Transcript Slide 1

Using High-Quality Vegetation Plot Datasets to Restore
Ecosystems: An Example from the Coastal Plain of the
Southeastern United States
M. Forbes
1
Boyle ,
Robert K.
1
Peet ,
and Thomas R.
2
Wentworth
1 University of North Carolina, Dept. of Biology; 2 North Carolina State University, Dept. of Plant Biology
Abstract
Introduction
The success of restoration efforts is typically based on the successional trajectory of a site to its
predefined target condition. Varying processes can determine the target condition, but a poor
description of the target within the region of study may ultimately hinder the restoration effort.
Reference sites of undisturbed, late-seral-stage ecosystems are needed to quantitatively define
regional (> km²) community types, which can then be used to successfully determine target
conditions for restoration efforts. Over the past twenty years, the Carolina Vegetation Survey
(CVS) has acquired high-quality biotic and abiotic records from over 6,000 vegetation plots
distributed across North and South Carolina, United States of America (US). Recently, we used
these data to assess the ability of an existing classification, the US National Vegetation
Classification Standard (NVCS), to define target conditions for restoration. We focused on nonalluvial wetland community types in the southeastern Coastal Plain of the Carolinas, a region
largely defined by its complex distribution of unique wetlands. Over half of the association-level
vegetation units currently recognized by the NVCS for the Coastal Plain of the Carolinas are
wetlands. These systems vary considerably in terms of species richness, geomorphology, soil,
hydrology, and disturbance history. We classified 184 plots containing 601 species using
hierarchical cluster analysis techniques and then used non-metric multidimensional scaling to
define groups based on soils and geomorphology. Our analyses indicated a need for substantial
revision of the NVCS classification. Of the 56 associations recognized in the current study, 21 are
newly described, 30 have complex relationships with one or more previously-described
associations, and only 5 remain unchanged from their original NVCS delineations. We were unable
to assess 37 associations because we were lacking appropriate data.
The Carolina Vegetation Survey (CVS) is a multi-institutional collaborative research program established in
1988 to document the composition and status of the natural vegetation of North and South Carolina,
United States of America (US), for purposes of inventory, assessment of conservation status, and
monitoring of environmental impacts. Data collection methods are standardized and are widely
employed across the southeastern US (Peet et al. 1998). Over the past 20 years, CVS has acquired high
quality, quantitative records from over 6,000 vegetation plots distributed across the Carolinas. These
data have allowed for the first time an accurate documentation of the natural communities and species
habitats from this region (Figure 1).
Ultimately, we designed a hierarchical framework to guide restoration efforts using the newly
described classification units. Once defined, these classification units were mapped and identified
to a biogeographic region. Regardless of seral-stage, restoration sites were delineated into specific
classification units by identifying environmentally stable discriminants. Classification models were
used to select the appropriate composition of species within a restoration site. Finally, monitoring
plots were established to ensure that the successional trajectory of a restored site would reach its
predefined target condition. This hierarchical framework provided a necessary platform to filter a
rigorous community classification dataset into a usable structure that restoration managers could
use to guide successful and ecologically meaningful restoration efforts.
In May 2005, CVS initiated a partnership with the North Carolina Ecosystem Enhancement Program
(EEP)—a state agency created for the purpose of wetland habitat restoration—with goals designed to
benefit the two organizations in terms of generation, management, and distribution of vegetation data
from both natural and mitigation sites. The CVS data management system has been designed to ensure
accurate collection, reporting and transmission of vegetation monitoring data from field collection to raw
datasheets to insertion into the CVS-EEP data archive, and then accurate reporting to EEP and future
contractors.
The degree to which restoration efforts are deemed successful is typically based on the successional
trajectory of a site to its predefined target condition. Brinson and Rheinhardt (1996) emphasized the
critical roles of reference wetlands in compensatory mitigation, whereby successful restoration involves
establishment of goals (restoration targets), development of templates for the targets, and tracking the
success rate of the endeavor. Information gathered from CVS database can be used to define site-specific
reference conditions (Figure 2). EEP contractors can use these data to establish restoration targets
specific to designated project sites. Furthermore, quantitative assessment from these high-quality
vegetation plots can provide a better understanding of lower-order community structure and
composition for the US National Vegetation Classification Standard (NVCS) (Federal Geographic Data
Committee 2008).
