To generate recent European season stratified maps

To generate recent European season-stratified maps of VC (Jan. 1, 2006–Dec. 31, 2015), daily temperature observations (minimum, maximum, and mean) from the E-OBS 12.0 dataset were used for each location gridded at 0.25×0.25° (about 25×25km at the equator) latitude and longitude (Haylock et al., 2008). This daily VC calculation included interpolating DTR based on daily observations, then aggregated over the decade by season (Winter: December–February; Spring: March–May; Summer: June–August; Autumn: September–November). The seasonal averaged VC for the recent decade were displayed as maps for Europe for each season for both vectors and compared to a recent survey of vector distribution (European Centre for Disease Prevention and Control (ECDC), 2015) for areas known to have Aedes activity according. To show seasonality of VC over a year, decade averages of VC for each month was displayed as a function of the month in a year. For the recent decade, 13 cities were chosen to compare seasonality of DEP. Ten European cities were selected to represent most of the European continent with different temperature zones from the north — Stockholm (latitude=59.3) to the south — Málaga (latitude=36.7) within the continent and Madeira (latitude=32.7) outside the continent. Madeira is an autonomous region of Portugal having a dengue outbreak in 2012, and for convenience in this paper, we will use the name of city for all the nine cities and Madeira region. Three reference cities from tropical and sub-tropical regions outside Europe were chosen for comparison. Colombo and Singapore are located in Asia close to the equator and display high dengue endemicity (Gubler, 2011; Gubler and Clark, 1995) despite political and financial investment in dengue control. By Kinase Inhibitor Library Miami, located in North America with a sub-tropical climate with more similarity in environmental and social economic conditions to some of Southern Europe, has reported autochthonous dengue transmission occasionally, which typically does not develop into large scale epidemics (Theiler et al., 1960). Diurnal temperature range is known to affect the competence of dengue vector Aedes aegypti (Lambrechts et al., 2011). Inclusion of DTR in modeling DEP for Ae. aegypti using relative VC (rVC) has shown a great difference comparing without DTR (Liu-Helmersson et al., 2014), especially in the cold to temperate climate zones in Northern hemisphere. This is because rVC (VC) depends on DTR strongly, both the peak intensity and the position. When DTR increases from 0°C to 20°C, the peak height of rVC reduced from 1.37 to 0.47day; the peak position of VC reduces from 29°C to 20°C. Therefore, in models including DTR, temperate climate zones with larger DTR will have greater DEP, while tropical areas with less DTR will have lesser DEP than estimated by models using mean temperature alone. This is particularly relevant to Europe, where DTR is greater than tropical areas. From the Climate Research Unit (CRU) online database, time series (CRU-TS 3.22) of gridded (0.5×0.5 degrees) monthly averages of daily temperature observations (minimums, maximum, and mean) were obtained for Europe for the period January 1, 1901 to December 31, 2013 (Jones et al., n.d.). Given the importance of DTR to temperate European climate, in all the VC calculations, diurnal temperature range (DTR) was included. DTR was reconstructed using a representative daily temperature for each 30min through a piece-wise sinusoidal function based on the monthly average of daily minimum, maximum, and mean observations for each location (same temperature for each day of the month in each 0.5×0.5 grid – See Supplementary information S3 for details). To illustrate the combined effect of DTR and mean temperature to VC, heat maps were generated for both Aedes vectors (Fig. S2 (d) & (d) in the Supplementary information). To show the seasonal window and its change over time, a 30-year average was used for each monthly averaged VC at three periods – the beginning of the 20th century (1901–1930), at the turn from 20th to 21st century (1984–2013), and the end of the 21st century (2070–2099). Future VC was calculated using projected climate under four greenhouse gas emission pathways (RCP2.6, RCP4.5, RCP6.0 and RCP8.5) (Weyant et al., 2009) based on CMIP5 (Taylor et al., 2011; Warszawski et al., 2014) atmosphere-ocean general circulation models. For each emission pathway, CMIP5 temperature datasets (min, max, mean resolution 0.5×0.5°) were used (Taylor et al., 2011; Warszawski et al., 2014). The VC was calculated for each of the five global models (NorESM1-M, MIROC-ESM-CHEM, IPSL-CM5A-LR, HadGEM2-ES and GFDL-ESM2M) and then averaged over the five models (Taylor et al., 2011; Warszawski et al., 2014). We used these models as an ensemble to form a multi-model mean, with the intention of providing results that are based on greater consensus. The four RCP scenarios describe the possible range of radiative forcing of greenhouse gases in the year 2100 (+2.6, +4.5, +6.0, and +8.5Wm, respectively) (Weyant et al., 2009). VC calculations were aggregated by decade to show the trends in DEP over two centuries. A selection of the outputs of these projection-based VC calculations was also mapped for RCP2.6 and RCP8.5 for both species to show the changes in DEP across Europe under scenarios of greater or lesser emission mitigation.