Disentangling compound effects of changing disturbance and regeneration across temperate forest landscapes
Data files
Oct 09, 2025 version files 9.33 MB
-
processed_data.zip
2.42 MB
-
README.md
23.20 KB
-
simulation_output.zip
6.89 MB
Abstract
Global change alters many ecological processes simultaneously. Yet, the interactive and compound effects of these changes remain difficult to quantify. Here, we conducted an extensive experiment in silico to address the question of how changes in disturbance and regeneration interactively alter temperate forest landscapes across three continents. Interaction effects between changing disturbance and regeneration processes amplified ecosystem change by an order of magnitude compared to the effects of changes in individual drivers. Interaction effects were context-dependent. At low to intermediate disturbance rates (<1 % yr^-1^), high rates of regeneration buffered effects on forest structure. In contrast, the interaction of high disturbance and regeneration rates amplified changes in forest composition. We conclude that the consequences of changing disturbance and regeneration need to be assessed jointly to understand their outcomes. Our findings highlight the importance of interaction effects of simultaneously changing ecological processes in shaping the future of forest ecosystems.
Description of the data
We used the individual-based forest landscape and disturbance model iLand to investigate the response of three protected temperate forest landscapes on three continents to changes in disturbance and regeneration processes under climate change.
The simulation outputs were processed in R to derive the final datasets for analysis.
Input files ("simulation_input.zip", Zenodo Supplemental Information)
Input files needed to replicate the simulations for the three landscapes in iLand (https://iland-model.org/) are available in the Zendodo repository associated with this data repository.
Folder name (n=3): "iLand_" [landscape; BGD: Berchtesgaden, STK: Shiretoko, GTE: Grand Teton]
- project files (.xml): one for running scenarios without climate change ("baseline"), one for running scenarios with climate change ("change")
- database: databases for climate and tree species
- GIS: spatial input data
- init: files for initializing tree vegetation
- lip: light interference pattern files
- scripts*: scripts detailing atmospheric CO2 values (1 for reference runs, 1 for climate change, .txt), JavaScript code for simulating generic disturbance sequences via the iLand base management module (.js), and input files for the management module ("nested_mgmt_X.csv", n=80)
- .sh-files (n=2): shell scripts for running all simulations (one per landscape and climate scenario)
- .xml-files (n=2): project files for initialising the model (one per landscape and climate scenario)
*The csv files for the 80 unique disturbance sequences per landscape are not all uploaded due to the file size. Only one input file for reference disturbance size and frequency ("nested_mgmt_1_1_1.csv") is provided for each landscape. The other files can be requested from the corresponding author, Christina Dollinger (christina.dollinger@tum.de).
Extensive model documentation on iLand can be found at https://iland-model.org and the full model source code at https://github.com/edfm-tum/iland-model. Additionally, the iLand book features in-depth tutorials for running simulations in iLand using the uploaded data (e.g., Chapter 3, https://iland-model.org/iland-book/).
Data files
Raw simulation outputs ("simulation_output.zip", Dryad Data)
For each simulation (n=7,680), two output files (.sqlite, .csv) containing data on forest composition and structure, as well as simulated disturbance, were generated.
Since the full set of output files has a total size of 130+ GB, only the files for one simulation are uploaded as an example (Berchtesgaden landscape, baseline climate, replicate 1, reference levels of disturbance size, disturbance frequency, seed production, and sapling growth). The other files can be generated using the model input files provided in the Zenodo Supplemental Information repository ("simulation_input", see above) or requested from the corresponding author, Christina Dollinger (christina.dollinger@tum.de).
