Full text of DOI:10.1672/0277-5212(2000)020[0581:EONWIM]2.0.CO;2

WETLANDS, Vol. 20, No. 4, December 2000, pp. 581–587
᭧
2000, The Society of Wetland Scientists
EVALUATION OF NATIONAL WETLAND INVENTORY MAPS IN A HEAVILY
FORESTED REGION IN THE UPPER GREAT LAKES
Gregory M. Kudray1,2 and Margaret R. Gale1
1 School of Forestry and Wood Products
Michigan Technological University
Houghton, Michigan, USA 49931
2 Present Address:
Ecological Inventory and Analysis
RR 1, Box 492
Chassell, Michigan, USA 49916
Abstract: National Wetland Inventory (NWI) maps are widely used in the United States but have not been
independently evaluated in the Great Lakes region nor in forested areas with level topography. Field data
from 148 plots in the Hiawatha National Forest ecological classification and inventory program were com-
bined with an additional review to evaluate NWI mapping accuracy. NWI maps were over 90% accurate in
identifying uplands and jurisdictional wetlands. All nonforested wetlands were identified correctly. Uplands
were correctly identified 96.9% of the time. The lowest level of accuracy, 90.7%, was achieved in identifying
forested wetlands. The most common error was the NWI classification of wetlands on the AuGres soil series,
a somewhat poorly drained upland soil that often occurs in complexes with wetland soils in the region.
Forested wetlands with a cover type similar to adjacent uplands were also a source of error on NWI maps.
The already high accuracy of NWI maps could be improved by the mapping of wetland–upland complexes,
a development corresponding to the increased mapping of hydric-nonhydric soil complexes in area soil
surveys. The continued refinement of regional lists of hydrophytic vegetation is supported by indicator status
discrepancies between an extensive Hiawatha National Forest database and the current NWI list for the
region.
Key Words: map accuracy, maps, National Wetland Inventory, plant indicators, soils
INTRODUCTION
Golet 1995) but are nevertheless used as a major tool
in regulatory activities such as the ‘‘Swampbuster’’
provision of the 1985 Food Security Act (Wilen and
Bates 1995).
The National Wetlands Inventory (NWI) maps and
associated digital data are produced by the United
States Fish and Wildlife Service (USFWS) and com-
prise a readily available nationwide data source that is
used by local, state, and federal agencies, as well as
by private industry and other organizations (Wilen and
Pywell 1981, Wilen and Bates 1995). User surveys
have documented over 100 different uses of the NWI
maps, including regulatory-based activities (Wilen and
Bates 1995). Uses have included comprehensive re-
source management plans, environmental impact as-
sessments, facility and corridor siting, oil spill contin-
gency plans, natural resources inventories, and habitat
surveys. The wide availability and use of the NWI
products have generated considerable interest and
sometimes controversy (Gillis 1996, Tiner 1997a) in
the ability of these maps to accurately identify and
delineate wetlands. Part of the problem is that the NWI
maps were never intended to map wetlands as regu-
lated by various state and federal laws (Cowardin and
The number of wetlands that meet the regulatory
definition of a wetland on NWI maps is unknown
(Stolt and Baker 1995). Additionally, wetlands can be
missed on NWI maps for various reasons (Tiner
1997a). Remote sensing, the primary tool used to iden-
tify and map wetlands in the NWI, cannot easily detect
certain wetlands with cover types similar to surround-
ing uplands, or easily distinguish drier-end wetlands.
Small or linear wetlands are often not mapped; neither
are most farmed wetlands. The minimum mapping unit
(mmu) size on NWI maps has depended on the aerial
photographs available. High-altitude photographs at a
scale of 1:80,000 were used from 1975 to the early
1980s, with a mmu unit size of 7.5–12.5 ha for early
NWI maps (Tiner 1990, Tiner 1997a). In 1980, the
U.S. Geological Survey (USGS) initiated a national
program of acquiring 1:58,000 scale color infrared
(CIR) photographs. The larger scale and CIR emulsion
581

582
WETLANDS, Volume 20, No. 4, 2000
has enabled a current mmu of 0.5–1.2 ha for forested
wetlands and a mmu of 0.5 ha for non-forested ponds
and pothole marshes.
ognized by the NWI as problem areas with a printed
disclaimer on each NWI map.
