ArticlePDF Available

Forecasting the spatial and seasonal dynamic of Aedes albopictus oviposition activity in Albania and Balkan countries

PLOS
PLOS Neglected Tropical Diseases
Authors:

Abstract and Figures

The increasing spread of the Asian tiger mosquito, Aedes albopictus, in Europe and US raises public health concern due to the species competence to transmit several exotic human arboviruses, among which dengue, chikungunya and Zika, and urges the development of suitable modeling approach to forecast the spatial and temporal distribution of the mosquito. Here we developed a dynamical species distribution modeling approach forecasting Ae. albopictus eggs abundance at high spatial (0.01 degree WGS84) and temporal (weekly) resolution over 10 Balkan countries, using temperature times series of Modis data products and altitude as input predictors. The model was satisfactorily calibrated and validated over Albania based observed eggs abundance data weekly monitored during three years. For a given week of the year, eggs abundance was mainly predicted by the number of eggs and the mean temperature recorded in the preceding weeks. That is, results are in agreement with the biological cycle of the mosquito, reflecting the effect temperature on eggs spawning, maturation and hatching. The model, seeded by initial egg values derived from a second model, was then used to forecast the spatial and temporal distribution of eggs abundance over the selected Balkan countries, weekly in 2011, 2012 and 2013. The present study is a baseline to develop an easy-handling forecasting model able to provide information useful for promoting active surveillance and possibly prevention of Ae. albopictus colonization in presently non-infested areas in the Balkans as well as in other temperate regions.
This content is subject to copyright.
RESEARCH ARTICLE
Forecasting the spatial and seasonal dynamic
of Aedes albopictus oviposition activity in
Albania and Balkan countries
Cle
´ment Tisseuil
1
, Enkelejda Velo
2
*, Silvia Bino
2
, Perparim Kadriaj
2
, Kujtim Mersini
3
,
Ada Shukullari
4
, Artan Simaku
2
, Elton Rogozi
2
, Beniamino Caputo
5
, Els Ducheyne
6
,
Alessandra della Torre
5
, Paul Reiter
7¤a
, Marius Gilbert
1¤b
1Spatial Epidemiology Lab. Universite
´Libre de Bruxelles, Brussels, Belgium, 2Control of Infectious
Diseases Department, Institute of Public Health, Tirana, Albania, 3National Veterinary Epidemiology Unit,
Food Safety and Veterinary Institute, Tirana, Albania, 4Department of Biology, Faculty of Natural Sciences,
University of Tirana, Tirana, Albania, 5Department of Public Health and Infectious Diseases, University of
Rome “Sapienza”, Rome, Italy, 6European Economic Interest Group—European Agro-Environmental Health
Geographic Information Systems, Zoersel, Belgium, 7Insects and Infectious Disease Unit, Institute Pasteur,
Paris, France
These authors contributed equally to this work.
¤a Current address: Independent Researcher, Divonne-les-bains, France.
¤b Current address: Fonds National de la Recherche Scientifiques, Brussels, Belgium.
*keladikolli@yahoo.com
Abstract
The increasing spread of the Asian tiger mosquito, Aedes albopictus, in Europe and US
raises public health concern due to the species competence to transmit several exotic
human arboviruses, among which dengue, chikungunya and Zika, and urges the develop-
ment of suitable modeling approach to forecast the spatial and temporal distribution of the
mosquito. Here we developed a dynamical species distribution modeling approach forecast-
ing Ae.albopictus eggs abundance at high spatial (0.01 degree WGS84) and temporal
(weekly) resolution over 10 Balkan countries, using temperature times series of Modis data
products and altitude as input predictors. The model was satisfactorily calibrated and vali-
dated over Albania based observed eggs abundance data weekly monitored during three
years. For a given week of the year, eggs abundance was mainly predicted by the number
of eggs and the mean temperature recorded in the preceding weeks. That is, results are in
agreement with the biological cycle of the mosquito, reflecting the effect temperature on
eggs spawning, maturation and hatching. The model, seeded by initial egg values derived
from a second model, was then used to forecast the spatial and temporal distribution of
eggs abundance over the selected Balkan countries, weekly in 2011, 2012 and 2013. The
present study is a baseline to develop an easy-handling forecasting model able to provide
information useful for promoting active surveillance and possibly prevention of Ae.albopic-
tus colonization in presently non-infested areas in the Balkans as well as in other temperate
regions.
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 1 / 16
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Tisseuil C, Velo E, Bino S, Kadriaj P,
Mersini K, Shukullari A, et al. (2018) Forecasting
the spatial and seasonal dynamic of Aedes
albopictus oviposition activity in Albania and Balkan
countries. PLoS Negl Trop Dis 12(2): e0006236.
https://doi.org/10.1371/journal.pntd.0006236
Editor: Pattamaporn Kittayapong, Faculty of
Science, Mahidol University, THAILAND
Received: July 20, 2017
Accepted: January 12, 2018
Published: February 12, 2018
Copyright: ©2018 Tisseuil et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
files.
Funding: This study was partially funded by EU
grant FP7-261504 EDENext and is catalogued by
the EDENext Steering Committee as EDENext399
(http://www.edenext.eu) and Institute of Public
Health, Tirana, Albania. The contents of this
publication are the sole responsibility of the
authors and do not necessarily reflect the views of
the European Commission. The funders had no
Author summary
The Asian tiger mosquito Aedes albopictus, originating from Asia, in the last decade has
spread in many regions in Europe and US. Beside the nuisance problem causing to the cit-
izens during the day, this species has raised public health concern, due to its strict associa-
tion with humans and anthropic habitats, its expanding distribution and its capacities to
transmit several human arboviruses. We developed a spatio-temporal model of Ae.albo-
pictus dynamics that helps to understand the biology and the ecology of the species in rela-
tion to environmental factors, and to inform efficient control strategies. Here we
developed a dynamical species distribution modeling approach at high spatial (over the
Balkans) and temporal resolution (weekly scale), enabling to link oviposition activities
and climatic conditions across different time periods to forecast the potential future ovi-
position activities of Ae.albopictus in unknown locations or identify target areas and peri-
ods of highest activities. Extrapolating Ae.albopictus abundance over the Balkan region
may help to identify habitat suitability where the species has never been reported so far.
The temperature-related predictors remain the most determinant predictors among all
candidate predictors e.g land cover and rainfall. The model provides useful information
for promoting active surveillance on Ae.albopictus and assessing the risk of exotic arbovi-
rus transmission in temperate regions.
Introduction
In the last decade, the increasing spread of the Asian tiger mosquito, Aedes albopictus, in
Europe and US has raised public health concern, as the species is involved in the transmission
of several human arboviruses among which dengue, chikungunya and Zika [1,2,3,4]. The tiger
mosquito is arrived in Europe in the seventies, probably through cargo transportation from
China [5]. The first record of this species in Europe was reported from Albania in 1979,
although it is quite possible that the species was already present in mid-1970s, at least two
decades before the species was first detected in Italy in 1991[5]. Nowadays, Ae.albopictus is
widespread and commonly found in Albania, even in tiny isolated villages and sites in high
altitude including beech forest up to 1200m. In other European countries, Italy reported Ae.
albopictus up to 600m altitude[6] and the species was found in Switzerland, France or Spain
and widely distributed in Balkan countries such as Croatia, Montenegro and Serbia [(http://
ecdc.europa.eu/en/healthtopics/vectors/vector-maps/Pages/VBORNET_maps.aspx) [7]
To mitigate the potential impact of the mosquito in transmitting human diseases, efforts
were made to better understand the biology and the ecology of the species. The role of the
environmental factors in the spread and temporal dynamics of Ae.albopictus has been investi-
gated in both laboratory and field work conditions [8]. Temperature is shown to be a crucial
driver for Ae.albopictus activity at different levels, from adult abundance [9] and oviposition
activity [10], to eggs incubation [11] and eggs hatching [12,10,11]. More specifically, laboratory
studies showed to which extent the strength of thermal conditions, their starting period and
duration, could impact spawning and embryogenesis [11]. Some other environmental factors
related to land cover, have been shown to be statistically associated with high habitat suitability
for Ae.albopictus larval breeding site, through the landscape structure [13].
To date, several studies have focused on modeling the spatial distribution of Ae.albopictus
at different spatial scales, from global [14], to continental for Europe, [15,16] or country and
regional scales e.g., for Japan, [17] for Northern Italy [18,19]. However, as most classical spe-
cies distribution modeling approaches, those studies provide a static picture of Ae.albopictus
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 2 / 16
role in study design, data collection and analysis,
decision to publish, or preparation of the
manuscript.
