USGS:Biology Biological Informatics Program Vegetation Characterization Program Standardized National Vegetation Classification System 5.0 A Standard National Vegetation Classification System

5.0 A Standard National Vegetation Classification System

5.1 Characteristics of the National Vegetation Classification System

5.1.1 Based on Existing Vegetation

The national vegetation classification system focuses on existing vegetation rather than potential natural vegetation, climax vegetation, or physical habitats (see Section 3). The vegetation types covered in the classification range from the short-lived to relatively stable and persistent plant communities. The classification includes natural, seminatural, modified, and cultural vegetation. The temporal and spatial variation in communities is an intrinsic property of the vegetation itself and, therefore, critical to the protection of biodiversity and landscape dynamics. Not restricting the classification to stable vegetation types ensures the units are appropriate for inventory and site description, and provide the level of detail required to build ecological and landscape models.

5.1.2 Combined Physiognomic Floristic Classification Approach

The national terrestrial vegetation system is hierarchical and combines physiognomy at the highest levels of the hierarchy and floristics at the lowest levels. This classification approach was chosen to allow the characterization of vegetation patterns at multiple scales. The combined physiognomic floristic system allows identification of units from both a divisive ("top-down") and agglomerative ("bottom-up") approach. The top-down approach allows the use of physiognomic distinctions to help map vegetation, to stratify sampling, and, where floristic information is lacking, to delimit vegetation units. The bottom-up approach requires that plot sampling and floristic analysis are the primary means for defining communities. Where physiognomy is variable, the bottom-up approach can also be used to help determine the important physiognomic distinctions.

The basic unit of the vegetation classification is the "community element" which is defined as an individual plant association or repeating complex of plant associations. These associations have definite floristic composition, uniform physiognomy, and represent uniform habitat condition (see Flahault and Schroter 1910). The community element concept is similarly related to the plant association concept used in the Zurich-Montpellier tradition (see above). These floristic units are characterized as patterns of co-occurring species that recur either in space or time under similar environmental conditions.

In the field, community elements are recognized as structurally and floristically homogeneous stands of vegetation that occur in a relatively uniform environmental setting. As a result of the individual species distribution patterns (the continuum concept) and the environmental complexity across the landscape, there is considerable variation within a community type across environmental gradients and the landscape. The vegetation communities can be defined as homogeneous stands of vegetation on the ground, but individual occurrences of a particular plant association will vary in species compositions and structure.

The floristic units are arranged under a hierarchy based on physiognomic characteristics of their dominant vegetation. This physiognomic hierarchy is a modification of UNESCO (1973) and Driscoll, et al. (1984), and utilizes the physical form of the dominant vegetation to organize the floristic units (see below).

5.1.3 Role of the environment

An underlying assumption of national vegetation classification system is that vegetation is the best and most easily measured assimilator of complex environmental and historical site conditions. Although the classification units are defined by vegetation only, the concept of a community as an ecological unit includes all the biological and physical diversity associated with that specific vegetation type. For example, a herbaceous woodland "serpentine barren" plant community (scientific name: Pinus [virginiana, rigida]/Schizachyrium scoparium alliance) actually describes the unique geologic setting in which it is found, the rare insects associated with the vegetation, and the fire disturbance history that maintains the community.

The community elements of the national vegetation classification system are related to a set of environmental factors rather than to a particular site. This ensures a consistent ecological meaning for the community level of the classification across a broad geographic range. Environmental parameters are measured with the floristic units to develop this correlation with the ecological reality. When the classification is mapped across a site, the distribution of community elements provides a basis for interpreting the ecological and land use processes across the landscape.

5.2 Description of the Levels of the Terrestrial Vegetation Classification Hierarchy

The national vegetation classification system has seven levels. The top level of the hierarchy identifies whether the community is terrestrial, aquatic, or subterranean. For the classification of natural and seminatural terrestrial vegetation, the next four levels describe physiognomic characteristics, and the last two levels describe the floristics. The levels are

5.2.1 System Level

The top division of the classification hierarchy separates vegetated communities (Terrestrial System) from those of unvegetated deep-water habitats (Aquatic System) and unvegetated subterranean habitats (Subterranean System). The Terrestrial System of the national hierarchy is very inclusive. It includes the vegetation of uplands, the emergent and rooted submerged vegetation of lakes, ponds, rivers, and marine shorelines, and the sparsely vegetated and nonvegetated communities. In relation to Cowardin et al. (1979), this system includes those portions of the palustrine, lacustrine, riverine, estuarine, and marine systems that have rooted vegetation.

