United States
Environmental Protection
Agency
Environmental Research
Laboratory
Corvallis OR 97333
Research and Development
EPA-600/S3-82-037 Feb. 1983
Project Summary
Vegetative Delineation of
Coastal Salt Marsh Boundaries:
Evaluation of Methodology
H. Peter Eilers, Alan Taylor, and William Sanville
This research compares six vegeta-
tive methods for determining West
Coast coastal salt marsh boundaries.
A common data set consisting of 22
transects from 13 Oregon and Wash-
ington coastal salt marshes is used in
the evaluation. Multiple occurrence.
joint occurrence, and the 5% technique
require prior plant classification into
salt marsh, upland, and nonindicator
species; cluster analysis and the simi-
larity indices require no initial clas-
sification. Close agreement between
the methods suggests that plant pre-
classification and cover value deter-
minations may be unnecessary to
develop vegetative boundary loca-
tions. Examples of each method as
applied to specific coastal salt marsh
data are presented. Also included is an
upland/wetland plant indicator classi-
fication list derived from a consensus
of several West Coast plant specialists.
This Project Summary was devel-
oped by EPA's Environmental Re-
search Laboratory, Corvallis, OR. to
announce key findings of the research
project that is documented in a
separate report of the same title fsee
Project Report ordering information at
back).
Introduction
The Federal Water Pollution Control
Acts of 1972 and 1977 require that
wetlands receive special consideration
prior to any alteration. Justification for
this legislation is based both on research
and empirically derived values. Coastal
salt marshes are important for export of
energy-rich organic detritus and dis-
solved organic carbon, buffers against
shoreline erosion, improvement of
water quality, different stages in
migratory and endemic animal life
histories, plant habitat, and aesthetics.
To implement the Federal Water
Pollution Control Acts, it is necessary to
identify wetlands and determine wet-
land/upland boundaries. Boundary
determination is often difficult because
it frequently involves an ecotone
interdigitating between upland and
wetland. In order to identify the most
appropriate methods for boundary
determination, EPA funded five research
projects to evaluate boundary location
techniques and provide a generalized
wetland plant species list.
This report: (1) evaluates methods
used by the five research projects; (2)
presents alternative methods; (3) rec-
ommends the best approach to wetland
boundary delineation based on vegeta-
tion; and (4) provides appropriate plant
lists and computer software to apply
these methods to Pacific Coast salt
marshes. The evaluation is based on
salt marsh vegetation data, but it
appears that the methods can be appl ied
to other wetlands. Vegetation is only
one of several boundary delineation
alternatives. Soils and hydrology are
important considerations; the ideal
approach would probably combine
these with vegetation.
Methods
The vegetation methods evaluated
range from those based on indicator
-------
plant species lists to quantitative
ana ytical techniques such as cluster
ana ysis.
fnaicator Species
Boundary determination by the indi-
cate r species method is largely empirical
and based on changes in plant species.
Boundaries are sited where vegetation
shifts occur. The actual decision depends
primarily on the expert judgment and
taxf.nomic skill of the census taker or
data interpreter. Because the deter-
mination tends toward subjectivity, this
method may suffer under rigorous
scrutiny.
FN2 Percent
Tliis method is similar to that of the
indicator species but uses cover values
to quantify results. Plant cover values
are taken from quadrats along a
wetland/upland transect. Some varia-
tion in data interpretation may occur,
but the boundary delineations are
generally made at the point where five
percent of the vegetation is either
upland, proceeding from upland to
wetland, or five percent wetland,
proceeding from wetland to upland.
Results are generally presented graph-
ically (Figure 1 is a graphical presenta-
tion of results from thfc six quantitative
methods applied to the common set).
Joint Occurrence
This numerical technique requires
plant preclassification into marsh,
upland, and non-indicator categories. It
is based on a ratio of joint occurrences
to independent occurrences. This
technique surfers if the plant distribution
is patchy. To partially correct for this
pioblem, a standardized cumulative
index is computed and values an
plotted to determine boundary location
Multiple Occurrence
Weighting coefficients are assignei
to plants preclassified as in the join
occurrence method and a multiph
occurrence score is computed fo
quadrats along the gradient. Th<
quadrat scores are plotted and thu
boundary location determined where ;
predetermined numerical shift occurs.