Competing interests: The authors have declared
that no competing interests exist.
activity, by statistically relating the occurrence [14,15] or eggs abundance [20] of the species to
some environmental spatial factors for some fixed period of the year. By contrast, a few studies
have either accounted for spatio-temporal variables [21,22,23] or developed dynamical models
of Ae.albopictus biological activity [24], enabling to link oviposition activities and climatic
conditions across different time periods [25]. The use of dynamical model provides a prag-
matic solution for public health policies to forecast the potential future oviposition activities of
Ae.albopictus or identify target areas and periods of highest activities. This might be used, in
return, to develop efficient scenarios to mitigate the impact of the tiger mosquito in the spread
of several human diseases among which dengue, chikungunya and Zika.
Hereby, we present a three-year study of the abundance of Ae.albopictus eggs at different
altitudes on Dajti mountain in central Albania. We developed a novel dynamical species distri-
bution modeling approach at high spatial and temporal resolution. Our major goal is to
develop a statistical forecasting model as simple as possible in terms of implementation for
non-statistician users, with relatively low computer-time calculation and flexibility for extrapo-
lating projected results to varying spatial and temporal scales. Our approach contrast with the
many species distribution models published previously by allowing spatially and temporally
explicit predictions. The three main objectives are: i) calibrating and validating a forecasting
approach based on independent data to assess the ability of the model to project the potential
future oviposition of Ae.albopictus in unknown locations; ii) ensuring that the strength of
environmental drivers on mosquito spawning, as fitted by our model at large spatial and time
scale, is consistent with literature review; iii) extrapolating our forecasting approach and proj-
ect the spatio-temporal oviposition activity of Ae.albopictus over the Balkans between 2009
and 2012 at high spatial (1km
2
resolution approximately) and temporal resolution (weekly
scale).
Methods
Forecasting model framework
The forecasting modeling framework was based on five major steps summarized hereafter.
First, the entomological and environmental data were collected and stored into a spatial-tem-
poral database for data analysis. Second, a calibration step aimed at selecting the best combina-
tion of predictors and statistical models that best fit the observed spatio-temporal patterns of
eggs abundance. Third, the ability of the forecasting model to project unforeseen future events
in new locations was tested on independent validation dataset. Four, the forecasting model
projections were extended to the entire Balkan countries (Albania, Montenegro, Macedonia,
Serbia, Kosovo, Greece, Croatia, Slovenia, Bulgaria and Rumania) at high spatial (one km) and
high temporal (week) resolution between 2009 and 2013. The overall modeling framework was
built upon Generalized linear models (GLM; [26]) to facilitate the ecological understanding of
model behavior while minimizing computer-time calculations and favoring the replication of
model to similar ecological systems.
Dataset
Entomological survey. Oviposition activity of Ae.albopictus was monitored in 26 sites
across Albania by ovitraps, i.e. black cylindrical vessels (9 cm high, 11 cm in diameter with an
overflow hole at 7 cm from the base) filled with ~300 ml tap-water with no attractants, and
internally lined with heavy-weight seed germination paper [27]. No specific permissions were
required for these sites/activities. Landowners gave permission to conduct the studies on their
properties. The field studies did not involve endangered or protected species.
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 3 / 16
A first set of 16 sites located in the center of Albania (Tirana-Dajti mount, 19 55’51.2"Eo, 41
21’34.5"No, across a 154–1559 meter altitude gradient, (Fig 1A) was used for model
calibration.
Monitoring started in April 2011 and continued on a weekly basis until December 2013, by
5 ovitraps/site (i.e.80 ovitraps/week). A second set of 10 sites located in the Northern part of
Albania (Malesia e Madhe-Vermosh 19 41’26"E, 42 35’22"N, along an altitude gradient ranging
from 41 to 1282 meters, (Fig 1B) was used for model validation.
In the latter 10 sites, monitoring was carried out in August-September 2012 every two
weeks by 5 ovitraps/site and the number of eggs/week was calculated by dividing the total
number of eggs collected over the two weeks by two.
Eggs laied on germination paper were counted and identified to species level based on their
color, size, shape and surface sculpting [28].
Environmental data. Climate and landscape data expected to influence Ae.albopictus
ecological dynamic were used to derive relevant predictors in the model [21,22,23]. The mean
land surface temperature (LST) were extracted from Modis data, by averaging night and daily
LST data, then rescaling data to the weekly resolution using spline interpolation to match the
sampling unit of the study. Altitude data were based on the Shuttle Radar Topography Mission
database (SRTM; [29]). For each site, the dominant class of Corine land-cover 1km resolution
raster [30] was extracted among: artificial surface, agriculture, forest, wetlands or water body.
Fig 1. Location of the sampling sites. a) 16 sites used for model calibration are located in Middle Albania (Tirana-
Dajti mount); b) 10 sites used for model validation are located in North Albania (Malesia e Madhe–Vermosh).
Information sources are open source information from: http://asig.gov.al/ https://landsatlook.usgs.gov/viewer.html.
https://doi.org/10.1371/journal.pntd.0006236.g001
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 4 / 16
Forecasting model structure
The spatial and temporal sampling unit of study was defined at the site and week scale, respec-
tively i.e. the forecasting model aimed to predict the total abundance of Ae.albopictus eggs/
site/week.
The structure of the forecasting model was made of three components; namely the ’Core’,
’Init’, and ’Max’ components. While the Core component is the key feature, making the direct
link between the spatial and temporal variability of environmental predictors and the abun-
dance of Ae.albopictus eggs/site/week, the ‘Init’ and ‘Max’ component defines the initial and
maximal condition values inside which the Core modeling component was allowed to run.
The Core model. Ten predictors expected to influence eggs abundance were used in the
models. We aimed to develop a parsimonious model using a relatively limited number of vari-
ables with a plausible biological causal relationship underlying the predictions. The variables
included: i) three biological predictors referring to the respective eggs abundance recorded at
each three weeks before the sampling; ii) four climate predictors including the mean tempera-
ture recorded at the sampling week and at each of the past three weeks of sampling, in order to
take into account the role of climate variables during the larval development in affecting A.
albopictus adult abundance and survival [23]; iii) two geophysical predictors related to Corine
land-cover and altitude; iv) one seasonal predictor denoting the week of the year the sample
was recorded.
Considering eggs abundance as count data, Poisson (hereafter referred as ’poisson’) or neg-
ative binomial (hereafter referred as ’nb’) errors distribution families were assumed in the
modeling setup. In addition, simple GLMs (hereafter referred as ’glm’) as well as more com-
plex types of model enable to account for the presence of large number of zeros in the database
(>50%) were also tested; namely hurdle (hereafter referred as ’hurdle’) and zero-inflated
(hereafter referred as ’zeroinf’) models [31]. It is worth to note that zero-inflated models are
based on a zero-inflated probability distribution i.e. a distribution that allows for frequent
zero-valued observations. So, the zero inflated model fit simultaneously two separate regres-
sion models. On one hand a logistic or probit model that predicts the probability of being a
non-zero count, and on the other hand a model that predicts the size of that count. In total, six
models derived from the combination between the three types of model and the two distribu-
tion families were compared to each other; i.e. glm-poisson, glm-negbin, hurdle-poisson, hur-
dle-negbin, zeroinf-poisson, zeroinf-negbin.
The Init and Max models. The Init modeling component defined initial condition values
to allow the Core component to initialize projections from any week of the year and any loca-
tion, and to populate the core model with these values to generate projections. This allows pro-
jections to be made in any location where the predictor covariate data is available. Similarly,
the Max modeling component defined maximal condition values at each step of the forecasting
process, preventing the Core model to project some excessive or unreliable eggs abundance
values above those maximal condition values.
Based on the calibration dataset, the mean and maximum weekly eggs abundance over the
three years of sampling was calculated at each site. The Init and Max models were setup indi-
vidually, by regressing the mean (for the Init model) and the maximum (for the Init model)
weekly eggs abundance against the mean altitude and the week of the year (as a second polyno-
mial degree). GLMs models with Poisson and negative binomial distribution families were
tested.
Models selection. The procedure used to select the best combination of predictors and
statistical models was inspired from [32]. For each sub-model (i.e. Core, Init and Max model),
each statistical model was first built using the full set of potential predictors, hereafter referred
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 5 / 16
as ’full model’. Then, the ’full’ model that best fulfilled the following five criteria was selected:
1) normality of residuals; 2) homogeneity of residuals; 3) absence of strong spatial and tempo-
ral autocorrelation patterns in the residuals as quantified by the experimental semi-variogram;
4) dispersion parameter as close as possible of value 1; 5) Akaike information criterion as small
as possible (AIC). Finally, a stepwise procedure was applied to the best ’full’ model to derive
the most parsimonious ’final’ model i.e. the model which maximizes the model goodness-of-fit
based on AIC criteria while minimizing the number of predictors. All predictors were trans-
formed to normality and scaled to zero mean and unique variance. All models were calibrated
using the calibration dataset.
