Climate Change and Aedes Vectors: 21st Century Projections for Dengue Transmission in Europe

Abstract

Warming temperatures may increase the geographic spread of vector-borne diseases into temperate areas. Although a tropical mosquito-borne viral disease, a dengue outbreak occurred in Madeira, Portugal, in 2012; the first in Europe since 1920s. This outbreak emphasizes the potential for dengue re-emergence in Europe given changing climates. We present estimates of dengue epidemic potential using vectorial capacity (VC) based on historic and projected temperature (1901-2099). VC indicates the vectors' ability to spread disease among humans. We calculated temperature-dependent VC for Europe, highlighting 10 European cities and three non-European reference cities. Compared with the tropics, Europe shows pronounced seasonality and geographical heterogeneity. Although low, VC during summer is currently sufficient for dengue outbreaks in Southern Europe to commence-if sufficient vector populations (either Ae. aegypti and Ae. albopictus) were active and virus were introduced. Under various climate change scenarios, the seasonal peak and time window for dengue epidemic potential increases during the 21st century. Our study maps dengue epidemic potential in Europe and identifies seasonal time windows when major cities are most conducive for dengue transmission from 1901 to 2099. Our findings illustrate, that besides vector control, mitigating greenhouse gas emissions crucially reduces the future epidemic potential of dengue in Europe.

Keywords: Aedes aegypti; Aedes albopictus; Climate change; Dengue; Temperature; Vectorial capacity.

Fig. 1

Season stratified maps of VC for Europe for Ae. aegypti (i), Ae. albopictus (ii), and in those areas having recently established and/or introduced Aedes vectors (iii) (European Centre for Disease Prevention and Control (ECDC), 2015; Wilder-Smith et al., 2014b). VC was calculated for each day of the period (Jan. 1, 2006–Dec. 30, 2015) and then seasonally aggregated over the decade. Winter: December–February; Spring: March–May; Summer: June–August; Autumn: September–November. DTR was included and m_max = 1.5. E-OBS 12.0 daily gridded (0.25 × 0.25°) temperature datasets were used (Haylock et al., 2008). The gray colored areas in this figure (iii) are those having unknown Aedes activity or for which survey information was unavailable (European Centre for Disease Prevention and Control (ECDC), 2015). The threshold value of 0.2 day^− 1 is marked with an arrow on the yellow portion color bar.

Fig. 2

Seasonality of VC for 13 selected cities for Ae. aegypti (a) and Ae. albopictus (b). VC was averaged over the recent 10-year period (2004–2013) for each month of the year. DTR was included and m_max = 1.5 where m is the female vector to human population ratio. CRU-TS3.22 monthly gridded (0.5 × 0.5°) temperature data (Jones et al., n.d.) were used.

Fig. 3

Season stratified maps of VC for Europe of the last decade of this century (2090–2099) under the greenhouse gas emission pathways RCP2.6 (i & iv) and RCP8.5 (ii & iii) for two Aedes vectors. The maps show the ten-year ensemble mean of five projection model-based VC calculations grouped by season (Taylor et al., 2011). Winter: December–February; Spring: March–May; Summer: June–August; Autumn: September–November. DTR was included and m_max = 1.5. Temperatures from five different global models (CMIP5 (Taylor et al., 2011, Warszawski et al., 2014)) were used as input for the projection and had original resolution of 0.5 × 0.5°. The threshold value of 0.2 day^− 1 is marked with an arrow on the yellow portion of color bar.

Fig. 4

Seasonality comparison in VC among ten European cities over two centuries for Ae. aegypti (A) and Ae. albopictus (B). A 30-year averaged VC was plotted as a function of month for 3 different periods: Past (Fig. 3i), Current (Fig. 3ii), Future (Fig. 3iii–vi) under four different projected climate scenarios or emission pathways (RCP). DTR was included and m_max = 1.5. CRU-TS3.22 (Jones et al., n.d.) and CMIP5 (Taylor et al., 2011) gridded (0.5 × 0.5°) temperature data were used. For each emission scenario, VC was averaged over five different global models (CMIP5 (Taylor et al., 2011, Warszawski et al., 2014)).

Fig. 5

Transmission intensity and seasonal time window of dengue epidemic potential in 10 European cities for (A) Ae. aegypti (B) Ae. albopictus. Intensity was defined as the averaged VC over the highest consecutive 3-months for each decade. Transmission window was defined as the number of months when the decade averaged VC was over the threshold value (0.2 day^− 1). Historical temperatures (CRU-TS3.22 (Jones et al., n.d.)) were used from 1901 to 2009. From 2011 to 2099, two emission pathways (CMIP5 (Taylor et al., 2011, Warszawski et al., 2014)) were evaluated: RCP2.6 (i & ii) and RCP8.5 (iii & iv).