STUDY OBJECTIVES
 Utilize high-resolution vegetation data from the CVS database in
order to assess the ability of the NVCS to define target conditions
for restoration
Figure 1. Distribution of established plots by the Carolina Vegetation Survey as of 2007 within North
and South Carolina, US. Orange dots represent plots occurring in nonalluvial wetland communities of
the southeastern US Coastal Plain (green shading).
Vegetation –
Environmental
Plot Database
Create higher-order levels based on:
1) physiognomic level
2) upland vs. wetland
3) open vs. closed forest
Quantitative assessment of vegetation – environmental
datasets (e.g. cluster analysis, ordination, indicator species
analysis, discriminant analysis)
Identify classification units based on similar vegetation, abiotic conditions
Map units onto biogeographic region
 Improve upon classification scheme of southeastern US Coastal
Plain nonalluvial wetlands
 Design a hierarchical framework to guide restoration efforts using
newly defined classification units
Provide restoration managers
localized, high-resolution targets
for restoration
Improve upon lower-order levels
within the National Vegetation
Classification Standard
Figure 2. Flow-diagram depicting how plot data from the Carolina Vegetation Survey database are organized
to produce high-resolution community classification with quantitative descriptions and defined
environmental settings.
Results
The southeastern Coastal Plain of the United States is home to a diverse array of nonalluvial wetlands.
Distribution of these systems is related to the presence of poorly drained soils, topographical
depressions, and the abundance of peat accumulations, while differing fire regimes maintain species
structure. Clockwise from upper left: Depression pond hardwood forest, bay forest, Atlantic white
cedar forest, and low pocosin.
Approach
I.
II.
Classified 184 plots containing 601 species using hierarchical cluster analysis (BrayCurtis, flexible beta) and non-metric multidimensional scaling. Mapped clusters
into higher-order hierarchical bins (Vegetation of the Carolinas 2008) and used
qualitative techniques to map individual plots to nearest NVCS-defined
Association(s). A three-level scale was used to assess the degree to which these
units corresponded to an Association:
1) Poor: these communities are newly described
2) Complex: these communities could be mapped to at least one other
Association
3) Good:
these communities corresponded well with current
Association descriptions
Used Indicator Species Analysis to identify diagnostic species of unique
classification units (NVCS Associations). Compared diagnostic species with those
specified in the existing unit definition. Discriminant analysis was used to
determine critical abiotic variables between clusters.
III. Designed a hierarchical framework to guide restoration efforts using the newly
described classification units.
BAY FORESTS / NONRIVERINE
SWAMPS / HARDWOOD FLATS
PONDS & MARSHES /
AQUATIC VEGETATION
POCOSINS /
PONDPINE WOODLANDS
Figure 3. Hierarchical cluster analysis of Carolina Coastal Plain plots obtained by the Carolina
Vegetation Survey (Distance Measure: Bray – Curtis, Linkage Method: Flexible Beta, beta = -0.25).
CVS plots from nonalluvial wetlands of the Carolina Coastal Plain formed classification units based on
similar vegetation patterns arising from hydrologic and soil gradients (Figure 3). We determined there
was a substantial need to revise the current NVCS classification of these systems based on the lack of fit
between our quantitative data and qualitative descriptions of NVCS vegetation associations. Of the 93
associations recognized as nonalluvial from the study area, 21 are newly described, 30 have complex
relationships with one or more previously described association, and 5 remain unchanged from their
original NVCS delineations. We were unable to assess 37 associations because we were lacking necessary
plot data. Community descriptions of eight of these associations are given in Figure 4.
