You are here

Harvard Forest Data Archive

HF210

Pre-Colonial and Modern Tree Data from Nine Northeastern States 1620-2008

Related Publications

Data

Overview

  • Lead: Jonathan Thompson, Charles Cogbill
  • Investigators:
  • Contact: Jonathan Thompson
  • Start date: 1620
  • End date: 2008
  • Status: completed
  • Location: Northeastern States
  • Latitude: +39.76 to +47.15
  • Longitude: -80.47 to -67.02
  • Elevation: 3 to 858 meter
  • Taxa: Abies (fir), Acer (maple), Betula (birch), Carya (hickory), Castanea (chestnut), Fagus (beech), Fraxinus (ash), Juglans (walnut), Juniperas (cedar), Larix (tamarack), Liriodendron (tulip), Magnolia (magnolia), Nyssa (blackgum), Ostrya (hornbeam), Picea (spruce), Pinus (pine), Platanus (sycamore), Populus (poplar), Prunus (cherry), Quercus (oak), Tilia (basswood), Tsuga (hemlock), Ulmus (elm)
  • Release date: 2013
  • Revisions:
  • EML file: knb-lter-hfr.210.3
  • DOI: digital object identifier
  • Related links:
  • Study type: historical
  • Research topic: historical and retrospective studies; regional studies
  • LTER core area: disturbance
  • Keywords: abundance, history, maps, region, species composition
  • Abstract:

    The northeastern United States is a predominately-forested region that, like most of the eastern U.S., has undergone a 400-year history of intense logging, land clearance for agriculture, and natural reforestation. This setting affords the opportunity to address a major ecological question: How similar are today’s forests to those existing prior to European colonization? Working throughout a nine-state region spanning Maine to Pennsylvania, we assembled a comprehensive database of archival land-survey records describing the forests at the time of European colonization. We compared these records to modern forest inventory data and described: (1) the magnitude and attributes of forest compositional change, (2) the geography of change and (3) the relationships between change and environmental factors and historical land use. We found that with few exceptions, notably the American chestnut, the same taxa that made up the pre-colonial forest still comprise the forest today, despite ample opportunities for species invasion and loss. Nonetheless, there have been dramatic shifts in the relative abundance of forest taxa. The magnitude of change is spatially clustered at local scales (less than 125-km) but exhibits little evidence of regional-scale gradients. Compositional change is most strongly associated with the historical extent of agricultural clearing. Throughout the region, there has been a broad ecological shift away from late successional taxa, such as beech and hemlock, in favor of early- and mid-successional taxa, such as red maple and poplar. Additionally, the modern forest composition is more homogeneous and less coupled to local climatic controls.

  • Methods:

    Study Region

    We examined changes in forest composition within a sample of colonial townships distributed throughout a nine state region of the northeastern U.S., ranging latitudinally from northern Maine (47d 30m N) to southern New Jersey (39d 30m N) and longitudinally from western Pennsylvania (67d W) to the Atlantic Ocean (80d 30m W). The study area encompasses 4.33 × 105 km2 and spans nine physiographic provinces, primarily in the Appalachian Highlands Division, but also extending into portions of the Atlantic and Interior Plains. The region’s rolling topography is interrupted by several discrete sub-mountain ranges belonging to the greater Appalachian Chain. Several major rivers including the Hudson, Connecticut, Merrimack, Susquehanna, and Penobscot form low-elevation valleys across the study area. Much of the present geology of the region was shaped by the last glaciation (c. 20,000 ybp). The region is influenced by a range of climatic conditions; annual mean temperatures range from 3 to 10 deg C (mean Jan temp = -6 deg C; mean July temp = 19 deg C), and average annual precipitation ranges from 79 to 255cm. The study area includes four USFS designated ecoregions, defined based on a broad set of ecologically-relevant attributes, which we used to stratify our samples.

