Mosquitoes can transmit viruses that cause Zika, dengue and several other tropical diseases that affect humans. Zika virus usually causes mild symptoms, but it is thought that infection during pregnancy can lead to brain abnormalities, including microcephaly, where babies are born with an abnormally small head. Recent studies have shed light on how the Zika virus spread from Africa to reach South America, the Caribbean and North America. However, much less is known about the ecological factors that contribute to the spread of the virus within towns, cities and other local areas.

In 2015, Brazil was struck by an outbreak of the Zika virus that led to an international public health emergency. Lourenço et al. used a mathematical model to explore the local conditions within Feira de Santana (a major urban center in Brazil) that contributed to the outbreak. The model took into account numerous factors, including temperature, humidity, rainfall and the mosquito life-cycle, which made it possible to reconstruct the history of the virus over the past three years and to make projections for the next decades.

It revealed that most of the infections occured during 2015, with approximately 65% of the population infected. The incidences of new infections declined in 2016, as increasing numbers of local people had already been exposed to the virus and became immune. Temperature and humidity appeared to have played a critical role in sustaining the mosquito population carrying the Zika virus.

Further analysis suggests that the risk of Zika virus causing microcephaly is very low – only 0.3–0.5% of the pregnant women in Feira de Santana who were infected with Zika gave birth to a baby with the condition. What therefore makes Zika a public health concern is the combination of a low risk with very high infection rates, which can affect a large number of pregnancies.

This study will help researchers and policy makers to predict how the Zika virus will behave in the coming years. It also highlights the limitations and successes of the current system of surveillance. Moreover, it will help to identify critical time periods in the year when mosquito control strategies should be implemented to limit the spread of this virus. In future, this could help shape new local strategies to control Zika virus, dengue and other diseases carried by mosquitoes.

The first cases of Zika virus (ZIKV) in Brazil were concurrently reported in March 2015 in Camaçari city in the state of Bahia (Campos et al., 2015) and in Natal, the state capital city of Rio Grande do Norte (Zanluca et al., 2015). During that year, the epidemic in Camaçari quickly spread to other municipalities of the Bahia state, including the capital city of Salvador, which together accounted for over 90% of all notified Zika cases in Brazil in 2015 (Faria et al., 2016a). During this period, many local Bahia health services were overwhelmed by an ongoing Chikungunya virus (CHIKV, East Central South African genotype) epidemic, that was first introduced in 2014 in the city of Feira de Santana (FSA) (Nunes et al., 2015; Faria et al., 2016b). The role of FSA in the establishment and subsequent spread of CHIKV highlights the importance of its socio-demographic and climatic setting, which may well be representative for the transmission dynamics of arboviral diseases in the context of many other urban centres in Brazil and around the world.

On the $1^{st}$ February 2015 the first ZIKV cases were reported in FSA, followed by a large epidemic that continued into 2016. The rise in ZIKV incidence in FSA coincided temporally with an increase in cases of Guillain-Barré syndrome (GBS) and microcephaly (Faria et al., 2016a), with an unprecedented total of 21 confirmed cases of microcephaly in FSA between January 2015 and May 2017. There is wide statistical support for a causal link between ZIKV and severe manifestations such as microcephaly (Rubin et al., 2016; de Araújo et al., 2016; Soares de Araújo et al., 2016; Honein et al., 2017; Brasil et al., 2016; de Oliveira et al., 2017), and the proposed link in 2015 led to the declaration of the South American epidemic as an international public health emergency by the World Health Organization (WHO) in 2016; the response to which has been limited to vector control initiatives and advice to delay pregnancy in the affected countries (WHO, 2016b; WHO, 2016a). With few cohort studies published (Honein et al., 2017; Brasil et al., 2016) and the lack of an established experimental model for ZIKV infection (Aman and Kashanchi, 2016; Dowall et al., 2016), modelling efforts have taken a central role for advancing our understanding of the virus’s epidemiology (Chowell et al., 2016; Ferguson et al., 2016; Bogoch et al., 2016; Nishiura et al., 2016; Zhang et al., 2016; Perkins et al., 2016; Messina et al., 2016). In particular, our knowledege on parameters of public health importance, such as the basic reproduction number ($R_0$), the duration of infection (Ferguson et al., 2016), attack and reporting rates (Kucharski et al., 2016), the risk of sexual transmission (Maxian et al., 2017; Gao et al., 2016; Moghadas et al., 2017) and birth-associated microcephaly (Bewick et al., 2016; Perkins et al., 2016) has advanced significantly from studies using transmission models. Climate variables are critical for the epidemiological dynamics of Zika and other arboviral diseases, such as dengue (Lourenço and Recker, 2014; Feldstein et al., 2015; Kraemer et al., 2015; van Panhuis et al., 2015) and chikungunya (Poletti et al., 2011; Mourya et al., 2004; Salje et al., 2016). Although these have also been previously addressed in mapping and/or modelling studies (e.g. (Bogoch et al., 2016; Zhang et al., 2016; Perkins et al., 2016; Messina et al., 2016)), their effects as ecological drivers for the emergence, transmission and endemic potential of the Zika virus, especially in the context of a well described outbreak, have not yet been addressed in detail.

In this study, focusing on an urban centre of Brazil (Feira de Santana), we explicitly model the mosquito-vector lifecycle under seasonal, weather-driven variations. Using notified case data of both the number of suspected Zika infections and confirmed microcephaly cases, we demonstrate how the combination of high suitability for viral transmission and a low detection rate resulted in an extremely high attack rate during the first epidemic wave in 2015. The rapid accumulation of herd-immunity significantly reduced the number of cases during the following year, when total ZIKV-associated disease was peaking at the level of the country. Projecting forward we find that the demographic loss of herd-immunity together with the frequency of reintroduction will dictate the risk of reemergence and endemic establishment of Zika in Feira de Santana. The conclusions of this study should be transferable to major urban centres of Brazil and elsewhere with similar climatic and demographic settings.

On the $1^{st}$ February 2015 the first Zika virus (ZIKV) case was reported in Feira de Santana (FSA). Weekly cases remained very low for the following two months, adding up to just 10 notified cases by the end of March that year (Figure 1A). A rapid increase in the number of cases was observed in April, coinciding with Micareta, a local carnival-like festival that takes place across the urban centres of Bahia. The epidemic peaked in July 2015, which was followed by a sharp decline in notified cases over the next 1–2 months. This first epidemic wave was followed by a significantly smaller outbreak in 2016, peaking around March, and an even smaller outbreak in 2017 with no discernable epidemic peak.

Figure 1

Download asset Open asset

Confirmed (and monthly aggregated) microcephaly (MC) cases were absent by November 2015, after which a small epidemic was observed with peak counts in January 2016. We found a time lag of 5–6 months (20–24 weeks) between the first reported Zika epidemic wave and the MC peak in case counts. This coincides with previous observations suggesting a link between the development of neurological complications in newborns and ZIKV infection during the second trimester (Faria et al., 2016a). We note that our lag may be offset by around 1–4 weeks, however, since the date of MC cases in our dataset represents the date of diagnostic confirmation, which is usually done postpartum.

Overall, the epidemic behaviour in FSA was in sharp contrast with trends observed in notified cases across Brazil (BR) as a whole, for which the second epidemic in 2016 was approximately 6 times larger than the one in 2015 (Figure 1A), suggesting the Bahia state as a focus point in the emergence and initial spread of ZIKV in Brazil (Faria et al., 2016c; Faria et al., 2016a). Nonetheless, a clear temporal synchronization between country level and FSA case counts could be observed.

The age distribution of notified ZIKV cases in FSA suggested a higher proportion of cases between 20 and 50 years of age, but with no discernible differences between the two epidemics (Figure 1B, top panel). However, when corrected for the expected number of cases assuming an equal risk of infection per age class, we found the number of cases within this age group to be closer to most other groups (incidence rate ratio, IRR, close to 1, Figure 1B, bottom panel). The per capita case counts within the youngest age class (<1 years) appeared higher than expected, with an IRR significantly above 1 and also higher in 2016 (IRR = 4.4, 95% CI [2.8, 7.0]) than in 2015 (IRR = 1.95, 95% CI [1.5, 2.6]), possibly indicating biased reporting and/or health care seeking with increased awareness of the disease. There was also a consistent trend towards reduced IRR in the elderly (>65 years), although with significant uncertainties. Finally, a small increase in IRR could be detected in the 20–34 year olds, which could potentially be a signature of sexual transmission in this age group (Gao et al., 2016; Carlson et al., 2016; Foy et al., 2011; Turmel et al., 2016; Yakob et al., 2016; Maxian et al., 2017; Moghadas et al., 2017). At this stage and without more detailed data it was not possible to ascertain whether these findings indicated age-related risk of disease, age-dependent exposure risk or simply notification biases in particular age groups, however.

The spatial distribution of total notified cases for BR highlighted the expected clustering of ZIKV cases within the Bahia state by the end of 2015 as well as the wider geographical range by July 2016 (Figure 1C). We speculate that the difference in geographical range could explain the higher number of cases observed during the 2016 epidemic at the country level. This, on the other hand, did not explain why the second epidemic in FSA was nearly 7 times smaller than the first and with only sporadic cases in 2017. To answer this question and to obtain robust parameter estimates of ZIKV epidemiological relevance we utilised a dynamic transmission model, which we fitted to notified case data and local climate variables of FSA within a Bayesian framework (see Materials and methods).

