As a result of extreme weather conditions associated with anthropogenic climate change, fire regimes are expected to continue to change in the boreal forest over the 21st century and beyond. Consequently, changes in ecological attributes like stand composition, tree density and forest carbon stock can be expected. In the present study, we used an adjusted version of the CanFIRE model to project long-term (1971–2100) changes in burn rates, fire severity and fire-induced shifts in vegetation composition in response to anticipated scenarios of climate change, in the black spruce-feather moss subdomain of Western Quebec. The model provides decadal-scale estimates of the immediate physical effects of fire on forest communities by computing expected fire behavior and the resulting ecological effects. Changes in species composition of the forest is also computed based on mechanisms of succession in natural forest communities and fire-mediated vegetation transitions. Projections suggest an increase in potential burn rates across the study area under future weather conditions and also an overall reduction in percent tree mortality and total fuel consumption. This reduction is caused by negative feedback from vegetation composition that shifts to less-fire prone states. Although common forest communities will remain the same in the studied subdomain until 2100 (recurrence dynamics), significant losses of productive area (LPA) are projected, particularly in forest management units rich in forest communities dominated by black spruce or jack pine, as a result of regeneration failure due to very short intervals between successive fires. While remaining similar under moderate (RCP4.5) and high-end (RCP8.5) warming scenarios in all forest management units, LPA will vary from 25 to 36% of the percent cover by 2100 compared to 1970. These results provide insights to policy makers and land managers, and they attract attention to the pressing need to adjust management practices in the context of climate change.