Bioclimate information for the ecosystems model was provided by the generation of isobioclimate classes for the conterminous United States (Cress and others, 2009a) using the Rivas-Martínez methodology (Rivas-Martínez 2004). This methodology is based on the concept of establishing a quantifiable classification system which would closely relate the distribution of vegetation to climatic parameters and indices (Rivas-Martínez and others, 1999). This method first establishes bioclimatic indices calculated from various ranges of temperature and precipitation data, compares these indices to defined thresholds, and finally applies sets of decision rules to identify the climate classes. The complete climate classification is done in four levels: macrobioclimates, bioclimates, thermotypes, and ombrotypes. Thermotypes, which represent thermoclimatic belts, are identified using the positive annual temperature (Tp) thresholds or the compensated thermicity index (Itc) thresholds (Rivas-Martínez 2004, Rivas-Martínez and others, 1999, 2004). Ombrotypes, which represent ombroclimatic belts, are based on the ombrothermic index (Io) which is calculated as a function of both the total positive precipitation and temperature (Rivas-Martínez 2004, Rivas-Martínez and others 1999, 2004). For this national implementation the source data used for establishing the bioclimatic indices was Daymet. Daymet temperature and precipitation data was developed from 18-years (1980-1997) of climatological records and is available at a spatial resolution of 1 kilometer (Thornton, 1997). This implementation of the Rivas-Martinez methodology resulted in the generation of four climate layers for the conterminous United States: macroclimates, bioclimates, thermotypes, and ombrotypes.

However, the biophysical stratification approach used for the ecosystems modeling effort required a single climate layer. Therefore, the final step in this implementation was the generation of a single isobioclimate based on some utilization of the four Rivas-Martinez climate classifications. Based on a series of investigations the decision was made that in order to achieve the required climate variations, especially in the upper mid-west and north-west, while still working with a manageable number of classifications, the best choice for this final isobioclimate dataset would be a combination of the thermotypes and ombrotypes. After this process was performed the final isobioclimates layer was generated that provides the required amount of climate variation, and contains 127 unique thermotype-ombrotype combinations.

The isobioclimates map shows ombrotypic regions (dry/wet gradients) for each thermotypic (warm/cold) region.

Land surface forms for the conterminous United States were also generated as part of the ecosystems modeling effort (Cress and others, 2009b). After extensive investigation into various land surface form methodologies, the decision was made to use the methodology developed by the Missouri Resource Assessment Partnership (MoRAP). MoRAP made modifications to Hammond's (1964a, 1964b) land surface form classification, which allowed the use of 30-meter source data and a 1-km2 window for analyzing the data cell and its surrounding cells (neighborhood analysis) (True, 2002; True and others, 2000). While Hammond's methodology was based on three topographic variables, slope, local relief, and profile type, MoRAP's methodology uses only slope and local relief (True, 2002). Using the MoRAP method, slope is classified as gently sloping when more than 50 percent of the area in a 1-km2 neighborhood has slope less than 8 percent, otherwise the area is not gently sloping, that is, steeply sloped. Local relief, which is the difference between the maximum and minimum elevation in a neighborhood, is classified into five groups: 0-15 m, 15-30 m, 30-90 m, 90-150 m, and >150 m. The land surface form classes are derived by combining slope and local relief which creates eight landform classes: flat plains (gently sloping and local relief <= 15 m), smooth plains (gently sloping and 15 m < local relief <= 30 m), irregular plains (gently sloping and 30 m < local relief <= 90 m), escarpments (gently sloping and local relief > 90 m), low hills (not gently sloping and local relief <= 30 m), hills (not gently sloping and 30 m < local relief <= 90 m), breaks/foothills (not gently sloping and 90 m < local relief <= 150 m), and low mountains (not gently sloping and local relief > 150 m). However, in the USGS application of the MoRAP methodology, an additional local relief group was used (> 400 m). As a result, low mountains were redefined as not gently sloping and 150 m < local relief < 400 m, and a new land surface from class, high mountains/deep canyons, was identified as not gently sloping and local relief > 400 m. The final application of the MoRAP methodology was preformed using the USGS 30-meter National Elevation Dataset (NED), and an existing USGS slope dataset that had been derived by calculating the slope from the NED in Universal Transverse Mercator (UTM) coordinates in each UTM zone, and then combining all of the zones into a national dataset.

As a final step, an additional class called drainage channels was derived independently from the other land surface form classes and was used to identify wet and dry river channels, which were not specifically represented by any of the other classifications. This class was based on the two Andrew Weiss's slope position classes that represented areas positioned lower than the surrounding areas, valley and lower slope (Weiss, 2001; Jenness, 2006). The USGS applied Weiss's algorithm to the 30-meter NED data using a 1-km2 neighborhood analysis window and then the resultant drainage channel class was added to the land surface forms dataset. Thus, the final land surface forms dataset contains ten classes: flat plains, smooth plains, irregular plains, escarpments, low hills, hills, breaks/foothills, low mountains, high mountains/deep canyons, and drainage channels.

