A TECHNICAL SUMMARY OF WETLAND
TCFSTOR ATTON COSTS T1SJ THF
1\ JL-j A ivii, X. x v/ x N V—- vy C A X JL ^1 XXX Xj
CONTINENTAL UNITED STATES
%
Dennis M. King and Curtis C. Bohlen
University of Maryland System
Center'for Environmental and Estuarine Studies
Cheseapeake Biological Laboratory
P.O. Box 38, Solomons, Maryland 20688
January 1994
University of Maryland, CEES Technical Report UMCEES-CBL-94-048, April 1994.
. Research funded under:
Environmental Protection Agency Cooperative Agreement No. CR-818227-CI .
and
Department of Energy Contract No. DE-AC22-92MT92006
EPA 230-R-94-023

-------
Technical Summary of Wetland Restoration Costs
1
INTRODUCTION
Most activities that impact wetlands are regulated by federal law under
section 404 of the Clean Water Act. State wetland protection laws are also in
effect in many states. These laws require anyone who proposes activities that
could adversely impact wetlands to obtain a permit. In what has become
known as "sequencing/' permit seekers must show that they have avoided
wetland impacts to the maximum extent practicable, that they have mini-
mized any unavoidable wetland impacts, and that they have or will mitigate
any remaining wetland impacts through wetland creation, restoration, or en-
hancement projects. The research described in this report deals with the last
step of this wetland permitting process. In particular, it focuses on the cost of
providing compensatory mitigation for wetland impacts that are permitted
under the Section 404 program and comparable state programs.
Historically, the level of mitigation required for permit approval was
determined on an ad hoc basis through negotiation between permit seekers
and regulators. Mitigation requirements typically were far below replacement
levels. In 1988, a broadly based and influential wetland policy forum that was
convened to explore wetland policy alternatives recommended a "no-net-
loss" goal for federal wetland policy (Conservation Foundation 1988). The
goal called for a halt in the net loss of wetland resources, not only by
restricting activities that harm wetlands, but also by expanding activities that
increase wetlands and wetland functions. The federal government officially
espoused this goal in 1990, in a Memorandum of Agreement between the
Army Corps of Engineers and the Environmental Protection Agency, which
spelled out wetland sequencing and compensation procedures
(Environmental Protection Agency and Department of the Army 1990). The
no-net-loss goal and the Memorandum of Agreement increased both the
significance and attention given to wetland mitigation within the federal
wetland regulatory scheme. This new focus on wetland mitigation as a means
of achieving the "no net loss" goal has changed the context within which the
cost and performance of wetland creation, restoration, and enhancement
projects are evaluated.
Because achievement of the no-net-loss goal for wetlands depends, in
part, on the success of wetland mitigation, much recent research has focused
on criteria for evaluating the performance of mitigation projects, the devel-
opment of design standards and engineering techniques for mitigation pro-
jects, and exploration of methods to maximize the likelihood that mitigation
projects will succeed. The research summarized in this report complements
this ongoing research by evaluating the factors that contribute to the cost of
designing and implementing successful wetland creation and restoration pro-
jects. Our research also illustrates the range of costs for historical wetland mit-
igation projects and provides estimates of the costs associated with pre-con-

-------
Technical Summary of Wetland Restoration Costs
2
, struction, construction, and post-construction tasks related to modern wet-
land restoration projects.
METHODS
Wetland creation and restoration costs vary tremendously. The survey
of existing sources of information on costs on which this report is, in part,
based, found projects with costs ranging from a low of $5 per acre to a high of
$1.5 million per acre. This wide range of costs reflects the equally wide range
of wetland types, site characteristics, project goals, and project design and
construction standards that characterized these projects. Typical projects
ranged from the simple reflooding of drained agricultural land to complex
projects involving careful engineering of surface and groundwater flows,
extensive excavation and grading, hand planting and seeding, and long-term
site monitoring and maintenance.	-
Given this wide range of projects, it would be misleading to simply
provide an average cost estimate for wetland restoration. It is no more useful
to think about the average cost of restoring an acre of wetland than to con-
sider the average cost of restoring a damaged automobile. In both cases what
is being restored is important—a Rolls Royce vs. a VW, a prairie pothole vs. a
mangrove swamp. Our results, however, show that costs also depend on
what features are damaged and how badly; and how fast, perfect, and perma-
nent the repairs need to be. Accordingly, the approach toward data collection
and analysis taken in this report was based on the understanding that
aggregating cost data compiled for very different projects could mask
important differences and produce misleading results. Wherever possible,
aggregation of dissimilar projects was avoided in order to emphasize, rather
than, ignore, important wetland-specific, site-specific, and project-specific
differences.	. •
Cost estimates for approximately 1,000 historical wetland creation,
restoration, and enhancement projects were examined, including records of
projects carried out in 44 states over the past 25 years. These historical
estimates (hereafter the "secondary database") were collected primarily from
secondary sources, including published sources in the trade and .technical
press, as well as from unpublished databases from public and nonprofit
agencies. These records were supplemented by detailed engineering and cost
profiles for a smaller set of 90 wetland creation and restoration projects from
10 states (the "primary database"). In most cases the site selection and project
design characteristics for the 1,000 projects in the secondary database were
unknown. In contrast, siting and project design characteristics for the 90
engineering and cost profiles from the primary database were known in
detail.
Unfortunately, the large secondary database permitted only limited
ability to classify projects on the basis of wetland, site, or project
characteristics. Available cost estimates were often accompanied only by brief

