Fire-adapted pineland vegetation of Florida:

A framework for inventory, management and restoration.

 

 

Introduction

 

Fire-dependent pineland vegetation once dominated the uplands of the Southeastern Coastal Plain from southern Virginia south to the tip of Florida and west to eastern Texas.  The natural communities that occupied this landscape, as well as many of the plants and animals that inhabited them, are now rapidly disappearing.  Economic development has removed the natural vegetation from much of this region, particularly from the finer-textured soils that readily support agriculture.  In addition, fire suppression has transformed most of the remaining natural vegetation to the point that it no longer resembles the vegetation present when European settlers first arrived.  The consequence is that we are on the verge of irrevocably losing much of the biotic diversity of the Southeastern United States.  Many of the communities that were maintained by fire, and the species that occupied them, will be permanently lost if action is not taken almost immediately.   The situation is particularly critical in Florida, which contains the greatest diversity of natural, fire-maintained pinelands, as well as the greatest diversity of organisms dependent on these systems. 

 


Tropical rainforests are often claimed to be the world's most species-rich plant communities.  However, richness can be measured at many scales.  Whereas tropical rainforests are the most species-rich plant communities at scales of 100 m2 and larger, at scales between 0.001 and 10.0 m2, the most species-rich communities include the pine-dominated, fire-maintained  temperate grasslands and open woodlands of the Southeast (Peet et al 1990, Peet & Allard 1995; Figure 1).  Values of 25-30 species per 0.25 m2, and 40 per m2 are not uncommon in these ecosystems.  Moreover, at scales of 100 and 1000 m2, the highest values currently known from the temperate zone are from Southeastern longleaf pine woodlands (95 / 100 m2; 150 / 1000 m2).  Although not all fire-maintained pinelands of the Southeast support high levels of species packing, the exceptionally high species-packing that characterizes much of this vegetation clearly marks these communities as important in efforts to maintain biodiversity. In the absence of fire, the high species-packing, as well as the bulk of the unusual biota, is quickly lost as woody plants grow up and crowd out the species-rich herbaceous layer and the animals that depend on it.

 

Preservation of the fire-dependent biota of the Southeast will not be easy.  Detailed documentation and analysis of the range of variation is needed so that major portions are not overlooked in conservation planning.  Moreover, simply reintroducing fire into a landscape transformed by fire suppression is not sufficient for restoring fire-maintained systems; in many cases the original species are no longer even present.  A multi-step research program will be needed to develop management and restoration methods, and not all the steps can be anticipated in advance.  However, one of the first steps must be to create a site-based template; if we are to manage, restore or recreate the original pineland vegetation, we need to know the range of variation in the original vegetation and we need to know how that vegetational variation mapped onto local site conditions.  Clearly, the success of future efforts to manage or restore conservation lands representing fire-maintained pinelands in Florida and throughout the Southeast will hinge on the availability of comprehensive documentation of the original composition and structure to serve as the appropriate template.

 


Given the distinctiveness, biotic diversity, and original vast extent of fire-maintained pinelands in Florida, there is shockingly little detailed, consistent or comprehensive information available on the compositional variation of this vegetation. Available descriptions either use classes so broad as to be nearly meaningless (e.g., flatwoods, high pine, savanna) or concentrate on only a handful of plots from a single locality. Much of the best data is buried in unpublished dissertations  (e.g., Edminston 1963) and government reports (e.g., Clewell 1971, Wolfe et al. 1988), and these data have been collected in such varied ways as to preclude meaningful comparisons and compilations.  When Christensen (1998) reviewed the regional literature, he concluded that classification of flatwood vegetation was intractable, given available information.  Myers (1990), Abrahamson & Hartnett (1990), Stout & Marion (1993), Ware et al. (1993) and Peet & Allard (1993) all have recently attempted to review the literature on these vegetation types for Florida with markedly little success in providing a comprehensive and detailed overview.  I was particularly struck when I attempted to compile original data from across the range of longleaf pine for a preliminary region-wide classification, that effectively no quantitative published data are available for Florida (see Peet & Allard 1993).   The research I am proposing is designed to provide for the first time a quantitative, comprehensive description of the composition and structure of this fire-maintained pineland vegetation.