Communities of the Aquatic System lack rooted vegetation and are generally described as having fish, macroinvertebrates, algae, and corals. The Aquatic System includes the nonvegetated (faunal) and vegetated communities of the Cowardin et al. (1979) marine, estuarine, riverine, and lacustrine systems beyond the limits of rooted vegetation. The

Subterranean System includes terrestrial cave communities which are generally described using the dominant fauna.

There are different hierarchical divisions below each of the three systems. The hierarchy for the Terrestrial System is structurally complete. It has six levels, with four physiognomic levels (physiognomic class, physiognomic subclass, formation group, and formation) and two floristic levels (alliance and community element). The hierarchical levels of the Aquatic and Subterranean classification systems are in different stages of development, and the marine component is also near completion.

For the purpose of the NPS/NBS Vegetation Mapping Project, the Aquatic System (e.g., freshwater streams and rivers, lakes, reservoirs) will be classified and mapped at a coarser level of detail than the communities in the Terrestrial System (see Section 5.2.4.2 below).

5.2.2 Physiognomic Levels

The physiognomic portion of the national vegetation classification hierarchy is a modification of the UNESCO world physiognomic classification of vegetation (1973) and incorporates some of the revisions made by Driscoll et al. (1984) for the United States.

The UNESCO vegetation classification system uses physiognomy (outward appearance) and structure of the vegetation to define the units. It is intended to provide a comprehensive framework for the preparation of vegetation maps at a scale of 1:1 million or smaller. The system was designed to include all natural and seminatural vegetation, but "cultural" vegetation (wheat fields, vineyards, etc.) is not included.

The UNESCO physiognomic system was incorporated as the physiognomic base for the hierarchy for the following reasons:

• It is one of the few classification systems already in place that could be employed with relatively little research and development cost.
• It is already the product of an international group of experts. As a result, it is worldwide in coverage and a more readily acceptable product than local and single-authored systems. Parts or variants of the system are presently being used by different United States and international agencies.
• It is ecologically meaningful.
• It is a hierarchical system that was designed for classification and mapping at multiple scales.
• The structure of the system makes it open-ended; units can be added as needed.

5.2.2.1 Modifications to the UNESCO Hierarchy

The UNESCO system has now been modified and refined to provide greater consistency at all hierarchical levels and includes additional physiognomic types. Several limitations of the UNESCO hierarchy prevented an unmodified application to the national vegetation classification system. As an example, there was little supporting information to explain the criteria used to define each hierarchical level. In addition, the same criteria were used at different levels to define the units. Finally, there were several vegetation formations, such as wetlands, that were not adequately represented in the original UNESCO system.

In particular, the "subclass level" of UNESCO has been modified to better conform to the Federal Geographic Data Committee's standards for vegetation classification. The UNESCO system has also been adjusted by including more explicit hydrological modifiers at the formation level. The hydrologic modifiers introduced by Cowardin et al. (1979) were explicitly adopted since these have been used extensively to map wetlands across the United States (Appendix 9.3). The levels are outlined in the following sections. See Appendix 9.1 for a complete version of the hierarchy.

5.2.2.2 Physiognomic Class

The Physiognomic Class is based on the structure of the vegetation. This is determined by the height and relative percentage of cover of the existing tree, shrub, dwarf shrub, and herbaceous strata (Figure 1). This level has nine mutually exclusive classes:

5.2.2.3 Physiognomic subclass

The Physiognomic Subclass is determined by the predominant leaf phenology of classes defined by a tree, shrub, or dwarf shrub strata (evergreen, deciduous, mixed evergreen-deciduous), the average vegetation height for types defined by the herbaceous stratum (tall, medium-tall, short), and particle size of the substrate for sparsely vegetated and nonvascular communities (e.g. consolidated rocks, gravel/cobble, sand accumulations, mud flats).

Examples:

• Evergreen forest
• Deciduous forest
• Mixed Evergreen Deciduous forest
• Tall grassland
• Medium-tall grassland
• Short grassland
• Sparsely vegetated sand accumulations

5.2.2.4 Formation Group

The units for the Formation Group are based largely on a combination of climate, leaf morphology, and leaf phenology. In addition to climate and leaf characteristics, the groups for the sparse woody classes (i.e., sparse woodland, sparse shrubland, and sparse dwarf shrubland) are defined by the dominant lower stratum.

Examples:

• Temperate evergreen needle-leaved woodland
• Broad-leaved evergreen sparse shrubland with a dominant herbaceous stratum
• Polar short grassland

5.2.2.5 Formation

The Formation represents an ecological grouping of vegetation units based on broadly defined environmental factors such as elevation and hydrologic regime, and additional structural factors such as crown shape, and Examples

• Tropical or subtropical seasonal montane evergreen forest
• Seasonally/Temporarily flooded medium-tall grassland
• Needle-leaved evergreen woodland with rounded crowns
• Broad-leaved evergreen sparse shrubland with tall graminoids

5.2.3 Floristic Levels

5.2.3.1 Alliance

The Alliance is a physiognomically uniform group of plant associations (see Community Element below) sharing one or more diagnostic species (dominant, differential, indicator, or character), which, as a rule, are found in the uppermost strata of the vegetation (see Mueller-Dombois and Ellenberg 1974).

