
==== Front
One Health
One Health
One Health
2352-7714
Elsevier

S2352-7714(24)00214-3
10.1016/j.onehlt.2024.100888
100888
Research Paper
Unveiling spatial patterns of West Nile virus emergence in northern Greece, 2010–2023
Angelou Anastasia a
Schuh Lea b
Stilianakis Nikolaos I. bc
Mourelatos Spiros d
Kioutsioukis Ioannis kioutio@upatras.gr
a⁎
a Department of Physics, University of Patras, Greece
b European Commission, Joint Research Centre (JRC), Ispra, Italy
c Department of Biometry and Epidemiology, University of Erlangen-Nuremberg, Erlangen, Germany
d Ecodevelopment S.A., Thessaloniki, Greece
⁎ Corresponding author. kioutio@upatras.gr
05 9 2024
12 2024
05 9 2024
19 1008885 7 2024
3 9 2024
3 9 2024
© 2024 The Authors
2024
https://creativecommons.org/licenses/by/4.0/ This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
The Region of Central Macedonia (RCM) in Northern Greece recorded the highest number of human West Nile virus (WNV) infections in Greece, despite considerable local mosquito control actions. We examined spatial patterns and associations of mosquito levels, infected mosquito levels, and WNV human cases (WNVhc) across the municipalities of this region over the period 2010–2023 and linked it with climatic characteristics. We combined novel entomological and available epidemiological and climate data for the RCM, aggregated at the municipality level and used Local and Global Moran's I index to assess spatial associations of mosquito levels, infected mosquito levels, and WNVhc. We identified areas with strong interdependencies between adjacent municipalities in the Western part of the region. Furthermore, we employed a Generalized Linear Mixed Model to first, identify the factors driving the observed levels of mosquitoes, infected mosquitoes and WNVhc and second, estimate the influence of climatic features on the observed levels. This modeling approach indicates a strong dependence of the mosquito levels on the temperatures in winter and spring and the total precipitation in early spring, while virus circulation relies on the temperatures of late spring and summer. Our findings highlight the significant influence of climatic factors on mosquito populations (∼60 % explained variance) and the incidence of WNV human cases (∼40 % explained variance), while the unexplained ∼40 % of the variance suggests that targeted interventions and enhanced surveillance in identified hot-spots can enhance public health response.

Highlights

• The entomological and epidemiological “hot-spot” of the region is identified.

• Warm winter and spring and wet March cause an increase in the mosquito abundances.

• Higher temperatures in May and summer lead to an elevated number of human cases.

• Environmental factors account for over 40 % (60 %) of the WNVhc (MCPL) variability.

• By including last year's WNV circulation data, prediction accuracy increases by 13 %.

Keywords

West Nile virus human cases
Mosquito abundance
Infected mosquitoes
Spatial autocorrelation
Moran's I
Climatic factors
Abbreviations

RCM Region of Central Macedonia

WNV West Nile virus

WNVhc WNV human cases

WNF West Nile fever

WNND West Nile neuro-invasive disease

HSA Hellenic Statistical Authority

MCPL Measure of Culex Population Level (mosquitoes/km2): The average mosquito abundance of the month during which the maximum number of mosquitoes were recorded, the month directly preceding and following

MIMS Measure of infected mosquito samplings (%): The highest monthly percentage of infected mosquito samplings, in terms of tested samplings

WNVhcP100TH WNV human cases per 100,000 population

EODY National Public Health Organisation

AAW areas The sum percentage (%) of arable land, heterogeneous agricultural areas and inland wetlands.

GDD Growing Degree Days

GLMM Generalized Linear Mixed Model

MR2 Marginal R-squared

CR2 Conditional R-squared

MAE Mean Absolute Error

NMAE Normalized Mean Absolute Error

LULC Land Use - Land Cover
==== Body
pmc1 Introduction

West Nile virus (WNV) is a neurotropic virus belonging to the Flavivirus genus. The virus is maintained in a mosquito-bird-mosquito transmission cycle, where infected mosquitoes, predominantly Culex pipiens [1,2], transmit WNV to birds [3] and vice versa [4]. Once infected mosquitoes feed on humans or other mammals, WNV is transmitted [5,6]. However, the low level of viremia is not sufficient for viral transmission from humans to mosquitoes, rendering the human a dead-end host for WNV [7]. Upon infection, about 20 % of humans develop West Nile fever (WNF) and less than 1 % develop West Nile neuro-invasive disease (WNND), which can be lethal [8]. To date, there is no vaccine against WNV [9] and disease prevention is focusing on mosquito control and personal protective measures to restrict exposure to infected mosquitoes. Until the end of the 1990s, WNV was considered a minor risk for humans because of its sporadic occurrence [10]. However, over the last two decades there have been repeated local outbreaks across various regions worldwide, such as Europe, Africa, Asia, United States [[11], [12], [13], [14], [15]]. The WNV outbreak in Romania in 1996 with 393 WNVhc [16] and the detection of the virus in the Americas in late summer 1999 were messengers of the disease re-emergence [17]. In Greece, WNVhc were first recorded in 2010 (262 cases, 35 deaths), while two major WNV outbreaks occurred during the period 2010–2023 [18]: in 2018, 316 laboratory-confirmed WNVhc (50 deaths) were reported, and in 2022, 286 WNVhc were identified (32 deaths). Overall, Greece reported the highest number in WNVhc in 2022 in Europe, after Italy [19,20], where most WNVhc occurred in RCM in Northern Greece. Identification of areas of high WNV transmission and disease risk is highly relevant for disease prevention measures such as mosquito control activities [21]. However, observed fluctuations and occasional explosive numbers of cases in some years, render WNV risk mapping challenging [22].

