Permanent Archive of the data published in Jiang, Y, EB Rastetter, AV Rocha, AR Pearce, BL Kwiatkowski, GR Shaver. 2015. Modeling Carbon-Nutrient interactions during the early recovery of tundra after fire.. Ecological Applications 25:1640-1652.

Model Calibration
We used version IV of the Multiple Element Limitation (MEL) model (Rastetter et al. 2013, Pearce et al. 2015, Jiang et al. 2015a) to simulate C, N, and P cycles in the three arctic tussock tundra sites following fire. The study sites represent a gradient in burn severity (severe burn, moderate burn, and unburned) in the southern portion of the 2007 Anaktuvuk River fire scar on the North Slope of Alaska (Jones et al. 2009). Sites were selected by Rocha and Shaver (2011a, b), based on the two-band enhanced vegetation index (EVI2) (Rocha and Shaver 2009) and the normalized burn ratio (NBR) from the Moderate Resolution Imaging Spectroradiometer (MODIS) (Boelman et al. 2011). These three sites had similar weather and pre-fire surface greenness (i.e. EVI2) but substantially different plant mortality. The MBL MEL IV model is a mass-balance model that couples ecosystem C, N, P, and water fluxes and operates at a plot scale with a daily time step. Detailed descriptions of the MEL model can be found in Rastetter et al. (2013). The model source code, information on compiling and running MEL are available on the MBL MEL web site ()
Pearce et al. (2015) calibrated the MEL model to match annual C, N, P and water cycles on the North Slope of Alaska near Toolik Lake, Alaska, USA (68.63oN, 149.72oW), and then to examine the recovery of tussock tundra from thermal erosion events resulting in downslope displacement of the seasonally thawed surface soil and associated vegetation. Based on the parameterization in Pearce et al. (2015), Jiang et al. (2015a) modified the canopy phenology parameters to fit the length of growing season at the Anaktuvuk River fire sites. These simulations by Jiang et al. (2015a) focused on the first five years following fire. In this study we use the same routines and parameterization used by Jiang et al. (2015a) to conduct 200-year simulations in the three study sites.
 
Simulation protocol
Model input (driver) data was derived from field measurements at several nearby sites. Specifically, the growing season radiation and air temperature were obtained from three eddy flux towers (Rocha and Shaver, 2011a) and the non-growing season radiation data were obtained from the ARC LTER experimental site at Toolik Lake (Shaver et al. 1989), about 50 km southeast of the burned sites. The non-growing season air temperature and precipitation were obtained from the Upper Kuparuk Meteorological Station (Kane and Hinzman 2013), which was approximately 48 km southeast of the Anaktuvuk River unburned site. The atmospheric CO2 concentration was obtained from ice core and the National Oceanic and Atmospheric Administration (NOAA) measurements (Keeling and Whorf 2005).
To assess long-term changes in C and nutrient cycles in burned and unburned tundra, we simulated changes in the biogeochemistry of the three sites in response to increases in air temperature, CO2, N deposition and P weathering for the next 200 years. Because the future climate is unknown we ran simulations for a wide range of climate and nutrient additions. The first year of data for all simulations was calculated as the average of five years (2008-2012) daily field measurements. To develop a future climate data set over the next 199 years, we linearly increased air temperature, CO2 concentration, N deposition, and P weathering year by year. We increased air temperature so that the final year of the simulation (year 200) temperature increased by 0, 4, 6, 9, and 12oC relative to the first year of simulation. Similarly, we increased CO2 each year so that in the last year CO2 was 1, 2, 3, 4 and 5 times the value in 2008 (385.6 ppm).
Only limited information is available on deposition of atmospheric N for the North Slope of Alaska. To cover a wide range of uncertainties in N deposition estimated from empirical data and global-scale overview and test the influence of different N deposition rates, we linearly increased annual N deposition rate so that in the 200th year N input was 1, 2, 5, 10, and 20 times their values in 2008 (0.035 g N/m2/yr); with P weathering rate (1.28 ×10-4 g P/m2/yr in 2008) increased by the same proportions. However, even at these rates, external nutrient supply was very small relative to plant requirements. In 2008, the N deposition rate was less than 1% of the plant N uptake and the P weathering rate was less than 0.03% of the P uptake rate in the unburned tundra; therefore the plants depended almost exclusively on N and P recycled through the soil. We ran changes in nutrient supply in a factorial combination with 5 sets of N treatments and the 5 sets of P treatments separately, then a total of 5 simulations with both N and P increased by 1, 2, 5, 10, or 20 times their 2008 values. We also ran simulations with the concentrations of NH4+, PO43-, and DON in the soil held constant at saturating concentrations (1000 times their respective concentrations in the unburned tundra) to show what the post-fire recovery potential was if nutrients were not limiting. In this study, we did not take into account the limitation from secondary nutrients such as calcium (Ca) and magnesium (Mg) because we assumed that they did not play a large role due to the order of magnitude lower C:Ca or C:Mg in arctic plants.
Results
The MEL simulations for this study include daily output for 200 years for each of the three study sites for all treatments, over 600 simulations per site. Because of the number and length of simulations we store the results for only two days each year, day 216 (peak season) and day 365 (end of year). The output file are named based on the site and the treatment, according to the following convention: Site-x-y-z-w.out, where the x (CO2), y (air temperature), z (N inputs), and w(P inputs) represent the increase over the control treatment (for example, z=19 means 19 times of the control N inputs).This dataset is the long term archive of the results published in the paper.