The ability of the NWI to accurately identify wet-
lands depends on the landforms involved. The pub-
lished verification studies all occurred in predomi-
nantly upland landscapes located in the eastern United
States (Swartwout et al. 1981, Crowley et al. 1988,
Nichols 1994, Stolt and Baker 1995, K. Klemow and
M. Mohseni pers. comm.) with one midwestern ex-
ception in Nebraska (Kuzila et al. 1991). No verifi-
cation studies have been reported for the wetland-rich
upper Great Lakes area. The objective of our study
was to evaluate how well the NWI identified jurisdic-
tional wetlands in a landscape dominated by forested
wetlands, the most difficult wetland type to identify
from aerial photographs (Dahl 1992). The predominant
landform is a low and level glacial lake plain with
upland–wetland boundaries that are difficult to accu-
rately photointerpret (National Research Council
1995). The study area is mostly coniferous wetlands,
recognized as a difficult cover type to accurately pho-
tointerpret because of year-round canopy retention pre-
venting the observation of saturated soils (Tiner 1990).
This cover type is often similar to adjacent uplands,
presenting another difficulty to the photointerpreter.
Plot data from the Hiawatha National Forest ecosystem
mapping and inventory program were used to evaluate
NWI data. Since wetland boundaries need to be field-
verified for jurisdictional purposes, this study focused
on evaluating what the NWI attempts to achieve: an
inventory of wetlands at a scale and level of detail
achievable with the remote sensing methods used.
Several studies have evaluated the accuracy of NWI
maps in various regions (Swartwout et al. 1981, Crow-
ley et al. 1988, Kuzila et al. 1991, Nichols 1994, Stolt
and Baker 1995, K. Klemow and M. Mohseni pers.
comm.). One approach was to evaluate accuracy by
overlaying NWI maps with soils maps (Kuzila et al.
1991, K. Klemow and M. Mohseni pers. comm.). In
Nebraska, an overlay of a 1981 soil survey and a 1981
NWI map found 94.2% agreement in wetland and non-
wetland designations based on considering hydric soils
as wetlands (Kuzila et al. 1991). The soil survey clas-
sified more areas as wetlands than did the NWI. Kle-
mow and Mohsehi (K. Klemow and M. Mohseni pers.
comm.) found a very different result in Pennsylvania.
They found that 66% of the hydric soil areas were not
marked as wetlands on the NWI maps. In addition,
97% of subhydric soil areas, which they described as
often being jurisdictional wetlands, and 26% of open
water areas were left off NWI maps. The map overlay
approach has been criticized because of the lack of
field verification for either source and the difference
between hydric soil and NWI wetland designations
(Tiner 1997a). Discrepancies between hydric soil map-
ping and NWI wetland types can be due to the inclu-
sion of up to 40% of a nonhydric soil within a hydric
soil mapping unit. In addition, soil maps generally do
not distinguish between undrained hydric soils and
drained or recently filled hydric soils (Tiner 1997a).
Reasons for differences also include the fact that most
somewhat poorly drained soils are not jurisdictional
wetlands and that drier wetlands are difficult to pho-
tointerpret and are mapped conservatively by the NWI
(Tiner 1997a).
Other evaluation methods, however, have included
field verification. Swartwout et al. (1981) used this
technique and found NWI maps to be very accurate,
over 95% correct, in differentiating wetland and up-
land types. A similar accuracy level was determined
in Vermont (Crowley et al. 1988) and Maine (Nichols
1994). In Virginia, Stolt and Baker (1995) found that
91% of palustrine wetlands identified on NWI maps
met the criteria as a jurisdictional wetland. In evalu-
ating the effectiveness of the NWI maps to inventory
all the jurisdictional wetlands, they found that the NWI
often underestimated the size of wetlands; most wet-
lands in this area were linear features along water-
courses and difficult to delineate accurately. Small
wetlands with forest cover were often missed entirely.
