Aedes aegypti Suitability in Haiti

Distribution modelling under baseline climate and warming scenarios

Project Overview

This project models the distribution of Aedes aegypti in Haiti under baseline climate conditions and projects suitability under warming scenarios of +1.5°C, +2°C, +3°C, and +4°C. The goal is to provide spatially explicit estimates that can support vector control prioritisation and broader public health planning.

Aedes aegypti is an important vector of dengue, Zika, and chikungunya viruses, which have all been of public health importance in Haiti.

Key finding: At baseline, 25.9% of Haiti’s landscape exceeded our computed 10th percentile training presence threshold (0.552), corresponding to an estimated 3,956,000 people (about 21% of the population) living in areas at or above this threshold. Under +4°C of warming, more than 85% of Haiti’s population is projected to be in areas at or above 0.552 probability of presence.

Key Results

Baseline suitability and population exposure

25.9% of Haiti’s landscape at or above the 10th percentile training presence threshold at baseline
3.96M people estimated to live at or above the threshold at baseline

Warming level projections

Future suitability was projected under +1.5°C, +2°C, +3°C, and +4°C warming scenarios using screened CMIP6 global climate models. Across warming levels, suitability increased consistently, with strong expansion in the share of the population living in areas above the baseline threshold.

Spatial patterns of risk

Risk is not determined by suitability alone. Areas with high suitability, high population density and high deprivation are expected to represent the highest priority for control, including the Port-au-Prince metropolitan area.

About the Modeling Approach

We used ensemble boosted regression tree models trained with 60 presence locations from the VectorMap database and 19 WorldClim bioclimatic variables at approximately 1 km resolution. Pseudo-absence points were generated in a 1:1 ratio with presences, with locations constrained to at least 717 meters away from the nearest presence location. Thirty replicates were generated to support an ensemble prediction and to characterise uncertainty across models.

Model performance was high (mean AUC = 0.87), and temperature-related variables showed the highest relative influence.

Data Sources

Interactive Maps

The maps below are hosted as HTML Leaflet widgets. They support zooming, panning, and layer controls. The HTMLs are also downloadable from the github repository.

Baseline suitability predictions

Explore baseline predicted suitability for Aedes aegypti and related uncertainty layers derived from the ensemble.

View map

Commune summaries

View commune-level summaries of predicted suitability to support comparisons across administrative units.

View map

Risk index map

Explore an integrated risk representation that relates predicted suitability to human population exposure and deprivation metrics.

View map

Public Health Implications

Describing vector distributions is a foundational public health activity. In Haiti, this is especially important given resource constraints and the likelihood of substantial changes in suitability under climate change.

The projected expansion of suitability under warming highlights a global health equity challenge. These outputs provide spatially explicit estimates that can support prioritisation and planning, while also clarifying where uncertainty remains.