Coastal Plain hardwood flats
NVCS Associations:
Depression ponds
NVCS Associations:
1) Quercus michauxii - Quercus pagoda / Clethra alnifolia Leucothoe axillaris Forest
Assessment – Good
Description:
1) Dichanthelium wrightianum – Dichanthelium
erectifolium Herbaceous Vegetation
Assessment – Complex
Description:
This community occurs on seasonally to nearly permanently saturated flats over mineral
soils characterized with having a high water table. Canopy dominants include Quercus
michauxii, Quercus laurifolia, Quercus pagoda, Liquidambar styraciflua, Liriodendron
tulipifera, Acer rubrum, and Fraxinus pennsylvanica. Subcanopy trees include Carpinus
caroliniana var. caroliniana and Ilex opaca var. opaca. A dense shrub stratum is
composed of Arundinaria tecta and Vaccinium fuscatum. Herbs found in this community
are diverse and can include Dichanthelium dichotomum var. ramulosum, Woodwardia
areolata, Thelypteris noveboracensis, and Carex laxiflora.
This community association is usually found on middle topographic
positions of limesink ponds, and thus less influenced by saturation.
Dominant species found in these plots include Lachnanthes caroliniana,
Centella erecta, Panicum virgatum, and Pinus taeda. Although these
plots fit with the described geomorphology of the NVCS community
type, they lack the nominal species necessary for accurate
characterization. Limesink communities within the study area are in
need of further examination to enhance the NVCS understanding of
Coastal Plain ponds and marshes.
2) Quercus pagoda – Quercus michauxii – Quercus alba /
Arundinaria tecta – Sabal minor / Chasmanthium laxum
Forest
Assessment – Complex
Description:
This oak-dominated wet swale association is described from the Outer Coastal Plain of
South Carolina and Georgia. It is composed of both upland and bottomland species, and
probably experiences flooding only during severe (e.g. hurricane) rain events, and then
without significant overland flow. Canopy species diversity is high; species include
Quercus pagoda, Quercus michauxii, Quercus falcata, Quercus alba, and Quercus phellos.
Other canopy trees include Fraxinus americana, Acer floridanum, Carya ovata, and Fagus
grandifolia var. caroliniana. The subcanopy and shrub strata are sparse within this
association. Species include Ulmus alata, Ilex decidua var. decidua, and Liquidambar
styraciflua. Unlike the NVCS community description, these plots do not contain either
Arundinaria tecta or Sabal minor. Furthermore, these plots have much higher species
richness values than acknowledged by the described community type. Herbaceous
species found in these plots include Poa cuspidata, Chasmanthium sessiliflorum var.
sessiliflorum, Viola affinis, and Chasmanthium laxum.
2) Panicum hemitomon – Eleocharis
equisetoides – Rhynchospora inundata
Herbaceous Vegetation / Lyonia lucida Forest
Assessment – Complex
Description:
This community is found over outer Coastal Plain limesink ponds,
usually at lower topographic positions. The only dominant species
found across these plots is Panicum hemitomon. Other species found
include Cephalanthus occidentalis, Amaranthus cannabinus, Azolla
caroliniana, Rhynchospora inundata, Nymphaea odorata ssp. odorata,
Lachanthes caroliniana, Eriocaulon decangulare var. decangulare, and
Rhexia cubensis. Each sampled limesink pond in this NVCS-described
community type had its own unique vegetation composition. There
is a clear need for study of the broad range of limesink and other
temporarily ponded depressions of the Carolina Coastal Plain.
Conclusions
 The use of high-quality plot data from CVS improves
upon reference information, target design, monitoring,
and data management - analysis.
 Finer-scale classification levels of nonalluvial wetlands
of the southeastern US Coastal Plain need revision
within the NVCS.
