The western U.S. is projected to experience more frequent and severe wildfires in the future due to drier and hotter climate conditions, exacerbating destructive wildfire impacts on forest ecosystems such as tree mortality and unsuccessful post-fire regeneration. While empirical studies have revealed strong relationships between topographical information and plant regeneration, ecological processes in ecosystem models have either not fully addressed topography-mediated effects on the probability of plant regeneration, or the probability is only controlled by climate-related factors, e.g., water and light stresses. In this study, we incorporated seedling survival data based on a planting experiment in the footprint of the 2011 Las Conchas Fire into the PnET extension of the LANDIS-II model by adding topographic and an additional climatic variable to the probability of regeneration. The modified algorithm included topographic parameters such as heat load index (HLI) and ground slope and spring precipitation. We ran simulations on the Las Conchas Fire landscape for 2012–2099 using observed and projected climate data (i.e., RCP 4.5 and 8.5). Our modification significantly reduced the number of regeneration events of three common southwestern conifer tree species (piñon, ponderosa pine and Douglas-fir), leading to decreases in aboveground biomass, regardless of climate scenario. The modified algorithm decreased regeneration at higher elevations and increased regeneration at lower elevations relative to the original algorithm. Regenerations of three species also decreased on eastern aspects. Our findings suggest that ecosystem models may overestimate post-fire regeneration events in the southwest U.S. To better represent regeneration processes following wildfire, ecosystem models need refinement to better account for the range of factors that influence tree seedling establishment. This will improve model utility for projecting the combined effects of climate and wildfire on tree species distributions.

These outputs are from a set of simulations using the LANDIS-II (v7.0) model with a modified PnET Succession extension (based on v.4.1) and the Dynamic Fuels and Fire System (v3.0) extension. Our simulated landscape is the footprint of the 2011 Las Conchas fire in the Jemez Mountains, New Mexico. Observed climate data (Daymet) and downscaled climate projection data (access1.0.1, canesm2.1, cesm1-bgc.1 and hadgem2-es.1) forced with two emission scenarios (RCP4.5 and 8.5) from the CMIP5 models were used. We used contemporary fire conditions for simulating fire events (Keyser et al., 2020; Krofcheck et al., 2019). 

We ran 10 replicate simulations from 2012 to 2099. We summarized the data to calculate cumulative post-fire regenerations and biomass distribution as well as the relationship of regeneration with topographical variables such as elevation and aspect. 

The archived file includes an R script for creating figures with a Rdata file. 

The modified PnET extension scripts (C#), executable setup files for LANDIS-II (v7.0) with the modified PnET extension and Dynamic Fuels and Fire System (v3.0) extension setup files and one set of examples for HPC application (SLURM) are archived at Zenodo (https://doi.org/10.5281/zenodo.7535673). Docker images for the HPC application are available in the Docker Hub (Original algorithm: https://hub.docker.com/r/ecochang/landis_pnet_dffs_org_v2; Modified algorithm: https://hub.docker.com/r/ecochang/landis_pnet_dffs_mod_v2).