-------
Technical Summary of Wetland Restoration Costs
3
project descriptions (e.g., "PFOIA," "salt marsh," or "drain tile"). Few sources
provided detailed project descriptions, and they tended to be sources that
included data on few projects. Phone, mail, and in-person contacts with the
staff of agencies and organizations responsible for the bulk of the cost
estimates revealed that many of the source agencies no longer had, or had
never had, access to detailed project descriptions. Most of these agencies
collected data about wetland creation and restoration projects for which they
did not have day-to-day management authority. Record keeping about
historical mitigation projects, in general, appears to have been weak. As a
result, developing better profiles for projects in the secondary database would
be prohibitively expensive.
Because of the high variability in project cost and lack of detailed pro-
ject descriptions in the secondary database, only a limited understanding of
costs could have been developed on the basis of the secondary data alone. In
many cases, published cost estimates and those available from state and
federal agencies excluded some significant cost components or were associated
with projects that would not meet modern design or construction standards.
The more time consuming approach of working directly with wetland
restoration experts to develop the primary database was therefore adopted o
help make up for the weaknesses in the secondary data. The primary database
was built using standard cost-accounting procedures applied to detailed
engineering descriptions of known wetland creation and restoration projects.
All cost estimates in both the primary and secondary databases were
standardized in 1993 dollars prior to analysis. Whenever the data were
sufficiently detailed, projects were classified on the basis of location, site char-
acteristics, wetland type, and project objectives.
Primary Data
The primary data includes information on approximately 90 different
wetland restoration and creation projects. Subcontracted wetland restoration
specialists with experience in various parts of the United States supplied
detailed project descriptions in terms of specific preconstruction, construction,
and postconstruction tasks. Typical preconstruction tasks included hydrologic
monitoring, site surveys, and preparation of project plans; typical
construction tasks included excavation, grading, and planting; typical post-
construction tasks included site monitoring arid maintenance. Each project
task was then characterized in terms of input requirements (e.g., labor,
material equipment) required to complete the task. Per-task and overall
project costs were calculated by applying unit costs (e.g., wages, rents, prices)
with appropriate adjustments to cover overhead expenses.
Most project profiles were based on actual wetland creation or restora-
tion projects that were designed or constructed by the collaborating wetland
restoration specialists. However, some were based on projects that had been
bid or planned, but never built; or projects with which they were familiar for

-------
Technical Summary of Wetland Restoration Costs
4
other reasons. In some cases, hypothetical variations in site characteristics
were used to develop project profiles that reflected differences in site
conditions (e.g., steep slopes, poor site access, difficult hydrology, or the
presence of an endangered species).
Wetland creation and restoration projects were separated into eight
project categories for analysis. These categories were based on wetland
characteristics that affected the tasks required to achieve restoration success
rather than by conventional wetland classification criteria. Thus wetlands
dominated by shrubs were treated as part of the forested wetland categories,
because tree and shrub planting require similar equipment and have broadly
similar inputs even though they may have dissimilar functions and values.
The classification scheme is somewhat similar to the standard Cowardin et al.
(1978) wetland classification system. With a few adjustments, such as the
grouping of wetlands with trees or shrubs, it may be thought of as a simplified
version of Cowardin et al. The eight categories selected on the basis of re-
quired restoration tasks tend to reflect differences in hydrology and vegetation
structure. The eight categories include:
(1)	Aquatic Beds, consisting of tidal or nontidal communities of perma-
nently or nearly permanently submerged plants;
(2)	Complex Projects, incorporating three or more wetland types in a
single project;
(3)	Freshwater Mixed Projects, consisting of nontidal projects in which
both forested and emergent vegetation is produced;
(4)	Freshwater Forested Projects, establishing woody vegetation (forest or
shrub) in nontidal wetlands;
(5)	, Freshwater Emergent Projects; establishing emergent wetlands in non-'
tidal wetlands;	,
(6)	Tidal Freshwater Wetlands Projects, often consisting of mixed emer-
gent and woody vegetation;
(7)	Saltmarsh Projects and other marine or estuarine projects, establishing
wetlands dominated by emergent vegetation; and
(8)	Mangrove Projects, establishing mangrove communities.
Secondary Data
The secondary database contains over 900 records of costs for individ-
ual wetland creation, restoration, and enhancement projects and was devel-
oped from published and unpublished project reports, the general trade litera-

-------
Technical Summary of Wetland Restoration Costs
5
ture, arid databases collected from county, state, and federal agencies in the
contiguous 48 states. This database includes examples of wetland creation,
restoration, and enhancement used as mitigation, as well as wetlands con-
structed for water quality improvement, waterfowl habitat, and for other pur-
poses. Approximately half the records in the secondary database involve the
restoration or creation of wetlands on agricultural lands undertaken outside
of a mitigation context. Of the remaining cases, over 95% were mitigation
projects, and three-quarters were associated with mitigating road or highway
impacts to wetlands; the rest were non-agricultural projects undertaken out-
side a mitigation context (e.g., wetlands for stormwater management or
nutrient removal from sewage effluent). Records vary widely with respect to
the degree of detail about site and project characteristics, but all included the
general location of the project, project size, and overall project cost.
Data Analysis
Costs per acre of wetland projects decreased substantially with
increasing project size. This pattern, while of interest, complicates much of
the statistical analysis. Different categories of wetland projects have different
average sizes. Creation projects, for example, are typically smaller than
restoration projects; freshwater emergent wetland projects tend to be smaller
than projects producing forested wetlands; and agricultural conversion
projects tend to be larger than other projects. Because project costs vary with
size, a direct comparison of average cost per acre for different categories of
projects may be misleading. We used a standard statistical technique called an
analysis of covariance (ANCOVA) to develop equations that indicate how
project costs change as project size changes and to produce estimates by project
categories of per acre project cost adjusted for project size. Costs per acre data
were highly skewed. Accordingly, parametric statistical analyses (including
the ANCOVA) were carried out on Log10 transformed data.
In both the Primary and Secondary databases, there was an extremely
uneven distribution of cases within and among project categories. Freshwater
emergent wetland creation projects were abundant in our sample, for
example, while projects to restore beds of submerged aquatic plants were rare.
This pattern, which reflects both the frequency with which specific wetland
types are restored or created nationwide, and the vagaries of data collection,
complicated the statistical analyses by making certain statistical comparisons
impossible, and others difficult to interpret. Nonsignificant (p>0.10) and
nonestimable interaction terms were dropped sequentially from all analyses
of covariance. The results shown here (except where otherwise noted) reflect
the most complete analyses possible with the existing databases. Statistical
details of the Analyses of Covariance are given f.n Appendix A.
Reported results, except where otherwise noted, are based on
hypothesis tests with p<0.05.