 Implementation of a hierarchical framework is
necessary to bridge the gap between rigorous data
analysis and model-building procedures advocated
by academic institutions AND the results they generate
that can be used by restoration managers:
• Perform quantitative analysis on high-quality plot data to
determine classification units and to improve upon
existing classification schemas
• Map classification units within a biogeographic region
• Identify environmentally stable discriminants in order
to delineate restoration sites into defined classification
units
• Use species lists determined from classification models
to select the appropriate composition of species within
a restoration site
• Establish monitoring plots within restoration sites to
ensure that the successional trajectory of a restored
site would reach its predefined target condition
Bay forests
NVCS Associations:
Pocosins
NVCS Associations:
1) Gordonia lasianthus - Magnolia virginiana - Persea
palustris - Sphagnum spp. Forest
Assessment – Poor
Description:
1) Pinus serotina / Zenobia pulverulenta - Cyrilla
racemiflora – Lyonia lucida Wooded Shrubland
Assessment – Complex
Description:
This community is found on peaty seepage depressions on landforms with a high water
table. The canopy in all of these plots is dominated by a mixture of Gordonia lasianthus,
Persea palustris, and Magnolia virginiana var. virginiana. Other canopy species include
Acer drummondii, Acer rubrum, and Nyssa biflora. The shrub and vine component of
these plots is patchy to dense, and includes Lyonia lucida, Clethra alnifolia, Cyrilla
racemiflora, Ilex coriacea, Gaylussacia frondosa, Leucothoe axillaris, Smilax laurifolia,
Vitis rotundifolia var. rotundifolia, and Toxicodendron radicans var. radicans. The high
variability of the shrub and herb component of these plots warrants an examination of
the NVCS schema for this community type.
This typical high pocosin of the Atlantic Coastal Plain is composed of a
very dense shrub stratum, up to 3 meters tall. Typical shrub species
include Lyonia lucida, Zenobia pulverulenta, Cyrilla racemiflora, Ilex
coriacea, Ilex glabra, and Persea palustris. An open canopy of Pinus
serotina, Taxodium ascendens, and Nyssa biflora are characteristic of
these plots. The presence of these latter two species explains the
inconsistency between some of these plots and the NVCS description
of this community type.
Brinson, M.M., and R. Rheinhardt. 1996. The role of reference wetlands in functional
assessment and mitigation. Ecological Applications 6(1):69-76.
2) Cyrilla racemiflora – Zenobia pulverulenta
Shrubland
Assessment – Good
Description:
Peet, R.K., T.R. Wentworth, and P.S. White. 1998. A flexible, multipurpose method for
recording vegetation composition and structure. Castanea 63:262-274.
2) Magnolia virginiana – Persea palustris / Lyonia lucida
Forest
Assessment – Complex
Description:
This saturated forest represents an intergrade to a shrub-dominated high pocosin of the
Outer Coastal Plain of the southeastern United States. The open canopy of this
community is composed of both Acer rubrum var. trilobum and Magnolia virginiana var.
virginiana. The shrub stratum is dense and composed of many species. These include
Ilex coriacea, Lyonia lucida, Clethra alnifolia, and Ilex laevigata. These plots differ from
the NVCS community description by having a diverse shrub stratum and by having a
significant amount of Arundinaria tecta in the herb-shrub strata. These plots are also
significantly north of the documented range of this association. The next closest match
would be the Gordonia-type described above, though these plots completely lack
Gordonia sp.
This community is defined as a low pocosin of peat domes within the
outer Coastal Plain of North Carolina. A mixture of both evergreen and
deciduous species is found within this community, and canopy heights
are generally low (< 15m for the scattered trees and ~ 2m for the
dominating shrubs). In these plots, the open canopy is composed of
Pinus serotina, Gordonia lasianthus, Ilex laevigata, and Acer
drummondii. The extremely dense shrub stratum is composed of
Zenobia pulverulenta, Ilex coriacea, Smilax laurifolia, and Kalmia
cuneata.
Figure 4. Plots were identified to associations following the US National Vegetation Classification Standard (NVCS). The
‘association’ is defined as a group of plots having similar species composition, structure, and habitat. This is the lowest, finest
level of the NVCS hierarchy.
Citations
Federal Geographic Data Committee. 2008. National Vegetation Classification Standard
http://www.fgdc.gov/standards/projects/FGDC-standards-projects/vegetation
Acknowledgements
We would like to acknowledge all the many field and data workers who collectively
provided the plot data for this analysis. Other support, including funding, was provided
by the following institutions: