Fuel age, weather and burn probability in Portugal Paulo M. Fernandes A,B,C , Carlos Loureiro A,B , Marco Magalha ˜es B , Pedro Ferreira B and Manuel Fernandes B A Centro de Investigac ¸a ˜o e de Tecnologias Agro-Ambientais e Tecnolo ´ gicas (CITAB), Universidade de Tra ´s-os-Montes e Alto Douro, Apartado 1013, PT-5001-801, Portugal. B Departamento de Cie ˆncias Florestais e Arquitectura Paisagista, Universidade de Tra ´s-os-Montes e Alto Douro, Apartado 1013, PT-5001-801, Portugal. C Corresponding author. Email: pfern@utad.pt Abstract. The relative influence of the factors acting on burn probability, namely fuel and weather, is not well understood, especially in Europe. We use a digital fire atlas (1975–2008) and apply survival analysis to individual fires (1998–2008) to describe how burn probability changes with fuel age in Portugal. The typical fire return interval and median fire-free interval vary regionally from 23 to 52 and 18 to 47 years. Increase of the hazard of burning with time is generally near-linear, denoting moderate fuel-age dependency, as in some other shrub-dominated Mediterranean environments. Analysis of complete fire intervals resulted in shorter fire return interval and higher fuel-age dependency of burn probability than findings that included censored observations. Increasingly severe weather conditions either expressed through fire size or by extreme fire danger concurrently decreased fuel-age dependency and selected older fuels. The results are discussed from the viewpoints of fire suppression and fuel treatments. Additional keywords: fire frequency analysis, fire regimes, Mediterranean-type ecosystems, shrubland. Received 9 November 2010, accepted 24 September 2011, published online 9 March 2012 Introduction The spatial and temporal patterns of landscape fire spread are determined by a complex interplay between the biophysical setting and humans as agents of both fire ignition and suppres- sion. Fire regimes are determined by a mix of top–down (climate and weather) and bottom–up (ignitions and fuels) controls, which respectively are exogenous and endogenous to the system (Boer et al. 2008). Increased landscape-level accumulation and connectivity of flammable fuels and more frequent extreme weather events are thought to have promoted an increase in large fires in the western Mediterranean Basin over the last decades (e.g. Dı ´az-Delgado et al. (2004). A robust cause–effect relationship exists between weather conditions and fire activity in the Mediterranean Basin. Carvalho et al. (2008) and Camia and Amatulli (2009) were able to explain more than 80% of the monthly area burned variability from components of the Canadian Forest Fire Weather Index (Van Wagner 1987), at Portuguese and Mediterranean European levels. A strongly asymmetrical fire size distribution implies that a small number of fires accounts for most of the area burned: 35 to 85% of the annual burned surface in Portugal from 1984 to 2004 was due to fires .100 ha (de Zea Bermudez et al. 2009). These large fires, as shown by Pereira et al. (2005) and Hoinka et al. (2009), occur in association with abnormally hot and dry atmospheric conditions. Understanding the relative influences of the mechanisms governing fire incidence is a pertinent scientific question with important land-management implications. If fire occur- rence is time-dependent, then weather will play a minor role on fire incidence, which will be controlled by the existing fuel age mosaic and thereby could be mitigated by fuel treatments (Minnich 1983). The opposing view contends that fuel age is poorly related to burn probability in crown-fire ecosystems such as Mediterranean conifer forests and shrub- lands, and has gained momentum in recent years (Keeley 2002; Moritz et al. 2004; Keeley and Zedler 2009). Ignition density and drought (Bradstock 2010) and the likelihood and severity of extreme wind events (Moritz et al. 2010) should influence the trade-offs between fuel and weather as drivers of burn probability. Fire frequency analysis of spatially explicit fire history data has been used to better describe and understand how fuel age affects fire incidence (e.g. Moritz 2003), adding to the debate over the environmental drivers of fire incidence and the rele- vance of managing fuels. Contemporary fire history data are analysed in the present study with the objectives of determining (i) how fuel age affects the recurrence of fire in Portugal and across its ecoregions, and (ii) how weather influences the dependence of burn probability on fuel age. Our expectations were that fire frequency would vary regionally, and that the CSIRO PUBLISHING International Journal of Wildland Fire 2012, 21, 380–384 http://dx.doi.org/10.1071/WF10063 Journal compilation Ó IAWF 2012 www.publish.csiro.au/journals/ijwf Research Note