Supporting data for increasing fire activity reinforces shrub conversion in Southwestern US forests

Fire-exclusion in historically frequent-fire forests of the southwestern United States has altered forest structure and increased the probability of high-severity fire. Warmer and drier conditions, coupled with dispersal distance limitations are limiting tree seedling establishment and survival following high-severity fire. Post-fire conversion to non-forest vegetation can be reinforced by subsequent fire events. We sought to determine the influence of fire probability on post-fire vegetation development in a severely burned landscape in New Mexico, USA. We used LANDIS-II to simulate three fire probability scenarios—contemporary mean fire return interval (CMFRI), and 1.5 times and 2 times CMFRI—with contemporary climate. As fire probability increased, the mean size of the largest fires and the mean landscape fire severity increased. These changes in fire characteristics resulted in a net decrease in total above ground biomass and photosynthetic capacity on the landscape. Additionally, the distribution of individual species biomass shifted, with early successional species, especially those that resprout after fire, increasing as a fraction of total biomass with increasing fire occurrence. Continued increases in fire frequency are likely to favor resprouting species and result in a loss of forest biomass and ecosystem productivity in this southwestern forest landscape.

Methods
These data are outputs of a set of simulations using the LANDIS-II (v6.2) model with the PnET Succession extension (v.2.1.1) and the Dynamica Fuels and Fire System (v2.1)

extension. We used current climate data from Daymet and atmospheric CO2 data from the Mauna Loa observatory. We ran the model with three fire ignition probability scenarios; these are labelled 'Low or LO', 'Medium or MED', and 'High or HI'. Our study objective was to examine differences in post-fire vegetation recovery under

variable fire regimes. Our simulation area was the footprint of the 2011 Las Conchas fire in New Mexico, USA.

We ran 30 replicate simluations for each scenario; the length of each was 50 years. We summarized most data by calculating the mean and standard deviation of the 30 replicates. Mean and standard deviation for fire severity data were calculated for all years/replicates together. We present the data sets (raw and final) and code for creating figures in Keyser et al.

The parameter files for the PnET succession and Dynamic Fire and Fuels extensions are also included.