During the late summer 2008 a large epidemic of West Nile virus occurred in North-East Italy involving an area of more than 7,000 Kmsq in 3 Regions (Lombardia, Veneto, Emilia-Romagna). This was the largest epidemic ever recorded in the country. Following the first evidence in equine and birds an active entomological surveillance plan was started by the Emilia-Romagna Surveillance Group on Vectorial Disease. Mosquito collection was conducted by CO₂ baited traps specifically positioned in 78 sites. Totally in the period September 3 - October 23, more than 20,000 mosquitoes of which 53% Culex pipiens, 43% Aedes caspius, 2% Aedes albopictus, 1% Aedes vexans, 0.2% Anopheles maculipennis, 0.02% Culex modestus, 0.02% Culiseta annulata were analysed by RT-PCR.

Two pools of Cx.pipiens collected at the end of September in different localities tested positive for WN virus.

In the 2009 summer the surveillance was activated focusing on sentinel equine, residential birds (corvids) and mosquitoes. Mosquito collection was conducted in the period week 25-42 by CO₂ traps in 75 fixed stations with weekly to monthly periodicity.

Totally more than 150,000 mosquitoes were collected and analyzed.

Starting from the end of July WN positivity was detected in Culex pipiens pools, magpie (Pica pica), and equines in and around the WN interested area 2008.

No human cases were registered at the time of abstract submission (August 31).

The persistence of WN activity in the Po plain for two consecutive years indicates that the area is becoming suitable for WN establishment and possible endemicity, stressing the need to organize standard surveillance measures aimed to the early detection of WN activity and risk evaluation in public health.

 

Key words: arbovirus, surveillance, CO₂ trap, Culex pipiens, risk assessment

Distribution and density of Aedes albopictus population in urban areas are promoted by local scale environmental variables like vegetation, land use, premises condition, catch basins number and quality, etc.

Study of possible correlations between ovitraps monitoring data and environmental data at local scale can provide useful information to improve species distribution-abundance landscape-based models. These modelling approach is more appropriate than classic geostatistics  to model mosquitoes abundance in particular when datasets are less autocorrelated and not stationary.

The present study was carried out within the framework of the Emilia-Romagna  Ae.albopictus monitoring and control program with the aim to investigate the correlations between some environmental data (vegetation and Normalized Difference Vegetation Index calculated from QuickBird high resolution images, number of city premises and density of streets calculated from Emilia-Romagna Layers) and mean eggs density using data from 107 ovitraps positioned in the urban area of Bologna (8,608 ha) by standard criteria at a mean distance of 600 m, checked weekly during the favourable season 2009 (May-September). Logistic regression models are used to compare the habitat characteristics of positive versus negative samples in order to determine habitat suitability with a set of explanatory variables.

In this study we used high/low mosquito abundance rather than presence/absence as the response variable. The mean was used to split the dataset into high and low abundance.

“Buffering” GIS procedure was used to extract values of variables within circular area (50 and 100 m) around 107 study ovitraps.

We evaluated whether ovitraps data were autocorrelated (significant correlation between  data from closer ovitraps ) for each monitoring month and for the whole monitoring season.

Then Logistic regression model was applied to mean egg densities of four months (June, July, August and September) and for the whole monitoring season, to identify the variables influencing mosquito distribution-abundance.

To simplify the data analysis each variable was categorized into percentile.

Akaike’s Information Criterion (AIC) was used to rank candidate models for promoting creation of Ae.albopictus distribution models in Urban Areas and to produce smoothed maps of virus (Chikungunya and Dengue) transmission at high resolution.

 

Key words: logistic regression, QuickBird images, ovitraps, Aedes albopictus, NDVI