Compositional Groups and Ecological Complexes: A Method for
Alliance-Based Vegetation Mapping
Leonard Pearlstine1, Alexa McKerrow2, Milo Pyne3, Steve Williams2, and Stacy McNulty4
1Florida Cooperative Fish and Wildlife Research Unit, University of Florida, Gainesville
2North Carolina State University, Raleigh
3The Nature Conservancy, Chapel Hill, Tennessee
4Fish and Wildlife Information Exchange, Blacksburg, Virginia
The National Vegetation Classification (NVC) is a hierarchical classification based on a) vegetation structure and b) floristic composition that treats all existing terrestrial vegetation types in one system. This classification was developed by The Nature Conservancy (TNC) (Anderson et al. 1998) and has been adopted as the National Vegetation Classification Standard (NVCS) by the Federal Geographic Data Committee (FGDC 1997) with the Ecological Society of America acting as a review board. The National Gap Analysis Program, recognizing a need for a consistent classification across federal programs as well as within its own program, has supported the development of the NVC for several years. Mapping land cover to the alliance level (a physiognomically uniform group of plant associations sharing one or more dominant or diagnostic species) is a stated goal of the program.
Historically, land cover mapping from remotely sensed satellite data rarely exceeded one or two dozen classes, usually confined to Anderson (1972) Level II classification. Statewide mapping from Landsat TM imagery with critically limited time and budget constraints has been a challenging task. Vegetation in the southeastern U.S. is diverse and spatially complex, and the regional portion of the NVC reflects this. There are 680 alliances listed in the September 1998 version of the alliancelevel classification (Weakley et al. 1998) as either confident or probable in the 12 states of the Southeast (AL, AR, FL, GA, KY, LA, MS, NC, SC, TN, TX, VA). Concerns over how alliance-level mapping can be accomplished have led to innovative developments in vegetation community sampling with videography, new classification procedures exploring the use of decision rules, ancillary data sources, and image stratification, as well as stretching the limits of tools for data ordination and transformation. While the issue has been raised of the appropriateness of the alliance level of the NVC as a mapping goal, the availability of it as a national and federal standard outweighs any difficulties inherent in its internal complexity. In addition, the rapidly evolving nature of remote sensing technology necessitates that standards not be dictated by current technological limitations. Land cover mapping to the alliance level with a high degree of accuracy may not always be possible or practical in the first iteration, but we need a labeling system that will be compatible with future, more detailed mapping.
Here we present the Southeast Region’s proposed approach for labeling map units in a way that is consistent with the NVC but accommodates the fact that all alliances cannot be mapped from TM imagery and existing ancillary data. This labeling approach also provides the framework for scaling up from state to regional products in an ecologically meaningful way. The proposed approach has received the endorsement of National GAP and maintains compatibility with the NVC at the alliance level.
At the Southeastern GAP meeting in the spring of 1998, it became apparent that each state was using some form of alliance aggregation in their land-cover mapping. In order to develop a regional or national land-cover map, it will be necessary for individual state projects to come to a consensus on common labels for their map units. The labeling structure that emerged from those discussions resulted in the definition of two types of alliance aggregations. These are defined as:
1) Compositional Groups: a grouping of alliances with similar taxonomic composition and physiognomy, and
2) Ecological Complexes: a grouping of dissimilar alliances that are spatially and ecologically related on the landscape.
Compositional groups are composed of alliances that are spatially discrete but cannot be discriminated into separate classes because of spectral similarity. For example, in south and central Florida, freshwater marsh is prevalent, and cattails (Typha spp.) cover large areas. From satellite imagery and other available data sources, Typha domingensis and Typha latifolia cannot be separated. As a result, Florida is proposing a Cattail Marsh Community Group composed of three herbaceous cattail alliances. Similarly, Virginia, Kentucky, and Tennessee have each proposed compositional groups to accommodate mixed vegetation dominated by southern yellow pine.
Ecological complexes are distinguished from compositional groups in that it is the spatial closeness of the alliances that prevents discrimination based on satellite imagery. In North Carolina, pocosin wetlands are spatially heterogeneous with pine pond (Pinus serotina) woodlands being intermixed with evergreen shrublands dominated by sweet bay (Magnolia virginiana), gallberry (Ilex glabra), swamp red bay (Persea palustris), and titi (Cyrilla racemiflora), along with mixed titi-honeycup (Cyrilla racemiflora - Zenobia pulverulenta) shrublands in such close spatial proximity that they cannot be delineated separately at the resolution of Landsat TM. North Carolina has created a Pocosin Ecological Complex to encompass these structurally dissimilar but ecologically and spatially related alliances.
Notice that these groupings are not hierarchical levels that fit neatly between the formation and alliance. For instance, the Pocosin Ecological Complex aggregates evergreen woodland and evergreen and mixed shrub alliances which represent different formations in the National Vegetation Classification. It would be meaningless, therefore, to place the Pocosin Ecological Complex in the existing NVC structure. The Compositional Groups and Ecological Complexes stand beside the NVC, not within it. Because the groupings are aggregations of published NVC alliances, however, they can always be decomposed to the alliance level-as time and technology allow. They therefore maintain the integrity of the NVC structure.
Having the types of aggregates defined does not automatically mean that the classes will be consistent across the entire region. In the Southeast, the states have initially developed their lists of complexes and groupings independently, and it is now necessary to review and develop a logical organization that works for and across all the states in the region. At the Southeast GAP meeting it was decided to approach standardization by developing common lists for each ecoregion. While it is likely that a state will have unique compositional groups and/or ecological complexes, they should reflect distinct communities and not independently derived aggregated map labels.
Literature Cited
Anderson, J.R., E.E. Hardy, and J.T. Rouch. 1972. A land-use classification system for use with remote sensor data. U.S. Geological Survey Circular 671, Washington, D.C.
Anderson, M., P. Bourgeron, M.T. Bryer, R. Crawford, L. Engelking, D. Faber-Langendoen, M. Gallyoun, K. Goodin, D.H. Grossman, S. Landaal, K. Metzler, K.D. Patterson, M. Pyne, M. Reid, L. Sneddon, and A.S. Weakley. 1998. International classification of ecological communities: Terrestrial vegetation of the United States. Volume II. The National Vegetation Classification System: List of types. The Nature Conservancy, Arlington, Virginia.
FGDC. 1997. FGDC vegetation and classification information standards. U.S. Geological Survey, Reston, Virginia.
Weakley, A.S., K.D. Patterson, S. Landaal, M. Pyne, M. Gallyoun, and others, compilers. 1998. International classification of ecological communities: Terrestrial vegetation of the southeastern United States. The Nature Conservancy, Southeast Regional Office, Southern Conservation Science Department, Community Ecology Group, Chapel Hill, North Carolina.