Dynamical feature. The dynamical feature of the forecasting model was designed to
refine the accuracy of projections while integrating meaningful ecological information related
to the biological cycle of Ae.albopictus. Technically, this was implemented in 4 steps by: i) ini-
tializing the Core model at a given week of the year (Wt) using the Init Model by estimating
starting egg counts; ii) Output from the Init Model were used to derive the egg count predictor
variables to be included in the Core Model; iii) Core model projections were then calculated
while paying attention that they do not exceed Max model projections, otherwise they were
given the corresponding Max model projection value; iv) Core model outputs at time Wt were
used as egg count predictors at time Wt-1 to derive Core model projections at time Wt+1; v)
The model was run iteratively until the ending week of the year, by repeating step iii) and iv).
Forecasting model calibration and validation
The forecasting model was calibrated using the calibration dataset while validated using the
validation dataset. Both dataset were assumed to be spatially independent, so that the valida-
tion step provided a suitable assessment of the forecasting model ability to extrapolate projec-
tions to higher spatial extents. Since the validation data were only available in weeks 32, 34, 36
and 38 for year 2012, the goodness-of-fit assessment was made for this period only.
Model goodness-of-fit and uncertainty was evaluated throughout bootstrap approach
under 100 iterations. For each iteration, 70% of the calibration dataset was randomly sampled
(with replacement) to calibrate a single forecasting model. The projections derived from the
100 forecasting models were then assembled to calculate the 10
th
, 50
th
and 90
th
percentiles pro-
jections values.
The model goodness-of-fit was quantified by comparing the projected values with the vali-
dation observational values using root mean square error (RMSE) and Spearman rank correla-
tion coefficients index. The robustness of the forecasting model was discussed in terms of
weekly variability and by initializing the model at different weeks of the year, ranging from
one to five weeks before the first recording week.
Forecasting model extrapolation to the Balkans
Once validated, the forecasting model was extrapolated to Albania and its surroundings coun-
tries (Montenegro, Macedonia, Serbia, Kosovo, Greece, Croatia, Slovenia, Bulgaria, Rumania)
at 0.01 degree spatial resolution (WGS84 coordinate reference system), and at a weekly interval
from 2009 to 2013. Model projections are provided as supplementary information data S1
Dataset.
The forecasting model assumed stationarity, so that the relationships fitted between the
abundance of eggs and its predictors was supposed to remain stable beyond the spatio-tempo-
ral extent of model calibration.
All statistical analyses were performed under the R software environment (version 3.1.1)
using packages pscl [33], visreg [34], dplyr [35], and ggplot2 [36].
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 6 / 16
Results
Entomological observations
The entomological survey revealed the presence of Ae.albopictus even in very tiny and isolated
sites up to the altitude 1,200 m a.s.l. in Middle Albania (Dajti mount) and up to 1,054 m a.s.l.
in North Albania (Vermosh). The highest species abundance (i.e. 1028 eggs/ovitrap in average
w 28 (second week of July), 2012) was observed in park and gardens in urban, suburban and
rural areas below 550 m a.s.l and high abundance was also observed up to 700 m a.s.l. in both
the North and the Central transects (i.e. 377 eggs/ovitrap in average w31, 2013). While at alti-
tudes <160 m oviposition continued from May to early December (weeks 19–49, with peaks
in August and September), the activity declined with at higher altitudes and was observed only
from end of June to end of September (w25-w38) at >760 m a.s.l.
Although Ae.albopictus eggs represented >97% of mosquito eggs found, eggs of other mos-
quito species were also found in ovitraps. Ochlerotatus geniculatus eggs were found in almost
all stations along the transect in Dajti mount from May (w19) to beginning of September
(w36) at altitudes <140 m a.s.l. and from beginning of June (w24) to end of August (w35) at
higher altitudes. Anopheles sp. eggs were found in ovitrap located in the 333 m a.s.l. site in
Dajti mount (w25, 2012). Presence of Anopheles plumbeus adults was observed at 1257m a.s.l.,
end of June 2012 (w25). Culex pipiens egg rafts were found in the 962 m a.s.l. site in Dajti
mount in June-July 2012 (w25-28).
Model selection
The first model selection step consisted in identifying the most suitable statistical models for
the Core, Init and Max modeling components, independently. For the Core modeling compo-
nent, zero-inflated negative binomial model displayed the best goodness-of-fit results, with
particularly low AIC, RMSE and dispersion parameters values (zeroinf-nb; Table 1).
The best statistical model was selected based on the following criteria: normality, homoge-
neity and absence of strong spatial and temporal autocorrelation patterns in the residuals (Fig
2), minimizing dispersion, RMSE and AIC parameter values. All models were built upon the
calibration dataset.
In regards with the Init and Max modeling component, best results were obtained using
GLM statistical model with negative binomial distribution family (glm-nb; Table 1).
In addition, the analysis of Core model results revealed a satisfying normality, homogeneity
as well as an absence of strong spatial and temporal autocorrelation patterns in residuals (Fig
2).
Table 1. Statistical model selection for the Core, Init and Max modeling components, based on based on Akaike information criterion (AIC), Spearman correlation
coefficient (COR), root mean square error (RMSE) and dispersion parameter (DISP).
Statistical model Core Model Init Model Max Model
AIC COR RMSE DISP AIC COR. RMSE DISP. AIC COR. RMSE DISP.
Poisson model (glm-pois) 31174. 0.76 125 103.7 89731 0.82 97.6 72.12 150582 0.81 164.5 124.42
Negative binomial model (glm-nb) - 0.76 29274 7.9 10446 0.83 126.9 2.91 11627 0.81 231.7 2.61
Hurdle Poisson model (hurdle-pois) 190935 0.77 115 19.1 - - - - - - - -
Hurdle negative binomial model (hurdle-nb) 22227 0.79 404 0.76 - - - - - - - -
Zero-inflated poisson model (zeroinf-pois) 190935 0.79 115 19.32 - - - - - - - -
Zero-inflated negative binomial model (zeroinf-nb) 22207 0.79 376 0.74 - - - - - - - -
‘-’ Statistical models that have not been tested as candidate models.
https://doi.org/10.1371/journal.pntd.0006236.t001
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 7 / 16
Final Core, Init and Max modeling components were derived using AIC backward stepwise
variable elimination and results are shown in Table 2.
Model behavior
Based on the final Core model results, the main relationships between conditional effect of pre-
dictors and eggs abundance was shown in Fig 3. One can note that the temperature in the cur-
rent week, the lagged temperature in the weeks -2 and -3, and the land cover variables were
taken out of the significant predictor variables by the backward stepwise procedure.
The influence of egg abundance in the previous weeks decreased with time, with the stron-
gest effect of egg abundance at week-1, and the lowest but yet significant effect for egg abun-
dance at week-2 and week-3 (Table 2 and Fig 3). Temperature measured by the MODIS LST
signal in the previous week had positive influence on eggs abundance, while altitude displayed
a negative effect. Neither the land cover nor the seasonal predictors were selected in the final
Fig 2. Spatial and temporal experimental semi-variograms calculated from the final Core model residuals,
highlighting low autocorrelation in model residuals (the black line indicated the smoothed trend).
https://doi.org/10.1371/journal.pntd.0006236.g002
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 8 / 16
model. This suggests the seasonal eggs abundance variability was satisfactorily captured by cli-
matic and previous week egg abundance predictors.
Model validation
The forecasting model accuracy was evaluated using calibration and validation sites for year
2012 during the summer period (weeks 32–38). Results from the 100 bootstrap model outputs
are summarized in Fig 4.
Goodness-of-fit criteria included the residuals (i.e. the difference between the predicted
and observed abundance), root mean squared error (RMSE) and Spearman correlation values.
Since the validation data were only available for weeks 32, 34, 36 and 38 in year 2012, the over-
all goodness-of-fit assessment was performed for this period only.
Model accuracy was assessed in terms of weekly variability (Fig 4A) as well as by initializing
the model at different weeks of the year, ranging from one to five weeks before the first record-
ing week (Fig 4B).
Globally, goodness-of-fit metrics were relatively comparable between calibration and vali-
dation dataset, although validation metrics values were moderately lower than the calibration
ones e.g. residuals
calibration
140.37 vs residuals
validation
67.14, RMSE
calibration
310.41 vs
RMSE
validation
342.15, Spearman
calibration
0.72 vs Spearman
validation
0.66 (Fig 4). Those
results indicated the relatively good ability of the forecasting model to extrapolate out of its
spatial range of calibration.