-------
Technical Summary of Wetland Restoration Costs
6
RESULTS
Primary Data
Wetland Types
Analyses of the Primary Data by Analysis of Covariance and by
Kruskal-Wallis tests show that differences in the costs of restoring different
types of wetlands are not large relative to the differences in costs within any
one wetland category. This reflects the enormous differences in the site and
project design characteristics within project categories and. the fact that the
tasks and costs associated with restoring wetlands in different categories can
be quite similar. Median, mean, minimum, and maximum per acre creation
and restoration costs for eight categories of wetland projects are shown in
Figure 1.
Cost Per Acre
In 1993 $; excludes land costs)
$300.0
$250.0 --
s $200.0
<0
5
2 $150.0 --
g $100.0 --
o
$50.0 --
$0.0
mean
median
I
I , i
§
a- 03
<
x
o
a.
O
X
"U
5 US
Cl P
c	>•

®
>
2
u>
c
CO
c
o
3
O
a>
>
' c
O) o
< O
Wetland Type
Figure 1, Point estimates and ranges of project costs from the Primary
database for specific wetland types.
Table 1 displays summary cost statistics for each wetland category and
includes a percentage breakdown of estimated costs by project stage
(preconstruction, construction, and post construction) and by input category
(labor, equipment, materials, other). In general, construction costs constitute
between two-thirds and three-quarters of total project costs, although they are
somewhat higher for freshwater tidal wetlands. Labor costs tend to account

-------
Technical Summary of Wetland Restoration Costs
7
for the largest overall share of project costs, ranging from about one-third of
overall project costs to about three-quarters.
Table 1.	Cost Estimates and Cost Allocation by Task and by Input
Category (excludes land cost)
Project Type

Aquatic Bed Complex
FW
FW
FW
Tidal
Salt
Man-
Agric.



Mixed
Forest*
Emerg.
FW
Marsh
grove
Corn-**
Project Costs (Thousands)
Average
S19.5
$56.7
$25.3
$77.9
$48.7
$42.0
$18.1
$18.0
$1.0
Minimum
18.3
4.3
1.4
0.9
1.7
0.6
1.0
2.1
0.005
Maximum
21.7
258.8
65.8
248.4
170.6
92.6
43.6
42":8
20.8
Median
18.6
24.8
23.4
42.7
35.2
32.9
10.2
13.6
0.5
Sample Size
3
8
10
19
28
3
9
4
494
Breakdown by Tasks:
Preconstruction
17%
10%
5%
9%
13%
9%
16%
13%
0%
Construction
63
74
78
74
58
87
73
66
100
Postconstruction
20
16
17
18
28
4
11
21
0
Breakdown by Input Category:
Labor
58%
50%
74%
51%
63%
31%
52%
51%
45%
Materials
8
23
10
30
26
54
27
21
0
Equipment
34
14
16
18
9
14
20
28
55
Other
0
14
0
2
1
1
2
0
0
*	High end of range involves researching and restoring hydrology and planting; low end involves
restoring hydrology only.
** Cost breakdowns for agricultural conversions are based on a project consisting of hydrologic
modification without planting or formal plan development.
Project Types
Although the data do not show strong differences between the costs of
restoring different wetland types, they do show significant differences in per
acre costs between creation, restoration, and enhancement projects (see Figure
2).	Enhancement projects are less expensive than creation or restoration
projects by approximately a factor of three. For complex wetland projects
(those incorporating several wetland types or both estuarine and freshwater
components), the enhancement projects were similar in cost to creation and
restoration projects. There are also weak indications in the data that wetland
enhancement costs, on a per acre basis, may not decline as rapidly with
increasing project size as wetland creation and restoration projects (see Figure
3).
We found no significant difference between wetland creation and
restoration costs for many types of wetlands (this pattern was repeated in the
secondary data as well). This runs counter to the conventional wisdom that
restoration projects are less expensive than creation projects because of the

-------
Technical Summary of Wetland Restoration Costs	8'
ease with which wetland hydrology can be established in areas that once were
wetland.
Least Square Mean Costs of Creation,
Enhancement, and Restoration
30000
20000

10000

Creation	Enhancement	Restoration •
Project Type
Figure 2. Comparisons of predicted costs of creation, enhancement, and
restoration projects from the primary database.
Economies of Scale
The analysis of covariance revealed that (1) project size has a strong
effect on per acre project costs, and (2) project type (creation, restoration,
enhancement) also significantly affects project costs. The analysis of co-
variance confirmed, however, that any effects of wetland type on project costs
are hidden by the wide variability in project costs among projects within each
wetland type.
Figure 3 illustrates the inverse relationship between cost per acre and
project size for wetland mitigation projects in the primary database. The pre-
diction lines in the figure, (produced by the analysis of covariance), are given
by the following prediction equations:
(6)	Cost = 49742 * Size'03833 for wetland creation projects.
(7)	Cost = 3712* Size°:2m6 for wetland enhancement projects.
(8)	Cost = 43946 * Size'0'4684 for wetland restoration projects.
Because of the small sample of enhancement projects, the exponent in
equation (7) is not significantly different from zero, and the differences in

-------
Technical Summary of Wetland Restoration Costs
9
exponents among the three project types are marginally statistically
significant (size by project type interaction, P=0.531). A simpler and more
robust prediction relationship pools all three project types to give:
(9) Cost = 30706* Size^3596
For each 10% increase in project size, this relationship predicts that
costs per acre will decline by 3.4%. A doubling in project size results in a 22%
decrease in per acre costs.
Cost Per Acre: Primary Data
Creation (O)
Enhancement (~)
Restoration (Q)
" 100 -}	1	j			(	1
0.1	1	10	100	1000
Size (Acres)
Figure 3. Cost per acre of creation, restoration and enhancement projects
from the primary data.
Secondary Data
Limitations of the Secondary Database
Developing reliable statistical results from the secondary database, de-
spite the large number of observations, proved to be difficult. Without details
about each creation or restoration project, including site conditions before the
project was undertaken, budget constraints, project goals, and so forth, one
can draw only limited conclusions about project costs. Inconsistencies in how
costs were defined, measured, and reported for various projects complicated
interpretation of the data still further. These inconsistencies reflect the wide
range of purposes for which the data were originally collected by many indi-
viduals within private and governmental organizations.