 

Vegetation description and classification play a key role in many areas of conservation, land management and scientific research.  In effect, the classification identifies the relevant natural communities.  When a scientist places a specimen in a museum, he or she almost always attaches a label that contains critical information about the origin of the specimen.  To conduct biological or ecological field research without providing precise information on the ecological community or communities where the work was done is much like omitting the label from a museum specimen; critical information on ecological context or habitat will be missing.  Inventories of natural areas can be no better than the classification used to define the inventory units.  Managers of conservation lands require accurate and detailed descriptions of the vegetation attributes they need to preserve or recreate, but without a well-formulated vegetation classification and description, a quality template for management or restoration is not possible.  Widespread realization of these needs has led to broad collaboration among federal agencies, conservation organizations and professional ecologists to develop standards for vegetation classification (see Federal Geographic Data Committee 1997, Weakley 1997).  The work I propose will provide the ecological context and necessary foundation for future research and management on Florida's fire-dependent pineland ecosystems.

 


The focus of the proposed work is explicitly fire-dependent, pine-dominated vegetation and is intended to include primarily open woodland communities with herbaceous or low-shrub dominated understories that burn regularly with low-intensity ground fires.  In the somewhat ambiguous traditional terminology, these communities would include the areas of flatwoods, high pine, piney woods, savannas, some prairies, and some scrubby flatwoods.  By far the most conspicuous plants are the open-grown pines (longleaf, slash, and pond), though in some cases they are extremely sparse.  Scrub areas, such as those dominated by sand pine, will not be included as those areas are not so much fire-maintained as they are periodically fire disrupted.  The subtropical rocklands and flatwoods of the southern tip of the Florida (south of the terminus of the Lake Wales Ridge; see Synder et al. 1990) dominated by South Florida slash pine (Pinus elliottii var. densa) are beyond the scope of this study and will be the focus of a separate study using similar methodology conducted by Dr. William Platt and others.  (Inclusion of these southern areas in the present project would be desirable, but is not possible within the existing timeframe and budget constraints.)

 

Research context and long-term objectives

There is an urgent need to document the composition and structure of the fire-maintained pineland vegetation of the Southeast before these communities are lost to development and fire-suppression.  If this is not done soon, the opportunity will be lost forever.  This urgency has motivated me to undertake a series of studies to document the diversity of vegetation composition and structure throughout this little-known biome.  Thus far, I have completed the survey (roughly 425 samples and 600 species) and analysis for the Carolinas and much of Georgia.  Three manuscripts are nearing completion and should soon be submitted for publication.  The editors of both Castanea and American Midland Naturalist have actively encouraged submission of these papers to their journals.

 


I view journal publication of regional papers as merely stepping stones toward a comprehensive and synthetic monographic book on the ecology of the fire-maintained pinelands of the Southeast.  A prospectus and timetable for this book have been completed in collaboration with Dr. William Platt, and we are currently in the process of soliciting potential contributing authors and contacting potential publishers.  The prospectus is included as Appendix 1 (note that in the prospectus we there refer to fire-maintained pinelands generically as "savannas", a usage of the term much broader than that traditionally employed by biologists in the Southeast.)  The present proposal is designed to provide funding for the comprehensive inclusion of Florida ecosystems in our synthetic treatment.  Without this funding, much of the diversity of Florida pinelands would, of necessity, be omitted or receive only inappropriately cursory treatment.

 

Objectives & benefits

Despite the broad focus of my overall research program, the focus of the proposed research will be exclusively on the identification and description of the natural diversity of the fire-maintained pinelands of Florida.  This can be viewed in terms of the series of specific objectives and benefits listed below.

