Temporal dynamics of the dengue fever vector in different tropical climates, in Brazil: implications for surveillance
Brazil is a continental country extending from the equator to sub-tropical latitudes. Dengue fever, a mosquito-borne viral disease, is endemic in the equatorial/tropical region, from latitude 0 to 25 degrees. Tropical climate is not uniform, showing variations in precipitation regimens and temperature amplitude that affect the seasonality of mosquito abundance and dengue transmission. To model this variation, a two-year entomological survey was carried out in five Brazilian cities, with Köppen-Geiger's tropical climate types: Af, As, Aw/Am, Aw. The goal was to assess the coherence of the entomological indices generated by different traps and to assess the degree of weather dependence of Aedes aegypti in areas characterized by distinct climate regimens.
METHODS:
Surveillance schemes using four trap types (BG-Sentinel, Adultrap, Mosquitrap and ovitraps) were applied monthly in three areas of 1 km2 in five cities. Positivity and density mosquito indices per area were calculated as the proportion of positive traps or the mean number of captures per trap, respectively. These outcome variables were linear regressed against meteorological variables (time lagged temperature and precipitation) after log transformation. The best model was chosen by AIC.
RESULTS:
Mosquito seasonality was detected in all tropical climate types. As expected, the greatest seasonal variation occurred in climate Aw, which presents the lowest winter temperatures (< 18C). Surprisingly, though, seasonal variation is also observed in climates Af and As, with < 5 degrees of temperature variation year around, always maintaining good conditions for mosquito development. Despite the low temperature variation ,we found significant impact of temperature on mosquito abundance in these climates. Minimum or maximum temperatures at lags 1 to 3 weeks are strong positive predictors of mosquito abundance in all climates and should be used as variables for climate based early warning systems. Precipitation is a weaker but still important predictor.
CONCLUSIONS:
Temperature and precipitation are strongly associated with dengue vector dynamics in tropical climates and could be introduced in climate based early warning systems, but predictive models should be specifically fitted to different types of tropical climates.