Data from: Beyond resilience: Responses to changing climate and disturbance regimes in temperate forest landscapes across the Northern Hemisphere
Data files
Aug 30, 2024 version files 880.66 MB
Abstract
Climate change has profound impacts on forest ecosystem dynamics and could lead to the emergence of novel ecosystems via changes in species composition, forest structure, and potentially a complete loss of tree cover. Disturbances fundamentally shape those dynamics: the prevailing disturbance regime determines the inherent variability of a system, and its climate-mediated change could accelerate forest transformation. We used the individual-based forest landscape and disturbance model iLand to investigate the resilience of three protected temperate forest landscapes on three continents – selected to represent a gradient from low to high disturbance activity – to changing climate and disturbance regimes. In scenarios of sustained strong global warming, natural disturbances increased across all landscapes regardless of projected changes in precipitation (up to a seven-fold increase in disturbance rate over the 180-year simulation period). Landscapes with historically high disturbance activity had a higher chance of remaining resilient in the future, retaining their structure and composition within the range of variability inherent to the system. However, the risk of regime shift and forest loss was also highest in these systems, suggesting forests may be vulnerable to abrupt change beyond a threshold of increasing disturbance activity. Resilience generally decreased with increasing severity of climate change. Novelty in tree species composition was more common than novelty in forest structure, especially under dry climate scenarios. Forests close to the upper tree line experienced high novelty in structure across all three study systems. Our results highlight common patterns and processes of forest change, while also underlining the diverse and context-specific responses of temperate forest landscapes to climate change. Understanding past and future disturbance regimes can anticipate the magnitude and direction of forest change. Yet, even across a broad gradient of disturbance activity, we conclude that climate change mitigation is the most effective means of maintaining forest resilience.
README: Beyond resilience: Responses to changing climate and disturbance regimes in temperate forest landscapes across the Northern Hemisphere
https://doi.org/10.5061/dryad.pc866t1zt
Description of the data
We used the individual-based forest landscape and disturbance model iLand to investigate the resilience of three protected temperate forest landscapes on three continents – selected to represent a gradient from low to high disturbance activity – to changing climate and disturbance regimes.
Input files
All 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_project" [landscape]; separated with "_"
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
log: empty, for storing log files written during simulations
output: empty, for storing output files written during simulations
scripts: scripts detailing atmospheric CO2 values and wind events, code for wiritng user-specific output ("customagg")
bite (only Grand Teton): script file for running the Mountain Pine Beetle in the BITE module
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
For each simulation (n=300) multiple output files (.sqlite, .csv, .asc) containing data on forest composition and structure as well as simulated disturbances were generated.
Since the output files range from ~400-2500 MB in size per simulation, only the files for one simulation are uploaded as an example (Berchtesgaden landscape, baseline climate, replicate 1). The other files can be requested from the corresponding author, Christina Dollinger.
Main output files (.sqlite, n=300)
File name: [landscape] [scenario] [precipitationChange] [temperatureChange] [replicate].sqlite; separated with "_"
- Landscapes (n=3): bgd (Berchtesgaden), grte (Grand Teton), stoko (Shiretoko)
- Scenarios (n=2): baseline (no climate change simulated), change (climate change simulated)
- Precipitation change (n=2): wet, dry
- Temperature change (n=2): warm, hot
- Replicate (n=20): 1-20
Each .sqlite database contains multiple output tables (n=9-10, dp. on landscape):
Table name | Description |
---|---|
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 |
dynamicstand | User-defined outputs for tree aggregates for each stand and species. Written every 10 years. https://iland-model.org/Outputs?highlight=basal+area#dynamic_stand_output_by_species_RU |