Climate-driven vectorial capacity

The reliance on Aedes mosquitoes for transmission implies that the transmission potential of ZIKV is crucially dependent on temporal trends in the local climate. We therefore investigated daily rainfall, humidity and mean temperature data in FSA between 2013 and May 2017 (Figure 2A). The data showed erratic fluctuations in rainfall with sporadic episodes of intense rain but without a clear seasonal trend. Temperature, on the other hand, presented a much clearer seasonal signature with fixed amplitudes between 22 and 27 degree Celsius, peaking between December and May. Humidity showed an intermediate scenario and appeared correlated with periods of intense rainfall but negatively correlated with temperature.

Figure 2 with 3 supplements see all

Download asset Open asset

By fitting our climate-driven transmission model to the local climate and ZIKV case data (see Material and Methods and Figure 2B) we obtained parameter estimates for the mosquito lifespan as well as the viral extrinsic incubation period (EIP) for the same period. Mosquito lifespan and EIP are main drivers of vectorial capacity and both showed seasonal oscillations with median values of around 9 and 5 days, respectively (Figure 2—figure supplement 3), which are in line with ranges found in the literature (Trpis et al., 1995; Trpis and Hausermann, 1986; Andraud et al., 2012; Hugo et al., 2014 and Table 3). Importantly, there was a strong negative temporal correlation between these two variables, with periods of longer EIP coinciding with shorter lifespans and vice-versa. This negative relationship resulted in large temporal variations in vectorial capacity and thus seasonal oscillations in the daily reproductive numbers, $R_0$, with a median value of 2.7 in the period 2015–2017 (range 1.0–4.3, Figure 2—figure supplement 3), and 2.2 before 2015, peaking in the local summer months between December and April (Figure 2C). Importantly, $R_0$ remained above 1 for the entire period, indicating a high suitability for ZIKV in FSA. It should be noted that $R_0$ in this context is a time-dependent variable, i.e. $R_0(t)$, but out of convenience we simply refer to it as $R_0$.

Table 3

Parameter/Function	Values and ranges reported	References
Intrinsic incubation period	6.5, 5.9 days	[ 34, 50 ]
Human infectious period	4.7, 9.9 days	[ 34, 50 ]
Extrinsic incubation period	8.2, $<10$, $<7$ days	[ 34, 51, 84 ]
Attack rates	74, 50, 73, 94, 52%	[ 17, 26, 47, 28 ]
$R_0$	3.2, 2.5, 4.8, 2.05, 2.6–4.8, 4.3–5.8, 1.8–2.0	[ 17, 37, 38, 47, 64 ]
Observation rate	0.024, 0.06, 0.03, 0.11	[ 17, 37, 47 ]

We also looked at the relationship between each climatic variable and $R_0$ and case counts (Figure 2D and E, respectively). The transmission potential was strongly and positively correlated with temperature ($r^2=0.728$) and negatively with humidity ($r^2=0.26$). As expected, from the highly random patterns in the climate series, there was no correlation between $R_0$ and rainfall ($r^2=0.008$). In contrast, there was an opposite trend in the relationship between the climatic variables and case counts, with a positive correlation with humidity ($r^2=0.28$) and a negative correlation with temperature ($r^2=0.23$). As with $R_0$ there was only a weak observable trend in the relationship between rainfall and the number of Zika cases. It should be understood that this macroscopic analysis does not take into account the expected temporal lags due to mosquito development, incubation periods etc., so the purpose here was simply to identify a general qualitative relationship between climate, vectorial capacity and disease incidence.

Model fit and parameter estimates

Four parameters of public health importance were estimated by our MCMC framework: the date of introduction, the human infectious period, the human (intrinsic) incubation period, and the case observation rate (Table 4). The posterior for the introduction date showed a strong support for an introduction into FSA in early-mid December 2014 (estimated median: ${10}^{th}$ of December), i.e. around 7–8 weeks before the first notified case (Figure 3A). The estimated human infectious period was $\approx 6$ days (Figure 3C, median = 5.9, 95% CI [5.47–6.14]), which was very similar to the estimated incubation period (Figure 3D, median = 5.8, 95% CI [5.6–6.15]) and in line with previously estimated ranges for ZIKV (Table 3). In this context it is important to note that informative priors had been used for these 2 parameters (Figure 2—figure supplement 2.), and the posterior for the incubation period presented an adjustment of $\approx -0.5$ days relative to the proposed distribution from the literature.

Table 4

Notation	Description	Ranges / priors
$t_0$	time point of first case (in a human)	($\mathrm{\infty }$, $\mathrm{\infty }$)
$K$	aquatic carrying capacity	(0, $\mathrm{\infty }$)
$\eta$	linear factor for mosquito mortality	(0, $\mathrm{\infty }$)
$\alpha$	linear factor for extrinsic incubation period	(0, $\mathrm{\infty }$)
$\rho$	non-linear factor for effects of humidity and rainfall	(0, $\mathrm{\infty }$)
${\sigma }^h$	human infectious period	(0, 15)
${\gamma }^h$	human (intrinsic) incubation period	(0, 15)
$\zeta$	observation rate	(0, 1)

Figure 3 with 2 supplements see all

Download asset Open asset

Of particular interest here was the very low observation rate (Figure 3B), with a median of just under 0.004 (median = 0.0039, 95% CI [0.0038–0.0041]), which equates to less than 4 in 1000 infections having been notified during the epidemic in FSA. Although lower than other previously reported estimates, this would explain the relatively long period of low viral circulation before the epidemic took off in April 2015. That is, based on our estimates, there were around 2,700 Zika infections during the first 2 months, of which only 10 were notified. More importantly, when applying this rate to the total number of cases we found that by the end of the first epidemic wave around 65% (95% CI [57.0–72.9]) of the population in FSA had been infected by the virus. This high attack rate is not unusual for Zika, however, and is in general agreement with observations elsewhere (Table 3).

Future transmission potential for Zika virus

As illustrated by the cumulative attack rate in Figure 4A, and similar to estimates from other regions in the world (Table 3), nearly 65% of the population got infected by ZIKV by the end of 2015, which rose to over 75% (95% CI [76.9–84.3]) by the end of 2016. During the first wave most cases occurred off-season, here defined by our estimated daily reproductive number ($R_0$), while the second wave appeared much more synchronized with the period of high transmission potential. Notably, this temporal phenomenon has also been observed for the chikungunya virus (CHKV) when it was first introduced into FSA in 2014 (Faria et al., 2016b).

Figure 4

Download asset Open asset

The amassed accumulation of herd-immunity during the first wave resulted in a marked difference between the estimated basic reproductive number, $R_0$, and the effective reproductive ratio ($R_e$) by the end of 2015 (Figure 4A). This in turn might explain the marked reduction in Zika cases in FSA in 2016, at a time when the virus was infecting large numbers of individuals elsewhere in the country (Figure 1A,C). At the start of 2016, $R_e$ was estimated to be more than 3 times smaller than $R_0$, which increased to 5 by the beginning of 2017. Projecting into the future using average climate data for this region showed that the mean effective reproductive number is expected to remain low and close to 1 for the next few years, suggesting a very weak potential for ZIKV endemicity in the near future. In fact, the sporadic nature of Zika cases in 2017 strongly suggest that herd immunity in this region is at a sufficiently high level to prevent sustained transmission. Furthermore, during 2017, $R_e$ was on average less than 1 (mean: 0.62, range: 0.25–1.06), and we would therefore argue that the small number of cases (1.4% of 2015–2017) were mostly a result of small transmission chains, either from resonant transmission from the previous year, or from introduction events from nearby locations. Crucially, this would also explain why our ODE model matched both the dynamics and the sizes of the first two epidemic waves in FSA between 2015 and 2016 but failed to capture the small number of cases during 2017 (Figure 2B).

Without external introductions of infectious individuals (human or vector) our results predicted an epidemic fade-out by 2017, in accordance with the lack of notified cases after March 2017 (Figure 4A). We therefore projected ZIKV’s epidemic potential over the next two decades (until 2040) using stochastic simulations (see Material and Methods) while assuming different rates of viral introduction (Figure 4B,C). Our results showed that the potential for ZIKV to cause another outbreak or to establish itself endemically in FSA is strongly dependent on the frequency of re-introductions, whereby higher rates of external introductions might in fact help to sustain high levels of herd immunity, whereas infrequent introductions are more likely to result in notable outbreaks. That is, semi-endemic behaviour was only observed in simulations with low introduction rates (Figure 4B–C), as these scenarios strike a fine balance between a low number of new cases affecting herd-immunity levels and population turnover. In contrast, high introduction rates quickly exhaust the remaining susceptible pool, resulting in very long periods without epidemic behaviours.

Sensitivity to reporting and microcephaly risk

In effect, our estimated observation rate entails the proportion of real infections that would have been notified if symptomatic and correctly diagnosed as Zika. Based on the previously reported Yap Island epidemic of 2007 (Duffy et al., 2009), the percentage of symptomatic infections can be assumed to be close to 18%. Unfortunately, measures of the proportion of individuals seeking medical attention and being correctly diagnosed do not exist for FSA, although it is well known that correct diagnosis for DENV is imperfect in Brazil (Silva et al., 2016). We therefore performed a sensitivity analysis by varying both the proportions of infected symptomatic individuals seeking medical attention and the proportion of those being correctly diagnosed for Zika. Figure 5A shows that if any of these proportions is less than 10%, or both between 15–20%, our observation rate of 3.9 per 1000 infections can easily be explained.