Some of the wetlands identified were below the mmu
size recognized on the NWI maps. Small wetlands and
those obscured by dense forest cover have been rec-
STUDY AREA
The Hiawatha National Forest is about 400,000 ha
in area and is located in the eastern half of the Upper
Peninsula of Michigan. The Forest is divided into two
separate units, an eastern half and western half. All of
the work for this study was completed in the western
half (Figure 1). A regional landscape ecosystem clas-
sification (Albert 1995) places this area in two ecolog-
ical sub-subsections. Most of the study area is the Se-
ney sand lake plain sub-subsection with landforms of
lacustrine origin containing the largest expanse of wet-
lands in Michigan (Albert 1995). It is further charac-
terized as having very poorly or excessively drained
sand lake plains, transverse dunes, outwash with shal-
low paludified peatlands (many patterned), jack pine
barrens, and hardwood-conifer and conifer swamps
(Albert 1995). The Grand Marais sandy end moraine
and outwash sub-subsection contains the rest of the
study area. Albert (1995) describes this landscape as
having sandy end-moraine ridges and outwash aprons,
Lake Superior shoreline features, transverse dunes,

Kudray & Gale, EVALUATION OF NATIONAL WETLAND INVENTORY MAPS
583
Figure 1. Location of the study area in the Upper Peninsula of Michigan, USA.
sand spits with white pine-red pine forests, jack pine
barrens, red pine forests, northern hardwood forest,
and patterned peatlands. Forested, conifer-dominated
wetlands are the most common wetland vegetation
type; mixed conifer-deciduous forested wetlands are
also widespread.
wet Spodosols with an iron, aluminum and humus-
enriched B horizon. Entic and Typic Haplorthods are
the most common upland soils. Some of the area has
Aquic Eutroboralfs soils, formed in fine-textured la-
custrine deposits.
Histosols composed of incompletely decayed organ-
ic material are common wetland soils in the lowest
parts of topographic depressions. At the edges of or-
ganic filled basins, the soils typically transition into
mineral soils, Entic or Typic Haplaquods, which are
METHODS
Field work was conducted during a 1994 ecological
classification and inventory (ECI) program that
mapped approximately 20,000 ha in the Hiawatha Na-
tional Forest. The ECI data included information on
soil organic depth, depth to water table, depth to re-
doxymorphic features, and vegetation composition.
ECI types were determined with a key based on water
chemistry, depth to water table and soil and vegetation
indicators (Kudray 1995). Previous mapping had oc-
curred prior to the development of the ECI wetland
classification (Kudray 1995), with wetland and upland
types based on U.S. Department of Agriculture
(USDA) soil taxonomy drainage classes. To maintain
consistency with the earlier mapping, the same criteria
were used in the 1994 mapping. Upland types were
identified through the location of redoximorphic fea-
tures or water tables at least 60 cm or deeper in the
soil profile. Two main wetland groups, called landty-
pes, were recognized (Table 1). Landtype 80 included
Table 1. Hiawatha National Forest ecosystem classification sum-
mary at the landtype level. Included are wetland and near-wetland
types.
Jurisdictional
Landtype
Criteria for Inclusion
Wetland?
Landtype 60
Water table or soil redoxyi-
morphic features Ͼ60 cm
and Ͻ100 cm
No
Landtype 70
Landtype 80
Water table or soil redoxyi-
morphic features Ͻ60 cm,
soil organic horizon Ͻ30
cm
Soil organic horizon Ն30
cm
Variable
Yes

584
WETLANDS, Volume 20, No. 4, 2000
ECI Landtype. If the plot was coded upland for both
systems, the NWI map was considered correct for that
plot. If a plot was coded as Landtype 80 and also cod-
ed as a NWI wetland, the NWI map was considered
correct for that plot. Since Landtype 70 plots could be
either jurisdictional wetlands (Environmental Labora-
tory 1987) or uplands, these plots were identified as
needing further review. Other plots with ECI–NWI
coding that did not agree as to wetland–upland typing
were also identified as needing further review.
