Repeated fuel treatments fall short of fire-adapted regeneration objectives in a Sierra Nevada mixed conifer forest, USA
Data files
Nov 19, 2024 version files 19.71 MB
-
abundance_data.csv
1.53 MB
-
abundancemod.R
4.48 KB
-
dens.mod.txt
2.56 KB
-
FFS_all_scripts.R
10.62 KB
-
FFSseedling_data.rdata
563.13 KB
-
FFSseedling_Figures_and_Tables.Rmd
42.36 KB
-
occ.mod.txt
7.70 KB
-
occmod.R
7.35 KB
-
occupancy_data.csv
4.31 MB
-
postprocessing_functions.R
36.34 KB
-
README.md
5.19 KB
-
results_and_functions.rdata
13.19 MB
Abstract
Fire exclusion over the last two centuries has driven a significant fire deficit in the forests of western North America, leading to widespread changes in the composition and structure of these historically fire-adapted ecosystems. Fuel treatments have been increasingly applied over the last few decades to mitigate fire hazard, yet it is unclear if these fuels-focused treatments restore the fire-adapted conditions and species that will allow forests to persist into the future. A vital pre-requisite of restoring fire-adaptedness is ongoing establishment of fire-tolerant tree species, and both the type and reoccurrence of fuel treatments are likely to strongly influence stand trajectories. Here, we leveraged a long-term study of repeated fuel treatments in a Sierra Nevada mixed-conifer forest to examine the regeneration response of six native tree species to repeated application of common fuel treatments: prescribed fire, mechanical, mechanical plus fire, and untreated controls. Our objectives were to 1) quantify differences in forest structure and composition following repeated application of alternative fuel treatments that may influence the establishment environment and then 2) identify the stand structure and climate conditions influencing seedling dynamics. We found that both treatment type and intensity are highly influential in shifting forests toward more fire-adapted conditions and determining species-specific regeneration dynamics. Specifically, the conifer species tracked here increased in either colonization or persistence potential following repeated applications of fire, indicating fire may be most effective for restoring regeneration conditions broadly across species. Fire alone, however, was not enough to promote fire-adapted composition, with concurrent mechanical treatments creating more favorable conditions for promoting colonization and increasing abundances of fire-tolerant ponderosa pine. Yet, even with repeated fuel treatment application, establishment of fire-intolerant species far exceeded that of fire-tolerant species over this 20-year study period. Moreover, increasing growing season water stress negatively impacted seedling dynamics across all species regardless of treatment type and intensity, an important consideration for ongoing management under heightened climatic stress. While repeated treatments are waypoints in restoring fire-adapted conditions, more intense treatments via gap-creation or hotter prescribed fires targeting removal of fire-intolerant species will be necessary to sustain recruitment of fire-tolerant species.
Code and data for Nagelson et. al. 2024
All data (.rdata files) and code are available to replicate the analyses and generate figures and tables. All intermediary data products are also provided (e.g. JAGS outputs), which means that one can jump straight to figure and table generation by using “FFSseedling_Figures_and_Tables.Rmd”.
Scripts
FFSseedling_Figures_and_Tables.Rmd - R markdown document used to generate all tables and figures in manuscript. This is the only script necessary if the user is only interested in reproducing figures/tables or the statistical results defined under research objective 1.
Bayesian models and post-processing scripts
postprocessing_functions.R - post-processing code for abundance and occupancy models.
abundancemod.R - pre-processing and run abundance model.
occmod.R - pre-processing and run occupancy model.
FFS_all_scripts.R - model scripts written in JAGS syntax. These are called by abundancemod.R and occmod.R.
dens.mod.txt - JAGS script for abundance model generated by FFS_all_scripts.R and called in abundancemod.R
occ.mod.txt - JAGS script for occupancy model generated by FFS_all_scripts.R and called in occmod.R
Pre-packaged data (outputs from scripts listed above)
results_and_functions.rdata - packaged dataframes and functions created by postprocessing_functions.R. This includes outputs from JAGS models.
dens.fp.df()
andocc.fp.df()
are functions to extract draws from the posterior distribution for the density models and occupancy models. Used to createdens.mod.results
andocc.mod.results
.dens.mod.results
andocc.mod.results
are dataframes containing summaries of the posterior distribution for the density and occupancy models. They have the following columns: term (predictor variable in model), coeff.type (scaled or unscaled), data.type (continuous, categorical, or time), species (two-letter species code), estimate (mean of posterior), median (median of posterior), conf.low (lower bound of credible interval), conf.high (upper bound of credible interval), overlap0 (boolean for whether the credible interval overlaps zero), model (model type, occupancy or density).curves.fun()
is a function to plot random draws from the posterior distribution across the range of a chosen predictor. This is the main function used to create figures 3 and 4.occ.curves.y
andocc.curves.median
are the output fromcurves.fun()
.occ.curves.y
is the complete set of random draws from the posterior distribution.occ.curves.median
is just the median of the posterior.
FFS_seedling_data.rdata - pre-processed data for use in FFSseedling_Figures_and_Tables.Rmd. Datasets:
climate
= climate data for the length of the study, with columns: sample.year (year of sampled seedling data for which the climate variable applies), value (climate value, either precipitation climatic water deficit in millimeters), season (cool or warm), variable (cwd, climatic water deficit, or ppt, precipitation)light
= total transmitted radiation data with columns: year, comp (stand identifier), plot (plot identifier within compartment), TTR (total transmitted radiation, %).ba.all
= plot-level basal area with one record for each plot x timestep x species. Columns: PlotID (plot identifier), treatment (treatment type), source (data carpenter name), timestep, species (two-letter code), species.ba (species-level basal area in square meters/hectare), total.ba (total basal area in square meters/hectare).adapt.table.raw
= dataframe used to generate table 1occ.model.data
= hierarchical list of dataframes and vectors. Useful in some cases for figure generation but can generally be ignored if using the .Rmd file.
Raw data files
abundance_data.csv
- PlotID = unique plot identifier
- comp = stand identifier
- treatment = treatment type
- trt = numeric treatment type
- timestep = timestep identifier relative to initial treatment implementation
- species = two-letter code for species. BO=black oak, DF = douglas-fir, IC=incense-cedar, PP=ponderosa pine, SP=sugar pine, WF = white fir
- plotsize = size of sampled plot area in hectares
- seedlings.count = count of seedlings
- seedlings.ha = seedlings per hectare
- timestep.num = numeric timestep identifier
- year = year of sampling
- timesince = years elapsed since the most recent treatment
- trtnum = number of treatment applications
- species.ba = conspecific basal area in square meters per hectare
- total.ba = total basal area in square meters per hectare
- cool.cwd = cool season climatic water deficit in millimeters
- warm.cwd = warm season climatic water deficit in millimeters
- cool.ppt = cool season precipitation in millimeters
occupancy_data.csv
- ID = unique plot identifier, with added specificity for clustered subplots (e.g. 40-0002-1, 40-0002-2, etc.)
- occupancy = presence (1) or absence (0) of seedlings at that plot
- type = plot type, either CSP (clustered subplot) or FI (full inventory)
- plotsize = plot size in square meters
- all other column names are defined the same as in abundance_data.csv
Seedling data were collected in .004 hectare and 1m2 plots. Overstory tree data (used for calculating basal area) were collected in 0.04 hectare plots. Weather data was collected from an on-site weather station. Data were processed using the R statistical software.