Status and trends of dam removal research in the United States

Abstract

Aging infrastructure coupled with growing interest in river restoration has driven a dramatic increase in the practice of dam removal. With this increase, there has been a proliferation of studies that assess the physical and ecological responses of rivers to these removals. As more dams are considered for removal, scientific information from these dam‐removal studies will increasingly be called upon to inform decisions about whether, and how best, to bring down dams. This raises a critical question: what is the current state of dam‐removal science in the United States? To explore the status, trends, and characteristics of dam‐removal research in the U.S., we searched the scientific literature and extracted basic information from studies on dam removal. Our literature review illustrates that although over 1200 dams have been removed in the U.S., fewer than 10% have been scientifically evaluated, and most of these studies were short in duration (<4 years) and had limited (1–2 years) or no pre‐removal monitoring. The majority of studies focused on hydrologic and geomorphic responses to removal rather than biological and water‐quality responses, and few studies were published on linkages between physical and ecological components. Our review illustrates the need for long‐term, multidisciplinary case studies, with robust study designs, in order to anticipate the effects of dam removal and inform future decision making. WIREs Water 2017, 4:e1164. doi: 10.1002/wat2.1164
This article is categorized under: Water and Life > Conservation, Management, and Awareness
Engineering Water > Sustainable Engineering of Water
Water and Life > Stresses and Pressures on Ecosystems

Full text

Advanced Review
Status and trends of dam removal
research in the United States
J. Ryan Bellmore,
1
*Jeffrey J. Duda,
2
Laura S. Craig,
3
Samantha L. Greene,
4
Christian E. Torgersen,
4
Mathias J. Collins
5
and Katherine Vittum
2
Aging infrastructure coupled with growing interest in river restoration has driven
a dramatic increase in the practice of dam removal. With this increase, there has
been a proliferation of studies that assess the physical and ecological responses of
rivers to these removals. As more dams are considered for removal, scientific
information from these dam-removal studies will increasingly be called upon to
inform decisions about whether, and how best, to bring down dams. This raises a
critical question: what is the current state of dam-removal science in the United
States? To explore the status, trends, and characteristics of dam-removal research
in the U.S., we searched the scientific literature and extracted basic information
from studies on dam removal. Our literature review illustrates that although over
1200 dams have been removed in the U.S., fewer than 10% have been scientifically
evaluated, and most of these studies were short in duration (<4 years) and had
limited (1–2 years) or no pre-removal monitoring. The majority of studies focused
on hydrologic and geomorphic responses to removal rather than biological
and water-quality responses, and few studies were published on linkages
between physical and ecological components. Our review illustrates the need for
long-term, multidisciplinary case studies, with robust study designs, in order to
anticipate the effects of dam removal and inform future decision making. Published
2016. This article is a U.S. Government work and is in the public domain in the USA.
How to cite this article:
WIREs Water . doi: 10.1002/wat2.1164
INTRODUCTION
For millennia, humans have built dams on river
systems for navigation, irrigation, flood control,
and power generation. Although new dams are still
being built to meet the needs of society, particularly
in developing countries,
1
many dams are aging,
2
have
become hazardous, or are no longer fully serving the
functions for which they were designed. Although
dam failures are rare, they can be costly in terms of
property damage and loss of life.
3
Evolving safety
and environmental standards in the U.S. are also
making it more costly to maintain and repair aging
dams. Fifty years ago, during the peak of
government-sponsored dam construction in the U.S.,
a widespread movement calling for the removal of
dams would have seemed far-fetched. Today, how-
ever, over 1200 dams have been removed, and the
majority of these dams were removed within the last
two decades
4,5
(Figure 1(a)). Dam removal is now
considered as a viable option when the cost of keep-
ing a dam in place exceeds the expense of its removal,
particularly in locations where the possibilities for
river restoration are high (e.g., Duda et al.
6
). Given
the vast number (potentially >2,000,000)
7
and age
structure of U.S. dams (up to 80% over 50 years old
by 2020),
8
as well as shifting societal values, this
upward trend in dam removal is likely to continue.
*Correspondence to: jbellmore@fs.fed.us
1
U.S. Department of Agriculture, Forest Service, Pacific Northwest
Research Station, Juneau, AK, USA
2
U.S. Geological Survey, Western Fisheries Research Center, Seat-
tle, WA, USA
3
American Rivers, Philadelphia, PA, USA
4
U.S. Geological Survey, Forest and Rangeland Ecosystem Science
Center, Seattle, WA, USA
5
NOAA, National Marine Fisheries Service, Gloucester, MA, USA
Conflict of interest: The authors have declared no conflicts of inter-
est for this article.
Published 2016. This article is a U.S. Government work and is in the public domain in the USA.