-------
Technical Summary of Wetland Restoration Costs
10
Another difficulty arose because cost data proved much more abundant
for certain types of projects than for others. Many cost estimates for the
conversion of agricultural land to wetland for wildlife and waterfowl benefits,
and for wetland mitigation projects that, involved the creation of small to
medium size, freshwater, emergent wetlands have either been published, or
are readily available. Reported costs for most other kinds of wetland projects
were relatively rare.
Other problems (revealed by individuals' who had published or
provided cost estimates during phone interviews) reflect quirks of the
original data from which the secondary database was compiled. The four
main problem areas include:
(1)	joint Costs—Mitigation and Development. The providers of cost data
for some projects were unable to distinguish between restoration costs
and the costs of earth moving and landscaping associated with the con-
struction project that resulted in the need for mitigation; this was espe-
cially true for highway expansion projects. To the extent that this re-
sults in allocation of construction project costs to mitigation, it will re-
. suit in an overstatement of mitigation costs. If mitigation costs are er-
roneously allocated to the original project (less common, we believe),
mitigation costs will be underestimated.
(2)	Joint Costs—Mitigation and Permitting. In other cases, it was impossi-
ble for providers of cost data to distinguish between restoration costs
and the costs of engaging in the wetland permitting process itself; this
was especially true for large complex projects and mitigation banks. To
the extent that project costs are inflated by permitting costs, this would
overstate true project costs.
(3)	Differing Design/Precision Standards. The secondary database includes
wetland construction projects designed to improve water quality (e.g.,
treat sewage, storm water, farm runoff,- and acid mine drainage). Since
these projects involve substantial engineering effort, - and all siting,
design, and construction decisions for them are directed exclusively at
waste treatment, they might be expected to be especially expensive, as
wetland creation projects go. Actual costs of constructed wetlands
designed to improve water quality, however, were not statistically
different from costs of wetlands created or restored for mitigation, arid
all such projects were retained.
(4),	Non-Priced Project Inputs. The database also includes projects carried
out with participation of volunteers or with voluntary contributions of
land, expertise, or equipment. These projects were generally designed
to create or restore specific wetland functions (e.g., duck habitat), usu-
ally, but not always, through the conversion of agricultural land to wet-

-------
Technical Summary of Wetland Restoration Costs
11
land conditions. The cost estimates for projects that use volunteers of-
ten exclude the opportunity cost of contributed labor and other "in-
kind" contributions, and thus may under report true project costs.
The combination of very different sample sizes for various categories
of wetland projects/ inconsistencies in the descriptive information available
to us on each project, and differences among sources of data regarding how
costs were reported make detailed interpretation of data from historical
sources of information of limited value. Further attempts to improve the
secondary database through additional contacts with individuals who
provided or published the data on which it was based would be marginally
successful and would not be as cost effective or as useful as adding to the
primary database.
Agricultural Conversions vs. Other Projects
The secondary data consisted of almost equal parts agricultural conver-
sions to wetland carried out for wildlife enhancement purposes and projects
carried out for other reasons, mostly mitigation. The two groups of data were
very different. In general, agricultural projects (1) were significantly less ex-
pensive than the other projects, and (2) the cost per acre of agricultural con-
versions was less sensitive to project size than was the cost per acre of mitiga-
tion projects (Figure 4).	-
The relationships between project cost to project size found in the .
analysis of covariance correspond to a decrease in per acre costs of about 4.3%
and 22% respectively for agricultural and other projects in response to a dou-
bling of project size. The prediction equations for cost per acre that correspond
to these parameters are as follows.
(8)	Cost = 536.4 * Size'0 06219 for agricultural conversions, and
(9)	Cost = 30850 * Size'035199 for non-agricultural projects.
Thus a one acre agricultural conversion project typically costs just over
$500, while a one acre project that is not an agricultural conversion typically
cost about $30,000. Exact reasons for this difference in cost could not be deter-
mined as part of this study; however, the effect does not appear to be entirely
due to geography. Even when attention is restricted to those states (CA, KS,
MN, MT, OR, TX) in which data was available for both agricultural conver-
sions and other projects, the cost per acre of agricultural conversion projects
remain significantly lower, and less sensitive to project size than are other
projects (by analysis of covariance, pcO.OOQl for both comparisons)..
Unfortunately, the databases from which we drew information on agri-
cultural conversion projects for this report were inconsistent with respect to
the descriptive information they included, and thus further analyses of agri-
cultural conversion projects were impossible.

-------
Technical Summary of Wetland Restoration Costs
12
Cost Per Acre Of Agricultural Conversions
and Nonagricultural Wetland Projects
10
10
» 10 f
¥ 4
o 10 -1-
-------
Technical Summary of Wetland Restoration Costs
13
projects, however, were similar in cost. Project type affected per acre project
costs as well. In general, costs of wetland creation projects and mixed projects
decreased less with size than did costs of enhancement and restoration pro-
jects (see ANCOVA table in Appendix A for details).
In part, the results of this analysis of covariance may reflect the poor
quality of the secondary data. Descriptive information was unavailable for a
majority of projects in the database, and, by chance, for almost all wetland en-
hancement projects. To double check the legitimacy of the results of the full
analysis of covariance, we fit a simpler model, in which we separated projects
only by project kind. This allowed us to increase our total sample size to 367
projects, (309 creation projects, 28 enhancement projects, 16 restoration pro-
jects, and 14 mixed projects). The results of this analysis of variance were gen-
erally similar to those described above, and are shown in Figure 5. At larger
project sizes (greater than one acre), mixed projects and creation projects
tended to be more expensive, and less sensitive to project size than restora-
tion and enhancement projects. At smaller sizes, we had little data for any-
thing other than wetland creation projects, and thus comparisons of the dif-
ferent project types are inappropriate.
Cost Per Acre of Creation,
Enhancement, and Restoration Projects
Creation (~)
,/ Mixed *)
^ Restoration (•)
Enhancement (A)

107

106

- 10s




£
104
o

<
w
CO
O


Q.