Various studies, conducted across different locations and years, consistently demonstrated pronounced spatial clustering of Culex pool counts [23,24] or WNV infection in humans [[24], [25], [26], [27], [28], [29]] and provide extensive analyses of the environmental and climatic conditions related to mosquito abundance and consequently also WNVhc [[30], [31], [32], [33], [34]]. Particularly, higher temperatures during spring and summer were identified to promote WNVhc [32,[35], [36], [37], [38]], while rainfall, humidity, soil water content, and wind were reported to decrease mosquito abundance and the recorded number of WNVhc [35,[37], [38], [39]].

Utilizing a novel entomological and epidemiological dataset in the RCM during the 2010–2023 period, upscaled at municipality level (below NUTS3), we investigate the spatial distributions of mosquito levels, infected mosquito levels and WNVhc through the Moran's I indices to identify the “hot-spot” areas common at all three maps. Then, using climatic data from Copernicus aggregated at municipality and monthly-seasonal scale, we fit a Generalized Linear Mixed Model to identify the factors driving the observed mosquito levels, infected mosquito levels and WNVhc and ultimately quantify the fraction of the observed variability explained by only environmental factors, under the current state of surveillance and targeted actions.

2 Materials and methods

2.1 Data collection and pre-processing

2.1.1 The region of Central Macedonia

Our study focused on the RCM in Northern Greece (Fig. 1(left)) consisting of 38 municipalities, examining data at the municipal level. We obtained data on the local administrative division, land use, and area at the municipal level from the Hellenic Statistical Authority (HSA). Detailed information is available in S1 Appendix.Fig. 1 Position of the Region of Central Macedonia in Greece (left) and geographical distribution of the location of mosquito traps (red points) in the region (right). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 1

2.1.2 Mosquito network

As part of a regional WNV surveillance program, CO2 mosquito traps were placed throughout the RCM (Fig. 1(right)), positioned based on criteria such as proximity to human settlements, protection from direct sunlight and wind, and historical WNV detection [40]. Some mosquito samples were tested for WNV presence using RT-nested PCR in reference laboratories [19], with testing frequency varying from 0 to 100 % monthly. More information is presented in S2 Appendix.

2.1.3 Measure of Culex population level (MCPL)

We calculated the mean monthly Culex mosquito abundance from all traps within the same municipality, provided a minimum of two samples were available to avoid spurious peaks. Το reduce the data variability, we defined the Measure of Culex Population Level (MCPL) (mosquitoes/km2) as the average mosquito abundance of the month during which the maximum number of mosquitoes were recorded, the month directly preceding and following. We defined the mosquito's flight range as 500 m, often reported for Culex mosquitoes [41,42]. Only municipalities with 13 years (2011−2023) of mosquito measurements were included to ensure reliability.

2.1.4 Measure of infected mosquito samplings (MIMS)

Based on the average monthly mosquito abundance from all traps within a municipality, we calculated the monthly percentage of infected mosquito samples out of those tested. We defined the Measure of Infected Mosquito Samplings (MIMS) as the annual maximum percentage (%) of infected mosquito samples.

2.1.5 WNV human cases (WNVhc)

We obtained laboratory-confirmed WNVhc in RCM from the National Public Health Organisation (EODY), excluding cases with unknown exposure locations or dates. We calculated WNVhc per 100,000 population (WNVhcP100TH) using municipal population data. Reported WNVhc were assumed to represent the actual number, despite underreporting due to asymptomatic cases [43]. Detailed information is available in S3 Appendix.

2.1.6 Climatic data

Air temperature and precipitation were considered crucial climatic factors in WNV transmission [44]. Six-hourly air temperature data for RCM from 2010 to 2023 from ERA5 were used to calculate average daily, monthly and seasonal temperatures, and Growing Degree Days at specific month (GDDMM). GDDMM was defined as the number of days from January 1 until the end of month MM when the average daily temperature exceeded 14.3 °C, the threshold for virus development [45]. Daily precipitation data were also acquired from ERA5, from which total monthly and seasonal precipitation were calculated.