Questionable plots were further reviewed by relo-
cating these plots on NWI maps. Five plots were lo-
cated within 100 m of a wetland–upland boundary and
removed from the analysis since the objective was not
to evaluate fine resolution jurisdictional wetland
boundaries. The methods used were intended to cap-
ture broader scale NWI type misclassifications. The
plots were then examined using the ECI data, CIR
photos, and the local USDA soil survey (Berndt 1977).
If the ECI data and the soil survey clearly indicated
that the plot was a wetland or an upland, no additional
field visit was made. If the data were considered am-
biguous, the plot was revisited.
The nine plots that needed to be reexamined on the
ground were relocated by using the GPS and aerial
photographs. A routine wetland determination proce-
dure (Environmental Laboratory 1987) was then con-
ducted to determine if the plot met the jurisdictional
wetland requirement. This determination procedure
consists of a stepwise evaluation of wetland indicators.
If hydrophytic vegetation is not dominant, the area is
not a wetland. If the vegetation is inconclusive, further
steps evaluating the presence of wetland hydrology
and hydric soils are undertaken. All wetland indicators
were recorded during our evaluation.
Figure 2. Location of the NWI maps (Steuben SW, Steu-
ben NW, Corner Lk. NE. and Corner Lk. SE quadrangles)
that contained the study plots within the western half of the
Hiawatha National Forest, Michigan, USA.
all wetlands with an organic soil depth 30 cm or great-
er. Landtype 70 included all other types with a water
table or soil redoxymorphic features shallower than 60
cm. Landtype 70 could then be either a jurisdictional
wetland (Environmental Laboratory 1987) or an up-
land.
RESULTS AND DISCUSSION
Plot location Universal Transverse Mercator (UTM)
coordinates were recorded in the field using a Trimble
Pathfinder Basic Global Positioning System (GPS) re-
ceiver. The data were post-processed using differential
correction software to maximize locational accuracy to
a probable accuracy level of a few meters. Plots were
marked on CIR aerial photographs. Plot UTM coor-
dinates were plotted manually on NWI draft 1:24,000
scale maps and coded with either a NWI wetland type
designation or as upland. The NWI maps evaluated
were the Steuben SE, Steuben NW, Corner Lk. NE,
and Corner Lk. NW quadrangles (Figure 2). Although
final versions of the NWI maps became available dur-
ing the project, there were no differences between the
draft and final versions. A total of 148 plots comprised
this data set.
Of the 143 plots in the final data set, seven (5%)
were considered misclassified by the NWI based on
field verification through the Hiawatha National Forest
ecosystem mapping and additional review (Table 2).
Jurisdictional wetlands were correctly identified on
NWI maps at 93.7% of the sampled points. A high
level of accuracy was also achieved in differentiating
uplands from wetlands; only 2 of 64 (3.1%) NWI-
mapped upland points were true jurisdictional wet-
lands. All nonforested wetlands were correctly identi-
fied. The lowest level of accuracy was achieved in
identifying forested wetlands; however, they were cor-
rectly identified at an accuracy rate of 90.7%. The mis-
classified polygons were substantially larger in size
than the NWI target mmu size, which is .4 to 1.2 ha
for this region (Tiner 1997b). Tiner (1997a) states that
NWI maps tend to err more by omission than com-
For each plot, the NWI type was compared to the

Kudray & Gale, EVALUATION OF NATIONAL WETLAND INVENTORY MAPS
585
Table 2. Classification matrix comparing NWI maps to jurisdic-
tional wetlands identified through Hiawatha National Forest ECI
mapping and additional review.
Table 3. Typing summary of plots misclassified by the NWI
based on Hiawatha N.F. ECI mapping and additional review.
