Presettlement tree distributions and forest types of northeast Ohio, USA, mapped with species distribution models
Data files
Nov 29, 2024 version files 1.24 MB
-
Flinn_et_al._JVS_forest_type.txt
1.24 MB
-
README.md
1.41 KB
Abstract
European colonization radically transformed the landscapes of eastern North America. Understanding this legacy is vital to managing current ecological communities. To do so requires spatially explicit information about presettlement vegetation. Here we test the ability of species distribution models, which have rarely been used with historical data like land survey records, to generate useful predictions of presettlement tree distributions. These models also allow us to assess pre-disturbance vegetation-environment relationships.
Location: Cuyahoga County, Ohio, USA
Methods: Generalized linear models, generalized boosting models, random forests, and maximum entropy models related the distributions of 17 tree taxa to elevation, slope, aspect, and soil type, based on 4234 tree observations from circa-1800 surveys. Cluster analysis defined forest types and created a prediction map of forest types using random forests.
Results: This study generated high-resolution predictions of presettlement tree distributions. Fagus grandifolia and Carya spp. were predicted to have found suitable habitat in over half the area. Elevation was by far the most important predictor, followed by slope. Many taxa were more likely to occur at higher elevations, corresponding to the Allegheny Plateau, while others followed river valleys. Broadly, 51% of the county was predicted to support forest types with Fagus grandifolia and/or Acer spp., and 48% of the county was predicted to support forest types with Quercus spp., Carya spp., and/or Castanea dentata.
Conclusions: This study demonstrates that species distribution modeling with historical data provides insights into presettlement vegetation and vegetation-environment relationships. Our results reveal striking ecological patterns not apparent in today’s landscape, such as the sharp difference in vegetation between the Central Lowland and Allegheny Plateau. The maps created here offer a historical perspective that can inform conservation, education, and further research.
https://doi.org/10.5061/dryad.wwpzgmsv8
Description of the data and file structure
The accompanying code and data was used to generate distribution models and prediction maps for analyses described in Flinn et al. Raster versions of the maps included in the paper are included here.
Files and variables
File: Flinn_et_al._JVS_forest_type.txt
Description: Data file for forest type analysis described in Flinn et al.
Variables
- Unique_ID: unique identifier for survey lines
- POINT_X: Longitude coordinate (WGS84)
- POINT_Y: Latitude coordinate (WGS84)
- Group10: forest type code, with values assigned as follows: 1 = Quercus-Castanea, 2 = Fraxinus, 3 = Fraxinus-Acer-Tilia, 7 = Fagus-Acer, 70 = Acer, 75 = Quercus-Carya, 91 = Fraxinus-Ulmus, 92 = Quercus, 384 = Fagus, 390 = Tsuga, Picea, or Pinus
Code/software
Code can be read in the R statistical environment and includes all analyses and output presented in Flinn et al.
Map files from Flinn et al. are shared as .tif files that can be read be either proprietary (e.g., ArcGIS) or free (e.g., QGIS) GIS software.
Access information
Other publicly accessible locations of the data:
- NA
Data was derived from the following sources:
- NA
We analyzed historical survey data on 17 individual tree species and 10 forest types using species distribution models to create prediction maps of pre-settlement distributions.