    Pre-Colonial Data

    We used the relative abundance of witness tree taxa (i.e., proportion of each taxon) within proprietary towns as our metric of pre-colonial forest composition. Proprietary towns (hereafter "towns") were granted by the colonies and states to absentee individuals to encourage colonization and "improvement" of the land throughout the period spanning from just after English colonization (1620) to after the creation of the Erie Canal (1825). Towns were usually 6-miles square (approx. 100 km2) and regularly shaped. Within each town, individual lots were established and surveyed using witness trees (WT) as markers. The original sources of the WT data typically include proprietors' records, field books, manuscripts, maps and published records of town land surveys before colonization. Town lotting surveys are the authoritative source, but when unavailable or inadequate other sources containing contemporary tree data were used. The witness trees are thus a relatively objective sample of forest composition prior to colonization. The land surveyors used English colloquial names to describe the trees and while they were skilled naturalists they often did not discern individual species within a genus. To reduce taxonomic uncertainty and ensure consistency across surveys we classified all trees into widely represented genera, following Cogbill et al. 2002. While this introduces some species ambiguity into the groupings, it is unavoidable, since surveyors of pre-colonial witness trees often did not distinguish species within genera. We assembled available town land survey records within the region, totaling 1280 towns and 325,000 trees. WT data are publicly available from individual town halls and archives throughout the region. Approximately 55 percent of the WT towns had been utilized in previous published studies, each of which focused on a smaller region.

    Modern Data

    The modern tree data come from the USDA Forest Service Forest Inventory and Analysis (FIA) program. We used the FIA census, spanning 2003-2008. FIA plots were sampled at an intensity of one plot per 2400-ha throughout the study region. Each plot consists of four, 7.3-m fixed-radius subplots (totaling 168-m2), on which all trees greater than 1.3-m in height are identified to species and the dbh recorded. . FIA protocols and data are publically available online (http://apps.fs.fed.us/fiadb-downloads/datamart.html). We obtained coordinates for the inventory plots, which allowed us to pair FIA plots with the WT town in which they reside, from the US Forest Service pursuant to a Memorandum of Understanding #09MU11242305123 between the U.S. Forest Service and Harvard University. We excluded FIA plots that were not classed as "Forest" within the FIA Condition table or contained less than 10 trees greater than 12.5-cm dbh. In the remaining plots we excluded all trees less than 12.5-cm dbh to reduce the potential for bias against smaller trees within the pre-colonial data. (Based on the findings of Wang et al 2009, we explored a 20-cm dbh threshold for inclusion. This resulted in a large reduction in number of towns meeting the sample intensity criteria outlined below, with little qualitative change in our findings.) We excluded any towns with less than 2 qualifying FIA plots. We then binned the tree species into the same 20 taxa used for the WTs and calculated the relative abundance in each plot.

    Assessing the Sample Intensity

    The density of WTs and FIA plots within towns varied widely. We used an approach somewhat akin to rarefaction analysis [e.g. 38] to estimate the minimum density of FIA plots and WTs at which tree compositional diversity had been adequately sampled--i.e. to determine whether or not each town had sufficient tree data to include in our analyses. Our approach relied on the fact that tree diversity increases asymptotically with the addition of each new WT or FIA plot. By using bootstrap sampling and fitting Michalis-Menton (M-M) functions, we estimated the density of WTs and FIA plots at which the full complement of diversity was represented. This density was used as a minimum threshold for determining whether or not to include a town in the analyses.

    More specifically, our procedure for determining adequate FIA plot density was as follows: (1) We stratified the study region by ecoregions. (2) Within each ecoregion, we selected the most plot-dense towns, taking only those towns with at least 5 FIA plots and that were within the upper tercile of plot density (plots/km2). (3) For each town within this subset, we iteratively took 100 sets of bootstrapped samples, with each set consisting of a sample of one FIA plot, a sample of two FIA plots (sampled with replacement), and so on up to the total number of FIA plots in that town. (4) For each bootstrap sample within a set we calculated the mean Sørenson distance (see below) between the sample’s relative composition and the relative composition with all plots included. As the number of plots sampled increases, Sørenson similarity tends to initially increase sharply before slowly leveling off as composition stabilizes. (5) Accordingly, for each set of bootstrap samples we fit an M-M curve and recorded the asymptote, Smax¬. Smax is an estimate of the similarity between two random samples of the town’s forest when plot density is impossibly large and is typically slightly less than one, or complete similarity, due to variation introduced by bootstrap sampling. (6) We calculated a threshold plot density, Dmin, as the minimum plot density required to reach a given proportion of Smax . Since reaching Smax would require infinite plot density, we decided that the plot density required to reach 90 percent of Smax would be adequate to approximate a town’s true forest composition. (7) We averaged Dmin over all 100 sample sets from each town, and then over all towns within each ecoregion. This ecoregion grand mean, Dmin, was taken to be the ecoregion-wide threshold plot density necessary to capture a town's compositional diversity.