1.  I will identify the best remaining examples of fire-maintained pineland vegetation across the range of sites conditions and geographic regions of Florida (excepting only the southern tip).

2.  I will collect detailed quantitative samples representative of the range of fire-maintained pineland vegetation in the state using a standard sampling protocol so as to  assure maximal compatibility with other datasets.

3.  I will analyze the resultant data to develop a comprehensive classification and associated summary information for each type with respect to species composition, species diversity, woody plant population structure, and environmental and geographic situation.


4.  I will develop or expand community nomenclature in a fashion consistent with the developing National Vegetation Classification (Federal Geographic Data Committee 1997) so as to assure maximal utility to and application by government agencies and conservation organizations.  (Our work in the Carolinas and Georgia have identified nearly 70 community types, roughly half of which were new to and have subsequently been added to the National Classification; see Weakley 1997.  For illustrative purposes, the community types recognized in that work are listed in Appendix 2 using the nomenclature of the National Classification.)

5.  I will publish the results of this research in peer-reviewed journal articles.  These publications would ultimately provide the basis for the synthetic overview monograph described in Appendix 1.

 

Note that the product is a classification and description based on the best remaining sites we can identify, and is not an inventory of all the sites in Florida.  Classification is a necessary step prior to inventory; once the difficult step of classification has been completed, inventory can be conducted using more extensive and less intensive sampling than we propose for this project.

 

Proposed research (Methods)

Identification of sample locations


The most difficult component of this project will likely be identification of the best remaining examples of fire-maintained pinelands across the range of available site conditions.  I am not so presumptuous as to think that our group alone can locate and sample all variation in the fire-maintained pineland vegetation of Florida within what is effectively a two-year study.   The only way to approach complete coverage is through extensive consultation with regional experts.  Representatives of state and private conservation organizations, employees of federal agencies like the Forest Service and the military, private and corporate land managers, consulting foresters, and local academics all will be contacted and can be expected to contribute in substantive ways.  For many years I have been building a network of contacts for such purposes.  We will not be shy about contacting others; we know that this project cannot be done well if done in isolation.

 

One important mechanism to assure relatively complete coverage is to stratify sites by critical environmental factors.  For example, we expect to first stratify the state by physiographic units, probably using those defined by Cooke (1939), which, although not the most recent (see Randazoo and Jones 1997), have a much finer resolution than most other schemes.  We will attempt to cover the range of relevant soil series in each region.  Large blocks of land such as managed by the U.S. Forest Service, state agencies, the military, and the forest products industry will receive special.  Aerial photos will be consulted to identify large blocks of undeveloped and fire-maintained vegetation.  We anticipate that we will continue to enjoy success at obtaining permission to work on large private land holdings, industrial lands and military lands. 

 

In summary, I am confident that we can make substantial progress toward identification of the best remaining sites, and in the process provide the essential skeleton for an evolving classification and description, plus flesh out the more important pieces.  With this critical core in place, professional and amateur botanists and ecologists should be able to identify what has been missed and contribute substantially to the refinement of the work over the subsequent years.

 

 In addition to identifying candidate sites, recognition of what constitutes the best remaining vegetation is especially challenging.  We will use a strategy similar to that which we have used successfully in the Carolinas and Georgia.  An initial literature review (starting with Peet and Allard 1993, Abrahamson & Hartnett 1990, Myers 1990)  will identify the range of variation that might be expected.  Historical literature and especially photographs can be invaluable (e.g., Harper 1914, Schultz 1907, Kurz 1942). We will then consult widely as described above.   Our broad experience with fire-maintained pinelands provides an invaluable experience base.  We will examine patterns in species diversity, the presence of weedy species, vegetation texture, presence of indicator species like wiregrass, presence of old trees, and other attributes. Integrity of the ground layer will be emphasized over structure of the tree canopy.  The bulk of the biodiversity resides in the groundlayer, this portion of the vegetation is most sensitive to site conditions, and it the most difficult to restore following disturbance.  Moreover, most pine canopies have been modified by cutting, planting and other activities.

