RVF vector spatial distribution models: vector abundance

First published in EFSA Supporting Publications
22 Abril 2020
Type
External Scientific Report

Abstract

EFSA has commissioned the VectorNet consortium to undertake a series of spatial distribution models for seven potential mosquito vectors of Rift Valley fever virus, namely Aedesalbopictus, Aedescaspius, Aedes detritus, Aedesjaponicus, Aedesvexans, Culexpipiens and Culextheileri. The modelling used the distribution data held within the VectorNet archive (as of September 2019), updated by literature searches to acquire new records available since 2016. The modelling has been implemented in three phases: (i) data collection, collation and standardisation; (ii) spatial modelling for presence and absence, and the calculation of presence metrics at the country level to be compatible with the MintRisk utilities; and (iii) the spatial modelling of vector abundance, dependent on the data available. This document presents the results of the abundance modelling due for delivery in March 2020. Sufficient data were amassed to produce statistically reliable spatial models for all species except Ae. detritus. Data for abundance models were extracted and the abundance values standardised for sampling effort where possible. Additional corrections were implemented that attempted to standardise for trapping methods and life stages, using expert opinion to define the conversion factors. The models were implemented at 1 km resolution covering the whole of continental Europe, using standard modelling techniques (Boosted Regression Trees and Random Forest) implemented through the VECMAP software suite. The models were evaluated according to expert opinion and the degree to which they matched the presence/absence models. Whilst all models produced were statistically reliable ‐ so represented the input data effectively ‐ not all were judged to reflect the field situation, implying that the input data were not sufficiently complete or extensive to feed continental scale distribution models. Training data and outputs for selected models are supplied in ESRI compatible format.

Contact
ALPHA [at] efsa.europa.eu
doi
10.2903/sp.efsa.2020.EN-1847