West Nile virus transmission potential in Portugal

Abstract

It is unclear whether West Nile virus (WNV) circulates endemically in Portugal. Despite the country's adequate climate for transmission, Portugal has only reported four human WNV infections so far. We performed a review of WNV-related data (1966-2020), explored mosquito (2016-2019) and land type distributions (1992-2019), and used climate data (1981-2019) to estimate WNV transmission suitability in Portugal. Serological and molecular evidence of WNV circulation from animals and vectors was largely restricted to the south. Land type and climate-driven transmission suitability distributions, but not the distribution of WNV-capable vectors, were compatible with the North-South divide present in serological and molecular evidence of WNV circulation. Our study offers a comprehensive, data-informed perspective and review on the past epidemiology, surveillance and climate-driven transmission suitability of WNV in Portugal, highlighting the south as a subregion of importance. Given the recent WNV outbreaks across Europe, our results support a timely change towards local, active surveillance.

Fig. 1. Past evidence of WNV circulation in Portugal.

Summary of data sources (articles, national reports) on various types of evidence for the circulation of WNV in Portugal. a presents the official (district) divisions of Portugal, colored according to north and south areas. b presents the geographical region of each data source (latitude-longitude). Positioning is only approximate, given that some reports refer to regions of different areas. For data sources referring to large areas (e.g. districts), the capital city of the region was used. Not all sources are mapped, e.g. not reporting where positive samples were found. Mapped words represent the type of evidence reported, aggregated into two categories: molecular, for PCR, isolation; and serology, for HI, VNT, ELISA. c shows a timeline of the number of data sources per year. The color scale (on the right) represents the different types of samples reported, and refers to both panels b and c. Not all years between 1966 and 2020 had data sources. A detailed summary of data sources in Supplementary Table 1.

Fig. 2. REVIVE Culex pipiens counts in Portugal (2017–2019).

Borders represent Portuguese counties, the level at which Cx. pipiens data are reported. a–c and e–g Counties are colored according to the yearly observations (transformed to $lo{g}_{10}\left(counts+1\right)$) of adult and immature mosquitoes (respectively). d, h Counties are colored by the yearly mean observations (transformed to $lo{g}_{10}\left(mean\left(counts\right)+1\right)$) of adult and immature mosquitoes (respectively). Counties for which there were zero counts are coloured in white. Counties for which collections were not performed are in grey.

Fig. 3. Land types in Portugal (1992–2019).

a Shows the spatial distribution of different land types for the years 1992 (left) and 2019 (right). b is the percentage of land types over time for the south (left) and north (right) regions. In all panels, colours for land types are shown on the right. The North-South divide is the same as defined in the legend of Fig. 1 (boundaries shown on the map to the right).

Fig. 4. Monthly REVIVE Culex pipiens counts in Portugal versus estimated transmission potential index P (2017–2019).

a–d Present the monthly time series for $lo{g}_{10}\left(mosquitopopulationsize+1\right)$ (grey bars) and mean (colored lines) and standard deviation (coloured areas) for four districts: two from the south (a for Beja in purple, c for Faro in purple) and two from the north (b for Bragança in red, d for Viana do Castelo in blue). The map panel presents all districts, coloured by their location. e Shows the mean index $P$ for all districts 2009-2019 (north in grey and south in black). f Shows the seasonal index $P$ per month from e independently of year. g is the distribution of index $P$ across the period 2009–2019. For the estimation of the index $P$, the monthly mean of the climate variables was used per district. The North-South divide is the same as defined in the legend of Fig. 1.

Fig. 5. Estimated WNV transmission potential across Portugal (2016–2019).

Maps present summary statistics of estimated index $P$ (WNV transmission potential) at the spatial resolution available in the climate dataset. a, c, e and g Present the yearly median index $P$ per coordinate according to the presented color scale on the right. b, d, f and h Present the number of months during each year for which the estimated index $P$ per coordinate was above zero, according to the presented color scale on the right. Boundaries in dark grey represent districts.

Fig. 6. Long-term trends and periodicity of transmission potential of WNV and climate across different regions of Portugal.

a, b and c present the median of climatic variables among counties (points) within the north (blue) and south (orange) regions defined by the set of districts listed in the map subpanel (on the right). d Presents the same as a–c but for the index $P$ (transmission potential). e Presents the cross-wavelet transform of the median yearly transmission potential between north and south time-series. Significant periodic components are circled in white ($p-value<0.01$), where the heat map denotes power with peaks indicated by black dots (wavelet ridges) and mean power throughout time shown in the right subplot.