 

A few examples may serve to illustrate some of the criteria that are useful.  When we sampled the flatwoods on the east side of the Okefenokee Wildlife Refuge, we found that they looked altogether natural owing the recent burning, but that small-scale diversity was substantially lower than found on similar soils elsewhere.  In the end, we deduced that these systems were substantially degraded from years of fire suppression, and that recent control burns have not yet fully restored the system.  Experimental plots at Tall Timbers Fire Ecology Lab have been burned annually (and at other intervals) for decades, but contain no trace of wiregrass and are clearly an artifact of land use prior to the current management practices.  The oldgrowth longleaf stands of the Patterson Natural Area at Eglin Air Force Base have clearly been in place for a a very long time, but they lack any sign of wiregrass, despite the species occurring immediately to the north on Blackwater and Conecuh forests.  Here we must join with other puzzled researchers and admit that we simply cannot tell whether wiregrass was once part of that ecosystem.  On the other hand, we were informed by ecologists at the Savannah River Ecology laboratory that wiregrass was not an important component of the natural vegetation of the Savannah River Site.  In this case, our4 extensive investigation provided evidence that wiregrass had at one time been a local dominant, although today it is rare.

 

Vegetation sampling

We will use the Carolina Vegetation Survey (CVS) vegetation sampling protocol, a methodology designed to simultaneously facilitate long-term studies of vegetation, assessment of biodiversity, and inventory of contemporary vegetation composition and structure (Peet et al. 1998).  Variations on this protocol have been used to obtain in excess of 2800 vegetation samples in the Carolinas and Georgia, including 427 that represent the range of fire-maintained pinelands in those states.

 


The CVS methodology is designed to be flexible so as to accommodate as many sampling situations as possible, to effectively sample species composition over a broad array of spatial scales, and to produce results compatible with the majority of commonly used sampling schemes.  The methodology is built around 10x10 meter plots as the basic sampling module.  In an intensively sampled 10x10m module, two or more sets of nested quadrats are established, each radiating from a permanently marked corner and increasing in area with each quadrat by a factor of ten.  All woody stems above breast height in the module are recorded by diameter class.   Typically 10  modules are combined to form a standard 0.1 ha (1000 m2) plot, though local heterogeneity in vegetation sometimes requires plots as small as a single module.  In a standard 0.1 ha plot, there will typically be four central modules sampled intensively, and an additional 6 (to make an array of 2 x 5) sampled for woody stems and additional herbaceous species not encountered in the four intensive models.  The method has considerable built-in flexibility to allow for such contingencies as sampling of species with either very high or very low densities of stems at breast height.  Analysis can be conducted using samples scales ranging from 0.01 m2 to 1000 m2.   A full elaboration of the methodology is beyond the scope of this proposal, but a detailed description is available upon request or by anonymous ftp (panicum.bio.unc.edu/NC_Veg_Survey/sampling/castanea.rtf); our suite of data management and analysis software written in SAS is also available from this site).

 

  Soil samples will be collected from the upper 10 cm of the A horizon as well as the B horizon of each intensively sampled module. Soils will be dried, sieved, bagged, composited across the 4 modules, and sent Brookside Labs in New Knoxville, Ohio  for chemical and textural analysis. The compositing of samples across modules is required to minimize costs. Our previous work in similar systems in the Carolinas and Georgia has shown that, unlike soils derived from igneous rock, soils of coastal plain pinelands are spatially rather consistent in soil chemistry and texture.  This work has also shown strong correlations between soil attributes and vegetation composition, making the soil analysis a critical part of the proposed research.