Deciding whether to remove aging dams and
how to do so with minimal adverse impacts, however,
is complicated by uncertainties associated with
potential environmental benefits and detriments.
9–12
On one hand, dam removal can re-establish more
natural flows, water temperatures, and sediment
regimes,
10,13–15
and allow native organisms to recolo-
nize habitats that were formerly inaccessible.
16–18
On
the other hand, removing dams can expose and mobi-
lize large volumes of sediment and contaminants,
19–22
and facilitate the spread of invasive species.
23
From
an environmental perspective, deciding whether and
how to remove dams requires balancing these risks
and benefits,
24,25
which in turn, necessitates relatively
accurate predictions about how rivers will respond to
dam removal and how long they may take to recover.
Making these scientific predictions will require lever-
aging knowledge from expert opinion, conceptual
and quantitative models, and, most importantly,
from studies that empirically evaluate the physical
and ecological responses to dam removal. Over the
last two decades, as the practice of dam removal has
accelerated, so too has the scientific evaluation of
these projects. Understanding the extent to which this
scientific research can inform dam-removal decision-
making, however, will first require taking stock of the
quantity, quality, and character of these studies.
Although an earlier review of dam-removal science
was conducted by the Heinz Center in 2002,
11,26
the
practice and science of dam removal has proliferated
Number of dams
Data from National
Inventory of Dams
through 2013
0
1–2000
2001–4000
4001–6000
> 6000
Number of dams removed
Data from American
Rivers 1912–2014
Data from
Bellmore et al. (2015)
through 2014
0
1–25
26–50
51–75
> 75
Number of dam removals
with ≥1 publication
0
1–3
4–6
7–10
> 10
267
1
4
3
3
18
2
9
720
30
793
964
1594
29
447
3316 904 1078
118 5
1592
1114
1224
2241 5132
2439
3262
363
609
927
941
1552
1968
611
1483
2241
3630
3927
5119
1251
513
2511
2835
6374
4925
7310
1639
835 1737
491
380
48
95
1
21
10
5
316
2
5
5
40
131
35
61
60
23
32
4
20
131
20
11
26
6
3
8
3
4
2
7
7
2
3
3
7
2
1
2
1
1
1
9
3
20 27
38
23
19
5
133
1
3
2
2063
(a)
(b)
(c)
641
1490
734
825
336
236
FIGURE 1 |Distribution of dams in the contiguous U.S. (a), the number of dams removed (b), and the number of published dam removal
studies (c), by state. The number of dams from the National Inventory of Dams database does not accurately reflect all of the dams in the
U.S. (see text).
Advanced Review wires.wiley.com/water
Published 2016. This article is a U.S. Government work and is in the public domain in the USA.

since the time of these publications, and an updated
review of the ‘state of the science’is needed.
To address this need, we conducted an exten-
sive literature search to identify published studies
that contained empirical information associated with
dam removal. One of the main goals of this literature
search was to extract basic information from these
studies
27
(http://doi.org/10.5066/F7K935KT), and to
create an online visualization and analysis tool to
make this information readily available to both prac-
titioners and researchers (https://www.sciencebase.
gov/drip/). It was not our aim to review study find-
ings, but rather, to extract attributes that describe
the design of the study, the type of response metrics
monitored (physical, biological, water quality), and
the characteristics of the removed dams (e.g., dam
height, location, and removal date). In this article, we
use the information from this database, as well as a
database that contains information on the practice of
dam removal in the U.S.,
4
to address the following
questions:
1. What are the characteristics of dam removal
studies (number, location, and size of dams)
and how representative are these studies of all
dams that have been removed in the U.S.?
2. What physical, biological, and water-quality
responses are these studies measuring, over
what duration, and what types of study designs
are being employed?
3. Where are there gaps in the research that might
limit the ability of dam-removal science to inform
the practice of dam removal in the future?
DAM-REMOVAL DATABASES
USGS Dam-Removal Science Database
We identified dam-removal studies published through
31 December 2014 using ISI Web of Science, Google
Scholar, and the U.S. Geological Survey (USGS) Pub-
lication Warehouse. The following keywords and
phrases were queried (1): ISI Web of Science: (dam
AND removal*) AND (stream OR river), (2) Google
Scholar (Advanced Scholar Search; italics indicate the
search term used by Google): with the exact phrase =
‘dam removal’;with at least one of the words =‘
stream OR river’;where my words occur =‘any-
where in the article,’and (3) the USGS Publication
Warehouse: ‘dam removal.’These searches identified
6068 documents. To identify relevant citations from
this list, we first examined titles and abstracts. Cita-
tions that were not related to the science of dam
removal (e.g., studies of beaver dams) were flagged
as ‘irrelevant’; all other documents were considered
‘possibly relevant’(n= 586). Next, we examined the
full text of each document, with documents contain-
ing empirical information on the biotic or abiotic
responses to dam removal flagged as ‘relevant.’In
total, we identified 139 documents that contained
empirical information on biotic and abiotic responses
to dam removal in the U.S., and from these we
extracted information on (1): characteristics of the
dam and its removal (e.g., dam height, location, year
of removal) (2); physical, water-quality, and biologi-
cal response metrics that were measured; and (3) the
type of experimental design employed, as well as the
duration and frequency of sampling. A complete list
of the different metrics extracted from each docu-
ment is available in an online database
27
(http://doi.
org/10.5066/F7K935KT). We recognize that dam-
removal decision-making is based on social and eco-
nomic factors as well as on potential physical and
ecological responses
28–30
; however, an analysis of
social and economic factors was outside of the scope
of our literature review and this article.
American Rivers Dam-Removal Database
The American Rivers Dam Removal Database, main-
tained by the non-profit organization American Riv-
ers (http://www.americanrivers.org/), lists dams that
have been removed in the U.S. since 1912
(n= 1231).
4
American Rivers has collected these
data annually since 1999 by surveying dam-removal
practitioners, compiling information from state and
federal agencies, and reviewing media references to
dam removals (e.g., books, newspaper articles).
Records in the database include dam name, water-
body name, state, year of removal, and dam height.
The three primary criteria for inclusion in the data-
base are (1): the dam removal was intentional (i.e.,
directly caused by humans) (2); the full vertical
extent of the dam was removed over more than half
of the dam’s width; and (3) the dam was not later
rebuilt in the same location. There was no minimum
dam height for inclusion in the database. The data-
base may under-represent the actual number of
removals because of incomplete historical knowledge,
inadequate formal tracking or limited information
sharing by agencies, or disparities among states
regarding what constitutes a dam, and thus a dam
removal.
DATA ANALYSIS
We used information from the USGS Dam Removal
Science Database and the American Rivers Dam
WIREs Water Status and trends of dam removal research in the U.S.
Published 2016. This article is a U.S. Government work and is in the public domain in the USA.