Juris-
Wet- Hydro- dic-
land phytic tional
Number
of
Plots
Sampled Correct
Number
Percent
Correct
SCS mapped Hydric Hydro- Vegeta- Wet-
NWI Category
NWI type*
soil series
soil logy** tion*** land
All
Upland
Wetland
Forested wetland*
Non-forested wetland
143
64
79
54
25
136
62
74
49
25
95.1
96.9
93.7
90.7
100.0
Upland
Upland
PSS/EMB
PFO4/1Y
PFO4/1Y
PFO4B
Bruce
Yes
Yes
No
No
No
No
Yes
Yes
Yes
No
No
No
No
No
Yes
Yes
Yes
Yes
No
Yes
Yes
No
No
No
No
No
Roscommon
Au Gres
Au Gres
Au Gres
Au Gres
Tawas
No
Yes
* Forested wetland includes forested and forested/nonforested mixed
types.
PFO4/1B
* PSS/EMB ϭ Palustrine Scrub/Shrub–Emergent Saturated. PFO4/1Y ϭ
Palustrine Forested Needle-Leaved Evergreen–Broad-Leaved Deciduous
Saturated Semipermanent. PFO4B ϭ Palustrine Forested Needle-Leaved
Evergreen Saturated. PFO4/1B ϭ Palustrine Forested Needle-Leaved Ev-
ergreen–Broad-Leaved Deciduous Saturated.
** Wetland hydrology based on visual observation of soil saturation with-
in 30 cm.
*** Hydrophytic vegetation is present if more than 50% of the dominant
species in the trees, sapling/shrub, herb vegetation layers have an indicator
status of OBL, FACW, and/or FAC. Wetland indicator status if from the
North Central region of the national list of plant species that occur in
wetlands: 1996 national summary (Reed 1997).
mission: a conservative approach indicating that areas
mapped as wetlands are usually true wetlands. A more
common error would then involve the omission of true
wetlands. That was not the case here, although the er-
ror rate was still quite low. These misclassifications
represent more inclusive mapping of uplands or up-
land-wetland complexes as wetlands by the NWI.
NWI misclassification errors fell into two types: 1)
somewhat poorly drained uplands with a mixed coni-
fer-deciduous forest cover similar to nearby wetlands
and 2) wetlands with an unusual cover type more sim-
ilar to uplands than the majority of forested wetlands
in the area. The first error type was the most common,
occurring in 71% of the misclassified plots. Most of
these were on the same somewhat poorly drained but
nonhydric (U.S. Department of Agriculture Soil Con-
servation Service 1993) Au Gres soil type (Table 3).
Au Gres soils are sandy mixed, frigid Entic Endoa-
quods: somewhat poorly drained sandy soils charac-
terized by a leached gray albic horizon over a dark
reddish-brown spodic horizon.
soil). The remaining four were larger (Ͼ12 ha) areas
of somewhat poorly drained soils, three mapped as Au
Gres soil types and one incorrectly mapped as a hydric
organic soil (Berndt 1977). The correct NRCS soil
type based on the ground truth would also have been
an Au Gres soil. The forested cover type on all these
plots was similar to adjacent wetlands. The NWI
mapped four of these areas as palustrine forested nee-
dle-leaved evergreen dominated wetlands. The fifth
was mapped as a scrub/shrub–emergent, a misinterpre-
tation due to young tree regrowth after a recent timber
harvest on the site.