 


Sampling efforts will be stratified to cover a broad range of geographic and edaphic conditions.  The actual data collection is time consuming, in no small part because of the exceptional species richness of these systems.  Past experience suggests that we can anticipate collection of between 100 and 140 samples per season for a total of somewhere around 250 detailed samples for the two field seasons.  Our final dataset will be available to state agencies and conservation professionals.

 

Data management and analysis

We have developed a suite of SAS computer programs for error checking CVS data and for data summarization.  Once the data are entered and error checked, species nomenclature will be revised to conform to Kartesz 1994 to provide consistency across all our Southeastern pineland datasets.

 

Numerical classification programs will be used to develop an initial classification of vegetation samples for each major geographic region.  Past experience suggests that agglomerative classification using a Bray-Curtis similarity metric and either the Wards or Lance-Williams algorithm produces the most ecologically interpretable results, an observation confirmed by Belbin and McDonald (1993).  Two-way indicator species analysis (Hill 1979) may be used for exploratory analysis, but we agree with Belbin and McDonald that this method can produce misleading results.

 


We will use ordination techniques to relate compositional patterns to environmental and site variation, mostly though a combination of multidimensional scaling (Minchin 1987, 1990), detrended correspondence analysis (ter Braak 1987, Peet et al. 1988) and canonical correspondence analysis (ter Braak 1987).   Once the primary vegetation gradients have been recognized, the vegetation pattern and distribution of community types will be represented in terms of these gradients (as in Peet & Loucks 1977, Austin et al. 1990, Allen et al. 1991).  To facilitate the untangling of complex, interacting environmental factors, we will use a strategy of progressive fragmentation (Peet 1980, Peet and Allard 1995) where the dataset is broken into progressively smaller portions as critical environmental factors are identified.  The nonlinear interaction of environmental gradients makes this approach better than one of simply partialling out the variance attributable to particular factors (as in Carleton 1984, ter Braak 1987, Wiser et al. 1996).  Multiple discriminent analysis will be used to develop site-based templates for community management and restoration (see Van Lear & Jones 1987, Allen & Wilson 1991).

 

The classification will be reviewed in light of the gradient analysis results and refined if necessary.  The resultant classification units will then be compared with the existing National Vegetation Classification (Weakley et al 1997) being developed by the Nature Conservancy and the U.S Federal Geographic Data Committee (1997).  Nomenclature from the National Classification will be used where appropriate, and new units will be proposed for the National Classification where needed.  Previous work in the Carolinas and Georgia suggests somewhat over half of the community types recognized will be new to the National Classification (see Appendix 2 and Weakley et al 1997).  The composition and environmental setting of the communities recognized will be summarized in terms of composition, structure, diversity and environmental setting,

 

Project schedule                            

The project will be divided geographically and temporally into two phases.  The first phase will focus on the pinelands of the panhandle plus those of the northeast corner of the Florida.  The second phase will focus on the bulk of peninsular Florida  The field work for the first phase will be conducted during May-October 1999, and the field work for the second phase will be conducted during May-October 2000.  Each field work period will be preceded by a reconnaissance period and each will be followed by a period of data analysis and manuscript preparation.  Each of the two phases can be expected to produce three deliverable products: a list of sites selected, a complete dataset, and a manuscript for publication.  A comprehensive report will be provided at the end which will contain the first approximation of a synthetic treatment of fire-maintained pinelands of Florida and adjacent states.

 

The specifics of the schedule can be summarized as follows:

 

 

The specifics of the schedule can be summarized as follows:

(Revised September 2000)

 

(i  = initiate work; w  = work in progress;  c = work completion; p  = product receivable)

 

Fiscal Year  1999-00     2000-01    2001-02     2002-03   2003

                                    Quarter                 3  4    1  2  3  4    1  2  3  4    1  2  3  4   1  2

 

Phase 1 – Panhandle Florida

Study site & selection                             i    w   w  w  .    w   c       .    .     .    .    .    .     .    .

Vegetation sampling                              .    i     w  w  .    .     w  c   .    .     .    .    .    .     .    .