01
100
o

O


10

1
10
-3
10
-2
10
-1
1
10
100
10*
Size (Acres)
Figure 5. Cost per acre of wetland projects estimated from the secondary
data.
The prediction equations corresponding to the different project types
(averaged across all hydro logic classes) produced by this analysis of covariance
are given by the following equations:
(10) Cost = 33164 * Size"® 2*21 for creation projects
-------
Technical Summary of Wetland Restoration Costs
14
(11)	Cost = 13198 * Size-0 6495 for enhancement projects
(12)	Cost = 20212 * Size'010'6 for restoration projects
(13)	Cost = 39354 * Size-0'3814 for mixed projects
At one acre in size, enhancement projects are somewhat less expensive
on a. per acre basis than other projects, and the relative cost advantage over
creation projects increases as projects get bigger. At larger sizes, however, en-
hancement and restoration projects have similar predicted costs per acre.
Comparisons Between the Primary and Secondary Data
An analysis of covariance comparing the primary with the non-agricul-
tural portion of the secondary database showed that (1) the overall slope relat-
ing project size to per acre project costs are not different for the two databases,
and (2) projects in the primary database are more expensive than projects in
the secondary database. Projects from the secondary database have costs that
are typically only 56.3% of the costs of similar sized projects in the primary
data. Put another way, the projects from the primary database cost about 78%
more than the projects in the secondary data,
DISCUSSION
Economies of Scale
In both the primary and secondary databases, per acre project costs
declined with project size. The analysis presented here provides a glimpse
into scale issues and probably offers the most reliable available, estimates of
scale factors applied to wetland creation and restoration costs. The analysis
has not, however, measured true economies of scale because uneven sample
sizes and other problems, with the data prevented isolation of project size
from other project-related characteristics that change with project size.
Economies of scale, by definition, measure changes in unit cost as the
scale of production—number of units—changes. Economies of scale can re-
flect changes in how production takes place at different scales of production
(e.g., more mechanical production or more labor specialization). They can
not, however, be estimated reliably when there are significant changes in
what is being produced at different scales of production. Unfortunately, the
unevenly sized samples of different types and sizes of wetland projects, made
it impossible to fully isolate project size as the only cause of differences
among project costs. Small scale projects tend to include careful (and expen-
sive) grading and planting, while larger projects are usually carried out with
less precision and use less intensive planting methods. Furthermore, certain
types of projects (e.g., erosion control plantings) are likely to produce small
wetlands, while others (e.g., removing or building water control structures)
are likely to produce larger wetland areas. Thus small and large wetland pro-
-------
Technical Summary of Wetland Restoration Costs
15
jects almost always differ by more than just size. Accurate determination of
true economies of scale would require examination of the effects on per acre
project cost of project size alone, independent of confounding influences.
Scale effects on per acre project costs differ among categories of wetland
projects. Agricultural conversion projects, in particular, decreased in cost
much less rapidly with increasing project size than did mitigation projects.
We are uncertain why that should be so. It may reflect differences in the
economies of scale for component restoration tasks (like monitoring, or
planting of trees and shrubs) that seldom are incorporated into agricultural
projects. It may also reflect the wide range of approaches used to provide
mitigation, as compared with limited techniques commonly used to restore
agricultural land to wetland. It may simply reflect the role of volunteer and
in-kind contributions in agricultural conversions that are less commonly part
of mitigation projects. Or the difference may reflect the dynamics of
negotiations between wetland regulators and permit seekers, which played no
role in the agricultural projects.
Although different scale effects on project costs were observed for cre-
ation, restoration, and enhancement projects in the Secondary data, those dif-
ferences were statistically unstable, changing with apparently small changes
in how the data was analyzed. Since no similar patterns were found in the
Primary data, it would be premature to hypothesize what produced the differ-
ences. There are significant fixed costs associated with all but the most simple
kinds of restoration projects, so economies of scale most certainly exist for
most types of wetland creation and restoration projects. Further research on
the scale issue should probably wait until there are specific questions raised
that require information about economies of scale and justify the expense of
measuring and explaining them precisely.
Creation, Restoration, Enhancement
Conventional wisdom suggests that wetland enhancement should be
less costly on a per acre basis than wetland restoration, which in turn should
be less costly than wetland creation. The primary and secondary data both
support the hypothesis that wetland enhancement projects are less costly than
creation and restoration projects. However, we saw no evidence in the pri-
mary data that restoration is less costly than creation, and we found only
weak evidence of this pattern (it holds for large projects only) in the sec-
ondary data.
It has frequently been suggested that restoration projects should be less
expensive than creation projects primarily because of the relative ease with
which appropriate hydrology can be reproduced in an area that previously
was wetland. Creation and restoration projects in the primary database gener-
ally required similar tasks and subtasks to reach completion. Many restoration
projects required substantial expenditures for excavation and site preparation
that were similar to those required for wetland creation. In fact, no systematic
-------
Technical Summary of Wetland Restoration Costs
16
differences exist between creation and restoration projects in their allocation
of costs between background research, project planning and design, site
preparation, planting, monitoring, or maintenance (by Kruskal-Wallis test,
p>0.10). The surprisingly high costs of restoration projects relative to creation
projects may reflect the costs of working in or near existing wetlands,
including increased regulatory costs, and the added costs of working "in the
wet" (creation projects presumably can be worked dry). In addition, both
creation and restoration projects in historical mitigation markets were often
built under strong pressure to minimize costs, with only secondary regard for
quality. Because purchasers of mitigation services have been price sensitive,
the costs of the two mitigation alternatives may simply have begun to
converge because they serve the same market.-
Differences in per acre costs for creation, restoration, and enhancement
projects provide only a partial view of the actual costs of providing wetland
mitigation through each of these strategies. The appropriate mitigation ratio
and the risk of project failure need to be taken into account in the final cost
comparison. Ordinarily, a larger area of wetland would have to be enhanced
to provide compensation for an acre of lost natural wetland than would have