The seasonal variability in model performances were relatively low and stable across time,
although the model displayed noticeably better performances in weeks 32 and 38 than in
weeks 34 and 36 (Fig 4A). Similarly, the variability in model performances due to the different
initialization weeks of the year was relatively low as well (Fig 4B). This highlighted the overall
good forecasting capacity of the model.
Model projections
The forecasting model was successfully applied to high spatial (0.01 WGS84 decimal degrees)
and temporal (weekly) resolutions over the Balkans for the period 2009–2013. Summary statis-
tics (percentiles 10, 50 and 90) were derived from the 100 bootstrap model runs to assess the
Table 2. Parameters, standard deviation and significance for the best Core (zero inflated negative binomial model; zeroinf-nb), Init (GLM negative binomial
model; glm-nb) and Max models (GLM negative binomial model; glm-nb).
Core model (zeroinf-nb) Init model (glm-nb) Max model (glm-nb)
count model zero model
Coef. Sd. Sig Coef. Sd. Sig Coef. Sd. Sig Coef. Sd. Sig
Intercept 4.24 0.04  -0.22 0.09 -15.68 -1.32  -15.91 -1.39 
Egg count
Week-1
0.25 0.03  -1.59 0.21  - - - - - -
Egg count
Week-2
0.12 0.03  -0.65 0.15  - - - - - -
Egg count
Week-3
0.08 0.02 -0.4 0.12 - - - - - -
Temperature
Week-1
0.34 0.04  -0.87 0.16  - - - - - -
Altitude -0.35 0.05  -0.01 0.00  -0.01 0.00  -0.01 0.00 
Week of the year - - - - - - 1.37 -0.08  1.40 -0.08 
Week of the year
2
- - - - - - -0.02 0.00  -0.02 0.00 
,,, Predictors significance at the alpha level of 0.05, 0.01 and 0.001 respectively.
‘-’Predictors which are not included in the modeling setup.
https://doi.org/10.1371/journal.pntd.0006236.t002
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 9 / 16
Fig 3. Conditional effects of selected predictors (x-axis) on the egg abundances (y-axis) in the the final Core zero-inflated negative binomial model.
https://doi.org/10.1371/journal.pntd.0006236.g003
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 10 / 16
mean trend and uncertainty in the projected eggs abundance spatiotemporal patterns (Figs 5
and 6).
Results were averaged to derive a spatial distribution maps in eggs abundance for each year.
The spatial patterns being relatively similar between years, results for year 2012 are shown to
illustrate the main spatial trends (Fig 5). Globally, results displayed strong spatial heterogene-
ity, the southern coastal regions displaying higher eggs abundance than in northern and cen-
tral regions. Resulting maps for percentiles 10, 50 and 90 were very similar to each other. This
indicated that the spatial uncertainty related to model error was relatively low.
Fig 4. Goodness-of-fit assessment for the forecasting model applied to the validation and calibration sites, based on 100 bootstrap
model projections. a) according to the monitoring week of the year and b) according to the week of model initialization.
https://doi.org/10.1371/journal.pntd.0006236.g004
Fig 5. Annual mean percentiles of 10% (p10), 50% (p50) and 90% (p90) of Aedes albopictus eggs abundance projected by the forecasting model over Balkan
countries in year 2012. The darker the color the more projected abundance by the model.
https://doi.org/10.1371/journal.pntd.0006236.g005
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 11 / 16
The results were analyzed annually and seasonally for three different regions located in the
southern (region A), middle (region B) and northern (region C) parts of Balkans (Fig 6).
Although the three regions displayed different results in terms of magnitude, each region
displayed a strong similar seasonal signal. The projected annual peaks in eggs abundance gen-
erally occurred between the summer months of August and September i.e. approximately
from weeks 32 to 38. Importantly, the annual and seasonal uncertainty in projections was rela-
tively low since the mean percentiles 10 and 90 values were closed to the median value.
Discussion
Our results provide confidence in the ability of the here proposed forecasting model to project
the spatial and seasonal oviposition activity of Ae.albopictus over Albania and its neighboring
countries. Firstly, the goodness-of-fit indicators from the calibration and validation step were
satisfying. In regards with the Core modeling component, the use of zero-inflated model was
shown to outperform other GLMs models. In particular, zero inflated models have the interest
to account both for the skewed distribution (generally best modeled using negative binomial
distribution family) and for the high proportion of zero values in eggs samples. While similar
spatial modeling studies of Ae.albopictus generally apply negative binomial distribution family
to their model, they do not take into account high proportion of zero values [23,22,21]. Sec-
ondly, the spatio-temporal projections validated on the Northern sites of Albania, provides
confidence in the ability of model to extrapolate the spatio-temporal dynamics of eggs abun-
dance over the other Balkan countries. Some noise may however, have been added in this vali-
dation by the fact that the sampling interval was different in the training and validation data,
but given the fairly good goodness-of-fit metrics, we believe that this effect was low.
It is noteworthy that the geographical extrapolation exercise remains questionable over lon-
ger geographical distance, as far as the model has not validated using observational data from
the neighboring Balkan countries, which was not feasible in the frame of the present work, but
may become feasible in the near future as more information on the Ae.albopictus abundance
in the region are becoming available [37]. One should note, that in many cases, the validation
of spatial models is internal through the use of cross-validation, so fact of training and validat-
ing the model in entirely different areas goes beyond the state-of-the-art encountered in many
species distribution models applied to disease vectors. However, despite this limitation,
Fig 6. Mean annual and monthly Aedes albopictus eggs abundance projected by the forecasting model over the period 2009–2013 for the three regions located in
the southern (region A), middle (region B) and northern part (region C) of Balkan countries. The central statistics (percentile p50) and the associated error bars
(percentiles p10 and p90) of each barplot were calculated from 100 bootstrap models projections, averaged over the three regions of interest.
https://doi.org/10.1371/journal.pntd.0006236.g006
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 12 / 16
extrapolating Ae.albopictus abundance over the Balkan region may help to identify suitable
environmental niche where the species has never been reported so far.
From a biological point of view, our results are in line with the known literature [23,22,21]
and provide further insights on the predominant predictors of Ae.albopictus oviposition activ-
ity. The key predictor was related to antecedent oviposition activities from the previous week.
That is, the abundance of eggs for a given week is likely to generate a proportional abundance
of adults which will spawn, in return, a proportional abundance of eggs for the next-coming
week. Temperature from the previous week was the second most important predictors of eggs
abundance, presumably through the influence of temperature on adults spawning. Thus, these
results are consistent with the known biological activity of Ae.albopictus, [38] which in tem-
perate areas have developed the ability to induce photoperiodic egg diapause, allowing over-
wintering and further assisting its establishment in more northerly latitudes. [11] Altitude and
seasonal information related to the week of sampling was of minor influence on eggs activity.
It is worth to note that temperature-related predictors are seemingly correlated with both alti-
tude and seasonality, thus spatio-temporal times series of temperatures are likely to capture
some seasonal and altitudinal signal. Altitude can be a proxy for temperature, but may is also a
proxy for slope or insolation, which may impact oviposition activity. Land cover was not
shown to be a determinant factor influencing oviposition activity, however this effect might be
significant in some particular areas not covered by our sampling design e.g. proximity to water
bodies. Some other predictors could have been included to improve the model such as water-
related predictors that may strongly influence oviposition activity, [39] in particular the devel-
opment time from eggs immersion to the adult state. However, the satisfying predictive power
of our model confirms that the temperature-related predictors remain the most determinant
predictors among all candidate predictors.
The proposed forecasting model provide information useful for promoting active surveillance
and possibly prevention of Ae.albopictus colonization in presently non-infested areas in the Bal-
kans (e.g. Kosovo and Macedonia), as well as in high altitude areas, and could represent a helpful
instrument for assessing the actual risk of exotic arbovirus transmission in temperate regions.
Supporting information
S1 Dataset. Spatial and temporal model projections over the Balkans (Albania, Montene-
gro, Macedonia, Serbia, Kosovo, Greece, Croatia, Slovenia, Bulgaria, Rumania) at 0.01
degree spatial resolution (WGS84 coordinate reference system), and at a weekly interval
from 2009 to 2013.
(PDF)
Acknowledgments
The authors are grateful to Eng. Migel Ali, GIS expert at the Company Geo Consulting &
Invent Albania, who helps us to prepare the map specifically for this manuscript; and thank
the families in Albania, which allow us to carry the field work in their properties. Moreover we
thank Francis Schaffner, who provides the cylindrical vessel of black plastic container to per-
form the field collection of Ae.albopictus eggs and Erion Muhaxhiri, Viola Jani and Gjergji
Sino for technical assistance.
