UNCORRECTED PROOF Area (2007) 39.3, 000 – 000 1 5 10 15 20 25 30 35 40 45 50 A R E A 7 5 6 Operator: Huang Liping Dispatch: 04.06.07 PE: Sally Moore Journal Name Manuscript No. Proofreader: Song Laihui No. of Pages: 11 Copy-editor: Area Vol. 39 No. 3, pp. 000 – 000, 2007 ISSN 0004-0894 © The Authors. Journal compilation © Royal Geographical Society (with The Institute of British Geographers) 2007 Blackwell Publishing Ltd Multivariate analysis of landscape wildfire dynamics in a Mediterranean ecosystem of Greece Kostas D Kalabokidis*, Nikos Koutsias**, Pavlos Konstantinidis† and Christos Vasilakos‡ *Department of Geography, University of the Aegean, 81100 Mytilene, Greece Email: kalabokidis@aegean.gr **Department of Environmental and Natural Resources Management, University of Ioannina, 30100 Agrinio, Greece †Forest Research Institute, NAGREF, 57006 Vasilika-Thessaloniki, Greece ‡Department of Environmental Studies, University of the Aegean, 81100 Mytilene, Greece Revised manuscript received 27 March 2007 This paper focuses on spatial distribution of long-term fire patterns versus physical and anthropogenic elements of the environment that determine wildfire dynamics in Greece. Logistic regression and correspondence analysis were applied in a spatial database that had been developed and managed within a Geographic Information System. Cartographic fire data were statistically correlated with basic physical and human geography factors (geomorphology, climate, land use and human activities) to estimate the degree of their influence at landscape scale. Land cover types of natural and agricultural vegetation were the most influential factors for explaining landscape wildfire dynamics in conjunction with topography and grazing. Key words: Greece, forest fires, multivariate statistics, physical geography, human geography, GIS Introduction Spatiotemporal attributes are important characteristics in landscape and wildfire dynamics (see Barbour et al. 2005; Roloff et al. 2005). Spatial analysis of landscape wildfire may be from local to global scales, while temporal resolution can be either short- or long-term. Consequently, wildfire and vegetation dynamics have been analysed using Geographic Information Systems (GIS) as it offers an effective way to manage the spatial and temporal information (Chou 1992; Salas and Chuvieco 1994; Kalabokidis et al. 2002; Miller et al. 2003). Vegetation mapping for fire dynamics is com- plicated because the existence of similar plants will not necessarily result in similar wildfire behaviour. Wildfire behaviour potential is strongly correlated with the quantity, size, density, moisture and quality of vegetation as these determine the amount of fuel available for combustion (Andrews et al. 2003; Pyne et al. 1996). Vegetation interacts with topography and weather to create conditions of fire behaviour unique in time and space. Simulation modelling can be used to predict fire potential at broad spatial scales. The primary tool used to model fire behaviour at different landscapes is FARSITE (Finney 1998). This programme integrates geospatial fuel data, climatic data and physically-based modelling of fire behaviour (BEHAVE; Andrews 1986). How- ever, owing to the quantity and quality requirements of the data and to other constraints of working on wildfires (Peterson et al. 2005), application of statis- tical/empirical models (as in this research study) complements simulation methods for analysis of physical and human impacts on landscape wildfire dynamics.