The national Fire and Fire Surrogate study: Effects of fuel treatments in the western and eastern US after 20 years
Data files
Dec 20, 2024 version files 166.85 KB
-
EAP23-0533.csv
160.46 KB
-
README.md
6.39 KB
Abstract
The national Fire and Fire Surrogate (FFS) study was initiated more than two decades ago with the goal of evaluating the ecological impacts of mechanical treatments and prescribed fire in different ecosystems across the United States. Since then, 4 of the original 12 sites remain active in managing and monitoring the original FFS study which provides a unique opportunity to look at the long-term effects of these treatments in different regions. These sites include California (Blodgett Forest Research Station), Montana (Lubrecht Experimental Forest), North Carolina (Green River Game Land) and Ohio (Ohio Hills). Although regions differed in ecosystem type (e.g., conifer- vs. hardwood-dominated), the overall goals of the FFS study were to promote desirable, fire-adapted species, reduce fire hazard, and improve understory diversity. Our study uses multivariate techniques to compare how these desired outcomes were maintained over the last 20 years and discusses whether we would modify the original treatments given what we know now. Our findings indicate that mechanical treatments and prescribed fire can promote desired tree species, mitigate potential fire behavior by reducing fuels and retaining larger-sized trees, decrease tree mortality, and stimulate regeneration – effects that are still apparent even after 20 years. However, we also found that maintaining desired outcomes was regionally specific with western sites (California and Montana) showing more desirable characteristics under mechanical treatments, while the eastern sites (North Carolina and Ohio) showed more desirable characteristics after prescribed burning. The beneficial effects of treatment were also more apparent in the long term when sites followed up with repeated treatments, which can be adapted to meet new objectives and conditions. These findings highlight the FFS study as an invaluable resource for research and provide evidence for meeting long-term restoration goals if treatments can be adapted to ecosystem type, be maintained by repeated treatments, and can accommodate new goals by adapting treatments to changing conditions.
README: The national Fire and Fire Surrogate study: Effects of fuel treatments in the western and eastern US after 20 years
https://doi.org/10.5061/dryad.5tb2rbpdp
Description of the data and file structure
Summary
This README file is associated with the Ecological Applications article entitled "The national Fire and Fire Surrogate study: Effects of fuel treatments in the western and eastern US after 20 years" (EAP23-0533) and was prepared by Alexis A. Bernal on 12/16/2024. This README file describes the data used to 1) test whether various fire and fire surrogate treatments differed over time using multi-response permutation procedure (MRPP) and 2) visually compare whether desired outcomes in forest structure/composition differentiated treatments 20 years after the initial treatments were implemented using canonical discriminant analysis (CDA). For site-specific applications of treatments including the timing of treatments, please refer to Table 1 of the article. For site-specific details of data collection methods, please refer to Table 2 of the article.
Any questions about this data may be forwarded to the corresponding author at alexis_bernal@berkeley.edu.
Files and variables
File: EAP23-0533.csv
Description: Plot-level summaries of data that were used in the MRPP and CDA are included in EAP23-0533.csv. Each column represents a variable that was used in MRPP and CDA analysis and each row represents an observation (i.e., plot-level summary for an individual plot) taken either pre-treatment (i.e., timestep = pre) or post-treatment (timestep = post).
Variables
- site: Categorical variable indicating the state where data were collected. Can take on 4 unique values including California, Montana, Ohio, and North Carolina.
- unit: Categorical variable identifying the unit (i.e. replicate) where data were collected.
- plot: Categorical variable identifying the plot within a given unit where data were collected.
- treatment: Categorical variable indicating which fire and fire surrogate treatment was applied. Can take on 4 unique values including Control (i.e., untreated), Mech (i.e., mechanical treatment only), Fire (i.e., prescribed fire only), and Mech+Fire (i.e., mechanical treatment followed by prescribed fire).
- timestep: Categorical variable indicating when data were collected. Can take on 2 unique values including pre (i.e., before treatment occurred) and post (i.e., 17-22 years after treatment occurred).
- l_tph: Numerical live tree density ha^-1 of mature trees. This was either >15 cm diameter at breast height (DBH; 1.37 m above ground) for California or >10 cm DBH for Montana, Ohio, and North Carolina.