2.2 The local and global Moran index

We computed the Global Moran Index for the median values of mosquito abundance, mosquito positivity, and WNVhc each year (detailed in S4 Appendix). The index ranges from −1 to 1, with values near zero indicating no spatial autocorrelation. The Local Moran's Index enabled us to identify clusters (i.e. groups of municipalities) of high values (Hot-spots), revealing significant spatial patterns of WNV activity. The central location(s) within a cluster, i.e. those exhibiting high values and also having neighboring municipalities with similarly high values, form the core of a cluster.

2.3 Generalized linear mixed model

We fitted a Generalized Linear Mixed Model (GLMM) to analyze entomological and epidemiological data, considering fixed effects such as environmental factors and random effects attributed to municipalities. We investigated 47 predictor variables: monthly temperature and monthly total precipitation from January to August (16), seasonal temperature and seasonal total precipitation (8), maximum and minimum seasonal temperature (8), GDD from January to August (8), percentages of arable land, heterogeneous agricultural land, and inland wetlands, and the sum of these areas (defined as AAW) (4), and previous year's MCPL, MIMS, and WNVhcP100TH (3) to estimate their influence on MCPL, MIMS, and WNVhcP100TH (predicted variables). The study period was 2011–2023, excluding years with missing data. We ranked models by Marginal and Conditional R-squared (MR2 and CR2) and selected the best as the Poisson model with a log link function and different predictor for each outcome variable.

Leave one-year-out cross validation was also used for verification and identification of anomalous years. To evaluate the predictive ability of the GLMM, we calculated the Normalized Mean Absolute Error (NMAE) (MAE divided by the mean of the observed data). MAE=1n∑1nPi−Oi,where n represents the total number of observations, Pi the predicted value, and Oi the actual value of the ith observation (detailed in S5 Appendix).

3 Results

In this section, we analyzed the spatial distributions of MCPL, MIMS and WNVhc and we fitted a GLMM to identify the factors influencing these levels.

3.1 Spatial patterns

3.1.1 Mosquito levels

We calculated the median MCPL values for each of the 18 municipalities of interest (Fig. 2A). In 22 % (4 out of 18) of the municipalities, the median MCPL exceeded 400 mosquitoes/km2. Considering all 18 municipalities across 13 years of observations (234 counts), the annual MCPLs peaked in July (41 %, 97 out of 234) and August (31 %, 73 out of 234) (Fig. 2B). Using the Local Moran Index, we identified 4 “spatial autocorrelation hot-spot” municipalities with significant spatial autocorrelation with adjacent municipalities, located in the Western part (Fig. 2C), overlapping with areas of high median MCPLs. Spatial autocorrelations were significant throughout the study period at a 0.01 significance level, except for 2021 and 2019, indicating consistent clustering of median MCPLs (Table S5). Interestingly, the AAW areas per municipality (Fig. 2D) corresponds well with the recorded median MCPLs, suggesting that AAW areas potentially foster increased mosquito abundances (Pearson's correlation coefficient = 0.76, p-value<0.001).Fig. 2 (Α) Map of the median MCPL for each of the 18 municipalities of interest of the RCM for the period 2011–2023. The municipalities excluded from this study are highlighted in gray. (B) Counts of months with the highest annual MCPL across the 18 municipalities of the RCM for the period 2011–2023. (C) Local Moran's I cluster map showing spatial association of median MCPLs with 4 “spatial autocorrelation hot-spot” municipalities: Alexandria, Delta, Pella, and Halkidona. (D) Map of the percentage of AAW areas for each municipality.

Fig. 2

3.1.2 Infected mosquito levels

We investigated WNV circulation by assessing infected mosquito levels through MIMS. The period 2014–2017 was excluded due to the low number of mosquitoes tested and the absence of WNV infection. In 44 % (8 out of 18) of municipalities, the MIMS exceeded 10, while in the remaining 56 %, no infected mosquitoes were found (Fig. 3A), suggesting a bimodality across the RCM. Considering the 8 municipalities and the 9 years of samplings (72 counts), MIMS was primarily recorded in July and August with 46 % (33 out of 72) and 53 % (38 out of 72), respectively (Fig. 3B), matching the months of highest MCPLs. Using the Local Moran Index, we identified one municipality as a “spatial autocorrelation hot-spot”, showing significant spatial autocorrelation with its five adjacent municipalities (Fig. 3C). However, autocorrelation was significant in only 1 out of 9 years at a 0.01 significance level, indicating lower spatial interconnectedness for MIMS compared to MCPL (Table S5).Fig. 3 (A) Map of the median MIMS for each of the 18 municipalities of interest in the RCM for the period 2011–2023 (excl. 2014–2017). The municipalities excluded from this study are highlighted in gray. (B) Counts of months of MIMS for the 8 municipalities and 9 years in which infected mosquitos were sampled. (C) Local Moran's I cluster map showing spatial association of median MIMS with one “spatial autocorrelation hot-spot” municipality: Alexandria (in red). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