Figure 5

Download asset Open asset

Finally we investigated the sensitivity of our results with regards to the expected number of newborns presenting microcephaly (MC). Following the observation that virtually all reported MC cases were issued before the summer of 2016 and with a lag of 5–6 months (Figure 1A), we assumed that the vast majority of Zika-associated MC cases would have been a consequence of the first epidemic wave in 2015. We used the estimated attack rate of approximately 65% from 2015 (Figure 4A) and varied the local birth rate and the theoretical risk of MC to obtain an expected number of cases. In agreement with other reports (de Araújo et al., 2016; Cauchemez et al., 2016; Jaenisch et al., 2016; Johansson et al., 2016), our model predicted a relatively low risk for MC given ZIKV infection during pregnancy (Figure 5B,C). In particular, using a conservative total of 21 confirmed MC cases in FSA, i.e. rejecting suspected or other complications, we estimate an average risk of approximately 0.35% of pregnancies experiencing ZIKV infection. Including the 3 foetal deaths where ZIKV infections were confirmed during pregnancy, i.e. using a total of 24 cases, only increased the risk to 0.39%. More generally, based on the results from our fitting approach and using the average birth rates of FSA as guideline, we estimate that on average 3–4 MC cases are expected per 100 k individuals at 65% exposure to the virus.

Using an ento-epidemiological transmission model, driven by temporal climate data and fitted to notified case data, we analysed the 2015–2017 Zika outbreak in the city of Feira de Santana (FSA), in the Bahia state of Brazil and determined the conditions that led to the rapid spread of the virus as well as its future endemic and epidemic potential in this region. Given FSA’s high suitability for ZIKV mosquito-vectors and its particular geographical setting as a state commerce and transport hub, our results should have major implications for other urban centres in Brazil and elsewhere.

The pattern of reported ZIKV infections in FSA was characterized by a large epidemic in 2015, in clear contrast to total reports at the country-level, peaking during 2016. Most notably for FSA was the epidemic decay in 2016 and fadeout in 2017. In order to resolve whether this was due to a lower transmission potential of ZIKV in 2016/2017 in FSA, we calculated the daily reproductive number ($R_0$) between 2013 and 2017 but found no notable decrease in 2016. Interestingly, the maximum $R_0$ in that period was observed in the season 2015/2016, coinciding with El Niño (Golden Gate Weather Services, 2017) and thus in line with the hypothesis that this phenomenon may temporary boost arboviral potential (Caminade et al., 2017; van Panhuis et al., 2015). By fitting our model to weekly case data we also estimated the observation rate, i.e. the fraction of cases that were notified as Zika out of the estimated total number of infections. It has previously been reported that the vast majority of Zika infections go unnoticed (Table 3), which is in agreement with our estimates of an observation rate below 1%. Based on this, around 65% of the local population were predicted to have been infected by ZIKV during the first wave in 2015, which is in the same range as the reported Zika outbreaks in French Polynesia (66%) (Cauchemez et al., 2016) and Yap Island (73%) (Duffy et al., 2009). The accumulation of herd-immunity caused a substantial drop in the virus’s effective reproductive number ($R_e$) and hence a significantly lower number of cases during the second wave in 2016 and subsequent demise in 2017. In the context of FSA, it is possible that the high similarity of case definition to DENV, the concurrent CHIKV epidemic, and the low awareness of ZIKV at that time could have resulted in a significant number of ZIKV infections being classified as either dengue or chikungunya. Furthermore, based on our analysis, we would argue that the percentage of correctly diagnosed ZIKV infections and infected individuals seeking medical attention must have been exceptionally low (both lower than 20%).

The age structure of notified cases showed a higher than expected incidence risk ratio (IRR) for individuals under the age of 4 years and a lower than expected risk for individuals aged + 50 years. This contrasts the observation during the Zika outbreak on Yap Island in 2007, where all age classes, except the elderly, presented similar attack rates (Duffy et al., 2009). We note here, however, that the Yap Island analysis was based on both a retrospective analysis of historical hospital records and prospective surveillance (serology, surveys). It is therefore possible that the signatures amongst the youngest and oldest individuals in FSA may reflect deficiencies and/or biases in local notified data. Such signatures could emerge by both a rush of parents seeking medical services driven by a hyped media coverage or prioritization of child-care due to the emergence of microcephaly during the Zika epidemic and a small proportion of the elderly seeking or having access to medical attention. In fact, the increased risk in young children in 2016 may have been a result of increased awareness as well as the interventions by the WHO in the second year. We also found a small increase in IRR in the 20–34 years age group, particularly during 2016, which could be indicative of the small contribution of sexual transmission (Moghadas et al., 2017; Maxian et al., 2017). Most of these observations are speculative, however, and more detailed data will be required to fully understand these age-related risk patterns. For instance, initiatives such as the ZiBRA Project (ZIBRA, 2016; Faria et al., 2016c; Faria et al., 2017), which perform mobile and real-time sampling with portable genome sequencing, could prove to be essential for a retrospective and future analysis of the ZIKV epidemic in Brazil, especially in areas where high levels of herd-immunity will prevent large-scale circulation in the coming years (Ferguson et al., 2016).

The implicit consideration of climate variables as drivers of vector biology allowed us to ascertain the relative roles of temperature, humidity and rainfall for Zika’s basic and effective reproductive potentials ($R_0$ and $R_e$, respectively). Similar to other studies in temperate and tropical settings, we found that temperature, with its direct influence on mosquito lifespan, aquatic development and extrinsic incubation period, was the key driver of seasonal oscillations in the transmission potential (Lourenço and Recker, 2014; Mordecai et al., 2016; Mourya et al., 2004; Feldstein et al., 2015). Rainfall, on the other hand, only seemed to play a marginal role and we argue that it may be a relevant player for arboviral transmission mainly in tropical regions subject to intense rain seasons, such as areas in South East Asia (Cuong et al., 2011; Hii et al., 2012; Xuan et al., 2014). We also noted that the correlations between climatic variables and case counts were inverted when addressed against the transmission potential. For instance, while temperature was positively correlated with $R_0$ it was negatively correlated with Zika cases. This implies that the transmission potential is readily responsive to climatic variation but that the Zika epidemics in FSA showed a slight but expected delay in relation to the peak in transmission potential, with case numbers generally increasing after a stable period of maximum $R_0$, followed by epidemic peaks that tended to coincide with declining $R_0$. An interesting observation is that the 2015 epidemic peaked approximately 3 months after the estimated peak in the virus’s transmission potential, whereas there was much higher synchrony during the second wave in 2016. The same behaviour has been described for the CHIKV outbreak in FSA in 2014–2015 and which has been linked to highly discordant spatial distributions between the first two epidemics (Faria et al., 2016b). It is likely that similar spatial effects (Kraemer et al., 2017) were present in FSA’s ZIKV outbreaks. Unfortunately we did not have access to sufficiently detailed spatial data to explore this hypothesis further.

A phylogenetic analysis has proposed that the introduction of ZIKV into Bahia took place between March and September 2014, although without direct evidence for its circulation in FSA at that time (Naccache et al., 2016). Our estimated date of introduction showed support for a date in early-mid December 2014, a few months after the proposed introduction into Bahia and just over 7 weeks before the first case of Zika was notified in FSA. Similar periods between the first notification and estimated introduction often represent the time taken to complete one or more full transmission cycles (human-mosquito-human) before a cluster of cases is generated of sufficient size for detection by passive surveillance systems (Lourenço and Recker, 2014). The case data also shows a 2 months period after the first notification during which weekly case numbers remained extremely low. This long period was unexpected as persistent circulation of ZIKV could hardly be justified by the observed total of only 10 cases. Given our estimated observation rate, however, the number of ZIKV infections during this time could have amounted to over 2700 actual cases. In April, the number of cases increased rapidly, coinciding with the Micareta festival, which we argue may have played a role in igniting the exponential phase of the epidemic by facilitating human-vector mixing as well as a more rapid geographical expansion.

After calibrating our model to the 2015–2017 epidemic, we projected the transmission of ZIKV beyond 2017 using stochastic simulations and average climatic variables for this region. Without the possibility of externally acquired infections, local extinction was very likely by 2017 due to the high levels of herd-immunity. According to our study, Zika’s reproductive potential ($R_e$) reached its lowest point in 2017, and it is expected to remain low for the next couple of years, given the slow replenishment of susceptibles in the population through births. When explicitly modelling the importation of infectious cases our projections for the next two decades corroborated the conclusions of previous modelling studies that suggest a weak endemic potential for ZIKV after the initial exhaustion of the susceptible pool (Ferguson et al., 2016; Kucharski et al., 2016). However, our simulations also showed that the future epidemic behaviour is strongly dependent on the frequency of re-introductions, where sporadic and unpredictable epidemics could still be in the order of hundreds of cases. Furthermore, given our estimated observation rate for the 2015–2017 epidemic, passive surveillance systems are unlikely to fully detect the scale and occurrence of such small epidemics, missing their actual public health impact, and as such efforts should thus be placed to improve ZIKV detection and diagnosis in order to optimize the local reporting rates and potential for control.

Human sexual and vertical transmission of ZIKV is an important public health concern, especially within the context of potential Zika-associated microcephaly (MC) and other neurological complications in pre- and neonatals. With a total of over 10,000 live births in 2015 in FSA, our crude estimate for the risk of Zika-associated MC per pregnancy was below 4 cases per 100,000 individuals in a generalized population under an attack rate of 65%. As discussed elsewhere (Cauchemez et al., 2016), this risk is extremely low when compared to other known viral-associated complications, such as those caused by infections by cytomegalovirus (CMV) and the rubella virus (RV) (De Santis et al., 2006; Naing et al., 2016). It is therefore crucial to reiterate that what makes the ZIKV a public health concern is not necessarily the per pregnancy risk of neurological complications, but rather the combination of low risk with very high attack rates. Other studies have reported that the risk for complications during the $1^{st}$ trimester of gestation is higher than the one estimated here. For example, in the French Polynesia (FP) outbreak (Cauchemez et al., 2016), the risk associated with ZIKV infection during the $1^{st}$ trimester was 1%, while the overall, full pregnancy risk was 0.42%, similar to our FSA estimates. For the Yap Island epidemic, no microcephaly cases have been reported. With an estimated 24 births per 1000 females (census 2000 as in (Duffy et al., 2009)) and using an overall risk of approximately 0.4% per pregnancy, only 0–3 cases per 100,000 individuals would have been expected. However, the island’s small population size (7391 individuals (Duffy et al., 2009)) together with a general baseline of 0–2 microcephaly cases per 100,000 in many areas of the world (Johansson et al., 2016; Butler, 2016; EUROCAT, 2003) would explain the absence of reported cases. It is also important to consider that a variety of birth defects have been found to be statistically associated with Zika virus infection during pregnancy, of which MC is one possible outcome. While the risk for birth defects per pregnancy is consistently reported to be high, estimations for the risk of MC vary considerably. For example, recent clinical trials (Honein et al., 2017; Brasil et al., 2016) suggested that the risk of Zika-associated MC could be an order of magnitude higher than the estimate reported in this or other previous studies (Cauchemez et al., 2016; Duffy et al., 2009). At this stage it is not possible to explain these differences, but it is tempting to speculate that other factors must influence either the actual or estimated risk. For example, there could be diagnostic biases or differences between epidemiological and clinical studies. Alternatively, viral or host genetic background, as well as the pre-exposition to other arboviruses may influence the absolute risk experienced by local populations or cohorts.

Official notification of Zika infections in Brazil started on the $1^{st}$ of January 2016, although cases were reported in many other regions in Brazil during 2015. It is therefore plausible that the observation rate changed upon official guidelines and that the capacity to accurately diagnose and report Zika infections could have been lower in 2015 compared subsequent years. To explore this, we reran our fitting approach allowing for a possible change in the observation rate for 2016 and onwards (Figure 3—figure supplement 2) and found a similar observation rate for 2015 (0.0039 versus 0.0034) as well as a similar attack rate between the two model variants. However, the estimated observation rate for 2016 and beyond was $\approx 4$ times larger than for 2015, implying a positive change due to changes in the surveillance system. Nevertheless, only about 13–14 out of 1000 Zika cases were reported after the $1^{st}$ of January 2016. It is hard to discern where the positive changes took place, but we suggest the revised diagnosis guidelines may have increased the proportion correctly diagnosed while the proportion of symptomatic individuals visiting medical facilities did not change. It is also tempting to speculate that the 2015/2016 imbalance in reporting may have been a general phenomenon across Brazil. As described elsewhere, it is thus possible that FSA is a good example of states and urban centres that may have witnessed larger epidemics than reported in 2015 (de Oliveira et al., 2017). This, together with our conclusion that low MC risk with very high attack rates makes ZIKV a public health concern, could explain why most MC reports at the level of the country were in 2015 (de Oliveira et al., 2017), although for many regions the total reported number of ZIKV cases may have been surprisingly small that year.

There are certain limitations to our approach, many of which could be revisited when more detailed data becomes available. For example, we assumed homogeneous mixing between human and mosquito hosts but it is possible that spatio-temporal heterogeneities may have played a role in FSA. Furthermore, we have curated and integrated functional responses of key entomological parameters to temperature, rainfall and humidity variation, which were originally reported for dengue viruses. Our fitting approach is also dependent on notified case data and it is possible that the reported cases are not representative of the initial expansion of the virus, which may have thwarted the obtained posterior of the introduction date. Finally, our future projections for the endemic and epidemic potential of ZIKV are based on average climatic trends of past years and do not capture the occurrence of natural variation between years, in particular for years affected by major Southern American climate events, such as the El Niño (Caminade et al., 2017).

In this study we have addressed the local determinants of ZIKV epidemiology in the context of a major urban centre of Brazil. Our results imply that control and surveillance of ZIKV should be boosted and focused in periods of high temperature and during major social events. These factors could identify windows of opportunity for local interventions to mitigate ZIKV introduction and transmission and should be transferable to other areas for which both temperature data and community event schedules are available. We further confirm that the high transmission potential of ZIKV in urban centres can lead to the exhaustion of the local susceptible pool, which will in turn dictate the long-term epidemic and endemic behaviour of the virus. Depending on the rate of re-introduction, sporadic outbreaks are to be expected, although these will be unlikely to result in a notable increase in the number of microcephaly cases due to their limited sizes and low risk per pregnancy. Nonetheless, these local sporadic occurrences could still have important public health consequences, and we argue that much better diagnostics and reporting rates are required for local authorities to detect and respond to such events in the near future. Our integrated mathematical framework is capable of deriving key insights into the past and future determinants of ZIKV epidemiology and its findings should be applicable to other major urban centres of Brazil and elsewhere.

1. 1. Alto BW
   2. Juliano SA

   (2001) Precipitation and temperature effects on populations of Aedes albopictus (Diptera: Culicidae): implications for range expansion

   Journal of Medical Entomology 38:646–656.

   https://doi.org/10.1603/0022-2585-38.5.646

   • PubMed
   • Google Scholar
2. 1. Aman MJ
   2. Kashanchi F

   (2016) Zika Virus: A New Animal Model for an Arbovirus

   PLOS Neglected Tropical Diseases 10:e0004702–e0004703.

   https://doi.org/10.1371/journal.pntd.0004702

   • PubMed
   • Google Scholar
3. 1. Andraud M
   2. Hens N
   3. Marais C
   4. Beutels P

   (2012) Dynamic epidemiological models for dengue transmission: a systematic review of structural approaches

   PLoS One 7:e49085.

   https://doi.org/10.1371/journal.pone.0049085

   • PubMed
   • Google Scholar
4. 1. Bewick S
   2. Fagan WF
   3. Calabrese JM
   4. Agusto F

   (2016)

   Zika Virus: Endemic Versus Epidemic Dynamics and Implications for Disease Spread in the Americas

   bioRxiv 041897.

   • Google Scholar
5. 1. Bicout DJ
   2. Sabatier P

   (2004) Mapping Rift Valley Fever vectors and prevalence using rainfall variations

   Vector-Borne and Zoonotic Diseases 4:33–42.

   https://doi.org/10.1089/153036604773082979

   • PubMed
   • Google Scholar
6. 1. Bogoch II
   2. Brady OJ
   3. Kraemer MUG
   4. German M
   5. Creatore MI
   6. Brent S
   7. Watts AG
   8. Hay SI
   9. Kulkarni MA
   10. Brownstein JS
   11. Khan K

   (2016) Potential for Zika virus introduction and transmission in resource-limited countries in Africa and the Asia-Pacific region: a modelling study

   The Lancet Infectious Diseases 16:1237–1245.

   https://doi.org/10.1016/S1473-3099(16)30270-5

   • PubMed
   • Google Scholar
7. 1. Brady OJ
   2. Johansson MA
   3. Guerra CA
   4. Bhatt S
   5. Golding N
   6. Pigott DM
   7. Delatte H
   8. Grech MG
   9. Leisnham PT
   10. Maciel-de-Freitas R
   11. Styer LM
   12. Smith DL
   13. Scott TW
   14. Gething PW
   15. Hay SI

   (2013) Modelling adult Aedes aegypti and Aedes albopictus survival at different temperatures in laboratory and field settings

   Parasites & Vectors 6:351.

   https://doi.org/10.1186/1756-3305-6-351

   • PubMed
   • Google Scholar
8. 1. Brasil P
   2. Pereira JP
   3. Moreira ME
   4. Ribeiro Nogueira RM
   5. Damasceno L
   6. Wakimoto M
   7. Rabello RS
   8. Valderramos SG
   9. Halai U-A
   10. Salles TS
   11. Zin AA
   12. Horovitz D
   13. Daltro P
   14. Boechat M
   15. Raja Gabaglia C
   16. Carvalho de Sequeira P
   17. Pilotto JH
   18. Medialdea-Carrera R
   19. Cotrim da Cunha D
   20. Abreu de Carvalho LM
   21. Pone M
   22. Machado Siqueira A
   23. Calvet GA
   24. Rodrigues Baião AE
   25. Neves ES
   26. Nassar de Carvalho PR
   27. Hasue RH
   28. Marschik PB
   29. Einspieler C
   30. Janzen C
   31. Cherry JD
   32. Bispo de Filippis AM
   33. Nielsen-Saines K

   (2016) Zika Virus Infection in Pregnant Women in Rio de Janeiro

   New England Journal of Medicine 375:2321–2334.

   https://doi.org/10.1056/NEJMoa1602412

   • Google Scholar
9. Website

   1. Brazil BDMEP

   (1961) Banco de Dados Meteorológicos para Ensino e Pesquisa

   Accessed June 2017.

   http://www.inmet.gov.br/projetos/rede/pesquisa/
10. Website

    1. Brazil SINAN

    (2016) Sistema de Informação de Agravos de Notificação (national notifiable diseases information system)

    Accessed June 2017.

    http://portalsinan.saude.gov.br
11. Website

    1. Brazil ZG

    (2016) National guidelines for Zika virus pathology and diagnosis

    Accessed January 2017.

    http://portalsaude.saude.gov.br/images/pdf/2016/marco/07/Nota-Informativa-zika.pdf
12. 1. Butler D

    (2016) Zika virus: Brazil's surge in small-headed babies questioned by report

    Nature 530:13–14.

    https://doi.org/10.1038/nature.2016.19259

    • PubMed
    • Google Scholar
13. 1. Caminade C
    2. Turner J
    3. Metelmann S
    4. Hesson JC
    5. Blagrove MS
    6. Solomon T
    7. Morse AP
    8. Baylis M

    (2017) Global risk model for vector-borne transmission of Zika virus reveals the role of El Niño 2015

    PNAS 114:119–124.

    https://doi.org/10.1073/pnas.1614303114

    • PubMed
    • Google Scholar
14. 1. Campos GS
    2. Bandeira AC
    3. Sardi SI