Main output files (.sqlite, n=300)
File names: "output" [landscape] [climate scenario] "rep" [replicate] "size" [size modification level] "freq" [frequency modification level] "browsing" [sapling growth modification level] "fecundity" [seed production modification level].sqlite; separated with "_"
example file: "output_bgd_baseline_rep1_size1_freq1_browsing1_fecundity100.sqlite"
- Landscapes (n=3): bgd (Berchtesgaden), grte (Grand Teton), stoko (Shiretoko)
- Scenarios (n=2): baseline (no climate change simulated), hotdry (climate change simulated)
- Replicate (n=5): 1-5
- Size modification level (n=4): 1, 2, 5, 10
- Frequency modification level (n=4): 1, 2, 5, 10
- Sapling growth modification level (n=4): 1, 2, 5, 10
- Seed production modification level (n=4): 100, 50, 20, 10
Each .sqlite database contains four output tables:
| Table name | Description |
|---|---|
| dynamicstand | User-defined outputs for tree aggregates for each stand and species. Written in the year 0 ("2020"), 30 ("2050"), and 80 ("2100"). https://iland-model.org/Outputs?highlight=basal+area#dynamic_stand_output_by_species_RU |
| landscape | Output of aggregates on the level of landscape x species. Values are always aggregated per hectare. Written every year. https://iland-model.org/Outputs#Landscape_aggregates_per_species |
| landscape_removed | Output of aggregates of all removed trees due to 'natural' death, harvest, or disturbance per species and reason. All values are totals for the whole landscape. Written every year. https://iland-model.org/Outputs#Aggregates_of_removed_trees_due_to_death_harvest_and_disturbances_per_species |
| runinfo | Timestamp of the simulation start and iLand version information. |
For easier accessibility, the example SQLite database ("output_bgd_baseline_rep1_size1_freq1_browsing1_fecundity100.sqlite") has been unpacked and saved as individual CSV files (in the sub-folder "sqlite_as_csv").
Patch lists
File names: "output" "patches" [landscape] [climate scenario] "rep" [replicate] "size" [size modification level] "freq" [frequency modification level] "browsing" [sapling growth modification level] "fecundity" [seed production modification level].csv; separated with "_"
- Landscapes (n=3): bgd (Berchtesgaden), grte (Grand Teton), stoko (Shiretoko)
- Scenarios (n=2): baseline (no climate change simulated), hotdry (climate change simulated)
- Replicate (n=5): 1-5
- Size modification level (n=4): 1, 2, 5, 10
- Frequency modification level (n=4): 1, 2, 5, 10
- Sapling growth modification level (n=4): 1, 2, 5, 10
- Seed production modification level (n=4): 100, 50, 20, 10
The .csv file contains data on simulated disturbance patches:
| Column name | Description |
|---|---|
| year | Simulation year in which the patch was generated |
| id | Unique patch ID. First 1-3 digits: simulation year in which the patch was generated; following digit: disturbance agent; last 4 digits: number of patches for the simulation year |
| agent | Disturbance agent ("wind", "barkbeetle", "fire", "bite") |
| fraction | Fraction of basal area to be killed |
| before_ba | Total basal area of all trees present [m2 ha-1] |
| killed_ba | Total basal area that was killed [m2 ha-1] |
| ncells | Number of cells (resolution of 10x10 m) where the basal area was killed |
Processed data sets for analysis ("processed_data.zip", Dryad)
dyn.df (.csv)
Data frame used for Q2
| Column name | Description |
|---|---|
| climate | Climate scenario, "baseline" (no climate change) or "hotdry" (climate change, RCP 8.5) |
| rep | Replicate, 1:5 |
| size | Disturbance size modification level (n=4): 1, 2, 5, 10 |
| freq | Disturbance frequency modification level (n=4): 1, 2, 5, 10 |
| browsing | Sapling growth modification level (n=4): 1, 2, 5, 10 |
| fecundity | Seed production modification level (n=4): 100, 50, 20, 10 |
| identifier | Unique identifier for each of the 7,680 simulation runs |
| landscape | Landscape abbreviation (n=3): bgd (Berchtesgaden), grte (Grand Teton), stoko (Shiretoko) |
| area | Simulated forested area in hectares |
| dist.dyn | Mean yearly disturbance rate over the 80-year simulation period in % yr^-1 |
| name | Name of the forest response (n=3): structural change ("1. Structure\nBasal area decreased by >50 % from reference"), composition change ("2. Composition\nDominant species changed from reference"), and forest loss ("3. Remaining forest\nStem density dropping below 50 trees/ha") |
| value | Proportion of unchanged landscape per dimension of forest change (see column "name") |
