CSIRO PUBLISHING www.publish.csiro.au/journals/ijwf International Journal of Wildland Fire, 2006, 15, 375–388 A 229-year dendroclimatic-inferred record of forest fire activity for the Boreal Shield of Canada Martin P. Girardin A,C,F , Yves Bergeron B , Jacques C. Tardif C , Sylvie Gauthier A , Mike D. Flannigan D and Manfred Mudelsee E A Natural Resources Canada, Canadian Forest Service, Laurentian Forestry Centre, 1055 du PEPS, PO Box 10380, Stn. Sainte-Foy, QC G1V 4C7, Canada. B Groupe de recherche en écologie forestière inter-universitaire (GREFI), Université du Quebec à Montréal, CP 8888, Succ. Centre-Ville, Montréal, QC H3C 3P8, Canada. C Center for Forest Interdisciplinary Research (C-FIR), University of Winnipeg, 515Avenue Portage, Winnipeg, MB R3B 2E9, Canada. D Natural Resources Canada, Canadian Forest Service, Great Lakes Forestry Centre, 1219 Queen Street-East, Sault Ste. Marie, ON P6A 2E5, Canada. E Institute of Meteorology, University of Leipzig, Stephanstrasse 3, 04103 Leipzig, Germany, and Climate RiskAnalysis, Wasserweg 2, 06114 Halle, Germany. F Corresponding author. Email: martin.girardin@rncan.gc.ca Abstract. Six independent tree-ring reconstructions of summer drought were calibrated against instrumental fire data to develop a 229-year dendroclimatic-inferred record of fire activity (annual area burned and fire occurrence) on the Boreal Shield, Canada. As a means of validating the statistical reconstructions of the fire activity, a comparison was made with a stand age distribution derived from a regional time-since-last-fire map for an area located at the transition between the mixedwood and coniferous boreal forests of south-western Quebec. Calibration statistics indicated that 31% of the area burned variance and 45% of the fire occurrence variance could be accounted for by the six drought reconstructions. The verification statistics indicated a tendency for the statistical reconstructions of the fire activity to reproduce with confidence both high and relatively low frequency variations in fire. Episodes of succeeding years with important fire activity were estimated for 1789–1796, 1820–1823, 1837–1841, 1862–1866, 1906–1912, 1919–1922, 1933–1938, and 1974–1977. Also estimated were periods of reduced forest fire activity, particularly in the occurrence rate of extreme fire years, from c. 1850 to 1900 and again during the second half of the 20th century. Correlation analysis between the statistical reconstruction of the area burned and the stand age distribution suggested that both proxies shared similar information on the fire activity. Correlation maps, however, indicated that variability in the statistical reconstructions was not necessarily representative of fire activity in all regions of the Boreal Shield. Introduction Climate is always changing owing to a number of factors such as changes in the earth’s orbital, chemical composition of the atmosphere, solar variability, and volcanic activities (Bonan 2002). Thus, with a dynamic climate and a strong linkage between climate and forest fires, variations in histor- ical observations of fire activity due to changes in the climate are expected (Flannigan and Harrington 1988; Johnson 1992; Swetnam 1993). Empirical data and model simulations sug- gested important changes in fire activity across the Canadian boreal forest in a context of a rapid change of the global climate (Bergeron et al. 2004a; Flannigan et al. 2005). Already, the warming of the Northern Hemisphere, estimated at ∼0.6 ◦ C in the past century (Zhang et al. 2000; Folland et al. 2001; Houghton et al. 2001), has had a detectable influence on fire activity. According to analyses of forest stand age distributions, the lower fire activity since c. 1850 reported in many regions of the Canadian boreal forest could, in part, reflect the impact of a changing climate (Masters 1990; John- son and Larsen 1991; Larsen 1997; Bergeron et al. 2001, 2004a, 2004b; Tardif 2004). Large forest fires in boreal Canada are often associated with upper level longwave ridging at boreal latitudes (Skin- ner et al. 1999, 2002). These large and persisting blocking © IAWF 2006 10.1071/WF05065 1049-8001/06/030375