The wetland plots that were misidentified by the
NWI as uplands were both large (Ͼ50 ha) species-rich
With one exception, the existing soil survey (Berndt
1977) correctly identified the misclassified plots as a
hydric or nonhydric soil. While the NWI relies heavily
on soil surveys and reports for ground-truth informa-
tion (Dahl 1993), Tiner (1997b) cautions that hydric
soil mapping in soil surveys cannot be directly com-
pared with NWI map units. He suggests that the map-
ping of hydric soils on survey maps may tend to over-
estimate the occurrence of wetlands for several rea-
sons, including nonhydric inclusions and soil series
that were not designed to strictly adhere to the defi-
nition of a hydric soil. The Au Gres soil series that
was most commonly incorrectly mapped as wetland by
the NWI is a somewhat poorly drained soil that is now
often mapped in Upper Peninsula of Michigan soil sur-
veys in complexes with hydric soils (L. Carey, pers.
comm.). In this area with subtle topographic changes,
wetlands often occur in complexes with uplands and
forested wetlands with a tree cover dominated by as-
pen (Populus spp.) with a substantial conifer compo-
nent. This is a cover type similar to the surrounding
uplands and much less common in the study area than
conifer-dominated forested wetlands. These plots were
clearly wetlands and probably misinterpreted as up-
lands on aerial photographs due to the similarity of the
cover type to the surrounding upland types. The cor-
rect NWI type for these polygons should have been
PFO1/4B, Palustrine Forested Broad-leaved Decidu-
ous–Needle-leaved Evergreen Saturated.
Of the five upland plots misidentified as wetlands,
one was a small (6 ha) upland ridge in a predominant
wetland matrix mapped as an Au Gres soil type
(Berndt 1977) but field-verified during ECI mapping
as a Croswell series, (topographically a slightly higher

586
WETLANDS, Volume 20, No. 4, 2000
cannot be accurately mapped exclusively as upland or
wetland. As a future option, Tiner (1997b) suggests
that the NWI could map some areas as ‘‘potentially
supporting some wetlands.’’ This approach could be
usefully applied to regions with landforms similar to
our study area. The increased mapping of hydric-non-
hydric soil complexes in soil surveys is a correspond-
ing development.
The procedure used to determine wetland/upland
identity in this study is the current federal jurisdiction-
al wetland definition (Environmental Laboratory
1987), which is more restrictive and requires more
proof than other current or proposed wetland defini-
tions (e.g., Cowardin et al. 1979, Federal Interagency
Committee for Wetland Delineation 1989). The pri-
mary difference between the current federal jurisdic-
tional wetland definition and the USFWS definition
(Cowardin et al. 1979) is that a USFWS wetland can
be identified with a positive indicator for any one of
the three vegetation, soil, or hydrology parameters,
while all parameters must be positive to identify a ju-
risdictional wetland.
Since the NWI was never intended to directly map
jurisdictional wetlands and jurisdictional definitions
change over time, the characteristics of the upland
plots misclassified as wetlands were further examined.
Many wetland definitions require hydrophytic vegeta-
tion; three of the misclassified plots had hydrophytic
vegetation but lacked hydric soils or wetland hydrol-
ogy. It can be expected that in the continuum of eco-
systems, some may have one wetland factor and lack
others, but a further examination of the vegetation data
suggested that local variability in the indicator status
of wetland plants could be considered as the cause.
For example, in the North Central region, balsam fir
(Abies balsamea (L.) P. Mill.) has an indicator status
of FACW—facultative wetland, usually occurring in
plain with forested wetlands on subtle topography. The
National Research Council (1995) reported that ‘‘map-
ping wetlands in level landscapes, such as coastal or
glaciolacustrine plains, is less precise because bound-
aries are not as evident.’’ Dahl (1992) considers for-
ested wetlands the most difficult wetland type to iden-
tify from aerial photographs. Additionally, the domi-
nant wetland type in the study area, evergreen forested
wetland, is among the most difficult forested wetlands
to identify due to canopy retention (Tiner 1990). This
cover type occurs both in the wetland areas and the
surrounding uplands, a situation recognized by Tiner
(1990) to represent the biggest problem in detecting
forested wetlands. Despite these difficulties, the high
level of NWI classification accuracy in our area agrees
with studies in other regions (Swartwout et al. 1981,
Crowley et al. 1988, Nichols 1994, Stolt and Baker
1995). All of these studies have indicated that NWI
maps correctly identify wetlands at an accuracy level
of over 90%. Although regional evaluations of NWI
maps should continue due to the unique problems each
regional landscape presents in the photointerpretation
of wetlands, the growing body of evidence supports
the ability of NWI maps to accurately identify wetland
areas throughout the nation. The only work suggesting
that NWI maps do not accurately map wetlands (K.