3.1.3 Infected human levels

We calculated the medians of laboratory-confirmed WNVhcP100TH per municipality (Fig. 4A). One municipality recorded a particularly high median of 23 WNVhcP100TH. Overall, 26 of the 38 municipalities (68 %) had at least a median WNVhcP100TH of one. Each municipality recorded at least 1 WNVhc from 2010 to 2023 (Fig. 4B), while 18 municipalities (47 %) recorded WNVhc of at least one in more than 5 of the 10 years, demonstrating WNV endemicity in the RCM. Only 3 municipalities (8 %) in the Western region demonstrated high median WNVhcP100TH above 1 for all 10 years, indicating continuous high virus circulation. The first WNVhc were recorded between the 26th and 30th week, corresponding to late of June to early August (Fig. 4C), mostly in the Western RCM (Fig. 4D). In 6 of the 10 years, the first WNVhc was recorded in the 28th or 29th week, corresponding to middle of July. Using the Local Moran Index, we identified a “spatial autocorrelation hot-spot” cluster of four municipalities in the Western RCM with significant spatial autocorrelation with adjacent municipalities for WNVhcP100TH (Fig. 4E). However, significant autocorrelation was observed in only 4 out of 9 years at a 0.01 significance level (Table S5).Fig. 4 (A) Map of the median WNVhcP100TH in each municipality for the period 2011–2023 (excl. 2014–2017). (B) Geographical distribution of the number of years per municipality (1–10 years) with at least one recorded WNVhcP100TH during the period 2010–2023. (C) Number of years in which the first WNVhc occurred each week. (D) Map of the number of years (1–3 years) in which the first case was recorded per municipality. (E) Local Moran's I cluster map showing spatial associations with 4 “spatial autocorrelation hot-spot” municipalities: Alexandria, Veria, Naousa, Pella. The map concerns the association of the median value of WNVhcP100TH the period 2010–2023 (excl. 2014–2017).

Fig. 4

3.2 Spatial predictability

We combined the entomological and epidemiological datasets with climate and Land Use - Land Cover (LULC) data in a GLMM framework. Our findings (Table 1) indicated that a wet March combined with warm winter and spring in municipalities with a high proportion of AAW areas, are expected to cause a heightened MCPL (MR2 = 0.61), while the inclusion of a spatial random effect yields a CR2 of 0.99. Furthermore, an increased number of GDD05 combined with a warm January and summer, and a notable number of WNVhcP100TH in the previous year, are expected to lead to a rise in maximum mosquito positivity (MR2 = 0.57, CR2 = 0.95). Finally, a wet summer and March combined with a hot summer and May, a wet April, a high number of WNVhcP100TH in the previous year, and a high proportion of suitable LULC are expected to cause an increase in WNVhcP100TH (MR2 = 0.53, CR2 = 0.89). Therefore, according to the MR2, 53–61 % of the observed variability of MCPL, MIMS, and WNVhcP100TH could be explained by the selected inputs, while the remaining variability is accounted by unknown factors. The predictor variables employed at each GLMM had either no significant multicollinearity (MCPL) or moderate multicollinearity (MIMS, WNVhcP100TH) according to their variance inflation factor (VIF) [46]. The VIF of the selected predictor variables for MCPL, MIMS, and WNVhcP100TH were below 2, 2, and 4 respectively (medians: 1.14, 1.25, and 2.92).Table 1 The predictor variables employed in the GLMMs for MCPL, MIMS and WNVhcP100TH. The estimates are presented as incidence rate ratios (IRR), which are exponentiated values from the Poisson model. A 95 % confidence interval (CI) was used in all calculations.

Table 1MCPL	MIMS	WNVhcP100TH	
Predictors	Estimate	CI	p-value	Predictors	Estimate	CI	p-value	Predictors	Estimate	CI	p-value	
(Intercept)	190.41	89.73–404.1	<0.001	(Intercept)	1.75	0.35–8.81	0.04	(Intercept)	0	0.00–0.00	<0.001	
Twinter	0.89	0.88–0.90	<0.001	TJanuary	0.82	0.80–0.85	<0.001	TPMarch	1.15	1.13–1.16	<0.001	
TPMarch	1.03	1.02–1.03	<0.001	GDD05	0.91	0.90–0.92	<0.001	TPApril	0.94	0.92–0.95	<0.001	
TMay	0.58	0.53–0.64	<0.001	
Tspring	0.93	0.92–0.93	<0.001	Tsummer	1.33	1.26–1.41	<0.001	Tsummer	2.29	2.02–2.60	<0.001	
TPsummer	1.06	1.05–1.06	<0.001	
AAW areas	1.03	1.01–1.04	<0.001	WNVhcP100TH previous year	1	0.99–1.00	0.005	WNVhcP100TH previous year	0.99	0.98–0.99	<0.001	
AAW areas	1.02	1.01–1.03	0.002	
Marginal R2 / Conditional R2	0.61/0.99	Marginal R2 / Conditional R2	0.57/0.95	Marginal R2 / Conditional R2	0.53/0.89	

