i EXECUTIVE SUMMARY We use a surface-fire extinction model to ask whether differences among surface fuels in aspen and conifer stands in the Saskatchewan mixedwood boreal forest are sufficiently large to affect patterns of burning across the landscape (i.e., where fires spread and where they go extinct). The key variables in the extinction model are fuel-bed surface area (a measure of the loading of fine fuels) and fuel moisture (determiined by weather and live-to-dead fuel ratios). A larger

more »

... l-bed surface area means that fires should propagate at higher fuel moisture levels. Surface fuels were sampled in 56 stands spanning the range in upland fuel variability. Dead-fuel moisture was estimated from weather data and fuel-drying models for periods during large-areaburned years when large fires made runs. We assume that it is during these periods when the vast majority of forest area burns that fuel differences have the potential to cause large effects on landscape-level patterns of burning. All stands >30 years-since-last fire were predicted to carry a surface fire during large-area burned years when actual fires made runs. Aspen stands had lower fuel-bed surface areas than conifer stands but fuel moisture levels were low enough that the differences did not matter. In contrast, fires were not predicted to spread in many recently burned stands under any weather conditions. Our results imply that increasing aspen dominance on the mixedwood boreal landscape may not be effective as a strategy to encourage fire extinction and, thereby, reduce area burned and tree mortality in fires (Fechner and Barrows 1976 . When surface fires spread past trees, tree components (i.e., roots, boles, and crowns) are heated, causing cell mortality, tissue necrosis, and tree death. We used a heat-transfer model and a biophysical cell-survivorship model to derive quantitative predictions of cell survivorship in the boles of trees in fires of a range of residence times and for trees with a range of bark thicknesses. The cell-survivorship model was parameterized with live-bark cell survivorship data from aspen and spruce. We assume that survivorship of live bark cells closely approximates survivorship of adjacent vascular cambium cells. We found that live-bark cell survivorship in surface fires can be described by a relatively simple relationship between bark thickness and fire residence time. We expect two thresholds relative to bark thickness: complete vascular-cambium cell mortality would be expected in all fires for trees with bark thickness <6 mm while trees with bark thickness >20 mm are expected to be immune to vascular-cambium necrosis during surface fires. Our mechanistic approach could provide managers with a tool for predicting vascularcambium necrosis and tree death from bark thickness and fire residence time, variables that are readily estimated in the field or from models. ii ACKNOWLEDGEMENT