Cluster
This procedure uses floristic date
without preclassification into marsh
upland, and non-indicator categories
Several measures can be used t(
develop the cluster program; the author;
use the Bray-Curtis dissimilarity meas
ure. The resulting dendrogram shows
quadrat clusters forming at decreasing
Transect 0310
to
/.or
0.5
1
0.4
0.2
0.0
Joint Occurre
Five percent
0 2 4 6 8 10 12 14 16 18
TOO'
80
60-
40
20
0
0 2
6
n
l)
70 ' 72~ 14
76
7/0.0
700.0
30.0
50.0
: 70.0
1 60.0
> 50.0
40.0
30.0
20.0
70.0
0.0
Similarity ISJ
LTZ
ULM
0
2 4 65 10 12
16
Multiple Occurrence
-8.0
Ffgvre 1.
1O 12 14 16 18
2.0
7.5
7.6
1.4
1.2
7.0
0.5
0.6
0.4
0.2
Cluster
rTl
432
Marsh
121110191'51'413
Transition Upland
110.0
100.0
90.0
80.0
70.0
60.0
50.0
40.0
30.0
20.0
10.0
0.0
Similarity ISE
LTZ
ULM
0 2 4 6
8 ^10
12 14 16
Comparison of six boundary determination methods for a single data set. Abscissa is transect distance in meters.
LTZ - Lower Transition Zone Limit. ULM - Upper Marsh Limit.
-------
dissimilarity levels. Interpretation is
based on the cluster patterns, these
generally grouping into upland, wetland,
and transition zone.
Similarity ISJ and ISE
The ISJ index is based on adjacent
quadrat plant presence/absence data.
This procedure requires binary data and
is attractive because it uses no preclass-
ification. A modification of this tech-
nique using species quantities rather
than just presence/absence is also
evaluated. Both methods present the
results graphically and the upper limit of
marsh is located where the similarities
are comparatively low.
Results
The indicator plant list upon which
portions of this report are based was
determined by consensus of EPA
researchers and university botanical
authorities. It has received extensive
review but should not be considered a
final compilation.
Each method, except for indicator
species, was used to evaluate a common
salt marsh data set. A lower limit of the
transition zone and an upper limit of the
marsh (wetland) were calculated for 22
transects. The primary effort was
calculation of the upper marsh limit
because of its significance in jurisdic-
tional questions.
There was close agreement in bound-
ary locations using these six different
methods. The species classification
techniques (five percent, joint occur-
rence, and multiple occurrence) had a
high intragroup correlation, as did the
nonclassification techniques (Table 1).
Some variability appears to result from
two transects which did not include
sufficient upland and wetland; others
are attributed to variations in methods.
An important observation is that pres-
ence/absence yields results almost
identical with the species-oriented
techniques.
Discussion and
Recommendations
The methods fall into two general
categories: those which require plant
preclassification and those which do
not. Techniques requiring prior plant
classification may be inherently biased.
Nonclassification techniques rely more
on quantitative analyses and are not as
prone to this problem. An important
result is that the presence/absence
techniques seem to provide as valid a
result as many of the classification
techniques. A schematic is presentedto
guide the reader in adapting or deter-
mining techniques for a specific evalua-
tion. It is critical that, whichever method
is used, validation by a trained field
person be done. It is also advisable to
use either the cluster or similarity
technique in conjunction with those
based on species because their use
further quantifies the results and makes
them more objective.
This paper evaluates methods using
coastal salt marsh vegetation data. The
authors presume the methods to have a
broader application than solely salt
marshes. It is also essential to recognize
that environmental factors are critical to
boundary determinations. Additional
work must be done with soil and
hydrology prior to the selection of a
"best" technique. The vegetative meth-
ods will facilitate wetland boundary
determinations until a more compre-
hensive approach is possible.
Table 1.