Removal Database to graphically analyze (1) the
number of removed dams and studied dam removals
by state (2); the cumulative number of removals and
studied removals by year; and (3) the distribution of
dam heights for removals and studied removals. We
used the National Inventory of Dams (NID), a con-
gressionally mandated database updated every
2 years by the U.S. Army Corps of Engineers (http://
nid.usace.army.mil/), to compare the distribution of
all dams (Figure 1(a)) to those that have been
removed. Dam removal studies were categorized by
publication outlet, experimental design, and the
amount and duration of monitoring data available.
Studies were grouped into four experimental designs,
that were distinguished by the availability of spatial
and temporal reference sites (1): before-after-control-
impact (BACI), (2) before-after, (3) control-impact
(space-for-time), and (4) impact only (i.e., only post-
removal data collected). We recorded the types of
variables that were monitored, which for the pur-
poses of this manuscript were categorized into 15 dif-
ferent metric types (Web Table 1). These metrics
were then grouped into three broad categories (1):
physical metrics (e.g., channel morphology and
hydraulics), (2) biological metrics (e.g., fish and
invertebrates), and (3) water-quality metrics (e.g.,
water temperature, nutrients, and contaminants). To
evaluate associations between metrics for dam-
removal studies, we plotted pairwise co-occurrence
among different metrics using the Circos Software
package.
31
NATIONWIDE PATTERNS OF
DAM-REMOVAL RESEARCH
The distribution of existing dams is markedly differ-
ent from the distribution of dam removals (Figures 1
(a) and (b)). Regions with relatively large numbers of
dam removals include the upper Midwest and the
Atlantic and Pacific coasts, such as Wisconsin, Penn-
sylvania, and California (Figure 1(b)). The difference
between the distribution of existing dams and dam
removals may be attributable to the removal of small
dams (i.e., low-head and run-of-river dams, <2 m in
height), which represent the majority of dam
removals (Figure 2(a)) but do not meet NID inclusion
requirements (i.e., ≥25 feet in height and >15 acre-
feet in storage, or ≥50 acre-feet storage and >6 feet in
height) unless they are considered hazardous.
7,32
Dif-
ferences in socio-cultural history between states and
regions may also partially explain the differences
200
10
20
400
600
800
1000
1200
Number of dams
Year
(a)
Studies
1944 1949 1954 1959 1964 1969 1974 1979 1984 1989 1994 1999 2004 2009 2014
(b)
0
10
20
30
40
50
<2 >2–4 >4–8 >8–16 >16–32 >32–64
Dam height category (m)
Percentage
Dams removed
Dams removed with ≥1 study
FIGURE 2 |Compilation of dams removed (orange) and dams with at least one published study (blue) by: (a) cumulative frequency
distribution by year removed (exclusive of dams with no known date of removal), with a count of the number of dam removal studies published
each year inserted below the x-axis, and (b) relative frequency (percentage) in each dam height category. Data from American Rivers
4
and
Bellmore et al.
27
Advanced Review wires.wiley.com/water
Published 2016. This article is a U.S. Government work and is in the public domain in the USA.

