COS 99-7
Post-fire tree recruitment in the U.S. Northern Rockies: The influence of seed source proximity and environmental conditions
In the U.S. northern Rockies, low elevation mixed-conifer forests historically burned in low- and mixed-severity fire regimes. Legacies of past management practices, including historical logging and fire exclusion, combined with predicted climate warming are likely to contribute to changes in the frequency, intensity, and spatial patterning of fires in the future, potentially influencing post-fire patterns of seedling establishment. To anticipate how post-fire recovery may vary in low elevation mixed conifer forests, we quantified relationships among burn severity, environmental gradients, and post-fire tree recruitment.
We sampled post-fire seedling density and species composition at 183 sites that burned in 2000 and 2007 in central Idaho and western Montana. Sites were stratified by burn severity (unburned, low, moderate, and high), elevation (675 - 2200 m), and aspect. We used logistic and negative binomial regression to predict the probability and density of conifer regeneration 5 to 13 years post-fire as a function of fire legacies (burn severity, distance to nearest live trees, and time since fire), landscape features (elevation, aspect, and slope), and ecological conditions (vegetation and canopy cover, tree density). Forward-backward stepwise selection with AIC criteria was applied to choose between candidate models. ROC curves were used to evaluate model fit.
Results/Conclusions
Seedling densities varied widely across sites (0 – 127,500 seedlings/ha). Seedling presence (AUC = 0.829) was predicted by the average distance to the nearest live seed source (z = -4.123, p << 0.001) and mature tree density (z = 2.593, p = 0.010). Beyond 300 m from the nearest live seed source, the probability of seedling establishment was less than 20%. Distance to the nearest live seed source was the most significant variable predicting seedling presence for Douglas-fir, ponderosa pine, and grand fir; whereas, mature tree density (z = 4.148, p << 0.001), elevation (z = 4.215, p << 0.01) and time since fire (z = -1.969, p = 0.049) were the best predictors of lodgepole pine presence. Variables such as heat load and vegetation cover were important determinants of seedling densities.
Our results highlight the overarching importance of nearby live seed sources for post-fire regeneration, across broad gradients in climate. The size of high burn severity patches in mixed-severity wildfire regimes will likely strongly determine patterns of reestablishment. In low and moderate burn severity patches, where seed sources are close, seedling regeneration varies widely, suggesting that other landscape and climatic variables determine seedling recruitment.