Abstract: Fire-prone forests in the western US are experiencing increasing area burned by high-severity wildfire as a result of past fire-exclusion and ongoing climate change. Fire-exclusion, resulting from past land-use and fire suppression, has homogenized dry western forests and increased fuel loads. Forest restoration aims to decrease tree density and surface fuels using some combination of thinning and burning. In most forest types, tree density targets and composition are estimated from research that reconstructed historic structure and composition at a handful of sites. While these empirically based estimates are useful, they typically only provide representation for small areas and over short time periods (e.g. mid-1800s). To improve our understanding of forest dynamics over a larger landscape and longer time period, we parameterized the forest landscape model LANDIS-II with fire and climate data from the late-Holocene (~6000 yrs BP) for the upper Rio Grande watershed in northern New Mexico and southern Colorado. Reconstructed climate data over this period indicate that mean annual temperature was 1.9 ˚C lower and mean annual precipitation was 49 mm higher than late-20th century climate. Charcoal data indicate that fires were more frequent and larger in mid-elevation than in high-elevation forests.
To account for the lack information about species distributions and biomass density data over the Holocene, we constructed a landscape by distributing cohorts of every species evenly across the landscape. We ran simulations with the late-Holocene climate and fire data for 1000 years to allow the species to sort across this topographically diverse landscape and then used this distribution of species-specific age-cohorts as our initial communities layer for all subsequent simulations. We ran simulations over the late-Holocene through the 20th century to compare tree species and biomass distributions in the absence of fire-exclusion. We found that mid-elevation forests, which historically burned at higher frequency had biomass distributions similar to site-specific empirical reconstructions using tree-rings. However, we did find a wide range of biomass conditions across the landscape, reflective of the variability in fire and climate conditions. At high-elevations, we found a more heterogenous distribution of forest types and biomass densities than we interpolated using modern forest inventory data. These results suggest that the fire regime of higher elevation forests caused more heterogeneous conditions, which may have had a self-reinforcing effect that helped constrain fire size. Our results suggest the range of structural variability in both mid- and high-elevation forests may be greater than previously thought.