between the spatial distribution of dams and dam
removals. In New York, for example, the removal of
Fort Edward Dam in 1973 released PCB-
contaminated sediments downstream, causing sub-
stantial environmental impacts,
33
and may have led
to heightened caution regarding dam removal. In
contrast, a catastrophic dam failure in Pennsylvania
caused the Johnstown Flood of 1889
34
that killed
more than 2000 people and possibly improved public
perception of dam removal as a means to assure
human safety. These two adjacent states have similar
numbers of dams in the NID (1969 and 1522 in
New York and Pennsylvania, respectively) but have
very different numbers of dam-removal projects
(23 and 276, respectively). Although this particular
comparison is speculative, historical events of this
nature can influence public opinion, which may
result in institutional and regulatory differences
among states and associated differences in the moti-
vation and capacity to remove dams.
The distribution of dam-removal studies is simi-
lar to patterns of dam removal (Figure 1(b) and (c)).
The two states with the most dams removed, Penn-
sylvania and Wisconsin, also have the largest number
of studied dam removals (20 and 18, respectively).
Regions with the greatest numbers of studied dam
removals include the upper Midwest and Atlantic
and Pacific coasts, with few studied dam removals in
the interior of the conterminous U.S.. These regional
patterns of dam-removal research may be related to
the timing of removals relative to the initiation of
interest in studying the effects of removal. For
instance, dam removals in the interior of the U.S. that
are included in the American Rivers database largely
occurred prior to the first dam-removal publication
35
that we identified from the literature. It is also nota-
ble that there are several states where a large number
of removals have occurred, but our literature search
yielded few (e.g., California, New Hampshire) or no
(e.g., Maine, New Jersey) scientific studies that met
our criteria. This finding may be attributed to our
inclusion criteria, search methods, publication
latency, or an actual absence of studies.
The number of dam-removal studies has not
increased at the same rate as the number of dam
removals, particularly during the past three decades
(Figure 2(a)). Based on our literature review, by the
end of 2014, only 9% of all dam removals had been
scientifically evaluated. Interestingly, this percentage
is similar to that reported for river restoration efforts
in general.
36
This seemingly low percentage of stud-
ied dam removals may be explained, in part, by the
relative scarcity of published studies associated with
the removal of small dams. Dams less than 2 m in
height represented 28% of all dam removals in the
U.S., but they comprised only 12% of all studied
dam removals (Figure 2(b)), a pattern which could be
explained by the removal of large high-visibility
dams that attract more research funding and public
interest (or because scientific assessments are viewed
as unnecessary for smaller dam removals). For
instance, removal of the 12-m-high Marmot Dam in
Oregon—one of the larger dam removals to date—is
associated with 12 scientific studies. The slower over-
all rate of increase in dam-removal studies relative to
dam-removal practice may also be related to delays
between data collection and publication; for long-
term studies, there may be lags of several years
between removal and publication.
CHARACTERISTICS OF
DAM-REMOVAL STUDIES
Of the documents that we identified in our literature
search, more than 50% were peer-reviewed journal
articles (Figure 3(a)), followed by theses (24%), and
reports (16%; primarily by federal and state govern-
ment agencies). Non-peer-reviewed reports (‘grey lit-
erature’) from private consultants and local agencies
were generally not identified by the search engines
we employed. Google Scholar, for instance, generally
searches for documents with a clear list of authors,
which is frequently not included by local agencies
and consulting firms. Thus, although these docu-
ments may contain relevant information, their rela-
tive inaccessibility may reduce their utility for
informing dam-removal science and practice.
The BACI experimental design was the most
prevalent approach for studying the effects of dam
removal (36% of studies; Figure 3(b)). Although the
BACI design has limitations,
37
it is generally consid-
ered a robust approach for detecting responses to
unreplicated treatments.
38
Overall, approximately
80% of studies had temporal (before-after) and/or
spatial (control-impact) control sites. These control
sites provide a baseline from which to calculate the
direction and magnitude of dam-removal responses.
The necessity of a control, however, depends on the
type of question being addressed. Impact studies
(those without a control) can provide important
insights into rates of physical and ecological changes
after dam removal.
39
However, these studies do not
permit the interpretation of post-removal changes in
the context of pre-removal or control site conditions
(e.g., upstream conditions). Whether a study employs
a temporal and/or spatial control depends on funding
availability, sufficient advance notice of dam
WIREs Water Status and trends of dam removal research in the U.S.
Published 2016. This article is a U.S. Government work and is in the public domain in the USA.

removal, the existence of an appropriate control site
(e.g., McHenry and Pess
40
), and the types of research
questions being addressed.
Although 65% of studies monitored conditions
before the dam was removed, monitoring was gener-
ally short in duration (one or two years; Figure 3(c)).
The collection of longer-term pre-removal data can be
difficult because it is often hard to predict when dam
removal will occur. For many small dam removals in
particular, short planning phases prior to removal
may limit opportunities for data collection. We also
found that the duration of post-removal monitoring
was limited. Only 35% of studies had post-removal
monitoring for longer than 2 years, and only 5% for
longer than 5 years (Figure 3(c)). The scarcity of
longer-term monitoring (longer than 5 years) may be
caused by limited funding for data collection or lack
of interest by scientists.
41
In addition, many responses
to dam removal can happen relatively quickly (e.g.,
sediment erosion and deposition
42,43
), after which
there may no longer be motivation to continue
monitoring. In contrast, ecological responses to dam
removal may take decades to detect.
44,45
Studies on the effects of dam removal generally
measured physical responses more frequently than
biological and water-quality responses (Figure 4).
The top five monitored metrics were all physical, and
all were measured in over 30% of studies (Figure 4
(c)). Fish were the only biological metric that was
measured in more than 30% of studies, followed by
aquatic macroinvertebrates (19%) and riparian vege-
tation (13%; Figure 4(b)). No water-quality metrics
were measured in more than 20% of studies
(Figure 4(a)), and contaminants were only measured
in 6% of dam removal studies, even though this
information is regularly collected during the permit-
ting process.
46
Given the potential negative conse-
quences associated with contaminants (such as the
New York case study listed above
33
), it is likely that
reservoirs with elevated contaminant concentrations
were not prioritized for removal. We hypothesize
that physical metrics were measured more frequently
010 20 30 40
0 102030405060
0
5
10
15
20
25
30
35
40
Journal article
Thesis
Book
Report
Conference proceeding
Before–after–control–impact (BACI)
Before–after
Impact
Control–
impact
11023456782345678
(a) Publication type
(c) Number of years studied before and/or after dam removal
(b) Study design
Before After
Proportion of studies (percentage)Proportion of studies (percentage)
Proportion of studies
(percentage)
No pre-removal
data collected
Number of years with data collection
FIGURE 3 |Summary statistics of dam removal research showing proportion of studies by (a) type of publication, (b) type of study design,
and (c) the number of years of before/after dam removal data collection. Note that for study designs with no pre-dam removal data (i.e. control/
impact and impact categories) ‘years before’data are included as a zero (light gray bar).
Advanced Review wires.wiley.com/water
Published 2016. This article is a U.S. Government work and is in the public domain in the USA.