Using one-year-out cross-validation, we calculated the NΜΑΕ between predicted and observed MCPL, MIMS and WNVhcP100TH (Fig. 5A). For the entomological variable MCPL, the median NMAE was 0.4 and the inter-quartile range was 0.15, while only 2018 classified as outlier. For the epidemiological variables MIMS and WNVhcP100TH, the median NMAE was on the order of 0.91, with 2018, 2019 and 2012 classified as outliers, respectively. Outliers could be overall justified by the anomalous-low values of the variables in these years relative to the rest (Fig. S6).Fig. 5 (A) Comparison of MAE values across test years for MCPL, MIMS and WNVhcP100TH. The red dashed lines indicate the 25th and 75th percentiles, while the blue dashed line represents the median MAE value. Each year corresponds to the test set each time where all other years were used for training. (B) Geographical distribution of normalized mean absolute error (NMAE) between predicted and observed MCPL, MIMS, and WNVhcP100TH. The NMAE at each municipality corresponds to the median between all-years through a leave one-year-out cross validation approach. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 5

Excluding the epidemiological inputs in the GLMMs and relying only in the emerged environmental variables (climatic indices and land use), we found MR2 of 0.51 and 0.40 for MIMS and WNVhcP100TH, respectively. Thus, a considerable fraction of the epidemiological variability on the order of 40–50 % could be attributed to environmental factors alone highlighting their significant impact on mosquito-borne disease dynamics.

Fig. 5B showed the spatial distribution of the NMAE for MCPL, MIMS, and WNVhcP100TH. Focusing on the cores of the “hot-spot” areas in the Western region, the NMAE for MCPL varied from 0.29 to 0.39. The NMAE for MIMS was 1.03, highlighting higher prediction errors compared to MCPL. For WNVhcP100TH, NMAE ranged from 0.36 to 0.74, intermediate values in relation to MCPL and MIMS, indicating reliable predictions in this critical area.

4 Discussion

In this study, we used epidemiological and entomological data to investigate the spatial variability in the mosquito populations, the infected mosquito population, and human incidences in the RCM, an area of repeated WNV outbreaks. Therefore, employing the available entomological and epidemiological data of the period 2010–2023 at the municipality level, we investigated statistical associations that could provide insights on the factors involved on WNV spread.

In the RCM, at least one WNVhc was observed in all municipalities (except for the years 2014–2017), demonstrating the high endemicity of WNV. The analysis uncovered spatial patterns of increased WNVhc in the Western part, which may be due to the specific environmental conditions and/or the land cover and use. This region is characterized by water-intensive agricultural cultivation of products such as rice fields, which allow for conditions particularly prone to mosquito breeding and therefore increased risk of virus transmission.

Previous studies identified the Global [25,28] and the Local [23,24] Moran's I index as a useful tool to study spatial dependencies. In our study, the Global Moran's I of the MCPL is statistically significant in 85 % of years (all but 2019 and 2021), as expected since it is a climate dependent parameter and climate variables are spatially correlated. For the mosquitoes positivity the Global Moran's I was statistically significant only in 2020, while for annual WNVhc in four years, namely 2010, 2018, 2020 and 2021. The existence of spatial trends in annual WNVhc could be possibly explained by patterns of bird migration, but there are no data for the RCM to address this assumption.

Via the Local Moran cluster map, the municipality of Alexandria and its surrounding municipalities located in the Western part of the region were identified as a “spatial autocorrelation hot-spot” of the RCM. This is on line with the reported land usage of the municipality, which is mostly covered by drainage canals, rice fields and permanently irrigated land, ideal conditions for the development of mosquitoes. To better understand the identified patterns, a longer time series of entomological data for more municipalities is necessary.

Our modeling analysis highlights the impact of environmental factors on Culex population and WNV transmission, with the ability to predict 61 %, 51 % and 40 % of mosquito levels, infected mosquito levels, and WNVhc, respectively, using only climate features. This modeling approach indicates a strong dependence of the mosquito levels on the temperatures in winter and spring and the total precipitation in early spring, while virus circulation relies on the temperatures of late spring and summer. Many studies underscore the critical role of temperature [32,[35], [36], [37], [38],47] and precipitation [35,[37], [38], [39],48] as major predictors in mosquito development and virus transmission, supporting our findings. Incorporating previous year's virus circulation data enhances the predictive accuracy of the infected humans by 13 %, demonstrating the added value of historical epidemiological data in managing WNV risk. However, various other environmental and non-environmental factors may also influence virus transmission as still 40 % of the observed variability is accounted by unknown factors, highlighting the need for enhanced surveillance.