    (2015) Zika Virus Outbreak, Bahia, Brazil

    Emerging Infectious Diseases 21:1885–1886.

    https://doi.org/10.3201/eid2110.150847

    • PubMed
    • Google Scholar
15. 1. Carlson CJ
    2. Dougherty ER
    3. Getz W

    (2016) An Ecological Assessment of the Pandemic Threat of Zika Virus

    PLOS Neglected Tropical Diseases 10:e0004968.

    https://doi.org/10.1371/journal.pntd.0004968

    • PubMed
    • Google Scholar
16. 1. Cauchemez S
    2. Besnard M
    3. Bompard P
    4. Dub T
    5. Guillemette-Artur P
    6. Eyrolle-Guignot D
    7. Salje H
    8. Van Kerkhove MD
    9. Abadie V
    10. Garel C
    11. Fontanet A
    12. Mallet HP

    (2016) Association between Zika virus and microcephaly in French Polynesia, 2013-15: a retrospective study

    The Lancet 387:2125–2132.

    https://doi.org/10.1016/S0140-6736(16)00651-6

    • PubMed
    • Google Scholar
17. 1. Champagne C
    2. Salthouse DG
    3. Paul R
    4. Cao-Lormeau VM
    5. Roche B
    6. Cazelles B

    (2016) Structure in the variability of the basic reproductive number (R0) for Zika epidemics in the Pacific islands

    eLife 5:e19874.

    https://doi.org/10.7554/eLife.19874

    • PubMed
    • Google Scholar
18. 1. Chowell G
    2. Hincapie-Palacio D
    3. Ospina J
    4. Pell B
    5. Tariq A
    6. Dahal S
    7. Moghadas S
    8. Smirnova A
    9. Simonsen L
    10. Viboud C

    (2016) Using phenomenological models to characterize transmissibility and forecast patterns and final burden of zika epidemics

    PLoS Currents 8:.

    https://doi.org/10.1371/currents.outbreaks.f14b2217c902f453d9320a43a35b9583

    • PubMed
    • Google Scholar
19. 1. Cuong HQ
    2. Hien NT
    3. Duong TN
    4. Phong TV
    5. Cam NN
    6. Farrar J
    7. Nam VS
    8. Thai KT
    9. Horby P

    (2011) Quantifying the emergence of dengue in Hanoi, Vietnam: 1998-2009

    PLoS Neglected Tropical Diseases 5:e1322.

    https://doi.org/10.1371/journal.pntd.0001322

    • PubMed
    • Google Scholar
20. 1. de Araújo TVB
    2. Rodrigues LC
    3. de Alencar Ximenes RA
    4. de Barros Miranda-Filho D
    5. Montarroyos UR
    6. de Melo APL
    7. Valongueiro S
    8. de Albuquerque M
    9. Souza WV
    10. Braga C
    11. Filho SPB
    12. Cordeiro MT
    13. Vazquez E
    14. Di Cavalcanti Souza Cruz D
    15. Henriques CMP
    16. Bezerra LCA
    17. da Silva Castanha PM
    18. Dhalia R
    19. Marques-Júnior ETA
    20. Martelli CMT
    21. investigators from the Microcephaly Epidemic Research Group
    22. Brazilian Ministry of Health
    23. Pan American Health Organization
    24. Instituto de Medicina Integral Professor Fernando Figueira
    25. State Health Department of Pernambuco

    (2016) Association between Zika virus infection and microcephaly in Brazil, January to May, 2016: preliminary report of a case-control study

    The Lancet Infectious Diseases 16:1356–1363.

    https://doi.org/10.1016/S1473-3099(16)30318-8

    • PubMed
    • Google Scholar
21. 1. de Oliveira WK
    2. de França GVA
    3. Carmo EH
    4. Duncan BB
    5. de Souza Kuchenbecker R
    6. Schmidt MI

    (2017) Infection-related microcephaly after the 2015 and 2016 Zika virus outbreaks in Brazil: a surveillance-based analysis

    The Lancet 390:861–870.

    https://doi.org/10.1016/S0140-6736(17)31368-5

    • PubMed
    • Google Scholar
22. 1. De Santis M
    2. Cavaliere AF
    3. Straface G
    4. Caruso A

    (2006) Rubella infection in pregnancy

    Reproductive Toxicology 21:390–398.

    https://doi.org/10.1016/j.reprotox.2005.01.014

    • PubMed
    • Google Scholar
23. 1. Dorigatti I
    2. Cauchemez S
    3. Ferguson NM

    (2013) Increased transmissibility explains the third wave of infection by the 2009 H1N1 pandemic virus in England

    PNAS 110:13422–13427.

    https://doi.org/10.1073/pnas.1303117110

    • PubMed
    • Google Scholar
24. 1. Dowall SD
    2. Graham VA
    3. Rayner E
    4. Atkinson B
    5. Hall G
    6. Watson RJ
    7. Bosworth A
    8. Bonney LC
    9. Kitchen S
    10. Hewson R

    (2016) A Susceptible Mouse Model for Zika Virus Infection

    PLOS Neglected Tropical Diseases 10:e0004658.

    https://doi.org/10.1371/journal.pntd.0004658

    • PubMed
    • Google Scholar
25. 1. Duffy MR
    2. Chen TH
    3. Hancock WT
    4. Powers AM
    5. Kool JL
    6. Lanciotti RS
    7. Pretrick M
    8. Marfel M
    9. Holzbauer S
    10. Dubray C
    11. Guillaumot L
    12. Griggs A
    13. Bel M
    14. Lambert AJ
    15. Laven J
    16. Kosoy O
    17. Panella A
    18. Biggerstaff BJ
    19. Fischer M
    20. Hayes EB

    (2009) Zika virus outbreak on Yap Island, Federated States of Micronesia

    New England Journal of Medicine 360:2536–2543.

    https://doi.org/10.1056/NEJMoa0805715

    • PubMed
    • Google Scholar
26. Website

    1. EUROCAT

    (2003) European surveillance of congenital anomalies, final activity report 2002-2003

    Accessed January 2017.

    http://ec.europa.eu/health/ ph_threats/non_com/docs/eurocat_en.pdf
27. 1. Fagbami AH

    (1979) Zika virus infections in Nigeria: virological and seroepidemiological investigations in Oyo State

    Journal of Hygiene 83:213–219.

    https://doi.org/10.1017/S0022172400025997

    • PubMed
    • Google Scholar
28. 1. Faria NR
    2. Azevedo R
    3. Kraemer MUG
    4. Souza R
    5. Cunha MS
    6. Hill SC
    7. Thézé J
    8. Bonsall MB
    9. Bowden TA
    10. Rissanen I
    11. Rocco IM
    12. Nogueira JS
    13. Maeda AY
    14. Vasami F
    15. Macedo FLL
    16. Suzuki A
    17. Rodrigues SG
    18. Cruz ACR
    19. Nunes BT
    20. Medeiros DBA
    21. Rodrigues DSG
    22. Queiroz ALN
    23. da Silva EVP
    24. Henriques DF
    25. da Rosa EST
    26. de Oliveira CS
    27. Martins LC
    28. Vasconcelos HB
    29. Casseb LMN
    30. Simith DB
    31. Messina JP
    32. Abade L
    33. Lourenço J
    34. Alcantara LCJ
    35. de Lima MM
    36. Giovanetti M
    37. Hay SI
    38. de Oliveira RS
    39. Lemos PDS
    40. de Oliveira LF
    41. de Lima CPS
    42. da Silva SP
    43. de Vasconcelos JM
    44. Franco L
    45. Cardoso JF
    46. Vianez-Júnior J
    47. Mir D
    48. Bello G
    49. Delatorre E
    50. Khan K
    51. Creatore M
    52. Coelho GE
    53. de Oliveira WK
    54. Tesh R
    55. Pybus OG
    56. Nunes MRT
    57. Vasconcelos PFC

    (2016a) Zika virus in the Americas: Early epidemiological and genetic findings

    Science 352:345–349.

    https://doi.org/10.1126/science.aaf5036

    • PubMed
    • Google Scholar
29. 1. Faria NR
    2. Lourenço J
    3. Marques de Cerqueira E
    4. Maia de Lima M
    5. Carlos Junior Alcantara L

    (2016b) Epidemiology of Chikungunya Virus in Bahia, Brazil, 2014-2015

    PLoS Currents 8:.

    https://doi.org/10.1371/currents.outbreaks.c97507e3e48efb946401755d468c28b2

    • Google Scholar
30. 1. Faria NR
    2. Quick J
    3. Claro IM
    4. Thézé J
    5. de Jesus JG
    6. Giovanetti M
    7. Kraemer MUG
    8. Hill SC
    9. Black A
    10. da Costa AC
    11. Franco LC
    12. Silva SP
    13. Wu CH
    14. Raghwani J
    15. Cauchemez S
    16. du Plessis L
    17. Verotti MP
    18. de Oliveira WK
    19. Carmo EH
    20. Coelho GE
    21. Santelli A
    22. Vinhal LC
    23. Henriques CM
    24. Simpson JT
    25. Loose M
    26. Andersen KG
    27. Grubaugh ND
    28. Somasekar S
    29. Chiu CY
    30. Muñoz-Medina JE
    31. Gonzalez-Bonilla CR
    32. Arias CF
    33. Lewis-Ximenez LL
    34. Baylis SA
    35. Chieppe AO
    36. Aguiar SF
    37. Fernandes CA
    38. Lemos PS
    39. Nascimento BLS
    40. Monteiro HAO
    41. Siqueira IC
    42. de Queiroz MG
    43. de Souza TR
    44. Bezerra JF
    45. Lemos MR
    46. Pereira GF
    47. Loudal D
    48. Moura LC
    49. Dhalia R
    50. França RF
    51. Magalhães T
    52. Marques ET
    53. Jaenisch T
    54. Wallau GL
    55. de Lima MC
    56. Nascimento V
    57. de Cerqueira EM
    58. de Lima MM
    59. Mascarenhas DL
    60. Neto JPM
    61. Levin AS
    62. Tozetto-Mendoza TR
    63. Fonseca SN
    64. Mendes-Correa MC
    65. Milagres FP
    66. Segurado A
    67. Holmes EC
    68. Rambaut A
    69. Bedford T
    70. Nunes MRT
    71. Sabino EC
    72. Alcantara LCJ
    73. Loman NJ
    74. Pybus OG

