© The Nature Conservancy View/Download: Please click on the link(s) below and you will be given the choice of opening or saving to disk. JPG Download (JPG, 460 KB, Created January 01, 1999)

Ecological Land Units (ELUs) and STL Systems Raster Data

8/13/03, Charles Ferree, Conservation Science Support

Background

Conservation planning at any scale-regional, landscape level, or localrequires an understanding of patterns of environmental variation and biological diversity. This dataset was developed as a tool for assessing the biophysical character of landscapes, and for mapping the distribution and composition of community assemblages across those landscapes. Informed decisions on where to focus conservation efforts require such tools.

Data on biological distributions are very often inadequate to a large-scale analysis of biodiversity. The close relationship of the physical environment to ecological process and biotic distributions underpins the ecological sciences, and in the absence of suitable biological datasets, conservation science has recognized that physical diversity could be an acceptable surrogate for biological diversity. Research has repeatedly demonstrated especially strong links between ecosystem pattern and process and climate, bedrock, soils, and topography. This recognition led to the development of the ecological land unit, or ELU.

The ELU is a composite of several layers of abiotic information: elevation, bedrock geology, distribution of deep glacial sediments that mask bedrock's geochemical effects, moisture availability, and landform. An ELU grid of 30 meter cells was developed for the Saint Lawrence/Champlain Valley (STL) ecoregion. The ELU dataset describes the 'ecological potential' of the landscape, but carries no information about actual landuse or landcover in a region where human alterations to the landscape have everywhere affected the natural vegetation. The current dataset informs ELUs with landcover data, bringing them to earth by telling us what is actually on the ground. We may use this dataset to map ecological systems, which are dynamic assemblages of communities that occur in a mosaic on the landscape, and that are linked by shared ecological processes and environmental gradients. A brief discussion of each of the layers of information built into the current dataset follows.

Dataset content and development

Elevation classes:

Elevation has been shown to be a powerful predictor of the distribution of forest communities in the Northeast. Temperature, precipitation, and exposure commonly vary with changing altitude. We broke continuous elevation data for the STL ecoregion (from the National Elevation Dataset of the USGS) into discrete elevation classes with relevance to the distribution of forest types region-wide.

Meaningful biotic zones may well be defined with quite different elevation cut-offs in the northern and southern parts of the region, so class ranges necessarily approximate critical ecological values.



Table 1. Ranges for elevation classes.

Elevzone M (ft) Characteristic forest type

1000 0-243 (0-800) Oak, pine-oak, pine-hemlock, maritime spruce, floodplain

2000 243-518 (800-1700) Hemlock-N. hardwoods, N. hardwoods, lowland spruce-fir

3000 518-762 (1700-2500) Northern hardwoods, spruce-hardwoods

4000 762-1219 (2500-4000) Spruce-Hardwoods, Spruce-Fir

Bedrock geology and deep sediments

Bedrock geology strongly influences area soil and water chemistry. Bedrock types also differ in how they weather and in the physical characteristics of the residual soil type. Because of this, local lithology is usually the principle determinant of soil chemistry, texture, and nutrient availability.

Many ecological community types are closely related to the chemistry and drainage of the soils or are associated with particular bedrock exposures.

We grouped bedrock units on the bedrock geology maps of New York and Vermont into seven general classes designed to have particular relevance to vegetation distributions. We based our scheme on broad classification schemes developed by other investigators which emphasize chemistry and texture, and on bedrock settings that are important to many ecological communities, particularly to herbaceous associations. In some settings deep sediments of glacial origin mantle the bedrock. The consolidated bedrock of valleys of pro-glacial lakes, for example, may lie under many meters of fine lacustrine sediments, and deep coarse deltaic or outwash deposits often overlay the bedrock in pine barrens and sand plains in the northeast. In these settings it is the nature of the sedimentstheir texture, compactness, and moisture-holding capacity, their nutrient availability, their ability to anchor overstory trees in a wind disturbance--that is ecologically relevant, and not the nature of the underlying bedrock.

We used a USGS dataset of sediments of the glaciated northeast to identify such places The USGS map was compiled at a coarse scale (1:1,000,000), but we made the data smarter by informing it with our landform map. The landform layer was compiled at a much finer scale (the nominal scale of the digital elevation models used to construct the landforms is 1:24,000), and we allowed the deep coarse or fine sediments of the USGS dataset to be mapped only on those landforms on which they would naturally be expected to occur. In the case of sandy, coarse sediments, this would be in broad basin and valley/toe slope settings. We mapped fine clayey lacustrine or marine sediments in these same settings, and in addition on low convexities and lower sideslopes. The seven bedrock classes were numbered 100 through 700 (Table 2), and the coarse and fine sediments were numbered 800 and 900, respectively.

With the elevation, substrate, and landform layers, all the elements for assembling ecological land units, or ELUs, are in place. ELU code values for each cell in the region-wide grid are simply the summed class values for elevation zone, substrate, and landform for that cell. The last step in the assembly of the systems grid is the combination of ELUs with a grid of landcover data. The National Land Cover Dataset (NLCD: web site at http://www.epa.gov/mrlc/nlcd.html) was derived from Landsat-5 Thematic Mapper images for the conterminous United States, and is the only such dataset that covers the entire CAP ecoregion. We used elevation and landform information to clean up some systematic errors in the data (forested wetland pixels often appeared on northwest-facing sideslopes or slopecrests, for example), and grouped all human landuses into two classes, developed and agriculture. The 2-digit landcover codes (table 3) were multiplied by 10,000 and added to ELU codes the resulting 6-digit number is the systems code, and the grid value in the value attribute table (see a fragment from sys30cap.vat in Table 4 below).

Table 3. Landcover classes.

Class Description

11 Water

20 Developed

32 Quarry/mine/gravel pit

33 Open transitional

41 Deciduous forest

42 Coniferous forest

43 Mixed forest

80 Agricultural

91 Forested wetland

92 Emergent wetland

The systems grid comprises over 2500 unique combinations of landcover, elevation zone, substrate type, and landform. We added a sys30code item to the attribute table, and used it to construct a coding scheme that groups systems values into biophysical components. We conceive of these as building blocks for assembling and mapping ecological systems.

It is a simple matter to attach elevation distinctions to these systems units. For example, a dry oak-dominated ecosystem on hillsides of quartzitic rock or non-calcareous schist at 2700 feet and a similar but more mesic community at 1200 feet would both have a sys30code of 205 (see the attribute table for the sys30cap grid). Simply by adding the elevzone value to the sys30code, the two communities can be discriminated from one another. A sys30code of 4205 would then connote the higher elevation system, 2205 the lowland system. Versatility and flexibility in this coding scheme are key, because ecosystems will be defined and mapped differently-- that is, assembled from different combinations of biophysical elements-- in different ecoregions, and even in different parts of the same ecoregion.