than biological and water-quality metrics because the
focus in early dam-removal studies was on showing
that dams can be removed safely, with low risks
associated with flooding and sediment deposition. As
dam removals continue, however, the prevalence of
biological and water-quality monitoring are likely to
increase as the focus turns to understanding the
longer-term ecological recovery.
Most studies (85%) measured 5 or fewer of the
15 dam-removal monitoring metrics that we enumer-
ated (Figure 4(d)). This finding suggests that most stud-
ies evaluate a specific aspect of dam removal (e.g.,
sediment dynamics or fish) rather than the broader eco-
system response (although there are notable excep-
tions, e.g.,
47,48
). Moreover, co-occurrence analysis
among the dam-removal response metrics illustrated
that physical responses were not frequently measured
in conjunction with biological and water-quality
responses (Figure 5). Studies that measured a given
physical component of the system were much more
likely to measure another physical component of the
system than a biological or water-quality component.
Channel morphology, for instance, is strongly linked
to all other physical metrics in the circular plot
(Figure 5), but it is only weakly linked to biological
and water-quality metrics. Conversely, we found that
many biological and water-quality metrics were meas-
ured more frequently with physical metrics than met-
rics of the same category. These patterns of co-
occurrence may reflect assumptions about causation
associated with dam removal. For example, biological
and water-quality responses are assumed to be depend-
ent on physical responses, and physical responses are
frequently assumed to be independent of biological
and water-quality influences (although this is fre-
quently incorrect
49
). Holistic ecosystem studies may
also be difficult to identify in the literature because
physical, biological, and water-quality findings from
the same dam removal get published in separate jour-
nal articles. For example, Doyle et al.
50
described the
Temperature
Nutrients
Dissolved O2
Turbidity
Contaminants
Substrate size
Stage/discharge
Sediment
dynamics
Channel
morphology
Hydraulics
Fish
Macroinvertebrates
Riparian
Mussels
Algae/
aquatic plants
0 15304560 0 15 30 45 60 0 15 30 45 60
0
5
10
15
20
25
30
1234567891011
Number of metrics measured
Proportion of studies
(percentage)
Percentage of studies
(a) Water quality metrics
(d) Frequency distribution of metrics measured per study (in metric classes a-c, above)
(b) Biological metrics (c) Physical metrics
FIGURE 4 |Proportion of dam removal studies (
n
= 139) classified by type of response metrics measured (a–c) and the total number of
metrics measured per study (d).
WIREs Water Status and trends of dam removal research in the U.S.
Published 2016. This article is a U.S. Government work and is in the public domain in the USA.

geomorphic response of the Baraboo River to the
La Valle Dam removal in Wisconsin, whereas Stanley
et al.
51
described the macroinvertebrate response.
CHALLENGES AND OPPORTUNITIES
Our review illustrates that there is a growing body of
scientific literature that reports on the outcomes of
dam removal. These studies were undertaken in a
research environment often driven by independent
actors working across wide geographic areas, and
frequently with limited funding for research and
monitoring. Despite these conditions, scientists have
mobilized data-collection efforts to evaluate project
outcomes and to take advantage of novel research
opportunities. These efforts have helped focus and
define research questions and inform practice. For
example, in the Elwha River, Washington, decades of
experiments,
52
physical and numerical models,
53,54
and monitoring
43,55
have helped manage the removal
of two large dams storing 21 million m
3
of sedi-
ment.
56
Nevertheless, our review identified several
‘potential’gaps in the scientific study of dam removal
in the U.S. (Table 1). We emphasize ‘potential’
Biological — Water Quality
Biological — Physical
Biologicial — Biological
Water Quality — Water Qualit
y
Physical — Physical
Physical — Water Quality
Intra-category co-occurrence
Inter-category co-occurrence
Water
Quality = 57
Biotic
n = 79
Physical
n = 117
Number of Papers
M
u
s
s
e
l
s
R
i
p
a
r
i
a
n
A
l
g
a
e
/
S
u
b
s
t
r
a
t
e
s
i
z
e
W
a
t
e
r
A
q
u
a
t
i
c
p
l
a
n
t
s
i
n
v
e
r
t
e
b
r
a
t
e
s
F
i
s
h
C
o
n
t
a
m
i
n
a
n
t
s
N
u
t
r
i
e
n
t
s
T
u
r
b
i
d
i
t
y
D
i
s
s
o
l
v
e
d
O
2
D
i
s
c
h
a
r
g
e
H
y
d
r
a
u
l
i
c
s
m
o
r
p
h
o
l
o
g
y
S
e
d
i
m
e
n
t
C
h
a
n
n
e
l
t
e
m
p
e
r
a
t
u
r
e
M
a
c
r
o
-
d
y
n
a
m
i
c
s
FIGURE 5 |Pairwise co-occurrence patterns of different response metrics monitored in dam removal studies, assigned to physical, biological,
and water-quality categories (outer ring). Ribbons inside the circle connect metrics that were measured in the same papers. The base of each
ribbon has a width proportional to the number of studies in which that metric was monitored in conjunction with the metric at the other end. To
read the figure, start with a given metric of interest (e.g., fish), and compare the width of the different ribbons at base of the metric. Wider-based
ribbons connect to metrics that were frequently measured in conjunction with the selected metric, whereas narrow ribbons connect to metrics that
were less frequently measured in conjunction with the selected metric. Ribbon colors denote co-occurrence patterns both within and among
categories (see legend). Inset Venn diagram shows the number of studies per category, with the amount of overlap proportional to co-occurrence.
TABLE 1 |Gaps in the Science of Dam Removal
Only 9% of dam removals have been described in published scientific literature.
No dam removal studies exist in the central U.S., and many states have few studies relative to the number of removed dams.
There are few studies of the smallest dam removals (those less than 2 m in height) relative to the prevalence of their removal.
Monitoring is generally short-term (1–2 years) and often includes little or no data prior to dam removal.
Fewer studies report biological and water-quality responses to dam removal relative to physical responses (e.g., sediment and flow).
Few holistic ecosystem-level studies exist that attempt to measure linkages among physical, water-quality, and biological responses.
Advanced Review wires.wiley.com/water
Published 2016. This article is a U.S. Government work and is in the public domain in the USA.