The Alliance is roughly equivalent to the "cover type" of the Society of American Foresters (Eyre 1980), although it is not restricted to describing forest cover. The Alliance may be finer in detail than a cover type when the dominant species extend over large geographic areas and varied environmental conditions. The Alliance is also similar in concept to the "Series." Alliances, however, are described by the diagnostic species for all existing vegetation types, whereas series are restricted to climax types and are described by the primary dominant species (see Pfister and Arno 1980).

Examples

• Acer rubrum - Liquidambar styraciflua Forest Alliance
• Pseudotsuga menziesii Woodland Alliance
• Juniperus osteosperma Sparse Woodland Alliance

See Appendix 9.1 for the list of known Alliances in the United States.

5.2.3.2 Community Element

The Community Element is the finest level of the classification system. For the Terrestrial System, the community element is defined as an individual plant association or repeating complex of plant associations. These associations have definite floristic composition, uniform physiognomy, and represent uniform habitat condition (see Flahault and Schroter 1910, Third International Botanical Congress 1910). This basic concept has been used by most of the schools of floristic classification (Braun-Blanquet 1965, Westhoff and van der Maarel 1978). The plant association concept applies to existing vegetation regardless of successional status.

The definition of the community element can be clarified with the following points:

• "Habitat" refers to the combination of environmental conditions and ecological processes influencing the community.
• Uniformity of physiognomy and habitat conditions may include patterned heterogeneity (e.g., hummock/hollow).
• As a rule, community elements recur over the landscape.
• The scale of the community element varies. Among other factors, the variation is determined by the size and apparent homogeneity of the occurrences across the landscape, the amount of data that has been collected, and the interpretation of these data by the field experts.
• The community element may be composed of a complex of plant associations that constitutes a functioning ecological unit if the plant associations always occur together (e.g., prairie pothole).
• The terms "community element" and "plant association" are both used to refer to the community element.
• The community element is differentiated from the Alliance level by additional plant species, found in any stratum, which indicate finer scale environmental patterns and disturbance regimes. This level is derived from analyzing complete floristic composition of the vegetation unit when plot data are available. In the absence of a complete data set, approximation of this level is reached by using available information on the dominant species, indicator species, and environmental modifiers.

Examples

• Acer rubrum-Liquidambar styraciflua-Populus heterophylla Forest
• Pseudotsuga menziesii/Festuca idahoensis Woodland
• Juniperus osteosperma/Stipa comata Sparse Woodland

See Appendix 9.2 for examples of community elements organized under the classification hierarchy.

5.2.4 Cultural Land Cover

5.2.4.1 Agricultural Land Cover

The national vegetation classification system classifies agricultural land cover using the Federal Geographic Data Committee (FGDC) Vegetation Subcommittee's recommended system for Cultivated Vegetation (Table 3). The FGDC system is still under development, and the national classification system will evaluate further changes that may be made to these classes.

5.2.4.2 Urban Land Cover and Water

The national vegetation classification system presently classifies and maps urban land cover and water at a coarser level of detail than natural and seminatural vegetation types. The system employs the land use and land cover (LULC) classification system developed by Anderson et al. (1976) for attributing Urban and Water dominated land cover. Urban, or "built up" land, water classes are attributed at Level II of Anderson's system (see Table 4). Anderson's LULC system is a widely accepted system used throughout many federal, state and local agencies. It was developed for use with remote sensor data.

Table 4. Anderson's Land Use and Land Cover Classification System

This portion of the Anderson et al. system was adopted for the national vegetation classification system to map cultural land cover (Anderson et al. 1976).

5.3 Nomenclature Standards

Each Alliance and community element is assigned a name based on the scientific names of the diagnostic species that have a high degree of constancy. To ensure consistency of plant species nomenclature, the plant species names follow the standards developed by Kartesz (1994). Provisional community names are updated as additional information becomes available.

In the Alliance and community element names, plant species used in the name occurring in the same stratum are separated by the "-" symbol, and those occurring in different strata of the vegetation are separated by the "/" symbol (e.g., Quercus macrocarpa/Corylus cornuta-Corylus americana Woodland). In those cases where the diagnostic species are unknown or in question, environmental modifiers or broad vegetation or geographic modifiers are used as placeholders until the diagnostic species become known with more certainty (e.g., Pinus palustris-Pinus echinata/Schyzachyrium scoparium Serpentine Woodland).