- l_ba: Numerical live basal area (m^2 ha^-1) of mature trees. This was either >15 cm diameter at breast height (DBH; 1.37 m above ground) for California or >10 cm DBH for Montana, Ohio, and North Carolina.
- l_qmd: Numerical live quadratic mean diameter (cm) of mature trees. This was either >15 cm diameter at breast height (DBH; 1.37 m above ground) for California or >10 cm DBH for Montana, Ohio, and North Carolina.
- d_tph: Numerical dead tree density ha^-1 of mature trees. This was either >15 cm diameter at breast height (DBH; 1.37 m above ground) for California or >10 cm DBH for Montana, Ohio, and North Carolina.
- d_ba: Numerical dead basal area (m^2 ha^-1) of mature trees. This was either >15 cm diameter at breast height (DBH; 1.37 m above ground) for California or >10 cm DBH for Montana, Ohio, and North Carolina.
- p_ba: Numerical percent of basal area of preferred species of mature trees. Please refer to site-specific details in article for details on species included in this calculation.
- p_seed: Numerical seedling density ha^-1 of preferred species. Seedlings were tallied if height was less than 1.37 m above ground.
- np_seed: Numerical seedling density ha^-1 of non-preferred species. Seedlings were tallied if height was less than 1.37 m above ground.
- p_sap: Numerical sapling density ha^-1 of preferred species. Saplings were tallied if height was greater than 1.37 m above ground and <15 cm DBH (California) or <10 cm DBH (Montana, Ohio, and North Carolina).
- np_sap: Numerical sapling density ha^-1 of non-preferred species. Saplings were tallied if height was greater than 1.37 m above ground and <15 cm DBH (California) or <10 cm DBH (Montana, Ohio, and North Carolina).
- midstory_shrubs: Numerical percent cover of midstory shrubs. Only measured at North Carolina site (i.e., NA values for other sites).
- graminoids: Numerical percent cover of grasses.
- forb: Numerical percent cover of forbs.
- shrub: Numerical percent cover of shrubs.
- tree: Numerical percent cover of trees. Only measured at California and Ohio sites (i.e., NA values for other sites).
- vine: Numerical percent cover of vines. Only measured at North Carolina site (i.e., NA values for other sites).
- non_native: Numerical percent cover of non-native species. Only measured at Montana site (i.e., NA values for other sites).
- richness: Numerical count of all genera in understory vegetation.
- diversity: Numerical index from Shannon's H formula using genera in understory vegetation.
- evenness: Numerical index from Pielou's formula using derived diversity index.
- ground: Numerical load (Mg ha^-1) of ground fuels (litter and duff). Only measured at California, Montana, and North Carolina sites (i.e., NA values for Ohio).
- fine: Numerical load (Mg ha^-1) of fine fuels (1-, 10-, and 100-hr fuels). Only measured at California, Montana, and North Carolina sites (i.e., NA values for Ohio).
- cwd: Numerical load (Mg ha^-1) of coarse woody fuels (1000-hr+ fuels). Only measured at California, Montana, and North Carolina sites (i.e., NA values for Ohio).
- ptorch: Probability of torching (percent) estimated in the Fire and Fuels Extension of the Forest Vegetation Simulator (FFE-FVS). Only estimated for California and Montana sites (i.e., NA values for other sites).
- pmort: Potential mortality (percent) estimated in the Fire and Fuels Extension of the Forest Vegetation Simulator (FFE-FVS). Only estimated for California and Montana sites (i.e., NA values for other sites).
NA values not described above are measurements that were not recorded (i.e., missing values).