to be created or restored. The overall cost of mitigation using enhancement,
therefore, may not be lower than the cost using creation or restoration.
Furthermore, conventional wisdom suggests that restoration projects have a
much higher success rate than wetland creation projects. To the extent that
regulators actually hold permit seekers responsible for mitigation failures
(e.g., through bonds or other financial assurances), or require higher
mitigation ratios to account for risk of failure, restoration will often prove to
be a substantially less costly way to mitigate for wetland losses than creation.
Wetland Type
In both the primary and secondary databases, the per acre costs of wet-
land creation and restoration projects were relatively insensitive to the type
of wetland being created or restored, whereas ranges of per acre project costs
tended to be relatively wide within most wetland categories. It is unlikely that
wetland type plays an insignificant role in determining per acre project costs.
Site-specific and project differences, however, are apparently so important in
determining project costs that they mask whatever role wetland type alone is
playing.	¦
Wetland creation or restoration- projects can differ in cost either be-
cause the inputs required to carry out the projects differ, or because the permit
cost of those inputs differ. Costs of inputs, especially of labor, can vary sub-
stantially by region. Within the United States, however, regional differences
in the costs of inputs alone are unlikely to lead to variation in per acre project
cost by even as much as a factor of two. Much greater differences in per acre
costs arise because the inputs required to complete two projects may be very
different. Inputs may differ because the wetlands being produced are of sub-
-------
Technical Summary of Wetland Restoration Costs
17
stantially different types. Many projects producing dissimilar wetlands, how-
ever, have similar inputs, and many differences in inputs reflect differences
in the projects that are less obvious than target wetland type. Such differences
may arise for many reasons, including:
(1)	Regional differences in the types of projects typically carried out (e.g.,
Florida has numerous mangrove restoration projects; in Southern
California, wetland specialists have difficulty finding sufficient water
for wetland construction);
(2)	Differences in project design (e.g., planting densities, choice of water
control structures, size and pattern of variation in pool depth, use of
enhancement techniques like artificial snags or nest boxes); or
(3)	Site-specific factors (amount of excavation required, difficulty of access,
hydrologic conditions, etc.); and
(4)	Differences in project implementation (thoroughness of site-specific re-
search, use of hydrologic modeling, degree of post-construction moni-
toring, etc.).
Of these four reasons that inputs for wetland projects may differ, two—
numbers (1) and (2), above—reflect differences in the physical product being
produced, the target wetland. The other two reflect variation in how the wet-
land is produced, or in what is needed to produce the wetland.
In both the primary and secondary databases, the type of project being
carried out (creation, restoration, enhancement) has a stronger and more con-
sistent influence on overall project costs than does the type of wetland being
constructed. Costs apparently depend not so much on what you are produc-
ing, as on what you are doing to produce it. A wide range of dissimilar pro-
jects were lumped together in each wetland category. Required inputs for
those projects vary widely depending on project goals, pre-existing site condi-
tion, landscape context, regional environmental patterns, and local regulatory
standards, and as a result, the costs of those projects also vary widely.
It would require considerable additional effort to collect and verify
enough additional cost data to fully sort through the effect of wetland type on
average project costs. This effort would be of limited value unless specific
questions are identified that can only be addressed in this way. This study has
demonstrated that the per acre costs of apparently similar projects can differ
significantly, easily by a factor of five or ten, but that costs for individual
wetland projects can be forecast with acceptable precision if only a few basic
facts about the project and the restoration or creation site are known. Our
analysis suggests that cost adjustment factors based on simple indicators of
site conditions (volume of soil to be moved, amount to be disposed of off-site,
site access requirements, whether the site can be prepared and planted dry,
-------
Technical Summary of Wetland Restoration Costs
18
etc.) can'reduce cost-estimating error to. within acceptable bounds. When a
specific project is being evaluated/ and project-specific information is
available, a similar engineering cost-accounting framework is a, far more
reliable way to provide accurate cost estimates than relying on baseline cost
estimates.
Differences Between Primary and Secondary Data
Costs of wetland projects in the primary database were almost double
what they were for projects in the secondary data. The primary data represent
projects designed with a reasonable commitment to both cost and quality,
whereas the secondary data, collected from a wide range of historic sources,
reflects projects developed in a mitigation context where low cost projects
have been allowed often with little regard for quality (King and Bohlen 1994).
In this sense, the difference in project costs between primary and secondary
data provides a rough estimate of the costs of increased quality. The'secondary
data may faithfully represent patterns of project cost that have held in the re-
cent past. Relatively low cost projects, however, contributed significantly to
the poor success rates of historical mitigation efforts, which have in turn
resulted in new standards. The lower cost estimates drawn from our
secondary data are unlikely to fully reflect the costs of wetland creation,
restoration, and enhancement projects that will meet the standards of the fu-
ture.
Agricultural Conversions
Federal agencies involved in programs to restore converted agricul-
tural lands back to wetland (e.g., the USDA "Water Bank' Program, the
Department of the Interior Small Wetlands Acquisition Program, the
Department of Agriculture's Wetland Reserve Program) have previously
reported estimates of the cost of wetland creation and restoration. Although
the agricultural conversion projects carried out under the auspices of these
federal programs represent a significant portion of nationwide wetland
creation and restoration efforts, the costs of such projects are quite different
from costs of projects carried out for mitigation, for several reasons:
(1)	Agricultural conversion projects usually involve restoring altered hy-
drology (e.g., breaking drainage tiles or filling ditches), which is inex-
pensive and often successful. Such projects are simpler than projects
aimed at restoring structurally and biologically more complex wetlands
that occur with greater frequency outside the farm belt.
(2)	Agricultural conversions usually do not face the complications of
restoration and creation of wetlands in urban and suburban landscapes,
precisely where wetland losses and the associated needs for mitigation
are often the greatest.	• .
-------
Technical Summary of Wetland Restoration Costs
19
(3)	Many agricultural conversion projects are carried out with the help of
agency personnel and volunteers. The opportunity costs of labor and
