To calculate the intensity and duration of dengue transmissi

To calculate the intensity and duration of dengue transmission, a seasonality curve was generated by plotting the decadally averaged VC as a function of month for each of the 10 cities first and then for each decade over the two centuries. The intensity of DEP was estimated by averaging the VC over the highest three consecutive months in the seasonality curve. The duration of transmission season was estimated by the intersections of the seasonality curve with the line defining the threshold condition (VC=0.2 (day)). The differences between the two intersections gave the number of months that VC was over the threshold. This was repeated for each decade over the two centuries and for each of the 10 European cities. Over the two centuries decadally averaged VC were from year zero to nine for each decade except the two decades with 9-year period: 1901–1909, 2011–2019 due to that both CRU and CMIP5 data started from year one in the data set: 1901 and 2011. Sensitivity analyses were carried out and the results were included in the Supplementary Information, where section and figure numbers were marked with “S”. Using Monte Carlo simulations, 95% Credible Intervals (CI) (Pericchi and Walley, 1991) of VC were estimated for temperatures ranging from 10°C to 32.5°C. The variability in each of the six vector parameters was simulated assuming a normal random distribution. At each temperature under the random generation of parameters, VC was calculated based on Eq. (1). Repeating this process 1000 times for each temperature, the 2.5th and the 97.5th percentiles of the VC were estimated to give the values of VC±95% CI. Using the fitting functions (VC±95% CI vs. temperature) as the basic equations and temperature data as input, we estimated VC±95% CI for the ten cities over two centuries, including DTR (Section S5.1–5.2, Figs. S3–S4 and Table S2 in Supplementary information). In addition, sensitivity of VC to threshold values (VC*) and the maximum value of the female vector-to-human Phos-tag Biotin BTL-105 ratio (mmax) were also estimated based on Eq. (1). We compared three values: VC*=0.1, 0.2, 0.25day based on the infectious period of 4 to 10days. The results were expressed as dengue transmission duration over the rest of this century for two RCPs (Fig. S5). We varied mmax in three values: 1, 1.5, and 2, which was chosen to reflect partially the range of pupae-to-human population ratio (0.3 to 60) for Ae. aegypti (Focks et al., 2000). The results were expressed as the seasonality curves (averaged VC vs. months) over the recent decade for selected 13 cities (Fig. S6), and transmission intensity and duration over the two centuries for two RCPs (Figs. S7–S8) in the Supplementary information.
Results
Discussion Comparing to tropical and subtropical countries, Europe showed strong seasonality in DEP. No European city is projected to have year-round dengue epidemic transmission; the longest period would be eight months for Ae. aegypti and seven months for Ae. albopictus in Málaga by the end of the 21st century under RCP8.5. As temperature increases with time from 1970s onward, Central (especially Nice) and Northern Europe has shown great increase in transmission intensity during summer while Southern Europe has shown decrease (Fig. 4). This is due to the combined effect of mean temperature and DTR as shown in Fig. S2 (c)–(f) in the Supplementary information. Over time, the intensity and seasonal windows for DEP has increased and is projected to continue increasing. As a result, more cities will be over the VC threshold starting from the South and progressing to the North during this century. However, the rate of the increase depends on the emission pathway for both vectors, especially toward the middle of the century. The same trend is observed even when using the lower bound of VC (95% CI) for Ae. aegypti (see Supplementary information, Section S6.2). This implies a significant potential benefit, if policies for climate change mitigation are implemented such that future emissions more closely reflect RCP2.6.