Author Contributions
Conceptualization: Enkelejda Velo, Silvia Bino, Kujtim Mersini, Beniamino Caputo, Els
Ducheyne, Alessandra della Torre, Paul Reiter, Marius Gilbert.
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 13 / 16
Data curation: Cle
´ment Tisseuil, Enkelejda Velo, Perparim Kadriaj, Kujtim Mersini, Artan
Simaku, Els Ducheyne, Marius Gilbert.
Formal analysis: Cle
´ment Tisseuil, Enkelejda Velo, Marius Gilbert.
Funding acquisition: Silvia Bino, Alessandra della Torre, Paul Reiter, Marius Gilbert.
Investigation: Enkelejda Velo, Silvia Bino, Perparim Kadriaj, Kujtim Mersini, Ada Shukullari,
Artan Simaku, Elton Rogozi, Beniamino Caputo, Paul Reiter.
Methodology: Cle
´ment Tisseuil, Enkelejda Velo, Kujtim Mersini, Beniamino Caputo, Els
Ducheyne, Alessandra della Torre, Paul Reiter, Marius Gilbert.
Project administration: Enkelejda Velo, Silvia Bino, Alessandra della Torre, Paul Reiter.
Resources: Silvia Bino, Alessandra della Torre, Paul Reiter.
Software: Cle
´ment Tisseuil, Els Ducheyne, Marius Gilbert.
Supervision: Silvia Bino, Alessandra della Torre, Paul Reiter, Marius Gilbert.
Validation: Cle
´ment Tisseuil, Marius Gilbert.
Visualization: Cle
´ment Tisseuil, Enkelejda Velo, Kujtim Mersini, Marius Gilbert.
Writing original draft: Cle
´ment Tisseuil, Enkelejda Velo.
Writing review & editing: Alessandra della Torre, Marius Gilbert.
References
1. Medlock JM, Hansford KM, Schaffner F, Versteirt V, Hendrickx G, Zeller H, et al. A review of the inva-
sive mosquitoes in Europe: ecology, public health risks, and control options. Vector Borne Zoonotic Dis.
2012; 12(6):435–47. https://doi.org/10.1089/vbz.2011.0814 PMID: 22448724
2. Schaffner F, Medlock JM, Van Bortel W. Public health significance of invasive mosquitoes in Europe.
Clin Microbiol Infec. 2013a; 19:685.
3. Wong PS, Li MZ, Chong CS, Ng LC, Tan CH. Aedes (Stegomyia) albopictus (Skuse): a potential vector
of Zika virus in Singapore. PLoS Negl Trop Dis. 2013 Aug 1; 7(8).
4. Chouin-Carneiro T, Vega-Rua A, Vazeille M, Yebakima A, Girod R, Goindin D, et al. Differential Sus-
ceptibilities of Aedes aegypti and Aedes albopictus from the Americas to Zika Virus. PLoS Negl Trop
Dis. 2016 Mar 3; 10(3).
5. Adhami J. & Reiter P. Introduction and establishment of Aedes (Stegomyia) albopictus Skuse (Diptera:
Culicidae) in Albania. J Am Mosq Control Assoc. 1998; 14: 340. PMID: 9813831
6. Romi R, Toma L, Severini F, Di Luca M. Twenty years of the presence of Aedes albopictus in Italy–
From the annoying pest mosquito to the real disease vector. Eur Inf Dis. 2008; 2: 98–101.
7. http://ecdc.europa.eu/en/healthtopics/vectors/vector-maps/Pages/VBORNET_maps.aspx
8. Juliano SA and Alto BW. Temperature effects on the dynamics of Aedes albopictus (Diptera:Culicidae)
populations in the laboratory. J.Med Entomol. 2001; 38(4): 548–556. PMID: 11476335
9. Roiz D, RosàR, Arnoldi D, Rizzoli A. Effect of temperature and rainfall on activity and dynamics of host-
seeking Aedes albopictus females in Northern. Vector Borne Zoo. Dis. 2010; 10.
10. Toma L, Severini F, Di Luca M, Bella A, and Romi R. Seasonal patterns of oviposition and egg hatching
rate of Aedes albopictus in Rome. Journal of the American Mosquito Control Association. 2003; 19(1):
19–22. PMID: 12674530
11. Hawley WA. The biology of Aedes albopictus. J Am Mosq Control Assoc Suppl. 1988; 1:1–39. PMID:
3068349
12. Vitek JC, and Livdahl PT. Field and laboratory comparison of hatch rates in Aedes albopictus (Skuse).
Journal of American Mosquito Control Association. 2006; 22(4):609–614. https://doi.org/10.2987/8756-
971X(2006)22[609:FALCOH]2.0.CO;2
13. Vanwambeke SO, Somboon P, Harbach RE, Isenstadt M, Lambin EF, Walton C, et al. Landscape and
Land Cover Factors Influence the Presence of Aedes and Anopheles Larvae. J. Med. Entomol. 2007;
44(1): 133–144. PMID: 17294931
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 14 / 16
14. Benedict MQ, Levine RS, Hawley WA, & Lounibos P. Spread of the Tiger: global risk of invasion by the
mosquito Aedes albopictus. Vector Borne Zoonotic Dis. 2007; 7, 76–85. (https://doi.org/10.1089/vbz.
2006.0562) PMID: 17417960
15. Caminade C, Medlock JM, Ducheyne E, McIntyre KM, Leach S, Baylis M, et al. Suitability of European
climate for the Asian tiger mosquito Aedes albopictus: recent trends and future scenarios. J R Soc Inter-
face. 2012; 9(75):2708–17. https://doi.org/10.1098/rsif.2012.0138 PMID: 22535696
16. Fisher D, Thomas SM, Niemitz F, Reineking B, & Beierkuhnlein C. Projection of Climate suitability for
Aedes albopictus Skuse (Culicidae) in Europe under climate change conditions. Glob. Planet. Change.
2011; 78, 54–65. (https://doi.org/10.1016/j.gloplacha.2011.05.008)
17. Kobayashi M, Nihei N, Kurihara T. Analysis of northern distribution of Aedes albopictus in Japan by geo-
graphical information systems. J Med Entomol. 2002; 39(1):4–11. http://dx.doi.org/10.1603/0022-
2585-39.1.4. PMID: 11931270
18. Neteler M, Roiz D, Rocchini D, Castellani C, Rizzoli A. Terra and Aqua satellites track tiger mosquito
invasion: modelling the potential distribution of Aedes albopictus in north-eastern Italy. Int J Health
Geogr. 2011; 10:49. https://doi.org/10.1186/1476-072X-10-49 PMID: 21812983
19. Roiz D, Neteler M, Castellani C, Arnoldi D, Rizzoli A. Climatic factors driving invasion of the tiger mos-
quito (Aedes albopictus) into new areas of Trentino, northern Italy. PLoS One. 2011; 6(4):e14800.
https://doi.org/10.1371/journal.pone.0014800 PMID: 21525991
20. Albieri A, Carrieri M, Angelini P, Baldacchini F, Venturelli C, Zeo SM, Bellini R. Quantitative monitoring
of Aedes albopictus in Emilia-Romagna, Northern Italy: cluster investigation and geostatistical analysis.
Bulletin of Insectology. 2010; 63 (2): 209–216.
21. Roiz D, RosàR, Arnoldi D, Rizzoli A. Effects of temperature and rainfall on the activity and dynamics of
host-seeking Aedes albopictus females in Northern Italy. Vector Borne Zoonotic Dis. 2010; 10:811–
816. https://doi.org/10.1089/vbz.2009.0098 PMID: 20059318
22. Roiz D, Boussès P, Simard F, Paupy C, Fontenille D. Autochthonous Chikungunya Transmission and
Extreme Climate Events in Southern France. PloS Negl Trop Dis. 2015; 9:e0003854. https://doi.org/10.
1371/journal.pntd.0003854 PMID: 26079620
23. Manica M, Filipponi F, D’Alessandro A, Screti A, Neteler M, RosàR, et al. Spatial and Temporal Hot
Spots of Aedes albopictus Abundance inside and outside a South European Metropolitan Area. PLOS
Negl Trop Dis 2016. https://doi.org/10.1371/journal.pntd.0004758 June 22, 2016. PMID: 27333276
24. Erguler K, Smith-Unna SE, Waldock J, Proestos Y, Christophides GK, Lelieveld J, et al. Large-Scale
Modelling of the Environmentally-Driven Population Dynamics of Temperate Aedes albopictus (Skuse).