other contributions by these "unpaid" participants are sometimes in-
completely reported.
(4)	-Agricultural conversion projects, as we have used the term, include
only those projects carried out outside of a mitigation context. Costs of
complying with regulatory requirements (e.g., plant species composi-
tion or vegetative cover requirements) and costs of participating in
regulatory processes are therefore minimized.
High Cost Projects
The secondary database contained a few records of exceptionally high
costs, including one case of restoration costs near $1.5 million per acre.
However, limited investigation revealed that unusually high costs were usu-
ally pushed up by extremely small project size (under one-half acre) or by ex-
traordinary conditions at the restoration site (e.g., the need to blast through
granite to attain an acceptable elevation). In many cases the selection of ex-
traordinary sites appears to be the result of regulatory decisions, in particular,
the regulatory preference for on-site rather than off-site mitigation. There are
many reasons why on-site mitigation might be preferred to off-site mitiga-
tion, and we did not compare on-site and off-site alternatives to determine if
there were significant cost differences. However, there were clearly cases
where exceptionally large amounts of money spent on restoration may have
been better invested if siting decisions were based on a search for more favor-
able locations from the perspective of improving wetland or watershed func-
tions rather than strictly adhering to the regulatory preference for on-site
mitigation.
Cost Implications of Regulatory Involvement
The relatively low costs of (nonregulatory) agricultural conversions
and mitigation projects suggest that regulatory involvement itself may
increase the costs of wetland creation and restoration. Similarly, examination
of the few very high cost wetland projects in our database suggest that, even
within the mitigation context, regulatory policies and decisions affect project
costs. These findings, however, must be interpreted with some care, as
regulatory involvement affects not only the cost of creation, restoration, and
enhancement, but also the character of the services (design construction and
monitoring) provided and, ultimately, of the product (the wetland) produced.
Changes in Design
Agricultural conversion projects carried'out outside of a mitigation
context usually involve little more than restoration of pre-disturbance hydro-
logic conditions by destruction of the ditches arid drain tile used to artificially
-------
Technical Summary of Wetland Restoration Costs
20
drain land. Such an approach to wetland restoration, is considerably less rig-
orous than would prove acceptable to most regulatory agencies.
Most regulatory agencies are concerned about ensuring re-establish-
ment of wetland conditions as rapidly as possible, working to see specifi'c bi-
otic communities become established on site, and trying to minimize risks of
project failure. Accordingly, many agencies require more intensive manage-
ment of mitigation sites than they would require of a non-mitigation restora-
tion effort. In particular, mitigation projects must often meet specific perfor-
mance conditions (e.g., plant survival or vegetative cover) by a certain time
after construction. No such requirements are imposed on most wildlife en-
hancement projects.
Regulatory preferences for on-site mitigation may also increase the
need for site preparation. To the extent that wetland creation or restoration at
a specific location requires grading or hydrologic modification, costs will be
increased. Many mitigation projects incorporate site grading, hydrologic mod-
ification and extensive planting. For many of the projects in our .primary
database, these three activities represented a majority of project costs. These
tasks also require considerable design and planning effort, with its associated
costs.
Changes in Implementation
Even if regulatory involvement did not change the design and imple-
mentation of a wetland restoration, regulatory involvement may be expected
to increase project costs by imposing planning, documentation, and monitor-
ing requirements that alter how a specific project is designed and carried out.
Regulatory requirements for increased care and better documentation of
plans, construction, and other activities bear additional cost, and, one hopes,
carry some benefits in terms of reduced risks of failure and higher probability
of producing desired wetland functions.
Many agricultural conversion projects have essentially no design costs.
Existing structures (drain tile and ditches) are.simply eliminated, and the area
that floods becomes wetland. In contrast, the area of many mitigation projects
is often calculated precisely (often in square feet, not acres). Design details are
worked out long before any earth is moved or seedlings planted. Blueprints
are rare for agricultural conversions, while several iterations of blueprints are
routinely produced for mitigation projects before construction begins.
After construction, mitigation projects are more likely than agricul-
tural conversions to incorporate monitoring and follow-up practices such as
annual vegetation surveys, photographs recording site conditions, and as-
built project plans. Long-term maintenance activities and remedial actions to
correct undesirable developments are also more likely with mitigation pro-
jects.	"
-------
Technical Summary of Wetland Restoration Costs
21
Additional Costs
Undoubtedly, participation in the regulatory process has certain costs
that neither alter the final wetland project, nor affect how the project is im-
plemented. Many mitigation projects are the result of extensive negotiations
among the builder, his or her client, regulators,, and other interested parties.
The transaction costs associated with these negotiations and with regulatory
compliance in general can be substantial. Meetings with regulators, construc-
tion delays produced by permitting problems, and so forth, are costs of regula-
tory decision making itself that are unlikely to bear direct environmental
benefits.
BIBLIOGRAPHY
Conservation Foundation. 1988. Protecting America's Wetlands: An Action
Agenda. The Final Report of the National Wetlands Policy Forum. The
Conservation Foundation. Washington, DC.
Cowardin, Lewis M., V. Carter, F. C. Golet, and E. T. LaRoe. 1979. Classification
of Wetlands and Deepwater Habitats of the United States. FWS/OBS-
79/31, Office of Biological Services, Fish and Wildlife Service. U.S.
Government Printing Office: Washington, DC.
Environmental Protection Agency and the Department of the Army. 1990.
Memorandum of Agreement Between the Environmental Protection
Agency and the Department of the Army Concerning the
Determination of Mitigation Under the Clean Water Act § 404(b)(1)
Guidelines. June 6, 1990.
King, D. M., and C. C. Bohlen. 1984. Making Sense of Wetland Restoration
Costs. University of Maryland Center for Environmental and Estuarine
Studies Technical Report UMCEES-CBL-94-045], January 1994.
-------
Technical Summary of Wetland Restoration Costs
22
APPENDIX A: ANALYSIS OF COVARIANCE TABLES
The following analysis of covariance tables provide statistical details for
the conclusions presented in the main~ text. All analyses were performed on
logxo-transformed data. The tables show partial sums of square and F ratios,
testing the hypothesis that the particular source of variation is associated with
more of the variability in cost among projects than can be accounted for by
chance.
Table A.l. Analysis of Covariance for Primary Data
ANCOVA Table
Source ,
DF
Sum of .
Mean Square
F Ratio
Prob>F