because more studies are needed to explore the extent
to which these gaps may actually limit our ability to
draw inferences, interpret findings, or make deci-
sions. However, by explicitly highlighting these gaps,
we provide clear directions for future study.
By comparing dam removals that have been
studied to all dam removals, we found (1) that fewer
than 10% of removals have been studied, (2) almost
no studies have occurred across the central U.S.,
and (3) studies of small dam removals are under-
represented in the literature relative to the prevalence
of their removal. The extent to which these discre-
pancies limit predictive capacity and decision-making
will depend on the transferability of this scientific
information to other locations. For example, the low
proportion of studied removals only limits predictive
capacity if those removals are not representative of
the range of local and regional factors that control
responses to removal, such as channel slope, land
use, location of the dam in the watershed, proximity
to other dams, and the type and age of the dam.
24
Monitoring all removals is not feasible, and given the
upward trend in dam removal, the proportion of
studied removals is likely to further decline in the
future. In this context, limited resources for monitor-
ing will need to be focused on those removals that
provide the greatest power to inform decision mak-
ing. Additional analyses are needed to gauge the ‘rep-
resentativeness’of dam-removal research by
comparing the local and regional environmental con-
text of studied removals to dam removals without
studies, and dams (particularly older dams) that are
likely to be removed in the future.
Along with the challenge of appropriately dis-
tributing research efforts, there is also a need for
studies that are robust enough to evaluate both
short- and long-term ecosystem responses to dam
removal. Most of the studies identified in our review
only measured short-term responses, and often with
limited or no pre-removal data collection. This lack
of monitoring information could compromise our
ability to understand long-term ecological responses,
and to separate those responses from background
environmental variability. Although some responses
can happen relatively quickly following removal
(e.g., sediment transport
5,42
), river channels and adja-
cent riparian vegetation may continue to adjust for
several decades,
24,57
which in turn can have long-
term effects on the recovery of aquatic organisms
such as fishes.
58
The paucity of biological and water-quality
studies (in comparison to physical studies), as well as
holistic ecosystem studies, may also limit our under-
standing. Because river ecosystems are inherently
complex and interconnected,
59
physical, water-
quality and biological responses to removal may
interact in complex ways. For instance, fish responses
to dam removal may be directly linked to changes in
dissolved oxygen and water turbidity, and indirectly
influenced by sediment deposition on macroinverte-
brate prey.
60
A mechanistic understanding of these
complex linkages is critical for predicting how spe-
cific components of the systems will respond to dam
removal. This understanding will require holistic eco-
system studies that integrate multiple scientific disci-
plines (sensu Bushaw-Newton et al.
47
) and employ a
combination of experimental, observational, and
modeling approaches. Moreover, because river resto-
ration actions such as dam removal are ultimately a
societal decision, holistic evaluations that extend
beyond the environmental sciences to include social,
economic, and political systems may also be
necessary.
61–63
Based on these challenges, we have identified
several opportunities to conduct dam-removal science
to better inform management decisions. Although
our focus was on dam removal, the research gaps
that we identified are strikingly—but perhaps not
surprisingly—similar to those identified for the
broader field of river restoration.
36,63,64
Thus, the
opportunities that we propose for advancing dam-
removal science largely echo recommendations made
for river restoration in general.
First, we suggest that scientists work with dam-
removal practitioners to identify regional and
national sets of priority research questions to focus
science on research that will advance practice. As an
example, questions could be organized around com-
mon management concerns, such as the spread of
invasive species (fishes and riparian vegetation) or
the downstream effects of reservoir sediment ero-
sion.
65
Second, to address these priority questions,
there is a need for greater national- and regional-level
research coordination that can facilitate the alloca-
tion of limited resources more efficiently and in a
manner that provides stronger inference for applying
results across a broader range of river systems in dif-
ferent geographic locations (e.g., Lindenmayer and
Likens
41
). This national-level coordination may also
promote long-term monitoring and holistic
ecosystem-scale studies.
A third and final recommendation is to create a
centralized database to store dam removal science
information. Although this has been recommended in
the past, both for dam removal
66
and river restora-
tion in general,
36
a complete and centralized data-
base for up-to-date dam removal information
currently does not exist (but see the Clearinghouse of
WIREs Water Status and trends of dam removal research in the U.S.
Published 2016. This article is a U.S. Government work and is in the public domain in the USA.

