MVSE: An R-package that estimates a climate-driven mosquito-borne viral suitability index

Abstract

Viruses, such as dengue, Zika, yellow fever and chikungunya, depend on mosquitoes for transmission. Their epidemics typically present periodic patterns, linked to the underlying mosquito population dynamics, which are known to be driven by natural climate fluctuations. Understanding how climate dictates the timing and potential of viral transmission is essential for preparedness of public health systems and design of control strategies. While various alternative approaches have been proposed to estimate local transmission potential of such viruses, few open-source, ready to use and freely available software tools exist.We developed the Mosquito-borne Viral Suitability Estimator (MVSE) software package for the R programming environment. MVSE estimates the index P, a novel suitability index based on a climate-driven mathematical expression for the basic reproductive number of mosquito-borne viruses. By accounting for local humidity and temperature, as well as viral, vector and human priors, the index P can be estimated for specific host and viral species in different regions of the globe.We describe the background theory, empirical support and biological interpretation of the index P. Using real-world examples spanning multiple epidemiological contexts, we further demonstrate MVSE's basic functionality, research and educational potentials.

Keywords: community ecological modelling; community ecology; disease ecological modelling; disease ecology; microbial ecology; mosquito-borne Viral Suitability Estimator; mosquitoes; viruses.

Figure 1

Climate, priors and estimated index P for Recife and São Paulo. (a, c) Local humidity (magenta) and temperature (turquoise) time series per day. (b, d) Examples of MVSE informed priors: mosquito life span (blue) and incubation period (yellow); human incubation (purple) and infectious (green) periods. (e, f) Estimated index P per day with mean (red) and 95% confidence interval (grey). Priors were assumed to be the same for Recife and São Paulo. See Supplementary Information S2 Text for prior distributions

Figure 2

Seasonality of index P for Recife and São Paulo. (a, b) For each year coloured differently, the week with highest suitability is identified across all estimated index P solutions of that year (solutions’ mean and 95% CI in Figure 1). The frequency of each peak week shown across the 1000 simulations. (c,d) Two‐dimensional sensitivity of mean index P per humidity (x) and temperature (y) observed time point (day). Each point is a combination of observed climate variables, coloured according to the mean index P estimated (colour scale on the right). The white dots (over the black link) mark the mean humidity and temperature of each month over the period of the data (2005–2016); the floating circles with numbers identify each month's number

Figure 3

Spatiotemporal characterization of index P across Brazil. (a) Map presents the mean index P per pixel ($\approx 340{\text{Km}}^2$). Values coloured according to scale on the right. (b) Using the estimated index P of each pixel, with 12 points representing months, the month with highest index P is identified. Each pixel is coloured according to that month, with the colour scale represented in a circle. (c) Same as (a), but presenting the mean for selected months. In all maps, priors were assumed to be the same per pixel, as used for Recife and São Paulo. See Supplementary Information S2 Text for prior distributions. These solutions are made available as Supplementary Files

Figure 4

Correlation of index P and notified cases in nine Brazilian cities. Monthly mean time series of index P and dengue notified case data (period 2007–2012) are shown for nine cities (in order, per row, left to right): São Paulo, Recife, Feira de Santana, Salvador, Manaus, Fortaleza, Belém, Rio de Janeiro and Porto Seguro. The geographical location of these cities is shown in the maps of Figure 3. Pearson's correlation coefficient ($\rho$) is shown within each subplot. The coloured shaded areas standard deviation of the index P estimates. Six other cities that widen the geographical range of the examples are presented in Figure S5 in S1 Text. Sensitivity of $\rho$ (Pearson's) to two priors (biting rate and mosquito life span) are presented in Figures S6–S7 in S1 Text

Figure 5

Index P across Honduras and its relationship with mosquito‐borne viruses and Aedes aegypti suitability score in the capital city. (a1) Maps of Honduras showing estimated index P at 25 ${\text{km}}^2$ (selected months). Tegucigalpa (the capital city) and San Pedro Sula (city) are highlighted on the maps (circles and square respectively). The inset square presents a close‐up of the region of San Pedro Sula, in which the centre geopixel is the city exhibiting high index P values relative to its surroundings. (a2) Typical year index P (red) and Aedes aegypti suitability score (AaS, blue) per month at the country level with Pearson's correlation of 0.89. Index P at each month is the mean across all geopixels (e.g. maps in (a1)). AaS is the score of Figure 3 in Thézé et al. (2018). (b) Typical year index P (red) AaS (blue) per month at the city level with Pearson's correlation of 0.865. Index P at each month is the mean of P at each month between 2005 and 2013 when using temperature and humidity per day from a local weather station (shown in Figure S8 in S1 Text). AaS is the score of Figure 3 in Thézé et al. (2018). (c1) Same P and AaS as in (b), and Zika virus (ZIKV) incidence per month (yellow, for 2016). (c2) Same P and AaS as in (b), and dengue virus (DENV) incidence per month (green, for 2015). (c3) Same P and AaS as in (b), and chikungunya virus (CHIKV) incidence per month (magenta, for 2015). (c1–c3) Pearson's correlation for each pair of suitability measure vs incidence is presented within each subplot. In the legend of Pearson's correlation, M refers to AaS. (b‐c3)All variables are normalized to 0–1 by their maximum value for visualization purposes. Incidence variables are log10 before being normalized