Methods
Overstory measurements
All mature trees within plots at each site were tagged and recorded for status (i.e., alive, dead), species, and diameter at breast height (DBH; at height of 1.37 m). Although tree regeneration (i.e., seedlings and saplings) is typically considered an understory component, we included it in the overstory analysis to demonstrate the potential for future overstory recruitment of preferred or non-preferred tree species as well as fire hazard over time. In California, a minimum DBH of 15 cm was used to inventory mature trees within the entire plot (0.04 ha), while live seedlings (<DBH in height) and saplings (≥DBH in height and <15 cm DBH) were tallied by species in either a subplot (0.004 ha) or the entire plot depending on size. Montana, Ohio, and North Carolina had a minimum DBH of 10 cm to inventory mature trees. North Carolina measured mature trees in half of the 0.10- ha subplots while Montana and Ohio measured mature trees in all 0.10 ha plots. At these sites, live saplings (≥DBH in height and <10 cm DBH) and seedlings (<DBH in height) were tallied by species in subplots of 0.01 and 0.0001 ha, respectively. Ocular estimates of percent cover of grasses, forbs, and shrubs by species were recorded in subplots at all sites but varied in terms of size in California (0.004 ha), Montana (0.0001 ha), Ohio (0.0012 ha), and North Carolina (0.01 ha for shrubs > 1.37 m and 0.0001 ha for herbaceous vegetation and shrubs < 1.37 m). More details can be found in Table 2 of EAP23-0533.
Fuels measurements and potential fire behavior
Each site used a modification of planar intersect fuels protocols (Brown 1974), establishing 2 – 3 transects (11.3 – 20 m in length) at each plot (Table 2 of EAP23-0533). On each transect, duff and litter depths (cm) were measured, as well as tallies of 1-hr (woody material < 0.64 cm diameter), 10-hr (0.64 cm ≤ diameter < 2.54 cm), 100-hr (2.54 cm ≤ diameter < 7.62 cm), and 1000-hr+ (diameter ≥ 7.62 cm) timelag classes. For California and Montana, fire behavior predictions were generated using the Fire and Fuels Extension of Forest Vegetation Simulator (FFE-FVS) (Rebain 2015) using plot-scale, mature tree measurements (i.e., canopy fuels) and surface fuels data. In FFE-FVS, surface fuels were represented by either Scott and Burgan’s (Scott and Burgan 2005) modified 40 fuel models (California) or Anderson’s (Anderson 1982) 13 original fuel models (Montana). The default algorithm to assign fuel models and weights was used in Montana, while California used a more extensive model selection and weighting process (see Stephens et al. 2023 in this Special Feature). Potential fire behavior was assessed using two outputs generated from FFE-FVS including probability of torching (P-torch) and potential mortality (P-mort) which is expressed as a percentage of basal area killed (Rebain 2015). For Ohio and North Carolina, surface fuel models, which are a required input for FFE-FVS and many other fire behavior models, that represent hardwood ecosystems in this region of the US are less reliable than those offered in conifer-dominated forests in the west (Phillips et al. 2006). Therefore, fire behavior predictions were not generated for these sites.
Treatment outcome metrics
Measurements were aggregated into three groups: overstory, understory, and fuels/fire behavior at the plot-level for each site. To assess changes in overstory, understory composition, fuels, and fire behavior across treatments and time, we used plot averages for all measurements taken Pre-treatment and Post-20. For overstory structure, common metrics to describe mature overstory characteristics were calculated, including live and dead tree density (trees ha-1), basal area (m2 ha-1), percentage of basal area occupied by site-specific preferred species (e.g., pines, western larch, oaks, and hickory), and live quadratic mean diameter (QMD; cm). Potential recruitment into the overstory was calculated as live seedling and sapling density (ha-1) of preferred (same species above) and non-preferred species (species other than preferred) separately. Due to the limitations of hand-thinning in North Carolina, a substantial amount of midstory shrubs resprouted following mechanical treatments. Therefore, we also included percent cover of midstory shrubs for this site as an overstory metric.
For understory composition, percent cover by species was aggregated to the genus level due to inconsistencies in data collection, as well as aggregated by lifeform (graminoid, forb, shrub, tree, and vine) and non-native cover (Montana only). From genus percent cover, we also calculated three metrics to describe understory communities including Shannon-Wiener diversity index (Shannon and Weaver 1949), richness (total count of genera present), and Pielou’s index of evenness (Pielou 1966). We then aggregated fuel loads (Mg ha-1) for fine (1-, 10-, and 100-hr) fuels, coarse woody debris (CWD; 1000-hr+) and ground (i.e., litter and duff) fuels for California, Montana, and North Carolina, as well as plot-level fire behavior outputs P-torch and P-mort for California and Montana. Due to inconsistent data collection Pre-treatment, fuels data from Ohio are excluded from this dataset.