custom_tree1 | Data on trees (height >4 m) on patches in the first year post disturbance. Type: "customagg"-output (highly customizable). https://iland-model.org/dynamic+outputs#customagg_output |
custom_sapBig | Data on saplings (height between 1.3-4 m) on patches in the first year post disturbance. Type: "customagg"-output (highly customizable). https://iland-model.org/dynamic+outputs#customagg_output |
custom_sapSmall | Data on saplings (height <1.3 m) on patches in the first year post disturbance. Type: "customagg"-output (highly customizable). https://iland-model.org/dynamic+outputs#customagg_output |
sapling | Output of the establishment/sapling layer per resource unit and species. The output covers trees between a dbh of 1cm (height>1.3m) and the recruitment threshold (i.e. a height of 4m). Written every 10 years. https://iland-model.org/Outputs#Sapling_Output |
barkbeetle (only in Berchtesgaden) | Barkbeetle related outputs per year. The outputs are created after each year (or spread event) and contain information about bark beetle generations, spread and damage for the total landscape. Written every year. https://iland-model.org/barkbeetle+module#BarkBeetle_module_output |
wind (only in Berchtesgaden and Shiretoko) | Wind related output generated by event per year. Written every year. https://iland-model.org/wind+module#Wind_disturbance_module_output |
fire (only in Grand Teton) | Fire event aggregates per fire event. The output contains a row for each (ignited) fire event. Written every year. https://iland-model.org/fire+module#Fire_specific_outputs |
bite (only in Grand Teton) | The output provides annual statistics for each biotic agent simulated with the BITE module (in this study: Mountain Pine Beetle). Written every year. (https://iland-model.org/Outputs#Annual_summary_for_each_Bite_agent). |
customagg | Empty |
runinfo | Timestamp of the simulation start and iLand version information. |
For easier accessability, the example sqlite database ("bgd_baseline_wet_warm_1.sqlite") has been unpacked and saved as individual csv-files (in the folder "sqlite_as_csv.zip").
Patch lists (.csv, n=300)
Folder: "patches"
File name: patches [landscape] [scenario] [precipitationChange] [temperatureChange] [replicate].csv; separated with "_"
- Landscapes (n=3): bgd (Berchtesgaden), grte (Grand Teton), stoko (Shiretoko)
- Scenarios (n=2): baseline (no climate change simulated), change (climate change simulated)
- Precipitation change (n=2): wet, dry
- Temperature change (n=2): warm, hot
- Replicate (n=20): 1-20
The .csv file contains data on simulated disturbance patches:
Column name | Description |
---|---|
year | Simulation year in which the patch was generated |
dtype | Disturbance agent ("wind", "barkbeetle", "fire", "bite") |
id | Unique patch ID. First 1-3 digits: simulation year in which the patch was generated; following digit: disturbance agent (1: wind, 2: barkbeetle, 3: fire, 4: mountain pine beetle); last 4 digits: number of patch for the simulation year |
area | Patch size (1 ≙ 10x10 m) |
Patch grids (.asc, n=5400)
Folder: "patches/patchgrids"
File names: patchgrid [decade] [landscape] [scenario] [precipitationChange] [temperatureChange] [replicate].asc; separated with "_"
- Decades (n=18): 10-180 in steps of 10
- Landscapes (n=3): bgd (Berchtesgaden), grte (Grand Teton), stoko (Shiretoko)
- Scenarios (n=2): baseline (no climate change simulated), change (climate change simulated)
- Precipitation change (n=2): wet, dry
- Temperature change (n=2): warm, hot
- Replicate (n=20): 1-20
The .asc file contains a grid of the landscape showing the simulated disturbance patches of the last 20 simulation years (resolution 10x10 m). Cell value = Patch ID. A cell value of 0 means that the cell has been undisturbed for the last 20 simulation years.
Code/software files
Data preparation as well as all analyses were done using the R project for statistical computing version 4.2.2 (R Core Team, 2021). The following R scripts were used for analysis:
- 0_setup: reading in packages and landscape specific grids, defining variables
- 1_patchgrids: extracting data from the patchgrids (.asc)
- 3_disturbances: analysing the disturbance outputs
- 4_landscapeScale: analysing changes in structure and composition, calculating reorganization pathways (additional code scripts used for the appendix can be found in the zip-folder)
- 7_disturbancePerCell: combine data on reorganization patterns and disturbance history on cell-level (100x100 m)
- 9_randomForest: code to run the random forest model
Methods
The dataset was simulated using the individual-based forest landscape and disturbance model iLand (Model page).