    We followed a similar procedure for determining the adequate number of WTs necessary to capture the compositional diversity within a town, except that we iteratively sampled bins of 20 trees, as opposed to FIA plots, before fitting the M-M function. We also used only those towns with at least 100 WTs and that were within the upper tercile of tree density (WTs per km2).

    Results of Data Screening

    Of 1280 towns in the study area with witness tree data, 904 contained two or more forested FIA plots. Of these, 761 towns contained sufficient WT density to meet the sampling density threshold. Most towns with insufficient WT density were clustered in northern Maine. Within the 904 towns with sufficient WT data, 756 towns had sufficient FIA plot density. In all, 701 towns met the sampling density threshold for both the WT and FIA data. We used only these 701 towns in all subsequent analyses. The average density of WTs in the final sample was 1.77 trees per km2 (s = 2.52) and 252 trees/town (s = 357). The average density of FIA plots in the final sample was 0.27 plots per km2 (s = 0.10) and 4.26 plots per town (s = 2.44). The average density of FIA trees was 0.93 trees per km2 (s = 0.41) or 156 trees per town (s = 86).

    For more details, please see: Thompson, J. R., Carpenter, D. N., Cogbill, C. V., and Foster, D. R. (2013). Four Centuries of Change in Northeastern United States Forests. PLoS ONE, 8(9), e72540. doi:10.1371/journal.pone.0072540

  • Use:

    This dataset is released to the public under Creative Commons license CC BY (Attribution). Please keep the designated contact person informed of any plans to use the dataset. Consultation or collaboration with the original investigators is strongly encouraged. Publications and data products that make use of the dataset must include proper acknowledgement.

  • Citation:

    Thompson J, Cogbill C. 2013. Pre-Colonial and Modern Tree Data from Nine Northeastern States 1620-2008. Harvard Forest Data Archive: HF210.

Detailed Metadata

hf210-01: witness tree data

  1. unique.id: unique identification number for the town
  2. town.name: name of the town
  3. state: name of the state containing the town
  4. latitude: latitude of the centroid of the town (unit: degree / missing value: NA)
  5. longitude: longitude of the centroid of the town (unit: degree / missing value: NA)
  6. fia.plots: number of FIA plots within town boundary (unit: number / missing value: NA)
  7. trees.fia: number of FIA trees used to calculate relative abundance (unit: number )
  8. trees.wit: number of Witness Trees used to calculate relative abundance (unit: number )
  9. ashs.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  10. basswd.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  11. beech.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  12. birchs.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  13. blkgum.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  14. cedars.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  15. cherry.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  16. chsnut.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  17. cypres.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  18. elms.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  19. firs.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  20. hemlck.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  21. hickry.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  22. hornbm.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  23. magnol.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  24. maples.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  25. oaks.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  26. pines.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  27. poplar.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  28. spruce.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  29. sycmor.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  30. tamrac.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  31. tulip.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  32. walnut.wt: relative abundance of this taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  33. other.wt: relative abundance of other taxa in the Witness Tree data (unit: dimensionless / missing value: NA)
  34. ashs.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  35. basswd.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  36. beech.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  37. birchs.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  38. blkgum.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  39. cedars.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  40. cherry.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  41. chsnut.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  42. cypres.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  43. elms.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  44. firs.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  45. hemlck.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  46. hickry.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  47. hornbm.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  48. magnol.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  49. maples.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  50. oaks.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  51. pines.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  52. poplar.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  53. spruce.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  54. sycmor.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  55. tamrac.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  56. tulip.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  57. walnut.fia: relative abundance of the taxa in the FIA data (unit: dimensionless / missing value: NA)
  58. other.fia: relative abundance of other taxa in the FIA data (unit: dimensionless / missing value: NA)

hf210-02: witness tree GIS

  • Compression:
  • Format: zip
  • Type: zip