As a rule, the diagnostic species for Alliances are consistently present (constant) in the community elements within the Alliance and the diagnostic species for community elements are consistently present in occurrences of the community. There are, however, certain situations where a diagnostic species is not consistently present in community elements within an alliance or in occurrences of a community element. When this happens, the species that are not consistently present in the community element or occurrences are placed in parentheses. For example Pinus ponderosa-(Pinus flexilis) Alliance means Pinus ponderosa is present in most of the associations while Pinus flexilis is not.

Some alliances have also been documented in which the associations share two diagnostic species, but neither of the diagnostic species are consistently present in all associations. In this situation both of the diagnostic species names are put in parentheses. For example, Pinus (ponderosa-flexilis) Alliance means that both species are not necessarily present in all of the associations, but at least one of them is present.

5.4 Development of the National Vegetation Classification System

5.4.1 Development of the Floristic Classes (Alliances and Community Elements)

Development of the floristic classes (Alliances and community elements) is an iterative qualitative and quantitative process. The majority of the floristic units presently defined in the classification system are the result of rigorous qualitative assessments due to the lack of quantitative data. The long-term goal for the national classification system is the determination of all floristic units through the quantitative analysis of consistent plot data. Field data (species lists and environmental information) will be prioritized and collected over time in order to verify the classification of many provisional types.

5.4.1.1 Qualitative Assessment

Qualitative assessments of existing information are completed to identify and describe provisional community elements. This process includes the compilation of existing state classifications and vegetation information from the literature and other sources. The vegetation units are placed into the physiognomic hierarchy based either on qualitative or quantitative description of structure and species composition. Alliances and community elements are named and described based on the qualitative assessment of patterns of diagnostic species. Groups of ecologists are required to develop and review these classification units. Problematic classifications and high-priority elements are targeted as a focus for future data acquisition and quantitative analyses to refine the classification of these types.

5.4.1.2 Quantitative Analysis

The process of quantitative assessment of the floristic elements includes the compilation and assessment of existing stand table and summary data on the community element and related types across the entire range of occurrence. Collection of additional field data is often required to support a robust analysis of the community. The resulting classifications are then sent out for peer review by appropriate experts (federal, state, and academic ecologists). Throughout this process the goal is to ensure consistent quality control of the data and application of the quantitative techniques.

Stand and summary data appearing in journal articles and published and unpublished reports are used extensively for the development of community elements. For a reference to a particular plant association to be included in the analysis, its source must provide location information, description of methods, species lists and quantitative measure of species abundance values. Primary data are collected by the Natural Heritage network and other researchers on community types that are undersampled and of high priority. Data collection is carried out by Natural Heritage and Conservancy ecologists using a standard relevé methodology (Sneddon 1992, Bourgeron et al. 1991 and see above).

Compiled data are assembled into a single file and transformed mathematically to a common abundance scale. The element classification process is implemented using quantitative approaches of ordination, clustering, and correlation depending on the information available. Three multivariate analysis programs, TWINSPAN, DECORANA, and CANOCO, are particularly useful in examining the floristic patterns and their relationships to measured environmental variables (Hill 1979, Hill and Gauch 1980, ter Braak 1990).

Despite their utility in synthesizing large data sets, many of the analytical programs identify vegetation patterns that are statistically but not ecologically meaningful. The quantitative analysis to determine vegetation patterns must be carried out under the guidance and review of experts who have a practical understanding of the ecological relationships in the field.

5.4.1.3 Confidence Levels

Each community type is assigned a "confidence level" that is determined by the amount and type of information available and the analysis methods used to define it (Table 5). These confidence levels help to identify where additional information will be required for the refinement of the classification. As additional field data become available, the classification is updated and the confidence levels reevaluated.

5.4.1.4 An Example of the Development of Floristic Classes - Pine Barrens

To refine the classification of pine barrens communities and help identify conservation and management priorities, The Nature Conservancy initiated a classification and mapping project at the Waterboro barrens in York County, Maine. This project involved the collection of data on all communities in a single pine barren site and relating these data to the information available on pine barren communities at a regional scale.

Local (Intensive) Analysis of a Single Pine Barren Site

Waterboro is an expansive pine barren site which occurs on sandy, nutrient poor, outwash soils in southern Maine (Harris 1991). The mosaic of communities that occur within the site exhibit a wide range of composition and structure. This reflects the complex of climate, terrain, hydrology and historical factors present at the site.

