Data and code from: Species, structural, and functional diversity in relation to live aboveground biomass in mature and late-successional northeastern temperate forests
Data files
Sep 28, 2026 version files 1.29 GB
-
CFI_AGB_covars_092026.csv
71.55 KB
-
cfi_ba.csv
75.12 KB
-
CV.zip
1.22 GB
-
FinalModel.zip
78.03 MB
-
make_struct_df.R
4.27 KB
-
README.md
13.25 KB
-
sf_plots.csv
24.90 KB
-
struct_df.rds
106.30 KB
-
trait_obs.csv
9.07 KB
-
TRY_means.csv
1.52 KB
-
TRY_sd.csv
1.59 KB
Abstract
Code and data for reproducing the analysis in the manuscript “Species, structural, and functional diversity in relation to live aboveground biomass in mature and late-successional northeastern temperate forests.” Provided data include local functional trait observations of five leaf and stem traits (specific leaf area, leaf nitrogen content, leaf phosphorus content, leaf carbon to nitrogen ratio, and wood density) for 142 individual trees within late-successional, mixed hardwood forest plots and plot-level variables including structural complexity and live aboveground biomass. R code is provided to generate model inputs, apply forest biomass models, process model results, and assemble tables and figures in the manuscript. Further, R code is provided to implement a Bayesian hierarchical model integrating local functional trait observations as defined in the manuscript. Details on the data, its processing, and the alternative model definitions and structure can be found in the main text of the manuscript. Potential users are encouraged to consider collaboration with authors when useful and appropriate. Misinterpretation of data may occur if used outside the context of the original analysis. All data are made available in their current state. While significant efforts have been made to ensure data accuracy, complete accuracy cannot be guaranteed. Data may be updated periodically. It is the responsibility of the data user to check for updated versions of the data.
Dataset DOI: 10.5061/dryad.kprr4xh8v
Tree functional trait data and associated plot-level variables for reproducing the analysis in the manuscript, "Species, structural, and functional diversity in relation to live aboveground biomass in mature and late-successional northeastern temperate forests." Data include plot-level structure and biomass variables derived from continuous forest inventory data across 304 late-successional mixed hardwood plots in Massachusetts, USA, surveyed between 2000 and 2021. Data also include individual functional trait observations from 142 individual trees across 26 of those plots and summarized species-level mean trait values derived from the TRY plant database for species included in the analysis. These are the same variables used to predict aboveground biomass within the associated manuscript.
Description of the data and file structure
Continuous Forest Inventory plot data is provided in four comma-separated text files below. Data was derived from Massachusetts Continuous Forest Inventory (MA CFI) database including inventories from 2000-2021. Data include variables from 304 plots, and 626 distinct plot-inventory year combinations from seven state forests in western Massachusetts. Most plots were measured twice, 10-15 years apart. The rows of the data files correspond to unique plot-by-inventory year combinations (plotyr). The columns provide several plot-level variables derived from the MA CFI database. Metadata for each variable is provided below.
- CFI_AGB_covars_092026.csv: Contains model covariates and predictor variable (live aboveground biomass) for the models.
- plotyr: Unique identifier for each plot-by-inventory year combination
- PLOT: CFI plot number
- INVYR: Inventory year
- ba_plot_ac: Total live basal area per acre (ft2/ac)
- spp_div_exp: Shannon's effective species diversity
- snag.ac: Number of standing dead trees per acre at least 12.7 cm DBH
- snag.large.ac: Number of large standing dead trees (>50 cm DBH) per acre
- fdis: Functional dispersion index calculated using species mean trait values derived from the TRY database (no local trait information integrated)
- sw_perc: Proportion of softwood species composition by basal area
- dia.div.exp: Shannon's effective diameter diversity of live trees, binned by 12.7cm (5in)
- agb.plot: Plot-level total live aboveground biomass (mt/ac), calculated using species-specific Jenkins equations (Jenkins et al. 2003). Response variable in model analysis
- rel_sdi: relative stand density index used in model analysis (unitless). See Woodall et al 2005
- cfi_ba.csv: Contains relative basal area by species in each plot-by-inventory year combination
- plotyr: Unique identifier for each plot-by-inventory year combination
- PLOT: CFI plot number
- INVYR: Inventory year
- Relative abundance of each species in each CFI plot by basal area listed by the species common name in the following columns (unitless). Includes all 22 species that were included in the model analysis
- sf_plots.csv: Provides the name of the state forest where each CFI plot is located.