    (2017) Establishment and cryptic transmission of Zika virus in Brazil and the Americas

    Nature 546:406–410.

    https://doi.org/10.1038/nature22401

    • PubMed
    • Google Scholar
31. 1. Faria NR
    2. Sabino EC
    3. Nunes MR
    4. Alcantara LC
    5. Loman NJ
    6. Pybus OG

    (2016c) Mobile real-time surveillance of Zika virus in Brazil

    Genome Medicine 8:97.

    https://doi.org/10.1186/s13073-016-0356-2

    • PubMed
    • Google Scholar
32. 1. Feldstein LR
    2. Brownstein JS
    3. Brady OJ
    4. Hay SI
    5. Johansson MA

    (2015) Dengue on islands: a Bayesian approach to understanding the global ecology of dengue viruses

    Transactions of The Royal Society of Tropical Medicine and Hygiene 109:303–312.

    https://doi.org/10.1093/trstmh/trv012

    • PubMed
    • Google Scholar
33. 1. Ferguson NM
    2. Cucunuba ZM
    3. Dorigatti I
    4. Nedjati-Gilani GL
    5. Donnelly CA
    6. Basanez M-G
    7. Nouvellet P
    8. Lessler J

    (2016) Countering the Zika epidemic in Latin America

    Science 353:353–354.

    https://doi.org/10.1126/science.aag0219

    • Google Scholar
34. 1. Focks DA
    2. Daniels E
    3. Haile DG
    4. Keesling JE

    (1995) A simulation model of the epidemiology of urban dengue fever: literature analysis, model development, preliminary validation, and samples of simulation results

    The American Journal of Tropical Medicine and Hygiene 53:489–506.

    https://doi.org/10.4269/ajtmh.1995.53.489

    • PubMed
    • Google Scholar
35. 1. Foy BD
    2. Kobylinski KC
    3. Chilson Foy JL
    4. Blitvich BJ
    5. Travassos da Rosa A
    6. Haddow AD
    7. Lanciotti RS
    8. Tesh RB

    (2011) Probable non-vector-borne transmission of Zika virus, Colorado, USA

    Emerging Infectious Diseases 17:880–882.

    https://doi.org/10.3201/eid1705.101939

    • PubMed
    • Google Scholar
36. 1. Funk S
    2. Kucharski AJ
    3. Camacho A
    4. Eggo RM
    5. Yakob L
    6. Murray LM
    7. Edmunds WJ

    (2016) Comparative analysis of dengue and zika outbreaks reveals differences by setting and virus

    PLOS Neglected Tropical Diseases 10:e0005173.

    https://doi.org/10.1371/journal.pntd.0005173

    • PubMed
    • Google Scholar
37. 1. Gao D
    2. Lou Y
    3. He D
    4. Porco TC
    5. Kuang Y
    6. Chowell G
    7. Ruan S

    (2016) Prevention and Control of Zika as a Mosquito-Borne and Sexually Transmitted Disease: A Mathematical Modeling Analysis

    Scientific Reports 6:28070.

    https://doi.org/10.1038/srep28070

    • PubMed
    • Google Scholar
38. Website

    1. Golden Gate Weather Services

    (2017) El Niño and La Niña Years and intensities - based on oceanic Niño Index (ONI) - Golden Gate Weather Services

    Accessed June 2017.

    http://ggweather.com/enso/oni.htm
39. 1. Hii YL
    2. Zhu H
    3. Ng N
    4. Ng LC
    5. Rocklöv J

    (2012) Forecast of dengue incidence using temperature and rainfall

    PLoS Neglected Tropical Diseases 6:e1908.

    https://doi.org/10.1371/journal.pntd.0001908

    • PubMed
    • Google Scholar
40. 1. Honein MA
    2. Dawson AL
    3. Petersen EE
    4. Jones AM
    5. Lee EH
    6. Yazdy MM
    7. Ahmad N
    8. Macdonald J
    9. Evert N
    10. Bingham A
    11. Ellington SR
    12. Shapiro-Mendoza CK
    13. Oduyebo T
    14. Fine AD
    15. Brown CM
    16. Sommer JN
    17. Gupta J
    18. Cavicchia P
    19. Slavinski S
    20. White JL
    21. Owen SM
    22. Petersen LR
    23. Boyle C
    24. Meaney-Delman D
    25. Jamieson DJ
    26. US Zika Pregnancy Registry Collaboration

    (2017) Birth defects among fetuses and infants of us women with evidence of possible zika virus infection during pregnancy

    JAMA 317:59.

    https://doi.org/10.1001/jama.2016.19006

    • PubMed
    • Google Scholar
41. 1. Hugo LE
    2. Jeffery JA
    3. Trewin BJ
    4. Wockner LF
    5. Nguyen TY
    6. Nguyen HL
    7. Nghia leT
    8. Hine E
    9. Ryan PA
    10. Kay BH

    (2014) Adult survivorship of the dengue mosquito Aedes aegypti varies seasonally in central Vietnam

    PLoS Neglected Tropical Diseases 8:e2669.

    https://doi.org/10.1371/journal.pntd.0002669

    • PubMed
    • Google Scholar
42. 1. Jaenisch T
    2. Rosenberger K
    3. Brito C
    4. Brady O
    5. Brasil P
    6. Marques E

    (2016) Estimating the risk for microcephaly after Zika virus infection in Brazil

    Bulletin of the World Health Organization pp. 1–15.

    https://doi.org/10.2471/BLT.16.178608

    • Google Scholar
43. 1. Johansson MA
    2. Mier-y-Teran-Romero L
    3. Reefhuis J
    4. Gilboa SM
    5. Hills SL

    (2016) Zika and the Risk of Microcephaly

    New England Journal of Medicine 375:1–4.

    https://doi.org/10.1056/NEJMp1605367

    • PubMed
    • Google Scholar
44. 1. Kraemer MU
    2. Sinka ME
    3. Duda KA
    4. Mylne AQ
    5. Shearer FM
    6. Barker CM
    7. Moore CG
    8. Carvalho RG
    9. Coelho GE
    10. Van Bortel W
    11. Hendrickx G
    12. Schaffner F
    13. Elyazar IR
    14. Teng HJ
    15. Brady OJ
    16. Messina JP
    17. Pigott DM
    18. Scott TW
    19. Smith DL
    20. Wint GR
    21. Golding N
    22. Hay SI

    (2015) The global distribution of the arbovirus vectors Aedes aegypti and Ae. albopictus

    eLife 4:e08347.

    https://doi.org/10.7554/eLife.08347

    • PubMed
    • Google Scholar
45. 1. Kraemer MUG
    2. Faria NR
    3. Reiner RC
    4. Golding N
    5. Nikolay B
    6. Stasse S
    7. Johansson MA
    8. Salje H
    9. Faye O
    10. Wint GRW
    11. Niedrig M
    12. Shearer FM
    13. Hill SC
    14. Thompson RN
    15. Bisanzio D
    16. Taveira N
    17. Nax HH
    18. Pradelski BSR
    19. Nsoesie EO
    20. Murphy NR
    21. Bogoch II
    22. Khan K
    23. Brownstein JS
    24. Tatem AJ
    25. de Oliveira T
    26. Smith DL
    27. Sall AA
    28. Pybus OG
    29. Hay SI
    30. Cauchemez S

    (2017) Spread of yellow fever virus outbreak in Angola and the Democratic Republic of the Congo 2015-16: a modelling study

    The Lancet Infectious Diseases 17:330–338.

    https://doi.org/10.1016/S1473-3099(16)30513-8

    • PubMed
    • Google Scholar
46. 1. Kucharski AJ
    2. Funk S
    3. Eggo RM
    4. Mallet HP
    5. Edmunds WJ
    6. Nilles EJ

    (2016) Transmission Dynamics of Zika Virus in Island Populations: A Modelling Analysis of the 2013-14 French Polynesia Outbreak

    PLOS Neglected Tropical Diseases 10:e0004726.

    https://doi.org/10.1371/journal.pntd.0004726

    • PubMed
    • Google Scholar
47. 1. Lambrechts L
    2. Paaijmans KP
    3. Fansiri T
    4. Carrington LB
    5. Kramer LD
    6. Thomas MB
    7. Scott TW

    (2011) Impact of daily temperature fluctuations on dengue virus transmission by Aedes aegypti

    PNAS 108:7460–7465.

    https://doi.org/10.1073/pnas.1101377108

    • PubMed
    • Google Scholar
48. 1. Lampoudi S
    2. Gillespie DT
    3. Petzold LR

    (2009) The multinomial simulation algorithm for discrete stochastic simulation of reaction-diffusion systems

    The Journal of Chemical Physics 130:094104.

    https://doi.org/10.1063/1.3074302

    • PubMed
    • Google Scholar
49. 1. Lessler J
    2. Ott CT
    3. Carcelen AC
    4. Konikoff JM
    5. Williamson J
    6. Bi Q
    7. Kucirka LM
    8. Cummings DA
    9. Reich NG
    10. Chaisson LH

    (2016) Times to key events in Zika virus infection and implications for blood donation: a systematic review