A set of stereo aerial photos was obtained for the Waterboro Barrens site. Boundaries of vegetation units were delineated on the photos using the criteria of texture (smoothness or coarseness of the image), tonal contrast, and topographic location (Avery 1977), and these boundaries were transferred to a 1" by 500' topographical map. Three 10 m x 10 m plot samples were taken from representative areas within each vegetation type identified from photointerpretation. Particular attention was paid to the pitch pine - shrub oak vegetation types. Within each plot, one nested 5 m x 5 m quadrat was used to sample the understory vegetation and two 1 m x 1 m quadrats were used for sampling herbaceous vegetation. Information collected for each plot followed The Nature Conservancy standards reviewed above (e.g., species composition and abundance, soil texture, slope ). A preliminary community classification was developed from this information, and a community map of the site was produced.

Regional (Extensive) Analysis of Each Community Type

Plot data for each community type occurring within the Waterboro Barrens complex were then compared with data from similar communities across the region. The analysis of the pitch pine - scrub oak community, for examples, benefited from a large data set (224 samples) that was assembled from published (Olsvig 1980, Ollson 1979, Milne 1985, Patterson 1984, McIntosh 1959) and unpublished literature (Pesiri, Latham, Tucker, Seichab, Harris, State Natural Heritage Program field forms for Massachusetts, Connecticut, Pennsylvania, New Hampshire, and Rhode Island). Each sample was collected from a standard plot placed in a vegetation association exhibiting a canopy of pitch pine over an understory of scrub oak and contained lists and abundance of all species (bryophytes and lichens excluded). Each sample was transformed into a common, four-category abundance scale to normalize the data. An arranged species-by-sample table illustrating patterns of floristic association was produced using TWINSPAN (Hill 1979), and the floristic patterns were circulated widely among state ecologists for review. Based on the discussions and comments of the reviewers, the table was manually rearranged until an agreement was reached on the ecological meaning of the floristic associations.

For the pitch pine scrub oak communities, the relationships between the floristic patterns and the ecological variables were examined quantitatively using CANOCO. Although the only consistent environmental data available for every sample were latitude, longitude and elevation, the CANOCO analysis confirmed that these variables explained a large proportion of the variation in the data. This was reconfirmed through a DECORANA ordination followed by a nonparametric correlation (Spearmans rank) between the axis scores and the environmental variables. A reassessment of the existing pine barrens literature, in light of the new classification scheme, was also very useful in elucidating the relationships between vegetation and environment.

This process was repeated for other communities found within the Waterboro Barrens complex until each sample was classified within the national context. The overall species composition and diagnostic species, associated environmental variables, typical structure, and the range of physiognomic expressions were examined and the distribution, range, and global rank of each community was determined. The information was then used to refine the classification attributes on the vegetation map with the regional classification names.

5.4.2 Arranging the Floristic Units under the Physiognomic Levels of the Hierarchy

Once defined, the floristic units are fit into the physiognomic structure of the hierarchy based on their physiognomic expression across all stands. In some cases, communities may exhibit different physiognomic expressions without a concurrent shift in species composition. In these cases, the physiognomic group is determined by the most common expression of the community as opposed to a theoretically stable expression. Where floristic and physiognomic groupings do not correspond, precedence is given to the floristic relationships over the physiognomic structure. Types that present more than one physiognomic expression are cross-referenced in the hierarchy.

5.4.3 Adding New Vegetation Types Identified during the Course of the NPS/NBS Mapping Project

The present classification is a dynamic product that has been developed through the continuous review of literature, communications with local and regional experts, directed field examinations, and some quantitative analyses. All of the units have been derived through consistent application of classification rules using available vegetation data and associated environmental information. The NPS/NBS vegetation mapping project will provide a large amount of additional information that will allow further refinement of the classification. The classification system will evolve to reflect the growing body of knowledge concerning the biology, ecology, and geography of the different vegetation types. Many new vegetation types may be added to the classification and some current types may be split into new types, while others may be lumped together.

For new types to be added to the classification, they must contain significantly different biotic composition, be associated with different environmental conditions, and be documented to recur across the landscape. They also must be compared to information on related types from a rangewide perspective to ensure it is not a local variant of a community already classified. For the NPS/NBS vegetation mapping project, suspected new types will be evaluated, qualitatively or quantitatively assessed depending on the level of available data.

It is the responsibility of The Nature Conservancy to review the process for the classification of all new types. The recent designation of a special Ecological Society of America panel for Vegetation Classification should provide assistance in the formal review of new vegetation types in the near future.

5.4.3.1 Qualitative Assessment

When a vegetation type is discovered that may be new to the classification, it will be reviewed using the process described above for qualitative assessment of the floristic classes (Section 5.4.1.1). New data collected for the suspected new type and any existing plot data from the park will be assessed by The Nature Conservancy regional ecologists. The type will be placed under the classification hierarchy and compared to the information available for closely related types. If the type is still suspected to be new, it will be described by the regional ecology staff. This description will be circulated to the state Natural Heritage ecologists, other regional ecologists, and other experts. If the experts agree there is sufficient reason to believe the type is new, it will be named and assigned a confidence level of "3 weak" or "2 moderate" depending on the amount of available data.