| regen.dyn | Mean yearly regeneration rate over the 80-year simulation period in n yr^-1 ha^-1 (regeneration rate defined as the number of saplings recruited into the tree layer (height >4 m) |
| n_year | Total number of simulation years |
dyn effect.df (.csv)
Data frame used for Q3
| Column name | Description |
|---|---|
| rep | Replicate, 1:5 |
| size | Disturbance size modification level (n=4): 1, 2, 5, 10 |
| freq | Disturbance frequency modification level (n=4): 1, 2, 5, 10 |
| browsing | Sapling growth modification level (n=4): 1, 2, 5, 10 |
| fecundity | Seed production modification level (n=4): 100, 50, 20, 10 |
| landscape | Landscape abbreviation (n=3): bgd (Berchtesgaden), grte (Grand Teton), stoko (Shiretoko) |
| area | Simulated forested area in hectares |
| .dyn_baseline | Mean yearly disturbance rate over the 80-year simulation period in % yr^-1 for the run under the reference climate |
| name | Name of the forest response (n=3): structural change ("1. Structure\nBasal area decreased by >50 % from reference"), composition change ("2. Composition\nDominant species changed from reference"), and forest loss ("3. Remaining forest\nStem density dropping below 50 trees/ha") |
| baseline | Proportion of unchanged landscape per dimension of forest change (see column "name") for the run under reference climate |
| regen.dyn_baseline | Mean yearly regeneration rate over the 80-year simulation period in n yr^-1 ha^-1 (regeneration rate defined as the number of saplings recruited into the tree layer (height >4 m) for the run under reference climate |
| dist.dyn_hotdry | Mean yearly disturbance rate over the 80-year simulation period in % yr^-1 for the run under climate change |
| hotdry | Proportion of unchanged landscape per dimension of forest change (see column "name") for the run under climate change |
| regen.dyn_hotdry | Mean yearly regeneration rate over the 80-year simulation period in n yr^-1 ha^-1 (regeneration rate defined as the number of saplings recruited into the tree layer (height >4 m) for the run under climate change |
| effect | hotdry-baseline, i.e., the difference in proportion of landscape unchanged between the run under climate change and the run under reference climate, the so-called "climate effect" |
singleProcess.df (.csv)
Data frame used for Q1
| Column name | Description |
|---|---|
| landscape | Landscape abbreviation (n=3): Berchtesgaden, Grand Teton, Shiretoko |
| mod | Process modification level (n=4): 1, 2, 5, 10 |
| name | Name of the forest response (n=3): structural change ("1. Structure\nBasal area decreased by >50 % from reference"), composition change ("2. Composition\nDominant species changed from reference"), and forest loss ("3. Remaining forest\nStem density dropping below 50 trees/ha") |
| rep | Replicate, 1:5 |
| process | Name of the process (n=4): "Disturbance size", "Disturbance frequency", "Seed production decrease", "Sapling growth limitations" |
| value | Proportion of unchanged landscape per dimension of forest change (see column "name") |
The sub-folder "helper_files" contains a data frame ("rid.df.csv") with all Resource Units IDs (100x100 m cells) and their elevation, as well as xy-coordinates for each landscape, generated using simulation input data.
Code/software files
Code repository: https://github.com/CEDollinger/2nd_analysis
Data preparation as well as all analyses were done using the R project for statistical computing version 4.2.2 (R Core Team, 2024). The following R scripts were used for analysis:
- *0_setup: reading in packages and setting variables
- 1_readingInData: code for processing raw simulation outputs, not needed since all the processed data sets generated by the script can be found in the Dryad data repository (folder "processed_data")
- 2_figures: code for generating the figures
- 3_text_calculations: code for calculating values reported in the paper
*The sub-folder "processed_data/helper_files" contains a data frame ("rid.df.csv") with all Resource Units IDs (100x100 m cells) and their elevation, as well as xy-coordinates for each landscape generated using simulation input data. This file is needed for running "0_setup.R".
We used the individual-based forest landscape and disturbance model iLand to investigate the response of three protected temperate forest landscapes on three continents to changes in disturbance and regeneration processes under climate change. The simulation outputs were processed in R to derive the final datasets for analysis.