Temporal variations in the usefulness of normalised difference vegetation index as a predictor for Ixodes ricinus (Acari: Ixodidae) in a Borrelia lusitaniae focus in Tuscany, central Italy

Host-seeking ticks were collected during monthly dragging sessions from November 2004 through October 2006 in Tuscany, central Italy. Normalized difference vegetation index (NDVI), which was calculated from Landsat ETM+ 7 remote sensing data recorded in August 2001, was signiÞcantly correlated with numbers of host-seeking immature Ixodes ricinus L. (Acari: Ixodidae) during periods of relatively low rainfall such as summer 2005 (Spearman's p= 0.78, P < 0.001 for nymphs in July) and to a lower extent in springÐsummer 2006. In spring 2005, when rainfall was relatively high, the correlation was weak and not statistically signiÞcant. Generalized estimating equations (GEEs), taking into account repeated sampling of the same dragging sites, were used to model the effects of NDVI and season on counts of host-seeking I. ricinus nymphs. Seasonal variations of the effect of NDVI yielded a signiÞcant NDVI-by-season interaction in the Þrst year of the study (November 2004-October 2005), but not in the second year (November 2005-October 2006) when there was a 2.5-fold increase of the number of nymphs per 100-m dragging for every 0.1 unit increase in NDVI (95% conÞdence interval= 1.6, 3.0). Risk maps that were obtained based on GEE results conÞrmed that the predicted number of I. ricinus nymphs per 100 m was relatively homogeneous through the study area during the 2005 spring peak of activity. Conversely, in 2006, the predicted abundance of nymphs was greater in moist bottomland habitat (characterized by high NDVI) than in dry, typically Mediterranean, upland habitat.