The information generated on the new type will be disseminated from the Conservancy regional ecologist to all field ecology and photo-interpretation teams in each of the parks that could potentially contain the new type. The photo interpreters will incorporate the new type into their photo-interpretation keys at the park level (if this has not been done already). New types will be reviewed by the experts, classified and described before the final maps are produced for the park in which it was discovered.

5.4.3.2 Quantitative Assessment

If there is significant interest or need to quantitatively assess a new vegetation type or group of types, the process described in Sections 5.4.1.2 and 5.4.1.3 will be followed. The literature and other sources of data (including existing plot data from parks) will be searched, and all available stand table data will be compiled and assessed. Additional primary data will be collected where necessary. The entire data set will be analyzed, and the results will be reviewed by the experts on the type.

5.5 Ecological Considerations and Variability

Plant communities need to be recognized over the entire range of environmental variability (Austin 1991, Bourgeron et al., 1994). The identification of community elements is performed to identify distinct floristic patterns with a clear ecological meaning. Thus, an important step to the classification itself is the identification of ecological factors that determine the vegetation patterns. Vegetation types are characterized by the co-occurrence of individual species as constrained by environmental features (e.g., climate, geomorphic, and edaphic factors), the dynamics of biotic processes (e.g., immigration, emigration, competition), and disturbance. The relationship between these factors and the vegetation patterns is often complex. Interpretation of the ecological meaning of the units is completed, in part, through qualitative understanding of the ecology, and wherever possible the quantitative analysis of correlations between species and a set of environmental factors. To understand these relationships, the literature on community processes, plant demography, reproductive biology, physiology, geography, must be consulted.

5.5.1 Homogeneity

Although some ecologists have identified shortcomings in the restriction that plant associations must be defined from homogeneous units (Noss 1987), floristic and physiognomic uniformity has been generally widely accepted as a valid criterion in the definition of floristic units (Mueller-Dombois and Ellenberg 1974). The criterion for homogeneity is particularly important when sampling vegetation for use in numerical analysis. On the landscape, however, existing vegetation is often transitional in nature. As a rule the national vegetation classification system does not presently recognize transitional areas or ecotones between two types as distinct elements in the classification. Two special exceptions to the classification rules deserve explicit attention: community complexes and gradients.

5.5.1.1 Community Complexes

Communities often occur as a fine-grained mosaic of interrelated, but distinct, floristic associations. Classification of these community complexes can be problematic as many, especially those with intrinsic microtopographical variation, are inseparable in any definable or useful fashion. These situations may occur in both random unpatterned fashion or as small-scale patterned heterogeneity (e.g., hummocks and hollows in bog situations might share some species, but have largely different dominants). When these situations occur, the complexes of plant associations are defined as a single community element. In these cases the patterning is described as attributes of the community complex.

5.5.1.2 Gradients

The composition of most communities reflects the distribution of individual species over multiple environmental gradients (Austin and Smith 1989). Deciding the optimal place along the major gradients to partition the continuum of change is one of the fundamental questions of classification theory. While in some cases the data are naturally clustered, in others several possible divisions of the data are justifiable. The final choices as to where to draw the line between related communities are driven by interpretation of the patterns by field experts and the objectives of the research.

5.5.2 Disturbance

Disturbance processes have a profound influence on the character and composition of vegetation. Broad-scale natural disturbances such as hurricanes, fire, flooding, avalanches, and disease as well as chronic small-scale disturbances such as hydrologic variation, tree-fall, animal digging, and herbivory often explain the variations in existing vegetation better than many of the traditionally measured ecological factors. A number of anthropogenic disturbances, such as clearing, plowing, grazing, development, and nutrient enrichment, have also affected existing vegetation patterns. These anthropogenic disturbances may simulate natural disturbances, create entirely new disturbance regimes or alter natural disturbance regimes (e.g. fire suppression). Often only circumstantial evidence is available to estimate the disturbance regime associated with a particular vegetation type.

Some disturbances, whether natural or anthropogenic, can cause alterations in the structure and composition of an occurrence of a community. If the disturbance is severe enough to alter the structure and floristic composition of a community on the ground, the classification of that unit may change. Following a catastrophic fire, for example, a Jack Pine/Blueberry Forest (Scientific name: Pinus banksiana/Vaccinium spp. Forest) may become a Jack Pine (Northern Pin Oak)/Little Bluestem Sparse Woodland (Scientific name: Pinus banksiana (Quercus elipsoidales)/Schizachyrium scoparium Sparse Woodland).