Data compilation and management     .    .     .    i    w  w   .    .    w  c    .    .    .    .     .    .   

Data analysis                                           .    .     .    .    .   ,      .    .    w  w   c   .    .    .     .    .   

Subsection report preparation              .    .     .    .    .    .     .    .    i    w        w  w  c    .    .

 

Phase 2 – Peninsular Florida                                          

Study site identification and selection  .    .     .    .    i    w   w  .    w  w   c   .    .    .     .   

Vegetation sampling                              .    .     .    .    .    i     w  w  .    .     w  c   .    .     .    .

Data compilation and management     .    .     .    .    .    .     .    i    w  w   .    w  c   .     .    .

Data analysis                                           .    .     .    .    .    .     .    .    .    .     i    w  w  c    .    .   

Subsection report preparation              .    .     .    .    .    .     .    .    .    .     .    i    w  w   c   .

 

Final report preparation                         .    .     .    .    .    .     .    .    .    .     .    .    i    w   w  p 

                                                        


 

 

Literature cited

Abrahamson, W.G., and D.C. Hartnett. 1990.  Pine flatwoods and dry prairies.  Pages 103-149 in  Ecosystems of Florida. R.L. Myers and J.J. Ewel, editors. The University of Central Florida Press, Orlando.                                

Allen, R.B., R.K. Peet, and W.L. Baker. 1991.  Gradient analysis of latitudinal variation in Southern Rocky Mountain forests.  Journal of  Biogeography 18:123-139.

Allen, R.B. and J.B. Wilson. 1991.  A method for determining indigenous vegetation from simple environmental factors, and its use for vegetation restoration.  Biological Conservation 56:265-280.

Austin, M.P., A.O. Nicholls, and C.R. Margules. 1990.  Measurement of the realized niche; environmental niches of five Eucalyptus species.  Ecological Monographs 60:161-177.

Belbin, L.  and C. McDonald. 1993. Comparing three classification strategies for uswe in ecology.  Journal of Vegetation Science 4:341-348.

Carleton, T.J. 1984.  Residual ordination analysis: a method for exploring vegetation-environment relatioinships.  Ecology 65:469-477.

Christensen, N.L. 1998.  Vegetation of the southeastern coastal plain.  In M.G. Barbour and W.D. Billings, eds.  North American Terrestrial Vegetation, Second edition..  Cambridge University Press, Cambridge (in press).

Clewell, A.F. 1971.  The vegetation of the Apalachicola National Forest: an ecological perspective.  Contract 38-2249, Final Report. U.S. Forest Service, Atlanta, Georgia.

Cooke, C.W. 1939.  Scenary of Florida, as interpreted by a geologist. Florida Geological Survey Bulletin 17. 118 p.

Edminston, J.E. 1963.  The ecology of Florida pine flatwoods.  PhD Thesis, Univ. Florida, Gainesville, FL.

Federal Geographic Data  Committee Secretariat. 1997.  FGDC Vegetation classification and information standards (http://www.nbs.gov/fgdc.veg/).

Harper, R.M. 1914. Geography and vegetation of North Florida.  Florida State Geological Survey.  Sixth Annual Report 163-437.


Hill, M.O. 1979.  TWINSPAN - A FORTRAN Program for arranging multivariate data in an ordered two-way table by classification of the individuals and attributes.  Cornell University.  Ithaca, NY.

Kartesz, J.T. 1994.  Synonymized checklist of the vascular flora of the United States, Canada, and Greenland.  Second edition.  Timber Press, Portland, OR

Kurz, H. 1942.  Florida dune and scrub vegetation and geology. Florida Geological Survey. Geological Bulletin 23. 154 p.

Minchin, P.R.  1987.  An evaluation of the relative robustness of techniques for ecological ordination.  Vegetatio 69:89-107.

Minchin, P.R.  1990.  DECODA Version 2.01.  Anutech Pty. Lts., Canberra. vegetation of north-central Florida.  American Midland Naturalist 79:441-457.