PLoS ONE. 2016; 11(2): e0149282. https://doi.org/10.1371/journal.pone.0149282 PMID: 26871447
25. Medlock JM, Avenell D, Barrass I, & Leach S. Analysis of the potential for survival and seasonal activity
of Aedes albopictus (Diptera: Culicidae) in the United Kingdom. J. Vector Ecol. 2006; 31, 292–304.
(https://doi.org/10.3376/1081-1710(2006)31[292:AOTPFS]2.0.CO;2) PMID: 17249347
26. McCullagh P, & Nelder J A. Generalized linear models. Vol. 37. CRC press. 1989; pp. 511.
27. Velo E, Kadriaj P, Mersini K, Shukullari A, Manxhari B, Simaku A, et al. Enhancement of Aedes albopic-
tus collection by ovitrap and sticky adult trap. Parasites & Vectors. 2016; 9:223 https://doi.org/10.1186/
s13071-016-1501-x PMID: 27102015
28. Zamburlini R, Frilli F. Igiene Alimenti-Disifestazione & Igiene Ambientale. 2003; pg: 8–10.
29. Farr TG, Rosen PA, Caro E, Crippen R, Duren R, Hensley S. The Shuttle Radar Topography Mission,
Rev.Geophys. 2007; 45, RG2004, https://doi.org/10.1029/2005RG000183
30. EEA. Raster data on land cover for the CLC2006 inventory. Version 17. European Environmental
Agency. 2013. URL: http://www.eea.europa.eu/data-and-maps/data/corine-land-cover-2006-raster-3.
31. Zeileis A, Kleiber C, and Jackman S. Regression Models for Count Data in R. Journal of Statistical Soft-
ware. 2008; 27(8). URL http://www.jstatsoft.org/v27/i08/.
32. Zuur AF, Savaliev AA, & Ieno EN. Zero Inflated Models and Generalized Linear Mixed Models with R.
Highland Statistics Ltd. 2012.
33. Simon J. 2015. pscl: Classes and Methods for R Developed in the Political Science Computational Lab-
oratory, Stanford University. Department of Political Science, Stanford University. Stanford, California.
R package version 1.4.9. URL http://pscl.stanford.edu/
34. Breheny P, and Burchett W. 2014. visreg: Visualization of regression models. R package version 2.1–0.
http://CRAN.R-project.org/package=visreg.
35. Wickham H, and Francois R. 2015. dplyr: A Grammar of Data Manipulation. R package version 0.4.1.
http://CRAN.R-project.org/package=dplyr.
36. Wickham H. ggplot2: elegant graphics for data analysis. Springer New York, 2009.
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 15 / 16
37. Patsoula E, Beleri S, Vakali A, Pervanidou D, Tegos N, Nearchou A, et al. Records of Aedes albopictus
(Skuse, 1894) (Diptera; Culicidae) and Culex tritaeniorhynchus (Diptera; Culicidae) Expansion in Areas
in Mainland Greece and Islands. Vector Borne Zoonotic Dis. 2017 Mar; 17(3):217–223. https://doi.org/
10.1089/vbz.2016.1974 Epub 2017 Jan 11. PMID: 28075232
38. Briegel H, and Timmermann SE. Aedes albopictus (Diptera: Culicidae): Physiological Aspects of Devel-
opment and Reproduction. Journal of Medical Entomology. (2001): 38(4):566–571. http://dx.doi.org/
10.1603/0022-2585-38.4.566 PMID: 11476337
39. Cunze S, Kochmann J, Koch LK, Klimpel S. Aedes albopictus and Its Environmental Limits in Europe.
PLoS ONE. 2016; 11(9): e0162116. https://doi.org/10.1371/journal.pone.0162116 PMID: 27603669
Aedes albopictus oviposition activity in Albania and Balkan countries
PLOS Neglected Tropical Diseases | https://doi.org/10.1371/journal.pntd.0006236 February 12, 2018 16 / 16

Supplementary resource (1)

... Oviposition activity began in April (17-18 weeks) in 2018 and 2019 with 5.56% and 2.78% positive traps, respectively. The season of oviposition activity was completed in November (week [45][46] in 2017, with 11.43% positive ovitraps, and in 2018 and 2019 with 11.11% and 2.86% positive ovitraps, respectively. The dynamics of oviposition are shown in Figure S4, where the green color represents the samples with the lowest values and the red color the highest values for the total number of eggs, POI and MET. ...
... Regardless of the continental climate in Zagreb, same as in Novi Sad, Serbia, the beginning of oviposition activity in Novi Sad was different and started several weeks later (23rd-24th week) [44]. Temperature is considered the most determinant predictor of Ae. albopictus oviposition and its population growth in the Balkan countries [45], which influences the beginning of oviposition activity of these species [46]. It is evident that the temperature in Zagreb during the four-year study period had an influence on the Ae. ...
Article
Full-text available
Aedes albopictus and Aedes japonicus are invasive mosquito species that are causing great public concern. Aedes albopictus have successfully spread in Croatia, established in both the coastal and continental parts of the country, while Aedes japonicus is invading temperate climate areas. The invasive Aedes species are particularly attracted to the black plastic water containers and flower vases in cemeteries where they oviposit their eggs. Therefore, monitoring of this species was carried out in 12 cemeteries in Zagreb, using ovitraps with masonite strips as a substrate for oviposition. The monitoring was carried out from 2017 to 2020. The traps were inspected and the substrate was replaced every two weeks. This study showed that these two invasive species were present and very abundant in the cemeteries. In the case of Ae. albopictus, a higher population density and a greater number of occupied cemeteries were detected. This species was identified in all 12 cemeteries. Aedes albopictus was identified as the dominant species at all study sites. The spread of Ae. japonicus increased during 2018 in comparison to the previous year. Although this species was present in approximately 9% of the ovitraps, the observed population remained consistent throughout the course of the study. The findings indicate that cemeteries can be considered as significant public health hotspots, as the invasive Aedes mosquitoes tend to develop and reproduce in such environments. Consequently, the mosquito population of these two invasive species may only be reduced by applying integrated mosquito management measures, focused on the education of citizens.
... The procedure of temporal downscaling and spatial aggregation did not alter the observed seasonal pattern of Ae. albopictus egg abundance (Fig. 2). Moreover, the observed seasonal pattern well matched those reported in other studies 5,[55][56][57] or predicted by different modelling approaches 37,42,57 . ...
Article
Full-text available
Modelling approaches play a crucial role in supporting local public health agencies by estimating and forecasting vector abundance and seasonality. However, the reliability of these models is contingent on the availability of standardized, high-quality data. Addressing this need, our study focuses on collecting and harmonizing egg count observations of the mosquito Aedes albopictus, obtained through ovitraps in monitoring and surveillance efforts across Albania, France, Italy, and Switzerland from 2010 to 2022. We processed the raw observations to obtain a continuous time series of ovitraps observations allowing for an extensive geographical and temporal coverage of Ae. albopictus population dynamics. The resulting post-processed observations are stored in the open-access database VectAbundance.This initiative addresses the critical need for accessible, high-quality data, enhancing the reliability of modelling efforts and bolstering public health preparedness.
... It was introduced to Europe at the end of the 20th century (Lounibos, 2002;Scholte and Schaffner, 2007;, and its first occurrence was reported in Albania in 1979 (Adhami and Reiter, 1998). The species is now well-established and present even in tiny isolated villages in high-altitude (>1,200 m) environments (Tisseuil et al., 2018). Although the risk for pathogen transmission related to the autochthonous Aedes species is currently considered low, intensified vector control measures are needed to prevent disease outbreaks similar to those that occurred in various countries with unexpected local cases of chikungunya, dengue, and Zika (ECDC, Vermeulen et al., 2020). ...
... because temperature influences mosquito behaviors such as host-seeking, biting, and oviposition. (Roiz et al. 2011, Serpa et al. 2013, Stewart Ibarra et al. 2013, Costa et al. 2015, Estallo et al. 2015, Reinhold et al. 2018, Tisseuil et al. 2018, Gimenez et al. 2020, Benitez et al. 2021, de Souza et al. 2022) ...