    Bulletin of the World Health Organization 94:841–849.

    https://doi.org/10.2471/BLT.16.174540

    • PubMed
    • Google Scholar
50. 1. Li MI
    2. Wong PS
    3. Ng LC
    4. Tan CH

    (2012) Oral Susceptibility of Singapore Aedes (Stegomyia) aegypti (Linnaeus) to Zika Virus

    PLoS Neglected Tropical Diseases 6:e1792.

    https://doi.org/10.1371/journal.pntd.0001792

    • PubMed
    • Google Scholar
51. 1. Lounibos LP
    2. Escher RL

    (2008) Sex ratios of mosquitoes from long-term censuses of Florida tree holes

    Journal of the American Mosquito Control Association 24:11–15.

    https://doi.org/10.2987/5656.1

    • PubMed
    • Google Scholar
52. 1. Lourenço J
    2. Recker M

    (2013) Natural, persistent oscillations in a spatial multi-strain disease system with application to dengue

    PLoS Computational Biology 9:e1003308.

    https://doi.org/10.1371/journal.pcbi.1003308

    • PubMed
    • Google Scholar
53. 1. Lourenço J
    2. Recker M

    (2014) The 2012 Madeira dengue outbreak: epidemiological determinants and future epidemic potential

    PLoS Neglected Tropical Diseases 8:e3083.

    https://doi.org/10.1371/journal.pntd.0003083

    • PubMed
    • Google Scholar
54. 1. Madeira NG
    2. Macharelli CA
    3. Carvalho LR

    (2002) Variation of the oviposition preferences of Aedes aegypti in function of substratum and humidity

    Memórias do Instituto Oswaldo Cruz 97:415–420.

    https://doi.org/10.1590/S0074-02762002000300025

    • PubMed
    • Google Scholar
55. 1. Maxian O
    2. Neufeld A
    3. Talis EJ
    4. Childs LM
    5. Blackwood JC

    (2017) Zika virus dynamics: When does sexual transmission matter?

    Epidemics.

    https://doi.org/10.1016/j.epidem.2017.06.003

    • PubMed
    • Google Scholar
56. 1. Messina JP
    2. Kraemer MU
    3. Brady OJ
    4. Pigott DM
    5. Shearer FM
    6. Weiss DJ
    7. Golding N
    8. Ruktanonchai CW
    9. Gething PW
    10. Cohn E
    11. Brownstein JS
    12. Khan K
    13. Tatem AJ
    14. Jaenisch T
    15. Murray CJ
    16. Marinho F
    17. Scott TW
    18. Hay SI

    (2016) Mapping global environmental suitability for Zika virus

    eLife 5:e15272.

    https://doi.org/10.7554/eLife.15272

    • PubMed
    • Google Scholar
57. 1. Moghadas SM
    2. Shoukat A
    3. Espindola AL
    4. Pereira RS
    5. Abdirizak F
    6. Laskowski M
    7. Viboud C
    8. Chowell G

    (2017) Asymptomatic Transmission and the Dynamics of Zika Infection

    Scientific Reports 7:5829.

    https://doi.org/10.1038/s41598-017-05013-9

    • PubMed
    • Google Scholar
58. 1. Mohammed A
    2. Chadee DD

    (2011) Effects of different temperature regimens on the development of Aedes aegypti (L.) (Diptera: Culicidae) mosquitoes

    Acta Tropica 119:38–43.

    https://doi.org/10.1016/j.actatropica.2011.04.004

    • PubMed
    • Google Scholar
59. Preprint

    1. Mordecai E
    2. Cohen J
    3. Evans MV
    4. Gudapati P
    5. Johnson LR
    6. Miazgowicz K
    7. Murdock CC
    8. Rohr JR
    9. Ryan SJ
    10. Savage V
    11. Shocket M
    12. Ibarra AS
    13. Thomas MB
    14. Weikel DP

    (2016) Temperature determines Zika. dengue and chikungunya transmission potential in the Americas

    bioRxiv.

    https://www.biorxiv.org/content/early/2016/07/15/063735

    • Google Scholar
60. 1. Mourya DT
    2. Yadav P
    3. Mishra AC

    (2004)

    Effect of temperature stress on immature stages and susceptibility of Aedes aegypti mosquitoes to chikungunya virus

    The American journal of tropical medicine and hygiene 70:346–350.

    • PubMed
    • Google Scholar
61. 1. Naccache SN
    2. Thézé J
    3. Sardi SI
    4. Somasekar S
    5. Greninger AL
    6. Bandeira AC
    7. Campos GS
    8. Tauro LB
    9. Faria NR
    10. Pybus OG
    11. Chiu CY

    (2016) Distinct Zika Virus Lineage in Salvador, Bahia, Brazil

    Emerging Infectious Diseases 22:1788–1792.

    https://doi.org/10.3201/eid2210.160663

    • PubMed
    • Google Scholar
62. 1. Naing ZW
    2. Scott GM
    3. Shand A
    4. Hamilton ST
    5. van Zuylen WJ
    6. Basha J
    7. Hall B
    8. Craig ME
    9. Rawlinson WD

    (2016) Congenital cytomegalovirus infection in pregnancy: a review of prevalence, clinical features, diagnosis and prevention

    Australian and New Zealand Journal of Obstetrics and Gynaecology 56:9–18.

    https://doi.org/10.1111/ajo.12408

    • PubMed
    • Google Scholar
63. 1. Nishiura H
    2. Kinoshita R
    3. Mizumoto K
    4. Yasuda Y
    5. Nah K

    (2016) Transmission potential of Zika virus infection in the South Pacific

    International Journal of Infectious Diseases 45:95–97.

    https://doi.org/10.1016/j.ijid.2016.02.017

    • PubMed
    • Google Scholar
64. 1. Nunes MR
    2. Faria NR
    3. de Vasconcelos JM
    4. Golding N
    5. Kraemer MU
    6. de Oliveira LF
    7. Azevedo RS
    8. da Silva DE
    9. da Silva EV
    10. da Silva SP
    11. Carvalho VL
    12. Coelho GE
    13. Cruz AC
    14. Rodrigues SG
    15. Vianez JL
    16. Nunes BT
    17. Cardoso JF
    18. Tesh RB
    19. Hay SI
    20. Pybus OG
    21. Vasconcelos PF

    (2015) Emergence and potential for spread of Chikungunya virus in Brazil

    BMC Medicine 13:102.

    https://doi.org/10.1186/s12916-015-0348-x

    • PubMed
    • Google Scholar
65. 1. Otero M
    2. Solari HG
    3. Schweigmann N

    (2006) A stochastic population dynamics model for Aedes aegypti: formulation and application to a city with temperate climate

    Bulletin of Mathematical Biology 68:1945–1974.

    https://doi.org/10.1007/s11538-006-9067-y

    • PubMed
    • Google Scholar
66. 1. Perkins TA
    2. Siraj AS
    3. Ruktanonchai CW
    4. Kraemer MU
    5. Tatem AJ

    (2016) Model-based projections of Zika virus infections in childbearing women in the Americas

    Nature Microbiology 1:16126.

    https://doi.org/10.1038/nmicrobiol.2016.126

    • PubMed
    • Google Scholar
67. 1. Poletti P
    2. Messeri G
    3. Ajelli M
    4. Vallorani R
    5. Rizzo C
    6. Merler S

    (2011) Transmission potential of chikungunya virus and control measures: the case of Italy

    PLoS One 6:e18860.

    https://doi.org/10.1371/journal.pone.0018860

    • PubMed
    • Google Scholar
68. 1. Rossi MM
    2. Lopez LF
    3. Massad E

    (2015)

    The dynamics of temperature- and rainfall-dependent dengue transmission in tropical regions

    Annals of Biometrics & Biostatistics 2:1–6.

    • Google Scholar
69. 1. Rubin EJ
    2. Greene MF
    3. Baden LR

    (2016) Zika virus and microcephaly

    New England Journal of Medicine 374:984–985.

    https://doi.org/10.1056/NEJMe1601862

    • PubMed
    • Google Scholar
70. 1. Salje H
    2. Cauchemez S
    3. Alera MT
    4. Rodriguez-Barraquer I
    5. Thaisomboonsuk B
    6. Srikiatkhachorn A
    7. Lago CB
    8. Villa D
    9. Klungthong C
    10. Tac-An IA
    11. Fernandez S
    12. Velasco JM
    13. Roque VG
    14. Nisalak A
    15. Macareo LR
    16. Levy JW
    17. Cummings D
    18. Yoon IK

    (2016) Reconstruction of 60 years of chikungunya epidemiology in the Philippines demonstrates episodic and focal transmission

    Journal of Infectious Diseases 213:604–610.

    https://doi.org/10.1093/infdis/jiv470

    • PubMed
    • Google Scholar
71. 1. Schoolfield RM
    2. Sharpe PJ
    3. Magnuson CE

    (1981) Non-linear regression of biological temperature-dependent rate models based on absolute reaction-rate theory

    Journal of Theoretical Biology 88:719–731.

    https://doi.org/10.1016/0022-5193(81)90246-0

    • PubMed
    • Google Scholar
72. 1. Silva MM
    2. Rodrigues MS
    3. Paploski IA
    4. Kikuti M
    5. Kasper AM
    6. Cruz JS
    7. Queiroz TL
    8. Tavares AS
    9. Santana PM
    10. Araújo JM
    11. Ko AI
    12. Reis MG
    13. Ribeiro GS

    (2016) Accuracy of Dengue Reporting by National Surveillance System, Brazil

    Emerging Infectious Diseases 22:336–339.

    https://doi.org/10.3201/eid2202.150495

    • PubMed
    • Google Scholar
73. 1. Soares de Araújo JS
    2. Regis CT
    3. Gomes RG
    4. Tavares TR
    5. Rocha Dos Santos C
    6. Assunção PM
    7. Nóbrega RV
    8. Pinto DF
    9. Bezerra BV
    10. Mattos SD

    (2016) Microcephaly in north-east Brazil: a retrospective study on neonates born between 2012 and 2015

    Bulletin of the World Health Organization 94:835–840.

    https://doi.org/10.2471/BLT.16.170639

    • PubMed
    • Google Scholar
74. 1. Dickerson CZ

    (2007)

    Doctor of Philosophy

    The effects of temperature and humidy on the eggs of Aedes aegypti and Aedes albopictus in Texas, Doctor of Philosophy, College Station, Texas, Texas A and M University.