In contrast, some disturbance regimes may alter the structure or composition only moderately and the community may still fall within the range of acceptable variation for the type. Since most communities are identified by groups of diagnostic species rather than single diagnostic species, small-scale disturbances that cause minor changes in the floristic composition of the type are often not severe enough to change the classification of the unit. For example, selective logging techniques may extract Jack Pine from occurrences of a Jack Pine/Blueberry Forest. If the rest of the species composition of the community remain, the loss of only the Jack Pine may not be enough to consider the community as a different type.

5.5.3 Succession

Successional stages are treated like any other existing vegetation type. Once the structure and composition of a community reaches a stable state that is physiognomically and floristically different from its previous successional stage, it is considered a different community in the classification. In developing the classification, particular emphasis is placed on understanding how the species composition relates to a particular successional process.

Floristic analysis of many successional vegetation types can reveal that a type is an unusual physiognomic expression of an existing community element. In the eastern region, for example, the Quercus ilicifolia (shrub oak) thickets that develop in areas of frequent fire share an identical species composition with the Pinus rigida (pitch pine) Quercus ilicifolia (scrub oak) barrens with which they typically occur. Both of these types are considered subcommunities of the major community element, though only the fire-maintained type does not contain unique species.

5.6 The Relationship Between the National Vegetation Classification System and Other Classification Systems

The national vegetation classification system was developed with the knowledge that it would need to be related to other major classification approaches. Cross-references to other major classification systems are currently being developed. In the southeastern United States, The Nature Conservancy is completing the classification, description, and keys to the national forests. Included in the description of each type is a list of the Society of American Foresters (SAF) Covertype (1980) with which it is associated. For example, the Longleaf Pine/Little Bluestem-Blazing Star Woodland from the national classification (Scientific name: Pinus palustris/Schizachyrium scoparium-Liatris pycnostachya Woodland) would be found within the "Longleaf Pine" SAF covertype. Additional crosswalks that are being documented include the Kuchler Potential Natural Vegetation classification (1975), the Classification of Wetlands, and Deepwater Habitats of the United States (Cowardin et al. 1979), Brown, Lowe, and Pase (1980), and others.

5.6.1 An Example of Crosswalking: The Relationship between the Brown, Lowe, and Pase Classification and The National Vegetation Classification System

The Brown, Lowe, and Pase system was developed for use in the southwest, with special emphasis on Arizona. A later version was expanded to include all of North America (Brown et al. 1979, 1980). The mechanics of crosswalking the Brown, Lowe, and Pase (1980) classification system to the national vegetation classification system have been completed for all of the communities that occur on the Gray Ranch site in New Mexico.

The Brown, Lowe, and Pase classification and the national vegetation classification system combine physiognomy and broad climatic patterns in the upper levels of the hierarchy, though the factors may be treated at different hierarchical levels. For example, both systems separate wetlands from uplands, but the Brown et al. system does this at the sec ond level (Vegetation level) of the hierarchy whereas the national classification does so at the fourth level (Formation level). Both systems also identify coarse physiognomic classes such as forest, woodland, scrubland, etcetera. The Brown, Lowe, and Pase classification calls this level of the hierarchy the "Formation-type" and this is recognized as the "Class" level in the national classification.

The major difference between the two systems is that the Brown, Lowe and Pase classification recognizes a Regional Formation or Biome level which is based on "distinctive evolutionary history within a given formation." These biomes tend to be centered in particular geographic regions or provinces (Brown 1982). The national classification does not make such regional distinctions. The national classification, which is physiognomic at the highest levels, is geographic only to the extent that physiognomy reflects local ecological factors.

The Brown, Lowe, and Pase classification has two floristic levels, both which tend to be coarser in scale than the national classification. The Series level of the Brown, Lowe, and Pase generally represents the dominant species at climax and are often named by the dominant genus (i.e., Pine series). This level is much broader than the Alliance level of the n ational classification. The lowest level of Brown, Lowe, and Pase (1980) system, called the association, is generally identical to the Alliance level used in the national classification. For example, Brown, Lowe, and Pase (1982) describe a Juniperus deppeana association that is equivalent to the national classification's Juniperus deppeana Alliance. In a few cases, Brown, Lowe, and Pase divided the vegetation into associations that correspond to one, or a related group of associations, from the national classification. For example, the national classification contains a single Pinus ponderosa Alliance which has roughly forty associations within it. One of the associations within the Pinus ponderosa Forest Alliance is the Pinus ponderosa/Quercus gambelii association. Brown, Lowe, and Pase describe two associations: (1) The Pinus ponderosa association, which is nearly equivalent to the Pinus ponderosa Alliance in the national classification except that it does not include Pinus ponderosa/Quercus gambelii types; and (2) the Pinus ponderosa/Quercus gambelii association, which is equivalent to the Pinus ponderosa/Quercus gambelii association in the national classification (Table 6).