Dataset Files
The full dataset has been broken into two parts because of the number and size of the files. Part 1 contains MBL MEL executable, a model description file in word, and the input files to run the simulations. Part 2 contains the output files for all simulations and the files for running MEL IV in batch mode. Both Part 1 and Part 2 contain several different types of files. In Part 1 the comma separated ascii file included with the dataset is one of the many driver files used for the simulations. The variable descriptions below describe the driving variables in that file and all the driver files. In Part 2 the comma separated ascii file included with the dataset is one of the many output files from the simulations. The variable descriptions below describe the ourput variables in that file and all the output files. To access all the files in the dataset download both Part 1 and Part 2 zip files. The contents of the zip files are listed below. Note that the full download is large, over 700 MB for each part.

Part 1 MEL_AR_ClimFire_Part1.zip Contents
Anaktuvuk_drivers.rar - compressed file containing all the driver files for the simulations
AnaktuvukFireTussock_unburn.par - MEL IV parameter file calibrated to Anaktuvuk River unburned tussock tundra
AnaktuvukFireTussock_moderate.par - MEL IV parameter file calibrated to Anaktuvuk River fire moderately burned tussock tundra
AnaktuvukFireTussock_severe.par - MEL IV parameter file calibrated to Anaktuvuk River severely burned tussock tundra
MELIV1NEWallometry3-revised.rtf - Word description file used with MBL's CreateEquations program to create the MEL IV code.
MELIV_Anaktuvuk.exe - Windows 64 bit executable for running the MELIV model

Part 2 MEL_AR_ClimFire_Part2.zip Contents
Anaktuvuk_moderate_fert.bch - Batch file used with MELIV_Anaktuvuk_Batch.exe to run the moderate site simulations
Anaktuvuk_output.rar - compressed file containing all the output files for the simulations.
Anaktuvuk_severe_fert.bch - Batch file used with MELIV_Anaktuvuk_Batch.exe to run the severe site simulations
Anaktuvuk_unburn_fert.bch - Batch file used with MELIV_Anaktuvuk_Batch.exe to run the unburn site simulations
MELIV_Anaktuvuk_Batch.exe - Windows 64 bit executable for running the MELIV model in batch mode

This material is based upon work supported by the National Science Foundation under Grants # DEB-1026843, EF-1065587, and OPP-1107707. Any opinions, findings, conclusions, or recommendations expressed in the material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.