    • Google Scholar
75. 1. Tran A
    2. L'Ambert G
    3. Lacour G
    4. Benoît R
    5. Demarchi M
    6. Cros M
    7. Cailly P
    8. Aubry-Kientz M
    9. Balenghien T
    10. Ezanno P

    (2013) A rainfall- and temperature-driven abundance model for Aedes albopictus populations

    International Journal of Environmental Research and Public Health 10:1698–1719.

    https://doi.org/10.3390/ijerph10051698

    • PubMed
    • Google Scholar
76. 1. Trpis M
    2. Hausermann W

    (1986) Dispersal and other population parameters of Aedes aegypti in an African village and their possible significance in epidemiology of vector-borne diseases

    The American Journal of Tropical Medicine and Hygiene 35:1263.

    https://doi.org/10.4269/ajtmh.1986.35.1263

    • PubMed
    • Google Scholar
77. 1. Trpis M
    2. Häusermann W
    3. Craig GB

    (1995) Estimates of population size, dispersal, and longevity of domestic Aedes aegypti aegypti (Diptera: Culicidae) by mark-release-recapture in the village of Shauri Moyo in eastern Kenya

    Journal of Medical Entomology 32:27–33.

    https://doi.org/10.1093/jmedent/32.1.27

    • PubMed
    • Google Scholar
78. 1. Turmel JM
    2. Abgueguen P
    3. Hubert B
    4. Vandamme YM
    5. Maquart M
    6. Le Guillou-Guillemette H
    7. Leparc-Goffart I

    (2016) Late sexual transmission of Zika virus related to persistence in the semen

    The Lancet 387:2501.

    https://doi.org/10.1016/S0140-6736(16)30775-9

    • Google Scholar
79. 1. van Panhuis WG
    2. Choisy M
    3. Xiong X
    4. Chok NS
    5. Akarasewi P
    6. Iamsirithaworn S
    7. Lam SK
    8. Chong CK
    9. Lam FC
    10. Phommasak B
    11. Vongphrachanh P
    12. Bouaphanh K
    13. Rekol H
    14. Hien NT
    15. Thai PQ
    16. Duong TN
    17. Chuang JH
    18. Liu YL
    19. Ng LC
    20. Shi Y
    21. Tayag EA
    22. Roque VG
    23. Lee Suy LL
    24. Jarman RG
    25. Gibbons RV
    26. Velasco JM
    27. Yoon IK
    28. Burke DS
    29. Cummings DA

    (2015) Region-wide synchrony and traveling waves of dengue across eight countries in Southeast Asia

    PNAS 112:13069–13074.

    https://doi.org/10.1073/pnas.1501375112

    • PubMed
    • Google Scholar
80. 1. Wearing HJ
    2. Rohani P

    (2006) Ecological and immunological determinants of dengue epidemics

    PNAS 103:11802–11807.

    https://doi.org/10.1073/pnas.0602960103

    • PubMed
    • Google Scholar
81. Website

    1. WHO

    (2016a) Mosquito (vector) control emergency response and preparedness for Zika virus

    World Health Organization. Accessed June 2017.

    http://bit.ly/ZikaEmergencyPrepar
82. Website

    1. WHO

    (2016b) WHO Director-General summarizes the outcome of the Emergency Committee regarding clusters of microcephaly and Guillain-Barré syndrome

    World Health Organization. Accessed June 2017.

    http://bit.ly/HealthEmergency
83. Website

    1. WHO

    (2016c) World Health Organization - Brazil - Statistics

    World Health Organization. Accessed June 2017.

    http://www.who.int/countries/bra/en/.
84. 1. Wong PS
    2. Li MZ
    3. Chong CS
    4. Ng LC
    5. Tan CH

    (2013) Aedes (Stegomyia) albopictus (Skuse): a potential vector of Zika virus in Singapore

    PLoS Neglected Tropical Diseases 7:e2348.

    https://doi.org/10.1371/journal.pntd.0002348

    • PubMed
    • Google Scholar
85. 1. Xuan LTT
    2. Van Hau P
    3. Thu DT
    4. Toan DTT

    (2014) Estimates of meteorological variability in association with dengue cases in a coastal city in northern Vietnam: an ecological study

    Global Health Action 7:23119.

    https://doi.org/10.3402/gha.v7.23119

    • PubMed
    • Google Scholar
86. 1. Yakob L
    2. Kucharski A
    3. Hue S
    4. Edmunds WJ

    (2016) Low risk of a sexually-transmitted Zika virus outbreak

    The Lancet Infectious Diseases 16:1100–1102.

    https://doi.org/10.1016/S1473-3099(16)30324-3

    • PubMed
    • Google Scholar
87. 1. Yang HM
    2. Macoris ML
    3. Galvani KC
    4. Andrighetti MT
    5. Wanderley DM

    (2009) Assessing the effects of temperature on the population of Aedes aegypti, the vector of dengue

    Epidemiology and Infection 137:1188–1202.

    https://doi.org/10.1017/S0950268809002040

    • PubMed
    • Google Scholar
88. 1. Yasuno M
    2. Tonn RJ

    (1970)

    A study of biting habits of Aedes aegypti in Bangkok, Thailand

    Bulletin of the World Health Organization 43:319–325.

    • PubMed
    • Google Scholar
89. 1. Zanluca C
    2. Melo VC
    3. Mosimann AL
    4. Santos GI
    5. Santos CN
    6. Luz K

    (2015) First report of autochthonous transmission of Zika virus in Brazil

    Memórias do Instituto Oswaldo Cruz 110:569–572.

    https://doi.org/10.1590/0074-02760150192

    • PubMed
    • Google Scholar
90. Preprint

    1. Zhang Q
    2. Sun K
    3. Chinazzi M
    4. Pastore-Piontti A
    5. Dean NE
    6. Rojas DP
    7. Merler S
    8. Mistry D
    9. Poletti P
    10. Rossi L
    11. Bray M
    12. Halloran ME
    13. Longini IM
    14. Vespignani A

    (2016) Projected spread of Zika virus in the Americas

    bioRxiv.

    https://www.biorxiv.org/content/early/2016/07/29/066456

    • Google Scholar
91. Website

    1. ZIBRA

    (2016) Zika in Brazil Real Time Analysis Project

    Accessed June 2017.

    http://www.zibraproject.org/

Author details

1. José Lourenço

   Department of Zoology, University of Oxford, Oxford, United Kingdom

   Contribution

   Conceptualization, Data curation, Formal analysis, Methodology

   For correspondence

   jose.lourenco@zoo.ox.ac.uk

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-9318-2581

2. Maricelia Maia de Lima

   Laboratory of Haematology, Genetics and Computational Biology, FIOCRUZ, SalvadorBahia, Brazil

   Contribution

   Conceptualization, Data curation, Investigation, Writing—original draft, Writing—review and editing

   Competing interests

   No competing interests declared

3. Nuno Rodrigues Faria

   Department of Zoology, University of Oxford, Oxford, United Kingdom

   Contribution

   Conceptualization, Resources, Supervision, Investigation, Writing—original draft, Writing—review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-8839-2798

4. Andrew Walker

   Department of Zoology, University of Oxford, Oxford, United Kingdom

   Contribution

   Data curation, Funding acquisition, Project administration

   Competing interests

   No competing interests declared

5. Moritz UG Kraemer

   Department of Zoology, University of Oxford, Oxford, United Kingdom

   Contribution

   Conceptualization, Data curation

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-8838-7147

6. Christian Julian Villabona-Arenas

   Institut de Recherche pour le Développement, UMI 233, INSERM U1175 and Institut de Biologie Computationnelle, LIRMM, Université de Montpellier, Montpellier, France

   Contribution

   Conceptualization, Investigation, Methodology

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-9928-3968

7. Ben Lambert

   Department of Zoology, University of Oxford, Oxford, United Kingdom

   Contribution

   Conceptualization, Investigation, Methodology

   Competing interests

   No competing interests declared

8. Erenilde Marques de Cerqueira

   Centre of PostGraduation in Collective Health, Department of Health, Universidade Estadual de Feira de Santana, Feira de SantanaBahia, Brazil

   Contribution

   Conceptualization, Data curation

   Competing interests

   No competing interests declared

9. Oliver G Pybus

   Department of Zoology, University of Oxford, Oxford, United Kingdom

   Contribution

   Resources, Data curation, Supervision

   Competing interests

   No competing interests declared

10. Luiz CJ Alcantara

    Laboratory of Haematology, Genetics and Computational Biology, FIOCRUZ, SalvadorBahia, Brazil

    Contribution

    Conceptualization, Data curation, Funding acquisition, Investigation, Writing—original draft, Project administration, Writing—review and editing

    Competing interests

    No competing interests declared

11. Mario Recker

    Centre for Mathematics and the Environment, University of Exeter, Penryn, United Kingdom

    Contribution

    Conceptualization, Data curation, Investigation, Writing—original draft, Writing—review and editing

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0001-9489-1315

• 5,013

  Page views

• 783

  Downloads

• 53

  Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.