- PLOT: CFI plot number
- SITE_NAME: State forest name
- make_struct_df.R: R script that standardizes model covariates derived from CFI data files above.
- struct_df.rds: Database that provides the standardized model covariates (output of "make_struct_df.R").
- Fields 1-12 are the same as described in cfi_AGB_covars.csv above
- SITE_NAME: State forest name where CFI plot is located
- sf_id: Unique index for each state forest
- dia_div_s: Standardized Shannon's effective diameter diversity of live trees, binned by 12.7cm (5in). Used as a structural complexity variable in the model analysis
- snag.ha: Number of standing dead trees per hectare at least 12.7 cm DBH
- snag_den_s: Standardized number of standing dead trees per acre at least 12.7 cm DBH. Used as a structural complexity variable in the model analysis
- snag.large.ha: Number of large standing dead trees (>50 cm DBH) per hectare
- snag_large_s: Standardized number of large standing dead trees (>50 cm DBH) per hectare
- sw_s: Standardized proportion of softwood species composition by basal area. Used in the model analysis control for composition across mixed hardwood forest
- spp_div_s: Standardized Shannon's effective species diversity. Used as metric of species diversity in model analysis
- agb.ha: Plot-level total live aboveground biomass in metric units (mt/hectare), calculated using species-specific Jenkins equations (Jenkins et al. 2003)
- l_agb_ha: Log of total live aboveground biomass (mt/hectare) per plot, calculated using species-specific Jenkins equations (Jenkins et al. 2003). Response variable in the model analysis
Local functional trait data is provided in the comma-separated text file "trait_obs.csv." Individual leaf and stem trait observations were collected from individual trees in three subplots around the perimeter of 26 CFI plots. Plots where trait observations were collected are a subset of the plots listed in the data files above. For details on methods used for functional trait measurements, refer to the associated manuscript. The rows of the data file correspond to unique tree identification numbers (treeID), made up of a plot number, subplot number, and tree number. The columns provide species name and functional trait measurements for five traits. Metadata for each variable is provided below.
- plot: Plot number
- treeID: Unique identifier for each tree sampled
- Species: Common name of tree species
- cn_ratio: Leaf carbon to nitrogen ratio (by dry mass)
- wd: Wood density (dry mass/fresh volume; g/cm3)
- leaf_n: Leaf nitrogen concentration by dry mass (mg/g)
- leaf_p: Leaf phosphorus concentration by dry mass (mg/g)
- sla: Specific leaf area (fresh leaf area/dry mass; mm2/mg)
TRY Trait Database-derived species trait means and standard deviation for each species-trait combination used in the analysis is provided in two comma-separated text files below. Five traits of 22 tree species are included. TRY database query was cleaned for duplicates and other data anomalies before summarizing species mean trait values and standard error for each species-trait combination. Metadata for each variable is provided below.
- TRY_means.csv: Each row corresponds to a species (common name) included in the analysis. Each column corresponds to the overall species mean trait value for that species-trait combination derived by averaging species mean trait values reported in the TRY database.
- cn_ratio: Overall species mean leaf carbon to nitrogen ratio (by dry mass)
- leaf_n: Overall species mean leaf nitrogen concentration by dry mass (mg/g)
- leaf_p: Overall species mean leaf phosphorus concentration by dry mass (mg/g)
- sla: Overall species mean specific leaf area (fresh leaf area/dry mass; mm2/mg)
- wd: Overall species mean wood density (dry mass/fresh volume; g/cm3)
- TRY_sd.csv: Each row corresponds to a species (common name) included in the analysis. Each column corresponds to the calculated species-specific standard deviation between mean trait values reported in the TRY database.
- cn_ratio: Species-specific standard deviation between leaf carbon to nitrogen ratio (by dry mass) mean values
- leaf_n: Species-specific standard deviation between leaf nitrogen concentration by dry mass (mg/g) mean values
- leaf_p: Species-specific standard deviation between leaf phosphorus concentration by dry mass (mg/g) mean values
- sla: Species-specific standard deviation between specific leaf area (fresh leaf area/dry mass; mm2/mg) mean values
- wd: Species-specific standard deviation between wood density (dry mass/fresh volume; g/cm3) mean values
Model Code, Outputs, and Cross-Validation
FinalModel.zip folder: Structural complexity + Functional diversity (local) model code
- fit_final_mod.R: R script file that loads the model inputs and fits the structural complexity + functional diversity model integrating local individual trait observations ("final_mod.stan" file). It then does post-processing analysis to check for model convergence, calculate partitioned R2 values, and make figures and tables from the associated manuscript. The source file needed to run the R script is provided and described below.