Table 6. Example of a Crosswalk between the Brown, Lowe, and Pase Classification and the National Vegetation Classification System

The "< " and "=" symbols identify the relationship between the floristic units from each system.

5.7 Current Status of the National Vegetation Classification System

5.7.1 State Coverage

The data used to generate the communities in the national vegetation classification come from a wide variety of sources. The national classification is primarily based on communities described and tracked by individual state Natural Heritage programs. The combined expertise of these programs has contributed substantially to the generation of the national vegetation classification. The national classification currently covers all of the United States except Alaska and Hawaii, and work is underway to incorporate these states.

5.7.2 Regional Coverage

The information on most vegetation types identified in the state Natural Heritage program classifications has been synthesized to describe national elements. In addition, some elements have been derived from rigorous analysis at the regional level. The western, midwestern, eastern, and southeastern regions have now completed provisional regional clas sifications (Allard 1990, Bourgeron and Engleking 1993, Bourgeron and Engleking 1994, Faber-Langendoen 1993, Sneddon and Metzler 1992, Sneddon et al. 1994). The majority of the floristic units in these classifications are based on qualitative assessment of available data. Approximately 20 percent of the elements are the result of quantitative analysis (see Table 7). Each regional classification is now organized under the national vegetation classification hierarchy. There remains some redundancy in the Alliances and community elements listed in the regional classifications, as the evaluation of communities that occur in more than one region has not been fully completed.

The regions vary in the degree of refinement and the total number of community elements identified (Table 7). The variation among regions in the number of floristic units is due to differences in the amount of available community information, the diversity of habitats, and the overall geographic coverage among regions. The differences also reflect the classification approach adopted by the regions to develop their units and the levels of financial support for classification work. For example, the eastern region has recently been supported by the NBS Gap Analysis program to generate a list of all Alliances in the region. A comprehensive list of Alliances (126) was completed as a result of this project, though the list of community elements is not complete for this region. This region expects to have classified approximately 400 community elements upon completion. In contrast, the western region has worked primarily from the bottom up and has identified alliances by grouping known plant associations.

5.7.3 National Coverage

The number of units currently identified for each level of the classification hierarchy is presented in Table 8. The physiognomic levels of the hierarchy are still being tested and refined. Significant structural modification of the physiognomic levels of the hierarchy is not expected. The addition of several new formations is pending review by the group of national and regional ecologists which comprise the "national ecology team." The greatest fluctuation in the number of units identified under each level of the hierarchy is expected to be in the floristic levels. It is estimated that perhaps as many as 1,500 additional community elements may be identified as the classification is refined.

Up to this time, the approach to refine the national vegetation classification system has been prioritized to those types that have been identified as rare at the state level and then proceed to the more common types. As part of a project supported by the United States Fish and Wildlife Service, an initial survey of the rare communities of the conterminous United States has been completed (Grossman et al. 1994). Each of the 371 rare communities identified in the report has been placed into the national hierarchy and duplication among regions rectified. Descriptions have been written for each type and confidence levels assigned.

5.7.4 Gaps in the Classification

5.7.4.1 Gaps in State-Level Information

The degree of community information varies considerably among states. Some states lack classifications for their communities altogether, while others have classifications that are at a coarser level than the national vegetation classification standard. Others may have classifications but inventory efforts for communities have not been extensive. As a result, the national vegetation classification contains more information in some states than others. In the eastern region, additional information is needed from Maryland, Rhode Island, Virginia, and West Virginia to fill gaps in the national vegetation classification. Additional information is needed from Alabama, Georgia, and Texas to refine the southeastern portion of the national vegetation classification. In the midwestern region, the states of Iowa, North Dakota, and South Dakota require additional inventory and classification work to refine the national vegetation classification. Although additional community information from the states listed above is needed to refine the national vegetation classification, efforts have been made to supplement the information from the state Natural Heritage programs with information from other sources available for communities in these states. The National Park Service vegetation mapping project will allow additional information to be collected in many of these states. Many of the vegetation units identified in parks in these states will be treated as new types and will be fit into the national vegetation classification using the process for adding new classes described above.

5.7.4.2 Types Still in Need of Basic Work

Although the classification includes vegetation from all of the physiognomic classes (forests, woodlands, shrublands, etc.), there is a greater amount of information available for some vegetation classes than others. In general, more is known about the forest, woodland, and shrubland classes than about herbaceous and sparse woody classes (sparse woodland, sparse shrubland, sparse dwarf shrubland). Comparatively little is known about the sparsely vegetated communities. In addition, the degree of confidence associated with upland types in the classification is generally higher than for wetland types. The classifications for communities that occur as complexes or in zones are also in need of further work.