Klemow and M. Mohseni pers. comm.) should be ei-
ther considered a notable exception in the overall qual-
ity of NWI maps or reevaluated due to the map-com-
parison methodology used and the lack of ground-truth
verification. NWI maps cannot be evaluated directly
with hydric soils areas on soil survey maps for a va-
riety of reasons (Tiner 1997a), but using soil surveys
and hydric soils lists to provide additional information
can help identify areas that will require a more inten-
sive examination.
Despite the fact that the USFWS wetland definition
is an ecological, not a regulatory, definition and that
the NWI maps do not directly define jurisdictional
wetlands, it is obvious that NWI maps are used as tools
in many activities that do involve regulated wetlands.
It is important to know how accurately NWI maps
identify and delineate wetlands regulated by the fed-
eral definition (Environmental Laboratory 1987) but
only with the realization that the NWI program goal
is neither to set absolute boundaries on regulated wet-
lands nor to map wetlands smaller than their minimum
target size. Final regulatory wetland identification and
delineation must remain an on-the-ground activity.
Since small wetlands less than the current NWI mmu
unit size of about .5 ha (Tiner 1990) are not within
the scale of NWI mapping, an accuracy assessment
using no minimum size criterion is informational but
not a valid criticism of NWI products. A more appro-
priate question is how accurately the NWI identifies
wetlands (estimated probability 67%
Ϫ 99%) but oc-
casionally found in non-wetlands (Reed 1997). A re-
view of the Hiawatha N.F. ECI database indicated that
the species is just as likely to occur in uplands. Of the
1466 plots in the ECI database that could be consid-
ered wetlands based on soil or hydrologic character-
istics, balsam fir was found in 56%, but it also oc-
curred in 71% of 1258 upland plots. Some other spe-
cies commonly found also differed from the regional
list in wetland indicator status. This finding confirms
the National Research Council (1995) recommenda-
tion to continue the refinement of regional hydrophytic
vegetation lists and to rely on other indicators when
the vegetation community is near a hydrophytic–non-
hydrophytic threshold.
The characteristics of this study area make it a par-
ticularly difficult landscape to accurately photointer-
pret wetland types. Much of the area is a lacustrine

Kudray & Gale, EVALUATION OF NATIONAL WETLAND INVENTORY MAPS
587
Environmental Laboratory. 1987. Corps of Engineers wetland delin-
eation manual. U.S. Army Engineer Waterways Experiment Sta-
tion, Vicksburg, MS, USA. Technical Report Y-87–1.
Federal Interagency Committee for Wetland Delineation. 1989. Fed-
eral manual for identifying and delineating jurisdictional wetlands.
U.S. Army Corps of Engineers, U.S. Environmental Protection
Agency, U.S. Fish and Wildlife Service, and U.S. Department of
Agriculture, Soil Conservation Service, Washington, DC, USA.
Gillis A. M. 1996. Make sure you have a good map. Bioscience 46:
808–809.
wetlands within its operational limits and across vary-
ing regions, landforms, soils, and cover types.
ACKNOWLEDGMENTS
Grateful acknowledgment is given to the Hiawatha
National Forest and their personnel for the funding and
support of this work. Special thanks to Eunice Padley,
Janet Sabernagle, Kirsten Saleen, and Don Howlett of
the Forest Service for their long-term support of eco-
logical research in the wetlands of the Hiawatha Na-
tional Forest. Members of the field crew who partici-
pated in this research including Andrew Geffert, Neil
Wilkie, and several others also have our gratitude for
their many hours of dedicated work in difficult con-
ditions.
Kudray, G. M. 1995. Hiawatha National Forest (western half) land-
type and landtype phase key. May 1996 draft. Hiawatha National
Forest, Escanaba, MI, USA.
