ORIGINAL CONTRIBUTION Large-scale simulation of temperature-dependent phenology in wintering populations of Bactrocera oleae (Rossi) R. Petacchi 1 , S. Marchi 1 , S. Federici 2 & G. Ragaglini 1 1 Life Science Institute, Scuola Superiore Sant’Anna, Pisa, Italy 2 Regional Center of Applied Agrometereology (CAAR), Sarzana, SP, Italy Keywords geographical information system, insect modelling, landscape monitoring network, olive fruit fly bio-ecology, spatial statistics Correspondence Susanna Marchi (corresponding author), Life Science Institute, Scuola Superiore Sant’Anna Viale R. Piaggio 34, I-56025 Pontedera, Italy. E-mail: susanna.marchi@sssup.it Received: September 6, 2014; accepted: Octo- ber 31, 2014. doi: 10.1111/jen.12189 Abstract To implement Area-Wide Pest Management protocols at a regional scale (Liguria, northern Italy), egg deposition and adult flight of olive fruit fly, Bactrocera oleae, were monitored during 2009, 2010 and 2011. The conse- quence of complete generation in late winter – early spring was also examined. The reliability of a degree-day model was tested to simulate the insect cycle, starting from October oviposition and considering a 379.01°C cumulative degree-day (CDD) needed to complete development. The model was validated and then used to simulate olive fruit fly phenology in the region of Liguria, using a GIS approach and the agrometeorological network in the region. The output of the CDD model was mapped with two different spatialization modelling techniques, geostatistical autocorre- lation and regression correlation, and altitude, aspect and distance from the sea were assessed as elements of variability. The regression correlation model provided a more accurate indication of B. oleae diversity at the local scale than the geostatistical autocorrelation model and identified the high spatial climatic variability of Liguria. The potential application of the dis- tribution of days after oviposition and prediction error maps in support of pest management planning is discussed. Introduction Area-Wide Pest Management (AWPM) has been rec- ognized as one of the most effective solutions for reducing costs end environmental impacts of pest management both in agricultural and in forestry sys- tems (Koul et al. 2008). However, the efforts required for AWPM programmes can lead to increasing costs and inconsistent results, if monitoring networks are not properly implemented with information on popu- lation ecology, as well as the environmental condi- tions in which the pest and its hosts develop. The key issue in area-wide settings is the knowledge of the temporal and spatial distribution of the pest within the landscape (Sciarretta and Trematerra 2011; Castri- gnan o et al. 2012). Thus, relating insect distribution over time with environmental factors (i.e. climate and geography) and landscape traits (i.e. arrangement, connectivity and quality of habitat patches) is an essential part of sustainable pest management strate- gies (Thies et al. 2003). The combined use of spatial statistics and phenolog- ical models for predicting the geographical distribu- tion and temporal dynamics of insect populations may significantly increase the efficacy of AWPM pro- grammes (e.g. Liebhold et al. 1993). The development of GIS-based decision support systems has combined territorial and meteorological information with clear advantages. Previous studies (R egni ere 1996) have addressed the problem of how to use temperature-dri- ven simulation models in AWPM. The generalized approach consists in relating some arbitrary feature of the model output (e.g. the date of occurrence of a spe- cific developmental stage), by multiple regressions, to geographical factors by characterizing a set of target sites over the area of interest. Temperature input for the simulations is usually collected from a set of nearby weather stations, and the model output is used J. Appl. Entomol. 139 (2015) 496–509 © 2014 Blackwell Verlag GmbH 496 J. Appl. Entomol.