Overall, prediction errors have smaller year-to-year spread for mosquito abundance and WNVhc compared to mosquito infectivity, probably linked to the data quality in the latter case. Spatially, the median NMAE was 0.47, 1.41 and 1.1 for mosquito levels, infected mosquito levels, and WNVhc respectively, with lower median values at the core municipalities. Taken together the temporal and spatial errors, it is evident that although infected mosquitoes affect the causal pathway between mosquito abundance and infected humans, in their current form (MIMS) are of limited importance for public health.

Our research limitations include the lack of information for larviciding and other spraying actions to control mosquito population and WNV circulation. The discrepancies in the temporal and spatial patterns highlight the uncertainties arising from the use of small sample sizes.

The “spatially autocorrelated hot-spot” municipalities for MCPL, MIMS and WNVhc coincide, marking the expected causal pathway, while they possess increased predictability. The unexplained 40 % variance on the examined variables highlights the need for additional inputs, including larvae data (important for MCPL) and virus circulation information (important for WNVhc). The latter may include for example explicit information for culex positivity or proxy indicators for sentinel positivity.

5 Conclusion

Our research highlights the critical role of climate in shaping mosquito populations and WNV transmission in Central Macedonia. The identified spatial interdependencies, particularly in the Western part of the region, along with the strong influence of temperature and precipitation on mosquitoes and infected human levels, demonstrate the importance of climate-driven models for public health. The unexplained variance in WNV incidence suggests that targeted interventions, enhanced surveillance, and consideration of additional factors related to spillover processes are essential for improving WNV risk management and response.

Funding

This research has been co-financed by the European Regional Development Fund of the European Union and Greek national funds through the Operational Program Competitiveness, Entrepreneurship and Innovation, under the call RESEARCH – CREATE – INNOVATE (project code: Τ2ΕΔΚ-02070 ). This work was also co-financed from the EIC Horizon Prize “Early Warning for Epidemics”.

CRediT authorship contribution statement

Anastasia Angelou: Writing – original draft, Visualization, Validation, Software, Investigation, Formal analysis. Lea Schuh: Writing – review & editing, Methodology. Nikolaos I. Stilianakis: Writing – review & editing, Methodology. Spiros Mourelatos: Data curation. Ioannis Kioutsioukis: Writing – review & editing, Supervision, Resources, Project administration, Methodology, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A Supplementary data

Supplementary material.

Image 1

Data availability

The mosquito data used were provided from EcoDevelopmet S.A. under license, and so are not publicly available. Data are available from the authors upon reasonable request and with permission of EcoDev

Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.onehlt.2024.100888.
==== Refs
References