Dam Removal Information; http://calisphere.org/col-
lections/26143; accessed May 3, 2016). In particular,
relational databases are needed in order to query
dam removals and scientific studies relative to geo-
spatial information (e.g., digital elevation models,
land use, and hydrography). These databases would
make it possible to examine the biogeographic con-
text of dam-removal responses and to predict
responses to future removals. Part of these efforts
could also include identifying, analyzing, and inter-
preting relevant, but as yet unpublished data and
gray literature, and ensuring that this information is
readily available and effectively communicated to
practitioners.
It is our hope that the information collected
and analyzed in this article, although incomplete, can
begin to address these database needs. For instance,
the information contained within the USGS Dam
Removal Science Database
27
and the American Riv-
ers Dam Removal Database
4
is currently being trans-
lated into a dynamic and visual online database tool,
termed the Dam Removal Information Portal or
‘DRIP’(https://www.sciencebase.gov/drip/). This
database can currently be used to visualize the loca-
tion of dam removals included in the American Riv-
ers database, summarize basic information on
removed dams (e.g., dam height, year removed, met-
rics monitored), filter removed dams by different
attributes (e.g., types of studies available), and pro-
vide links to published studies. This database has
already been linked to the National Hydrography
Dataset (NHD), and, in the future, it will be linked
to other relevant geospatial data sets such as the
National Water Information System (NWIS).
CONCLUSION
Scientists will increasingly be called upon to make
predictions about how rivers will respond to dam
removal and how long they may take to recover.
Our analysis serves as an assessment of the current
state of dam-removal science—information that can
be used to inform the practice of dam removal.
Although we identified numerous gaps in the
research that might limit the application of science
to decision making, these challenges can be con-
fronted by articulating and prioritizing research
needs and questions, facilitating regional and
national coordination of research, and increasing
the accessibility and communication of dam-removal
science to researchers and managers, as well as the
public. Accomplishing these tasks will facilitate feed-
back between researchers and practitioners needed
to align science with management, and may help
shift the perception of dam removal from that of a
localized and opportunistic endeavor to a broader
and more strategic nation-wide adaptive manage-
ment experiment.
ACKNOWLEDGMENTS
This article was produced with support from the U.S. Geological Survey’s John Wesley Powell Center for Anal-
ysis and Synthesis. We thank Amy East, Chris Magirl, Jim Evans, Kathryn Ronnenberg, Stuart Lane and two
anonymous reviewers for their feedback on this manuscript, as well as our fellow Powell Center working group
participants for their insights about dam removal and the many discussions that sustained this project. We also
thank Jill Baron and Leah Colasuonno for logistical support at the Powell Center. Any use of trade, product,
or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government, the
authors, or their affiliations.
REFERENCES
1. ZarflC, Lumsdon A, Berlekamp J, Tydecks L,
Tockner K. A global boom in hydropower dam con-
struction. Aquat Sci 2015, 77:161–170. doi:10.1007/
s00027-014-0377-0.
2. Doyle MW, Stanley EH, Havlick DG, Kaiser MJ,
Steinbach G, Graf WL, Galloway GE, Riggsbee JA.
Aging infrastructure and ecosystem restoration. Science
2008, 319:286–287.
3. Brown CA, Graham WJ. Assessing the threat to life from
dam failure. Water Resour Bull 1988, 24:1303–1309.
4. American Rivers. American Rivers dam removal data-
base. 2014. Available at: http://www.americanrivers.
org/initiative/dams/projects/2014-dam-removals.
Accessed March 1, 2016.
5. O’Connor JE, Duda JJ, Grant GE. 1000 dams down
and counting. Science 2015, 348:496–497.
6. Duda JJ, Freilich JE, Schreiner EG. Baseline studies in
the Elwha River ecosystem prior to dam removal: intro-
duction to the special issue. Northwest Sci 2008, 82
(Special Issue 1):1–12.
Advanced Review wires.wiley.com/water
Published 2016. This article is a U.S. Government work and is in the public domain in the USA.

