Mapping Presettlement Forest Species with Witness Trees

By Michael P. Strager, Melissa Thomas-Van Gundy, and Michael M. Metz

During the late 1700s and early 1800s, the original surveyors of West Virginia used tree markers or other natural features as reference landmarks. These locations often referenced vegetation that existed before any European settlement and influence on the land. Such a vegetation dataset can provide a useful and interesting snapshot into the past to help understand the historical aboriginal forest. It could be possible to determine the species and mixes that grew and matured optimally in a natural, untouched environment. The US Forest Service (USFS) recognized an opportunity with this historic information to reconstruct vegetation patterns to help better manage the land today for sustainable forestry.

The goal of the witness tree project became the development of a GIS geodatabase of corner trees and other markers from the original survey tracts contained within the proclamation boundaries of the Monongahela National Forest in West Virginia.

Researchers at West Virginia University were also interested in exploring the ecophysical characteristics of a historic tree inventory to examine relationships to topographic roughness, moisture index, aspect, and landform. In addition, the site descriptions by species could be used to validate existing natural vegetation models or as an input to species prediction models. With the availability of this unique data and interest from both USFS and the university, the goal of the witness tree project became to develop a GIS geodatabase of corner trees and other markers from the original survey tracts contained within the proclamation boundaries of the Monongahela National Forest (MNF) in West Virginia.

A database of witness trees of the MNF was created using the ArcGIS georeferencing toolbar. The authors later incorporated the ArcGIS Spatial Analyst extension to determine terrain attributes. The Esri products were instrumental in the collection, mapping, and analysis of the witness trees by landform and physiographic subsection.

The first step in the authors' methodology was to carefully scan the historic maps with a large-format scanner. Next, the authors used the georeferencing toolbar to link common features on the historical maps to features they could register already in their spatially mapped layers. Fortunately, many maps included streams and tributary intersections that provided the necessary control tie points. By examining the link table within the georeferencing toolbar, they were able to make sure all their referenced maps had a root mean square error of less than .009 inch, which represents the difference between the original location and the new one determined from the transformation process.

The first step was to carefully scan the historic maps with a large-format scanner. Next, the georeferencing toolbar was used to link common features on the historical maps to features already registered already in the spatially mapped layers.

After the maps were referenced, the authors digitized and attributed all features using a consistent display scale of 1:5,000. Survey corner information that they wanted in the database included location; type of corner (tree, river bend, pile of rocks, stake, etc.); tree species (if used); and any other pertinent tree data, such as tree size.

Common names used for tree species had to be related to common names likely to be found in old survey records. Each polygon that represented the first parcel surveyed for the area was also given a point referenced by the individual locator number, generally in the center of the polygon. The deeds describing the parcels were used to determine survey corners on the maps and tally the tree species or other markers used to witness those corners in the attribute file. Along with tree species, the locator number, date of the deed or survey, and deed holder's name were recorded in the ArcGIS geodatabase.

The authors performed a spatial join in ArcGIS, then determined the landform of each corner. Landform data came from an MNF ecological classification system that was assigned during soil surveys. Landforms include ridge, saddle, shoulder, knob, bench, plateau, foothill, toe slope, gentle side slope, middle/back slope, steep side slope, cliff, cove, floodplain, newer terrace, older terrace, valley floor, flat, and valley. ArcGIS Spatial Analyst was also used to model and compare the soil-derived landforms to elevation-derived landforms.

Common names used for tree species had to be related to common names likely to be found in old survey records. Each polygon that represented the first parcel surveyed for the area was also given a point referenced by the individual locator number, generally in the center of the polygon.

The resultant frequency counts of species by landforms (by subsection) were analyzed for significance through contingency table analysis and chi-square likelihood ratio tests. Standardized residuals were calculated for each landform and species by subsection. The analysis was made on 15,591 corners representing 22,107 witness trees from deeds ranging from 1752 to 1900. Of the database, 24 percent of the corners date to the late 1700s. The greatest numbers of corners were established in the 1840s and 1850s at 17.8 percent and 29.3 percent, respectively.

Findings

Pre-European settlement ridge and valley forests were dominated by mixed oak, pine, American chestnut, and hickories on ridges, while the valley floors were dominated by white oak, sugar maple, pine, basswood, and hemlock. As compared to European settlement-era forests, current forests have less American chestnut, pine, and white oak and greater amounts of chestnut oak, northern red oak, scarlet oak, and red maple in the Ridge and Valley province and greater amounts of black cherry, red maple, and birch in the Allegheny Mountains province. Many of the authors' findings supported those also found by Abrams and McCay (1996), while the study used more landforms and subsections.

The deeds describing the parcels were used to determine survey corners on the maps and tally the tree species or other markers used to witness those corners in the attribute file.

Witness trees provide a glimpse into historical forest conditions for much of the eastern United States due to impacts of early settlement by Europeans and a history of near-complete forest clearing in many states at the turn of the 19th century. Because these descriptions represent a static point in time, they should not be considered the restoration end point or management goal. However, they are often the best source of information for the time of European settlement, offer clues to Native American influences (or absence of influences) on the landscape, and can inform restoration actions.

About the Authors

Michael P. Strager is with the Division of Forestry and Natural Resources, West Virginia University, Morgantown, West Virginia; Melissa Thomas-Van Gundy is with the USFS Timber & Watershed Laboratory & Fernow Experimental Forest, Elkins, West Virginia; and Michael M. Metz is with the Natural Resource Analysis Center, West Virginia University, Morgantown, West Virginia.

For more information, contact Michael Strager, associate professor, West Virginia University (e-mail: mstrager@wvu.edu).

Work cited: Abrams, M. D., and D. M. McCay, 1996. "Vegetation-site relationships of witness trees (1780-1856) in the presettlement forests of eastern West Virginia." Canadian Journal of Forest Research 26: 217-224.