1 Gray T.J. Webb C.E. A review of the epidemiological and clinical aspects of West Nile virus Int J Gen Med 7 2014 193 203 10.2147/IJGM.S59902 24748813
2 Marini G. Poletti P. Giacobini M. Pugliese A. Merler S. Rosa R. The role of climatic and density dependent factors in shaping mosquito population dynamics: the case of culex pipiens in northwestern Italy PLoS One 11 2016 e0154018 10.1371/journal.pone.0154018
3 Valiakos G. Touloudi A. Iacovakis C. Athanasiou L. Birtsas P. Spyrou V. Molecular detection and phylogenetic analysis of West Nile virus lineage 2 in sedentary wild birds (Eurasian magpie), Greece, 2010 Euro Surveill. 16 2011 19862 10.2807/ese.16.18.19862-en 21586266
4 Rochlin I. Faraji A. Healy K. Andreadis T.G. West Nile virus mosquito vectors in North America J. Med. Entomol. 56 2019 1475 1490 10.1093/jme/tjz146 31549725
5 Komar N. West Nile viral encephalitis Rev. Sci. Tech. 19 2000 166 176 10.20506/rst.19.1.1201 11189714
6 Colpitts T.M. Conway M.J. Montogomery R.R. Fikrig E. West Nile virus: biology, transmission, and human infection Clin. Microbiol. Rev. 25 2012 635 648 10.1128/CMR.00045-12 23034323
7 Calistri P. Giovannini A. Hubalek Z. Ionescu A. Monaco F. Savini G. Epidemiology of West Nile in Europe and in the Mediterranean basin Open Virol J 4 2010 29 37 10.2174/1874357901004010029 20517490
8 Ladbury G.A. Gavana M. Danis K. Papa A. Papamichail D. Mourelatos S. Population seroprevalence study after a West Nile virus lineage 2 epidemic, Greece, 2010 PLoS One e80432 2013 10.1371/journal.pone.0080432
9 World Health Organization Vector-borne diseases Available online https://www.who.int/news-room/fact-sheets/detail/vector-borne-diseases 2024 (accessed on February 2024)
10 Dauphin G. Zientara S. West Nile virus: recent trends in diagnosis and vaccine development Vaccine 25 2007 5563 5576 10.1016/j.vaccine.2006.12.005 17292514
11 Kramer L.D. Styer L.M. Ebel G.D. A global perspective on the epidemiology of West Nile virus Annu. Rev. Entomol. 53 2008 61 81 10.1146/annurev.ento.53.103106.093258 17645411
12 Papa A. Xanthopoulou K. Gewehr S. Mourelatos S. Detection of West Nile virus lineage 2 in mosquitoes during a human outbreak in Greece Clin. Microbiol. Infect. 17 2011 1176 1180 10.1111/j.1469-0691.2010.03438.x 21781205
13 Papa A. Xanthopoulou K. Tsioka A. Kalaitzopoulou S. Mourelatos S. West Nile virus in mosquitoes in Greece Parasitol. Res. 112 2013 1551 1555 10.1007/s00436-013-3302-x 23371497
14 Petersen L. Brault A. Nasci R. West Nile virus: review of the literature JAMA 310 2013 308 315 10.1001/jama.2013.8042 23860989
15 Young J. Haussig J. Aberle S. Pervanidou D. Riccardo F. Sekulić N. Epidemiology of human West Nile virus infections in the European Union and European Union enlargement countries, 2010 to 2018 Euro Surveill. 26 19 2021 10.2807/1560-7917.ES.2021.26.19.2001095
16 Tsai T.F. Popovici F. Cernescu C. Campbell G.L. Nedelcu N.I. West Nile encephalitis epidemic in southeastern Romania Lancet 352 1998 767 771 10.1016/S0140-6736(98)03538-7 9737281
17 Lanciotti R.S. Roehrig J.T. Deubel V. Smith J. Parker M. Steele K. Origin of the West Nile virus responsible for an outbreak of encephalitis in the northeastern United States Science 286 1999 2333 2337 10.1126/science.286.5448.2333 10600742
18 National Public Health Organisation (EODY) Annual reports about West Nile Virus (WNV) surveillance Available online https://eody.gov.gr/en/epidemiological-statistical-data/annual-epidemiological-data/ 2024 (accessed on May 2024)
19 Tsioka K. Gewehr S. Pappa S. Kalaitzopoulou S. Stoikou K. Mourelatos S. West Nile virus in Culex mosquitoes in Central Macedonia, Greece, 2022 Viruses 15 2023 224 10.3390/v15010224 36680264
20 European Centre for Disease Prevention and Control (ECDC) Annual reports about West Nile Virus (WNV) surveillance Available online https://www.ecdc.europa.eu/en/west-nile-fever/surveillance-and-disease-data/annual-epidemiological-report 2024 (accessed on May 2024)
21 Nasci R.S. Mutebi J.P. Reducing West Nile virus risk through vector management J. Med. Entomol. 56 2019 1516 1521 10.1093/jme/tjz083 31549724
22 Wimberly M.C. Davis J.K. Hildreth M.B. Clayton J.L. Integrated forecasts based on public health surveillance and meteorological data predict West Nile virus in a high-risk region of North America Environ. Health Perspect. 130 2022 8 10.1289/EHP10287
23 Giordano B.V. Kaur S. Hunter F.F. West Nile virus in Ontario, Canada: a twelve-year analysis of human case prevalence, mosquito surveillance, and climate data PLoS One 12 8 2017 e0183568 10.1371/journal.pone.0183568
24 Radojicic S. Zivulj A. Petrovic T. Nisavic J. Milicevic V. Sipetic-Grujicic S. Misic D. Korzeniowska M. Stanojevic S. Spatiotemporal analysis of West Nile virus epidemic in South Banat District, Serbia, 2017–2019 Animals 11 2021 2951 10.3390/ani11102951 34679972
25 Sugumaran R. Larson S.R. DeGroote J.P. Spatio-temporal cluster analysis of county-based human West Nile virus incidence in the continental United States Int. J. Health Geogr. 8 2009 43 10.1186/1476-072X-8-43 19594928