Myers, R.L. 1990.  Scrub and high pine.  Pages 150-193 in R.L. Myers and J.J. Ewel (editors). Ecosystems of Florida. Univ. Central Florida Press, Orlando, FL.

Peet, R.K. 1980.  Ordination as a tool for analyzing complex data sets.  Vegetatio  42:171-174.

Peet, R.K. and D.J. Allard. 1993.  Longleaf pine vegetation of the Southern Atlantic and Eastern Gulf Coast regions: A preliminary classification.  Proc. Tall Timbers Fire Ecology Conf. 18: 45-81.

Peet, R.K., R.G. Knox, R.B. Allen, and J.S. Case. 1988.  Putting things in order: the advan­tages of detrended correspondence analysis.  American Naturalist 131:924-934.

Peet, R.K. and O.L. Loucks. 1977.  A gradient analysis of Southern Wisconsin forests.  Ecology 58:485-499.

Peet, R.K., E. van der Maarel, E. Rosen, J., Willems, C. Norquist, and J. Walker. 1990.  Mechanisms of coexistence in species-rich grassland.  Bulletin of the Ecological Society of America 71:283.

Peet, R.K., T.R. Wentworth and P.S. White. 1998.   A flexible, multipurpose method of recording vegetation composition and structure.  Castanea (in press). (available via ftp as  panicum.bio.unc.edu/NC_Veg_Survey/Sampling/castanea.rtf

Randazzo, A.F. and D.S. Jones (eds.). 1997.  The geology of Florida. U. Presses of Florida, Gainesville, FL. 327 pp.

Schwarz, G.F. 1907.  The longleaf pine in virgin forest: a silvical study. J. Wiley & sons, NY. 135 p.


Snyder, J.R.,. A. Herndon and W.B. Robertson Jr. 1990.  South Florida Rockland.  Pages 150-193 in R.L. Myers and J.J. Ewel (editors). Ecosystems of Florida. Univ. Central Florida Press, Orlando, FL.

Stout, I.J. anad W.R. Marion. 1993.  Pine flatwoods and xeric pine forests of the Southern (lower) coastal plain.  Pages 373-446 in W.H. Martin, S.G. Boyce and A.C. Echternacht, eds.  Biodiversity of the Southeastern United States: Lowland terrestrial communities.  John Wiley and Sons, New York, NY.

ter Braak, C.J.F. 1987.  The analysis of vegetation-environment relationships by canonical correspondence analysis.  Vegetatio 69:69-77. 

Van Lear, D.H. and S.M. Jones. 1987.  An example of site classification in the Southeastern Coastal Plain based on vegetation and land type.  Southern Journal of Applied Forestry 11:23-28.

Ware, S., C. Frost and P.D. Doerr. 1993.  Southern mixed hardwood forest: the former longleaf pine forest.  Pages 447-493 in W.H. Martin, S.G. Boyce and A.C. Echternacht, eds.  Biodiversity of the Southeastern United States: Lowland terrestrial communities.   John Wiley and Sons, New York, NY.

Weakley, A.S., K.D. Patternson, S. Landaal and Mark Gallyoun. 1997.  International classification of ecological communities: terrestrial vegetation of the Southeastern United States (Draft of November 1997).  The Nature Conservancy, Southeast Regional Office, Chapel Hill, NC 

Wiser, S.K., R.K. Peet and P.S. White. 1996.  High-elevation rock outcrop vegetation of the Southern Appalachian Mountains.  J. Vegetation Science 7 (in press).

Wolfe, S.H., J.A. Reidenauer and D.B. Means. 1988.  An ecological characterization of the Florida panhandle. U.S. Department of Interior, Fish and Wildlife Service, FWS Biological Report 88(12).

Wharton, C.H. 1978.  The natural environments of Georgia.  Georgia Dept. Nat. Res.  Atlan­ta, GA.