- final_mod.stan: Stan source code for the Bayesian hierarchical model framework which integrates local individual functional trait information with existing TRY database species mean values to estimate plot-level functional diversity (functional dispersion, FDis) and make predictions on plot-level aboveground biomass. Calls to the source files are demonstrated in the fit_final_mod.R R script.
- mod_out.rds: Saved model output of posterior samples for all parameters derived from the structural complexity + functional diversity (local) model generated from running the model in the fit_final_mod.R script. The model output for the alternative models were generated separately.
- gen_part_res.stan: Stan source code that generates partial residual samples from the structural complexity + funcitonal diversity (local) model needed to create Figure 2 from the associated manuscript.
CV.zip folder: Cross-validation and model comparison of candidate models
Folders include model outputs, cross-validation and model comparison for 6 candidate models abbreviated as:
- func_div_db_mod= functional diversity (database) model
- funct_div_local_mod= functional diversity (local) model
- null_model= null model (only relative SDI)
- spp_div_mod= species diversity model
- struct_comp_func_div_local_model= structure +functional diversity (local) model
- struct_comp_mod= structural complexity model
- struct_comp_mod_v2= structural complexity model that includes quadratic term for diameter diversity (excluded from results as it did not improve model performance compared to simplified structural complexity model v1)
Specific descriptions of each folder within CV.zip are below:
- log_likelihood: Saved log-likehood output from all candidate models.
- lppd_scores: Saved lppd scores output from all candidate models.
- mcmc_out: Saved mcmc output from all 6 candidate models.
- model_comparison: R scripts make the table of elpd scores for all candidate models and compare groups of candidate models based on the difference in elpd scores and SD.
- full_vs_struct_comp.R: compares the structural complexity + functional diversity (local) model vs. structural complexity model.
- struct_vs_spp_vs_func.R: compares the species diversity vs. functional diversity (local) model vs. functional diversity (database) model.
- struct_comp_v1_versus_v2.R: compares two versions of the structural complexity models (v2 has a quadratic term for diameter diversity).
- make_elpd_table.R: makes a table of the elpd scores of all candidate models.
- models: Contains stan source code for the candidate models (except the structural complexity + functional diversity local model) with different combinations of structural complexity, functional trait diversity, and species diversity.
- run_scripts: R script file that loads the model inputs and fits all the candidate models (scripts call the Stan source files found in the "models" folder). It then does post-processing analysis to calculate the elpd score for each model.
- cv_func_div_db_mod.R: Fits the functional diversity-database model, which includes the functional diversity metric calculated using species-level mean trait values derived from the TRY database).
- cv_func_div_local_mod.R: Fits the functional diversity-local model which uses a Bayesian hierarchical framework to adjust species-mean trait values using local trait information for the functional diversity calculation.
- cv_null_mod.R: Fits the null model which includes only relative SDI as a predictor variable.
- cv_null_mod_sf.R: Fits the null model which includes only relative SDI as a predictor variable and partitions the data by state forest as separate folds for cross-validation.
- cv_spp_div_mod.R: Fits the species diversity model.
- cv_struct_comp_mod.R: Fits the structural complexity model.
- cv_struct_comp_mod_v2.R: Fits the structural complexity model with the quadratic term for diameter diversity.
- cv_struct_func_div_local_mod.R: Fits the structural complexity +functional diversity (local) model.
Sharing/Access information
Data and code were derived from the following sources:
- Kattge, J., G. B?nisch, S. D?az, S. Lavorel, I. C. Prentice, et al. 2020. TRY plant trait database -- enhanced coverage and open access. Global Change Biology 26:119--188.
- Massachusetts Continuous Forest Inventory (CFI) Database
- Laliberte, E., and P. Legendre. 2010. A distance-based framework for measuring functional diversity from multiple traits. Ecology 91:299--305.
- USDA NRCS. (2023). The PLANTS Database. National Plant Data Team. http://plants.usda.gov