26 Young SG, Jensen RR. Statistical and visual analysis of human West Nile virus infection in the United States, 1999–2008. Appl. Geogr., 34 (0) (2012), pp. 425–431, 10.1016/j.apgeog.2012.01.008.
27 DeGroote J.P. Sugumaran R. Ecker M. Landscape, demographic and climatic associations with human West Nile virus occurrence regionally in 2012 in the United States of America Faculty Publications. 3 2014 10.4081/gh.2014.13
28 Rocheleau J.P. Kotchi S.O. Arsenault J. Can local risk of West Nile virus infection be predicted from previous cases? A descriptive study in Quebec, 2011–2016 Can. J. Public Health 111 2020 229 238 10.17269/s41997-019-00279-0 32020540
29 Karki S. Brown W.M. Uelmen J. Ruiz M.O. Smith R.L. The drivers of West Nile virus human illness in the Chicago, Illinois, USA area: fine scale dynamic effects of weather, mosquito infection, social, and biological conditions PLoS One 15 5 2020 e0227160 10.1371/journal.pone.0227160
30 Chuang T.W. Hildreth M.B. Vanroekel D.L. Wimberly M.C. Weather and land cover influences on mosquito populations in Sioux Falls, South Dakota J. Med. Entomol. 48 3 2011 669 679 10.1603/me10246 21661329
31 Ferraguti M. Dimas Martins A. Artzy-Randrup Y. Quantifying the invasion risk of West Nile virus: insights from a multi-vector and multi-host SEIR model One Health 17 2023 100638 10.1016/j.onehlt.2023.100638
32 Marini G. Manica M. Delucchi L. Pugliese A. Rosa R. Spring temperature shapes West Nile virus transmission in Europe Acta Trop. 215 2021 105796 10.1016/j.actatropica.2020.105796
33 Giesen C. Herrador Z. Fernandez-Martinez B. Figuerola J. Gangoso L. Vazquez A. A systematic review of environmental factors related to WNV circulation in European and Mediterranean countries One Health 16 2023 100478 10.1016/j.onehlt.2022.100478
34 DeGroote J.P. Sugumaran R. Brend S.M. Tucker B.J. Bartholomay L.C. Landscape, demographic, entomological, and climatic associations with human disease incidence of West Nile virus in the state of Iowa, USA Int. J. Health Geogr. 7 2008 19 10.1186/1476-072X-7-19 18452604
35 Ferraccioli F. Riccetti N. Fasano A. Effects of climatic and environmental factors on mosquito population inferred from West Nile virus surveillance in Greece Sci. Rep. 13 2023 18803 10.1038/s41598-023-45666-3 37914706
36 Paz S. Malkinson D. Green M.S. Tsioni G. Papa A. Danis K. Permissive summer temperatures of the 2010 European West Nile fever upsurge PLoS One 8 2013 e56398 10.1371/journal.pone.0056398
37 Stilianakis N.I. Syrris V. Petroliagkis T. Pärt P. Gewehr S. Kalaitzopoulou S. Identification of climatic factors affecting the epidemiology of human West Nile virus infections in northern Greece PLoS One 11 2016 e0161510 10.1371/journal.pone.0161510
38 Lorenz C. De Azevedo T.S. Chiaravalloti-Neto F. Impact of climate change on West Nile virus distribution in South America Trans. R. Soc. Trop. Med. Hyg. 116 2022 1043 1053 10.1093/trstmh/trac044 35640005
39 Mavrakis A. Papavasileiou C. Alexakis D. Papakitsos E.C. Salvati L. Meteorological patterns and the evolution of West Nile virus in an environmentally stressed Mediterranean area Environ. Monit. Assess. 193 2021 227 10.1007/s10661-021-09011-3 33772423
40 Tsioka K. Gewehr S. Kalaitzopoulou S. Pappa S. Stoikou K. Mourelatos S. Papa A. Detection and molecular characterization of West Nile virus in Culex pipiens mosquitoes in Central Macedonia, Greece, 2019-2021 Acta Trop. 230 2022 106391 10.1016/j.actatropica.2022.106391
41 Reisen W.K. Lothrop H.D. Lothrop B.B. Factors influencing the outcome of mark-release-recapture studies with Culex tarsalis (Diptera: Culicidae) J. Med. Entomol. 40 6 2003 820 829 10.1603/0022-2585-40.6.820 14765659
42 Verdonschot P.F.M. Besse-Lototskaya A.A. Flight distance of mosquitoes (Culicidae): a metadata analysis to support the management of barrier zones around rewetted and newly constructed wetlands J. Vector Ecol. 39 1 2014 181 187 10.1111/j.1948-7134.2014.12083.x
43 Hadjichristodoulou C. Pournaras S. Mavrouli M. Marka A. Tserkezou P. Baka A. West Nile virus Seroprevalence in the Greek population in 2013: a nationwide cross-sectional survey PLoS One 10 11 2015 e0143803 10.1371/journal.pone.0143803
44 Wang Y. Pons W. Fang J. Zhu H. The impact of weather and storm water management ponds on the transmission of West Nile virus R. Soc. Open Sci. 4 2017 170017 10.1098/rsos.170017
45 Reisen W.K. Fang Y. Martinez V.M. Effects of temperature on the transmission of west nile virus by Culex tarsalis (Diptera: Culicidae) J. Med. Entomol. 43 2 2006 309 317 PMID: 16619616 10.1093/jmedent/43.2.309 16619616
46 O'brien, R.M A caution regarding rules of thumb for variance inflation factors Qual. Quant. 41 2007 673 690 10.1007/s11135-006-9018-6
47 Moser S.K. Barnard M. Frantz R.M. Scoping review of Culex mosquito life history trait heterogeneity in response to temperature Parasit. Vectors 16 2023 200 10.1186/s13071-023-05792-3 37316915
48 Albrecht L. Kaufeld K.A. Investigating the impact of environmental factors on West Nile virus human case prediction in Ontario Canada. Frontiers in public health 11 2023 1100543 10.3389/fpubh.2023.1100543 36875397