Lower Transition Zone Limit (LTZ) and Upper Limit of Marsh (ULM) as Determined by 6 Methods Applied to 22 Transects
from Frenkel et al. (1978). Limits Expressed as Distance (m) Along Transect Where Distance Increases from Marsh to
Upland
Joint Multiple
Five Percent Occurrence Occurrence
Cluster
Similarity Similarity
ISJ ISE
Transect
Number Location
Oregon
0105 Coquille Estuary
0208 Coos Bay
0301 Alsea Bay
0310 Alsea Bay
0402 Yaquina Bay
0407 Yaquina Bay
0704 Nehalem Bay
0706 Nehalem Bay
0710 Nehalem Bay
LTZ
11.0
16.5
9.0
—
—
4.5
1.0
10.5
—
ULM
14.5
19.5
15.5
13.0
19.5
19.5
11.0
13.0
16.0
LTZ
9.0
16.5
—
—
—
4.5
1.0
10.5
—
ULM
14.5
21.5
15.5
13.5
19.5
19.5
11.5
13.5
15.5
LTZ
11.5
—
1O.O
10.0
—
7.5
—
10.5
—
ULM
13.0
21.0
15.0
12.0
18.5
19.5
8.0
11.1
15.0
LTZ
9.0
—
9.0
9.0
13.5
1.5
7.0
10.5
—
ULM
14.5
19.5
15.5
13.5
19.5
19.5
15.5
15.5
15.5
LTZ
11.5
—
9.0
7.0
13.5
10.5
—
7.0
—
ULM
15.5
21.5
15.5
13.5
19.5
19.5
9.0
16.5
15.5
LTZ
12.5
—
9.O
9.0
13.5
10.5
—
12.5
—
ULM
14.5
21.5
15.5
13.5
19.5
19.5
90
16.5
15.5
ULM
Mean
14.4
20.8
15.4
13.2
19.3
19.5
10.7
14.4
15.5
ULM
S.D.
0.8
1.0
0.2
0.6
0.4
0.0
2.7
2.2
0.3
ULM
Range
2.5
2.0
0.5
1.5
1.0
0.0
7.5
5.4
1.0
Washington
0804
O8O8
0809
0910
1001
1103
1201
1606
1610
1611
1612
1703
1802
Willapa Bay
Willapa Bay
Willapa Bay
Willapa Bay
Willapa Bay
Grays Harbor
Grays Harbor
Thorndyke Bay
Thorndyke Bay
Thorndyke Bay
Thorndyke Bay
Snohomish Estuary
Oak Bay
14.5
—
15.0
84.5
256.0
105.5
18.5
—
—
9.0
—
—
—
15.5
—
22.5
57.5
265.0
146.0
19.5
—
6.0
12.5
21.5
7.5
26.0
14.5
—
—
—
—
/05.5
—
—
3.5
—
—
—
—
76.5
—
22.5
87.5
265.0
147.5
19.5
—
7.5
12.5
21.5
7.5
25.5
11.0
8.0
15.0
63.5
248,0
117.5
—
—
—
6.0
1.0
—
—
15.0
—
22.0
87.5
259.0
729.5
79.0
—
3.0
12.0
20.0
6.0
25.5
9.0
5.0
19.0
—
—
777.5
77.0
—
—
—
72.0
—
—
75.5
75.5
22.5
57.5
259.0
747.5
79.5
—
705
7O.5
23.5
37.5
25.5
9.0
—
20.5
65.0
—
777.5
77.0
—
—
4.5
—
—
70.5
75.5
_
22.5
57.5
249.0
747.5
79.5
10.5
10.5
10.5
12.0
31.5
25.5
9.0
—
79.0
65.0
—
95.0
770
—
—
4.5
12.0
—
79.5
15.5
—
22.5
87.5
249.0
7475
795
70.5
70.5
70.5
23.5
37.5
25.5
15.6
—
22.4
87.5
257.7
7443
794
—
8.0
77.4
20.3
793
25.6
0.5
—
02
0.0
72
73
0.2
—
3.7
7.0
4.3
73.4
0.2
7.5
—
0.5
0.0
76.0
78.0
0.5
—
7.5
2.0
77.5
25.5
0.5
U.S. GOVERNMENT PRINTING OFFICE: 1983 659-OI7/O899
-------
The EPA authors H. Peter Eilers, Alan Taylor, and William Sanville (also the
EPA Project Officer, see below) are with the Environmental Research Labora-
tory. Corvallis. OR 97333.
The complete report, entitled "Vegetative Delineation of Coastal Salt Marsh
Boundaries: Evaluation of Methodology," (Order No. PB 83-1.6? 441; Cost:
$ 10.00, subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield. VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Research Laboratory
U.S. Environmental Protection Agency
200 SW 35th Street
Corvallis. OR 97333
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Postage and
Fees Paid
Environmental
Protection
Agency
EPA 335
Official Business
Penalty for Private Use $300
US S pSviR^PROTECTiON
REGION t> LIBRARY
?30 S DfcAKRORn STKEfcJ
IL
------- |