Article
The transmission of Aedes-borne viruses is on the rise globally. Their mosquito vectors, Aedes aegypti (Linnaeus, Diptera: Culicidae) and Ae. albopictus (Skuse, Diptera: Culicidae), are focally abundant in the Southern United States. Mosquito surveillance is an important component of a mosquito control program. However, there is a lack of long-term surveillance data and an incomplete understanding of the factors influencing vector populations in the Southern United States. Our surveillance program monitored Ae. aegypti and Ae. albopictus oviposition intensity in the New Orleans area using ovicups in a total of 75 sites from 2009 to 2016. We found both Aedes spp. throughout the study period and sites. The average number of Ae. aegypti and Ae. albopictus hatched from collected eggs per site per week was 34.1 (SD = 57.7) and 29.0 (SD = 46.5), respectively. Based on current literature, we formed multiple hypotheses on how environmental variables influence Aedes oviposition intensity, and constructed Generalized Linear Mixed Effect models with a negative binomial distribution and an autocorrelation structure to test these hypotheses. We found significant associations between housing unit density and Ae. aegypti and Ae. albopictus oviposition intensity, and between median household income and Ae. albopictus oviposition intensity. Temperature, relative humidity, and accumulated rainfall had either a lagged or an immediate significant association with oviposition. This study provides the first long-term record of Aedes spp. distribution in the New Orleans area, and sheds light on factors associated with their oviposition activity. This information is vital for the control of potential Aedes-borne virus transmission in this area.
... Accidentally introduced in Europe in the 1970s, it is actually present in more than 20 countries of this continent [1][2][3]. Arrived in Northern Italy by tyre trade in 1990, A. albopictus is nowadays widespread and common in all Italian regions [4][5][6], mainly in human-made habitats within urban areas [7][8][9][10][11]. Due to the ongoing climate change, its distribution range is expected to further expand towards higher latitudes and altitudes [12][13][14]. ...
Article
Full-text available
The Asian tiger mosquito, Aedes albopictus, competent vector of several arboviruses, poses significant impact on human health worldwide. Although global warming is a driver of A. albopictus range expansion, few studies focused on its effects on homodynamicity (i.e. the ability to breed all-year-round), a key factor of vectorial capacity and a primary condition for an Aedes-borne disease to become endemic in temperate areas. Data from a 4-year monitoring network set in Central Italy and records from weather stations were used to assess winter adult activity and weekly minimum temperatures. Winter oviposition occurred in 38 localities with a seasonal mean photoperiod of 9.7 : 14.3 (L : D) h. Positive collections (87) occurred with an average minimum temperature of the two and three weeks before sampling of approximately 4°C. According to these evidences and considering the climate projections of three global climate models and three shared socio-economic pathways for the next three 20-year periods (from 2021 to 2080), the minimum temperature of January will increase enough to allow an all-year-round oviposition of A. albopictus in several areas of the Mediterranean Basin. Due to vector homodynamicity, Aedes-borne diseases could become endemic in Southern Europe by the end of the twenty-first century, worsening the burden on human health.
... It was introduced to Europe at the end of the 20th century (Lounibos, 2002;Scholte and Schaffner, 2007;Medlock et al., 2012), and its first occurrence was reported in Albania in 1979 (Adhami and Reiter, 1998). The species is now well-established and present even in tiny isolated villages in high-altitude (>1,200 m) environments (Tisseuil et al., 2018). Although the risk for pathogen transmission related to the autochthonous Aedes species is currently considered low, intensified vector control measures are needed to prevent disease outbreaks similar to those that occurred in various countries with unexpected local cases of chikungunya, dengue, and Zika (ECDC, 2019;Vermeulen et al., 2020). ...
Article
Full-text available
The pathogen transmitting Aedes albopictus mosquito is spreading rapidly in Europe, putting millions of humans and animals at risk. This species is well-established in Albania since its first detection in 1979. The sterile insect technique (SIT) is increasingly gaining momentum worldwide as a component of area-wide-integrated pest management. However, estimating how the sterile males will perform in the field and the size of target populations is crucial for better decision-making, designing and elaborating appropriate SIT pilot trials, and subsequent large-scale release strategies. A mark-release-recapture (MRR) experiment was carried out in Albania within a highly urbanized area in the city of Tirana. The radio-sterilized adults of Ae. albopictus Albania strain males were transported by plane from Centro Agricoltura Ambiente (CAA) mass-production facility (Bologna, Italy), where they were reared. In Albania, sterile males were sugar-fed, marked with fluorescent powder, and released. The aim of this study was to estimate, under field conditions, their dispersal capacity, probability of daily survival and competitiveness, and the size of the target population. In addition, two adult mosquito collection methods were also evaluated: BG-Sentinel traps baited with BG-Lure and CO2, (BGS) versus human landing catch (HLC). The overall recapture rates did not differ significantly between the two methods (2.36% and 1.57% of the total male released were recaptured respectively by BGS and HLC), suggesting a similar trapping efficiency under these conditions. Sterile males traveled a mean distance of 93.85 ± 42.58 m and dispersed up to 258 m. Moreover, they were observed living in the field up to 15 days after release with an average life expectancy of 4.26 ± 0.80 days. Whether mosquitoes were marked with green, blue, yellow, or pink, released at 3.00 p.m. or 6.00 p.m., there was no significant difference in the recapture, dispersal, and survival rates in the field. The Fried competitiveness index was estimated at 0.28. This mark-release-recapture study provided important data for better decision-making and planning before moving to pilot SIT trials in Albania. Moreover, it also showed that both BG-traps and HLC were successful in monitoring adult mosquitoes and provided similar estimations of the main entomological parameters needed.
Article
Full-text available
Originally from Southeast Asia, the tiger mosquito (Aedes albopictus) is now found almost everywhere in the world. Additionally, it spread throughout all of Northeastern Italy’s cities, including Trento, and settled in the Alpine regions. The invasion of mountain regions in the Alps above 600 meters, which is generally thought to be the upper limit of Ae. Albopictus’s range, is confirmed for the first time in this note. We observed it in the summer of 2023 between 615 and 708 meters above sea level in the Trento region’s surrounding mountains, specifically in the Montevaccino and Sopramonte localities. The mosquito was caught using ovitraps, and by raising the eggs to adulthood in a lab, the species identification was verified. This finding heightens worries about how tropical insect species, which are favored by global warming, may invade the mountains and affect tourism.
Article
Full-text available
Species distribution modeling (SDM) has become an increasingly common approach to explore questions about ecology, geography, outbreak risk, and global change as they relate to infectious disease vectors. Here, we conducted a systematic review of the scientific literature, screening 563 abstracts and identifying 204 studies that used SDMs to produce distribution estimates for mosquito species. While the number of studies employing SDM methods has increased markedly over the past decade, the overwhelming majority used a single method (maximum entropy modeling; MaxEnt) and focused on human infectious disease vectors or their close relatives. The majority of regional models were developed for areas in Africa and Asia, while more localized modeling efforts were most common for North America and Europe. Findings from this study highlight gaps in taxonomic, geographic, and methodological foci of current SDM literature for mosquitoes that can guide future efforts to study the geography of mosquito-borne disease risk. Graphical Abstract
Article
Full-text available
Mosquito-borne infections are increasing in endemic areas and previously unaffected regions. In 2020, the notification rate for Dengue was 0.5 cases per 100,000 population, and for Chikungunya <0.1/100,000. In 2019, the rate for Malaria was 1.3/100,000, and for West Nile Virus, 0.1/100,000. Spatial analysis is increasingly used in surveillance and epidemiological investigation, but reviews about their use in this research topic are scarce. We identify and describe the methodological approaches used to investigate the distribution and ecological determinants of mosquito-borne infections in Europe. Relevant literature was extracted from PubMed, Scopus, and Web of Science from inception until October 2021 and analysed according to PRISMA-ScR protocol. We identified 110 studies. Most used geographical correlation analysis (n = 50), mainly applying generalised linear models, and the remaining used spatial cluster detection (n = 30) and disease mapping (n = 30), mainly conducted using frequentist approaches. The most studied infections were Dengue (n = 32), Malaria (n = 26), Chikungunya (n = 26), and West Nile Virus (n = 24), and the most studied ecological determinants were temperature (n = 39), precipitation (n = 24), water bodies (n = 14), and vegetation (n = 11). Results from this review may support public health programs for mosquito-borne disease prevention and may help guide future research, as we recommended various good practices for spatial epidemiological studies.
Article
Full-text available
Annual entomological surveillance programs aiming to monitor mosquito populations and record presence and absence of mosquito species have been performed in Greece. We report, in this study, new records and expansion of Aedes albopictus in the islands of Lesvos (region of North Aegean), Crete (region of Crete), and the regional units of Rodopi in East Macedonia-Thrace. Furthermore, Culex tritaeniorhynchus was recorded for the first time in Arta (region of Epirus) in northwestern Greece.