Kuzila, M. S., D. C. Rundquist, and J. A. Green. 1991. Methods for
estimating wetland loss: the Rainbasin region of Nebraska, 1927–
1981. Journal of Soil and Water Conservation 46:441–445.
National Research Council. 1995. Wetlands Characteristics and
Boundaries. National Academy Press Washington, DC, USA.
Nichols, C. 1994. Map accuracy of National Wetlands Inventory
Maps for areas subject to Maine Land Use Regulation Commis-
sion Jurisdiction. U.S. Fish and Wildlife Service, Hadley, ME,
USA. Ecological Services Report R5–94/6.
Reed, P. B. Jr. 1997. National list of plant species that occur in
wetlands: 1996 National Summary. U.S. Fish and Wildlife Ser-
vice, St Petersburg, FL, USA.
LITERATURE CITED
Stolt, M. H. and J. C. Baker. 1995. Evaluation of National Wetland
Inventory Maps to inventory wetlands in the southern Blue Ridge
of Virginia. Wetlands 15:346–353.
Swartwout, D. J., W. P. MacConnell, and J. T. Finn. 1981. An eval-
uation of the National Wetlands Inventory in Massachusetts. p.
685–691. In Proceedings of the In-Place Resource Inventories
Workshop, Orono, ME, USA.
Tiner, R. W. 1997a. NWI maps—basic information on the nation’s
wetlands. Bioscience 47:269.
Tiner, R. W. 1997b. NWI maps: what they tell us. National Wet-
lands Newsletter 19:7–12.
Tiner, R. W. Jr., 1990. Use of high-altitude aerial photography for
inventorying forested wetlands in the United States. Forest Ecol-
ogy and Management 33/34:593–604.
U. S. Department of Agriculture Soil Conservation Service. 1993.
Hydric soils of Michigan. U.S. Department of Agriculture Soil
Conservation Service, East Lansing, MI, USA.
Wilen, B. O. and M. K. Bates. 1995. The U.S. Fish and Wildlife
Service’s National Wetland Inventory project. Vegetatio 118:153–
169.
Wilen, B. O. and H. R. Pywell. 1981. National Wetlands Inventory
map production. p. 189–190. In Proceedings of the In-Place Re-
source Inventories Workshop, Orono, ME, USA.
Albert, D. 1995. Regional landscape ecosystems of Michigan, Min-
nesota, and Wisconsin: a working map and classification. U.S.
Department of Agriculture, Forest Service, North Central Forest
Experiment Station, St. Paul, MN, USA. General Technical Report
NC-178.
Berndt, L. W. 1977. Soil Survey of Delta county and Hiawatha
National Forest of Alger and Schoolcraft counties, Michigan. U.S.
Department of Agriculture Soil Conservation Service and Forest
Service. Washington D.C., USA.
Cowardin, L. M. and F. C. Golet. 1995. U.S. Fish and Wildlife
Service 1979 wetland classification: a review. Vegetatio 118:139–
152.
Cowardin, L. M., V. Carter, F. C. Golet, and E. T. LaRoe. 1979.
Classification of wetlands and deepwater habitats of the United
States. U.S. Fish and Wildlife Service, Washington, DC, USA.
FWS/OBS-79/31.
Crowley, S., C. O’Brien, and S. Shea. 1988. Results of the wetland
study and the 1988 draft wetland rules. Agency of Natural Re-
sources, Division of Water Quality, Waterbury, VT, USA.
Dahl, T. E. 1992. Wetland status and trends—the link to remote
sensing. p. 221–230. In Proceedings of Remote Sensing & Natural
Resource Management, The Fourth Forest Service Remote Sens-
ing Applications Conference, Orlando, FL, USA.
Dahl, T. E. 1993. Use of hydric soils in national programs of the
U.S. Fish and Wildlife Service. p. 348. In Agronomy Abstracts,
American Society of Agronomy, Madison, WI, USA.
Manuscript received 12 November 1999; revisions received 18 May
2000; accepted 19 July 2000.