepwise selection selected only the GBM logistic model (results not shown); i.e., the GBM classification model may be redundant. In fact, if we used accuracy or AUC > 0.7 for model selection in this study [24,25], we would actually end up with no model for larval positive habitat prediction. This is a reminder that the combination of artificially selected weights/models may severely affect the results of the ensembled model.…”

“…We have noticed that very few studies have predicted the potential of larval positive habitats [19]. Since all aquatic habitats had water at the time they were surveyed, to identify predictors of aquatic habitats we artificially generated about 13,500 pseudo-habitats in randomly selected locations with known residential houses [25,36] (Table 1). Pseudo-habitats were selected at least 50m away from any known aquatic habitats.…”

“…Many models can be used to identify mosquito larval habitats [17,18,20,25]. There is no standard method for model selection in ensemble modeling.…”