Squares


Log(Size)
1
0.9383
0.9383
4.4075
0.0393
Wetland Type
7
3.1219
0.4460
2.0951
0.0551
Project Type
2
2.6330
1.3165
6.1843
0.0033
(Ire./Rest, vs Enhnc.
1
2.6318
2.6318 '
12.3631
0.0008
Create vs Restore
1
0.0042
0.0042
0.0198
0.8885
Log(Size)*Project Type
2
1.3026
0.6513
3.0595
0.0531
Model .
12
12.0979
1.0081
4.7359
0.0000
Error
71.
15.1146
0.2129


Total
83*
27.2123



Parameter Estimates

Slope
Std Error
Least Sq.
Std Error
N


Mean


Creation
-0.3833 ¦
0.10721
4.3742
0.08744 .

Enhancement
0.2086
'0.24937
3.7452
0.17701

Restoration
-0.4684
0.12848
4.2788
0.09211

Table A.2. Analysis of Covariance on the Secondary Data, Comparing
Agricultural Conversions with all Other Projects
ANCOVA Table
Source
DF
Sum of
Squares
Mean Square
F Ratio
Prob>F
log,0(acres)
Ag. Status
Ag. Status*log(acres)
1
1
1
23.051
239.244
11.345
23.051
239.244
11.345
52.061
540.339
25.624
0.0000
0.0000
0.0000
Model
Error
C Total
3
878
881
688.041
388.750
1076.792 •
229.347 •
0.443
517.984
0.0000
Parameter Estimates

Slope
Std Error
Least Sq.
Mean
Std Error
N
Ag Conversion
Other
-0.06279
-0.35798
0.04471
0.03744
2.67410
4.17241
0.03619
0.03912
485
387
-------
Technical Summary of Wetland Restoration Costs
23
Table A.3. Analysis of Covariance on the Secondary Data, Omitting
ANCOVA Table
Source
DF
Sum of
Squares
Mean
Square
F Ratio
Prob>F
Logio(Acres)
Project Type
Hydrology
Logio(Acre)*Project Type
1
3
2
3
7.7962
3.7112
3.9148
5.6072
7.79627
1.23707
1.95740
1.86909
15.1823
2.4091
3.8118
3.6398
0.0001
0.0693
0.0243
0.0142
Model
Error
Total
9
151
160
34.03871
77.54000
111.57871
3.78208
0.51351
7.3652
0.0000
Parameter Estimates , *

Slope
Std Error
Least Sq.
Mean
Std Error
N
Project Type
Creation
Enhancement
Restoration
Mixed
-0.157401
-2.757531
-0.443081
-0.324461
0.0676
0.84121
0.40529
0.28156
4.7915
4.9881
3.7105 '
4.80S6
0.14907
0.42915 .
0.38589
0.50280 .
140
3
13
5
Wetland System
Estuarme/Marine
Lacustrine/ Palustrine
Riparian**


4.3108
4.4901
5.5267
• 0.27620
0.22405
0.40328 *
16
140
5'
the small number of enhancement projects for which we had information on wetland system and project type.
Per acre costs for wetland enhancement projects declined at a relatively high rate in other analyses as well.
** Riparian projects include projects that focus on repairing, river and stream banks, or restoring
stream bottom communities to a more natural state.
Table A.4. Analysis of Covariance for the Secondary Data (Omitting
Agricultural Conversions), Reduced Model
ANCOVA Table
Source
DF
Sum of
Squares
Mean
Square
F Ratio
Prob>F
log( Acres)
Project 2
Project *log(Acre
1
3
3
13.8270
2.8677
5.2751
13.8270
0.9559
1.7583
27.2307
1.8826
3.4629
0.0000
0.1321 '
0.0165
Model
Error
C Total
7
359
366
63.5650
182.2906
245.8557
9.0807
0.5077
17.8834
0.0000

Slope
Std Error .
Least Sq.
Mean
Std Error
N
Project Type
Creation
Enhancement
Mixed
Restoration
-0.242127
-0.649531
-0.381426
-0.701555
0.04784 .
0.14003
0.24257
0.24991
4.446027711
. 3.920265920
4.477403264
4.089342538
0.04094
0.14705
0.24268
0.31347

-------
Technical Summary of Wetland Restoration Costs	24
Table A.5. Analysis of Covariance Comparing Costs from the Primary and
Secondary Data
ANCOVA Table
Source
DF
Sum of
Squares
Mean
Square
F Ratio
Frob>F
Log10(Acres)
Database
Log10(Acre)* Database
1
1
1
22.714623
2.583559
0.008275 -
22.714623
2.583559
0.008275
46.1314
5.2470
0.0168
0.0000
' 0.0224
0.8969 • ¦
Model
Error
Total
3
477
480
-47.58531
234.87003 .
282.45535
15.8618 '
0.4924
32.2138
Prob>F
0.0000
Parameter Estimates

Slope
Std Error
Least Sq.
Mean
Std Error
N
Logio(Acres)
Primary Data
Secondary Data
-0.365184
0.05377
4.5769
4.3333
0.07891
.0.03568,
"481
92
389
-------
{
-------