Climate change projected to increase rates of background tree mortality across eastern North America
Data files
Jun 29, 2026 version files 15.35 MB
-
BRT_models.zip
10.62 MB
-
Model_fitting.R
6.66 KB
-
Mortality_Dataset.csv
4.72 MB
-
README.md
1.87 KB
Abstract
Tree mortality patterns play a central role in forest health, terrestrial carbon dynamics, and biodiversity conservation. While changes in tree mortality patterns to recent climate anomalies have been documented, the effects of climate change on ambient or ‘background’ mortality rates in the absence of severe disturbance agents remain poorly understood. Here, we applied a machine learning approach to a large dataset of 24,576 forest permanent sample plots distributed across Canada and the United States, spanning 28.1°C in mean annual temperature and 1743.5 mm in annual precipitation, to study and predict the effects of climate on background mortality for nine of the most abundant tree species in eastern North America. We developed species-specific climate-mortality models while controlling for the interactive effects of stand development processes, CO2, and atmospheric pollutant SO4. We found that temperature ranked among the three strongest predictors of mortality rates and that warmer temperatures led to higher background tree mortality for most species. Furthermore, we predicted notable increases in tree mortality rates along the southern portions of species’ ranges in response to future warming. On average, mortality rates are predicted to rise by 0.2–0.7 %·year-1 for five of the nine species under SSP2-4.5 by mid-century (2041–2070). By examining climate-driven variation in background tree mortality at the continental scale, our study provides valuable insight on future forest dynamics under climate change.
Dataset DOI: 10.5061/dryad.zs7h44jq3
Description of the data and file structure
Mortality Dataset
This file contains the processed mortality datasets for nine common tree species in North America.
Model Fitting R Code
This file includes the R scripts used to fit species-specific boosted regression tree models.
Files and variables
File: Model_fitting.R
Description: The R scripts used to fit species-specific boosted regression tree models.
File: Mortality_Dataset.csv
Description: This file contains the processed mortality datasets for nine tree species.
Variables
- Species: tree species name (unitless)
- Mortality: plot level annual mortality rate (% per year)
- Age_plot: age of plot (unit: year)
- BA_plot: basal area of plot (unit: m2 per hectare)
- DBH_plot: mean diameter at breast height (DBH) of plot (unit: cm)
- SI_plot: Shannon index of plot (unitless)
- CMI: climate moisture index (unit: mm)
- MAT: mean annual temperature (unit: degrees Celsius)
- CO2: atmospheric CO2 (unit: ppm)
- SO4: SO4 wet deposition (unit: kg S per ha per year)
File: BRT_models.zip
This folder contains the nine fitted species-specific boosted regression tree models.
Code/software
All analyses and visualizations in this study were performed using R (version 4.5.2).
Access information
Other publicly accessible locations of the data:
- N/A
Data was derived from the following sources:
- The PSP dataset included data from the U.S. Forest Inventory and Analysis (FIA) program, covering thirty-one eastern US states, as well as Canadian PSP programs from seven provinces (Nova Scotia, New Brunswick, Québec, Ontario, Manitoba, Saskatchewan, and Alberta).