Article
Full-text available
The Asian tiger mosquito Aedes albopictus, native to South East Asia, is listed as one of the worst invasive vector species worldwide. In Europe the species is currently restricted to Southern Europe, but due to the ongoing climate change, Ae. albopictus is expected to expand its potential range further northwards. In addition to modelling the habitat suitability for Ae. albopictus under current and future climatic conditions in Europe by means of the maximum entropy approach, we here focused on the drivers of the habitat suitability prediction. We explored the most limiting factors for Aedes albopictus in Europe under current and future climatic conditions, a method which has been neglected in species distribution modelling so far. Ae. albopictus is one of the best-studied mosquito species, which allowed us to evaluate the applied Maxent approach for most limiting factor mapping. We identified three key limiting factors for Ae. albopictus in Europe under current climatic conditions: winter temperature in Eastern Europe, summer temperature in Southern Europe. Model findings were in good accordance with commonly known establishment thresholds in Europe based on climate chamber experiments and derived from the geographical distribution of the species. Under future climatic conditions low winter temperature were modelled to remain the most limiting factor in Eastern Europe, whereas in Central Europe annual mean temperature and summer temperatures were modelled to be replaced by summer precipitation, respectively, as most limiting factors. Changes in the climatic conditions in terms of the identified key limiting factors will be of great relevance regarding the invasive potential of the Ae. albopictus. Thus, our results may help to understand the key drivers of the suggested range expansion under climate change and may help to improve monitoring programmes. The applied approach of investigating limiting factors has proven to yield valuable results and may also provide valuable insights into the drivers of the prediction of current and future distribution of other species. This might be particularly interesting for other vector species that are of increasing public health concerns.
Article
Full-text available
Aedes albopictus is a tropical invasive species which in the last decades spread worldwide, also colonizing temperate regions of Europe and US, where it has become a public health concern due to its ability to transmit exotic arboviruses, as well as severe nuisance problems due to its aggressive daytime outdoor biting behaviour. While several studies have been carried out in order to predict the potential limits of the species expansions based on eco-climatic parameters, few studies have so far focused on the specific effects of these variables in shaping its micro-geographic abundance and dynamics. The present study investigated eco-climatic factors affecting Ae. albopictus abundance and dynamics in metropolitan and sub-urban/rural sites in Rome (Italy), which was colonized in 1997 and is nowadays one of the most infested metropolitan areas in Southern Europe. To this aim, longitudinal adult monitoring was carried out along a 70 km-transect across and beyond the most urbanized and densely populated metropolitan area. Two fine scale spatiotemporal datasets (one with reference to a 20m circular buffer around sticky traps used to collect mosquitoes and the second to a 300m circular buffer within each sampling site) were exploited to analyze the effect of climatic and socio-environmental variables on Ae. albopictus abundance and dynamics along the transect. Results showed an association between highly anthropized habitats and high adult abundance both in metropolitan and sub-urban/rural areas, with “small green islands” corresponding to hot spots of abundance in the metropolitan areas only, and a bimodal seasonal dynamics with a second peak of abundance in autumn, due to heavy rains occurring in the preceding weeks in association with permissive temperatures. The results provide useful indications to prioritize public mosquito control measures in temperate urban areas where nuisance, human-mosquito contact and risk of local arbovirus transmission are likely higher, and highlight potential public health risks also after the summer months typically associated with high mosquito densities
Article
Full-text available
Background In the last decades, Aedes albopictus has become an increasing public health threat in tropical as well as in more recently invaded temperate areas due to its capacity to transmit several human arboviruses, among which Dengue, Chikungunya and Zika. Enhancing the efficiency of currently used collection approaches, such as ovitraps and sticky traps, is desirable for optimal monitoring of the species abundance, for assessment of the risk of arbovirus transmission and for the optimisation of control activities. Findings Two sets of 4 × 4 Latin-square experiments were carried out in Tirana (Albania) to test whether modifications in ovitrap shape and size and in oviposition substrate would increase collections of Ae. albopictus eggs and whether hay-infusion would increase adult catches by sticky trap. Generalized Linear Mixed Models with negative binomial error distribution were carried out to analyse the data. Cylindrical ovitraps lined with germination paper yielded significantly higher egg catches than those exploiting either the (commonly used) wooden paddles or floating polystyrene blocks as oviposition substrates. No difference was observed between cylindrical and conical shaped ovitraps. Ovitraps and sticky traps baited with hay infusion yielded significantly higher egg and adult catches than un-baited ones. A significant relationship between ovitrap and sticky trap catches was observed both in the absence and in the presence of attractants, with ovitrap catches increasing more than sticky trap catches at increasing adult female densities. Conclusions This study provides grounds for optimisation of ovitraps and sticky traps as monitoring tools for Ae. albopictus by (i) supporting use of germination paper as most appropriate oviposition substrate; (ii) suggesting the possible use of stackable conical ovitraps for large scale monitoring; (iii) confirming the use of hay-infusion to increase egg catches in ovitraps, and showing that hay-infusion also significant increases adult catches by sticky traps.
Article
Full-text available
Background Since the major outbreak in 2007 in the Yap Island, Zika virus (ZIKV) causing dengue-like syndromes has affected multiple islands of the South Pacific region. In May 2015, the virus was detected in Brazil and then spread through South and Central America. In December 2015, ZIKV was detected in French Guiana and Martinique. The aim of the study was to evaluate the vector competence of the mosquito spp. Aedes aegypti and Aedes albopictus from the Caribbean (Martinique, Guadeloupe), North America (southern United States), South America (Brazil, French Guiana) for the currently circulating Asian genotype of ZIKV isolated from a patient in April 2014 in New Caledonia. Methodology/Principal Findings Mosquitoes were orally exposed to an Asian genotype of ZIKV (NC-2014-5132). Upon exposure, engorged mosquitoes were maintained at 28°±1°C, a 16h:8h light:dark cycle and 80% humidity. 25–30 mosquitoes were processed at 4, 7 and 14 days post-infection (dpi). Mosquito bodies (thorax and abdomen), heads and saliva were analyzed to measure infection, dissemination and transmission, respectively. High infection but lower disseminated infection and transmission rates were observed for both Ae. aegypti and Ae. albopictus. Ae. aegypti populations from Guadeloupe and French Guiana exhibited a higher dissemination of ZIKV than the other Ae. aegypti populations examined. Transmission of ZIKV was observed in both mosquito species at 14 dpi but at a low level. Conclusions/Significance This study suggests that although susceptible to infection, Ae. aegypti and Ae. albopictus were unexpectedly low competent vectors for ZIKV. This may suggest that other factors such as the large naïve population for ZIKV and the high densities of human-biting mosquitoes contribute to the rapid spread of ZIKV during the current outbreak.
Article
Full-text available
The Asian tiger mosquito, Aedes albopictus, is a highly invasive vector species. It is a proven vector of dengue and chikungunya viruses, with the potential to host a further 24 arboviruses. It has recently expanded its geographical range, threatening many countries in the Middle East, Mediterranean, Europe and North America. Here, we investigate the theoretical limitations of its range expansion by developing an environmentally-driven mathematical model of its population dynamics. We focus on the temperate strain of Ae. albopictus and compile a comprehensive literature-based database of physiological parameters. As a novel approach, we link its population dynamics to globally-available environmental datasets by performing inference on all parameters. We adopt a Bayesian approach using experimental data as prior knowledge and the surveillance dataset of Emilia-Romagna, Italy, as evidence. The model accounts for temperature, precipitation, human population density and photoperiod as the main environmental drivers, and, in addition, incorporates the mechanism of diapause and a simple breeding site model. The model demonstrates high predictive skill over the reference region and beyond, confirming most of the current reports of vector presence in Europe. One of the main hypotheses derived from the model is the survival of Ae. albopictus populations through harsh winter conditions. The model, constrained by the environmental datasets, requires that either diapausing eggs or adult vectors have increased cold resistance. The model also suggests that temperature and photoperiod control diapause initiation and termination differentially. We demonstrate that it is possible to account for unobserved properties and constraints, such as differences between laboratory and field conditions, to derive reliable inferences on the environmental dependence of Ae. albopictus populations.
Article
Generalized linear mixed models extend linear mixed models, or hierarchical linear models, to accommodate noncontinuous responses, such as binary responses or counts. Such models are useful when the data are clustered in some way, a canonical example in education being students nested in schools. Another important example is longitudinal data where repeated observations of the same students are nested within students. Another term for generalized linear mixed models is hierarchical or multilevel generalized linear models. The terms random coefficient models or random effects models are also often used for either linear or generalized linear mixed models. We start by fitting a particular kind of generalized linear mixed model, a logistic random intercept model, to data on reading proficiency of students nested in schools. We then discuss random coefficient models and models for different response types, conditional and marginal relationships, estimation and software, and assignment of values to random effects. We close by briefly considering some useful extensions to the models discussed here.