The topographic moisture potential for the conterminous United States was derived to help contribute substrate moisture regimes into the ecosystems model (Cress and others, 2009c). The method used to produce these topographic moisture potential classes was based on the derivation of ground moisture potential using a combination of computed topographic characteristics (CTI, slope, and aspect) and mapped National Wetland Inventory boundaries. This method does not use climate or soil attributes to calculate relative topographic moisture potential since these characteristics are incorporated into the ecosystem model though other input layers. All of the topographic data used for this assessment was derived from the USGS 30-meter NED including the National Compound Topographic Index. The CTI index is a topographically derived measure of slope for a raster cell and the contributing area from upstream raster cells, and thus expresses potential for water flow to a point. In other words CTI data are "a quantification of the position of a site in the local landscape", where the lowest values indicate ridges and the highest values indicate stream channels, lakes and ponds (USGS, 2003). These CTI values were compared to independent estimates of water accumulation by obtaining geospatial data from a number of sample locations representing two types of NWI boundaries, Freshwater Emergent Wetlands and Freshwater Forested/Shrub Wetlands. Where these shorelines (the interface between the NWI wetlands and adjacent land) occurred, the CTI values were extracted and a histogram of their statistical distributions was calculated. Based on an evaluation of these histograms, CTI thresholds were developed to separate periodically saturated or flooded land, mesic uplands (moderately moist), and uplands. After the range of CTI values for these three different substrate moisture regimes was determined, the CTI values were grouped into three initial topographic moisture potential classes. As a final step in the generation of this national data layer, the uplands classification was further broken down into either very dry uplands or dry uplands. Very dry uplands were defined as uplands with relatively steep, south-facing slopes, and identification of this class was based on the slope and aspect datasets derived from the USGS NED. The remaining uplands that did not meet these additional criteria were simply re-classified as dry uplands. The final National Topographic Moisture Potential dataset for the conterminous United States contains four classes: periodically saturated or flooded land (CTI >= 18.5), mesic uplands (12 <= CTI < 18.5), dry uplands (CTI < 12), and very dry uplands (CTI < 12, Slope > 24 degrees and 91 degrees <= Aspect <= 314 degrees).

The surficial lithology classes for the conterminous United States (Cress and others, 2010) were derived from the USGS map "Surficial Materials in the conterminous United States", which was based on texture, internal structure, thickness, and environment of deposition or formation of materials (Soller and Reheis, 2004). This original map was produced from a compilation of regional surficial and bedrock geology source maps using broadly defined common map units for the purpose of providing an overview of the existing data and knowledge. For the terrestrial ecosystem effort, the 28 lithology classes of Soller and Reheis (2004) were generalized and then reclassified into a set of 18 lithologies that typically control or influence the distribution of vegetation types (Kruckeberg, 2002).

The structural footprints for the conterminous United States were produced by spatially combining each of the major structural components of ecosystems described above. These ecosystem structure units characterize the abiotic (physical) potential of the environment, and represent distinct facets of the landscape with respect to biogeography, bioclimate, landform, lithology, and surface moisture potential.

This layer was developed from a spatial combination of each of the layers described above, and an Ecological Divisions of the United States data layer (Comer and others, 2003). NatureServe's North American Ecological Divisions, twelve biogeographic regions for the conterminous United States, was used to represent sub-continental landscapes reflecting both climate and biogeographic history (Comer and others, 2003). The final set of ecosystem structural footprints was generated from a spatial combination of all of the base layers, and then to minimize artifacts, the final footprints dataset was aggregated to a minimum mapping unit of approximately two acres, resulting in 49,168 unique structural footprints.

The U.S. Geological Survey (USGS), with support from NatureServe, has modeled the potential distribution of 419 terrestrial ecosystems for the conterminous United States (Sayre and others, 2009) using a classification developed by NatureServe (Comer and others, 2003). A biophysical stratification approach, developed for South America (Sayre and others, 2008) and now being implemented globally (Sayre and others, 2007), was used to model these terrestrial ecosystem distributions. This biophysical stratification approach is based on mapping the major structural components of ecosystems (landform, surface moisture potential, surficial lithology, bioclimate and biogeographic region) and then spatially combining them to produce a set of unique biophysical environments.

Terrestrial ecosystems for the conterminous United States were mapped by delineating physically distinct areas as the fundamental structural units ("building blocks") of ecosystems, and subsequently aggregating and labeling these structural footprints into the NatureServe classification. These structural footprints were developed from the union of several base layers including: biogeographic regions, isobioclimates (Cress and others, 2009a), land surface forms (Cress and others, 2009b), topographic moisture potential (Cress and others, 2009c), and surficial lithology (Cress and others, 2009d). Among the 49,168 unique structural footprint classes that resulted from the union, 13,482 classes met a minimum pixel count threshold (20,000 pixels), and were aggregated into 419 NatureServe ecosystems using a semi-automated labeling process based on rule-set formulations for attribution of each ecosystem.

The resulting ecosystems are those that are expected to occur based on the combination of the bioclimate, biogeography, and geomorphology. Where land use by humans has not altered land cover, natural vegetation assemblages are expected to occur, and these are described in the ecosystems classification. The map does not show the distribution of urban and agricultural areas—these will be masked out in subsequent analyses to depict the current land cover in addition to the potential distribution of natural ecosystems.