7. Graf W. Landscapes, commodities and ecosystems: the
relationship between policy and science for American
rivers. In: Sustaining Our Water Resources.
Washington, DC: National Academies Press; 1993,
11–42. doi:10.17226/2217.
8. U.S. Army Corps of Engineers. National inventory of
dams. 2013. Available at: http://nid.usace.army.mil.
Accessed April 20, 2014.
9. Baish SK, David SD, Graf WL. The complex decision
making process for removing dams. Environ Sci Policy
Sustain Dev 2002, 44:20–31.
10. Gregory S, Li H, Li J. The conceptual basis for ecologi-
cal responses to dam removal. BioScience 2002,
52:713–723.
11. Heinz Center. Dam Removal Science and Decision
Making. Washington, DC: The H. John Heinz III Cen-
ter for Science, Economics and the Environment; 2002.
12. Doyle MW, Harbor JM, Stanley EH. Toward policies
and decision-making for dam removal. Environ Manag
2003, 31:453–465.
13. Ward JW, Stanford JA. The serial discontinuity con-
cept of lotic systems. Dyn Lotic Ecosyst 1983,
10:29–42.
14. Stanley EH, Doyle MW. Trading off: the ecological
effects of dam removal. Front Ecol Environ 2003,
1:15–22.
15. Wohl E, Bledsoe BP, Jacobson RB, Poff NL,
Rathburn SL, Walters DM, Wilcox AC. The natural
sediment regime in rivers: broadening the foundation
for ecosystem management. BioScience 2015,
65:358–371.
16. Shafroth PB, Friedman JM, Auble GT, Scott ML,
Braatne JH. Potential responses of riparian vegetation
to dam removal. BioScience 2002, 52:703–712.
17. Hitt NP, Eyler S, Wofford JEB. Dam removal increases
American eel abundance in distant headwater streams.
Trans Am Fish Soc 2012, 141:1171–1179.
18. Pess G, Quinn T, Gephard S, Saunders R. Re-
colonization of Atlantic and Pacific rivers by anadro-
mous fishes: linkages between life history and the bene-
fits of barrier removal. Rev Fish Biol 2014,
24:881–900.
19. Doyle MW, Stanley EH, Harbor JM. Geomorphic ana-
logies for assessing probable channel response to dam
removal. J Am Water Resour Assoc 2002,
38:1567–1579.
20. Pizzuto J. Effects of dam removal on river form and
process. BioScience 2002, 52:683–691.
21. Evans JE, Gottgens JF. Contaminant stratigraphy of
the Ballville reservoir, Sandusky River, NW Ohio:
implications for dam removal. J Great Lakes Res
2007, 33:182–193.
22. Evans JE. Contaminated sediment and dam removals:
problem or opportunity? EOS 2015, 96:12–17.
doi:10.1029/2015EO036385.
23. Kornis MS, Vander Zanden MJ. Forecasting the distri-
bution of the invasive round goby (Neogobius mela-
nostomus) in Wisconsin tributaries to Lake Michigan.
Can J Fish Aquat Sci 2010, 67:553–562.
24. Hart DD, Johnson TE, Bushaw-Newton KL,
Horwitz RJ, Bednarek AT, Charles DF, Kreeger DA,
Velinsky DJ. Dam removal: challenges and opportu-
nities for ecological research and river restoration. Bio-
Science 2002, 52:669–681.
25. Stanley EH, Catalano MJ, Mercado-Silva N, Orr CH.
Effects of dam removal on brook trout in a Wisconsin
stream. River Res Appl 2007, 23:792–798.
26. Graf WL. Dam Removal Research: Status and Pro-
spects. Washington, DC: The H. John Heinz III Center
for Science, Economics and the Environment; 2003.
27. Bellmore JR, Vittum KM, Duda JJ, Greene SL. USGS
dam removal science database. US Geological Survey,
2015. http://doi.org/10.5066/F7K935KT. Accessed
June 1, 2015.
28. Born SM, Genskow KD, Filbert TL, Hernandez-
Mora N, Keefer ML, White KA. Socioeconomic and
institutional dimensions of dam removals: the Wiscon-
sin experience. Environ Manag 1998, 22:359–370.
29. Johnson SE, Graber BE. Enlisting the social sciences in
decisions about dam removal. BioScience 2002,
52:731–738.
30. Robbins JL, Lewis LY. Demolish it and they will come:
estimating the economic impacts of restoring a recrea-
tional fishery. J Am Water Resour Assoc 2008,
44:1488–1499.
31. Krzywinski M, Schein J, Birol I, Connors J,
Gascoyne R, Horsman D, Jones SJ, Marra MA. Cir-
cos: an information aesthetic for comparative geno-
mics. Genome Res 2009, 19:1639–1645.
32. Poff NL, Hart DD. How dams vary and why it matters
for the emerging science of dam removal. BioScience
2002, 52:659–668.
33. Shuman JR. Environmental considerations for asses-
sing dam removal alternatives for river restoration.
Regul River Res Manag 1995, 11:249–261.
34. McCullough D. The Johnstown Flood: The Incredible
Story Behind One of the Most Devastating Disasters
America Has Ever Known. New York: Simon and
Schuster Paperbacks; 1968.
35. Williams D. Effects of Dam Removal: An Approach to
Sedimentation. U.S. Army Corps of Engineers Hydro-
logic Engineering Center: Davis, CA; 1977.
36. Bernhardt ES, Palmer MA, Allan JD, Alexander G,
Barnas K, Brooks S, Carr J, Clayton S, Dahm C,
Follstad-Shah J, et al. Synthesizing U.S. river restora-
tion efforts. Science 2005, 308:636–637.
37. Underwood A. Beyond BACI: Experimental designs
for detecting human environmental impacts on tempo-
ral variations in natural populations. Aust J Mar
Freshwater Res 1991, 42:569–587.
WIREs Water Status and trends of dam removal research in the U.S.
Published 2016. This article is a U.S. Government work and is in the public domain in the USA.

38. Downes B, Barmuta L, Fairweather P, Faith D,
Keough M, Lake P, Mapstone B, Quinn G. Monitor-
ing Ecological Impacts: Concepts and Practice in
Flowing Waters. New York: Cambridge University
Press; 2002.
39. Downs PW, Singer MS, Orr BK, Diggory ZE,
Church TC. Restoring ecological integrity in highly
regulated rivers: the role of baseline data and analyti-
cal references. Environ Manag 2011, 48:847–864.
40. McHenry ML, Pess GR. An overview of monitoring
options for assessing the response of salmonids and
their aquatic ecosystems in the Elwha River following
dam removal. Northwest Sci 2008, 82 (Special Issue
1):29–47.
41. Lindenmayer DB, Likens GE. The science and applica-
tion of ecological monitoring. Biol Conserv 2010,
143:1317–1328.
42. Major J, O’Connor J, Grant G, Spicer K, Bragg H,
Rhode A, Tanner D, Anderson C, Wallick J. Initial flu-
vial response to the removal of Oregon’s Marmot
Dam. EOS 2008, 89:241–252.
43. East AE, Pess GR, Bountry JA, Magirl CS, Ritchie AC,
Logan JB, Randle TJ, Mastin MC, Minear JT,
Duda JJ, et al. Large-scale dam removal on the Elwha
River, Washington, USA: river channel and floodplain
geomorphic change. Geomorphology 2015,
228:765–786.
44. Pess G, McHenry M, Beechie T, Davies J. Biological
impacts of the Elwha River dams and potential salmo-
nid responses to dam removal. Northwest Sci 2008, 82
(Special Issue 1):72–90.
45. Beechie TJ, Sear DA, Olden JD, Pess GR,
Buffington JM, Moir H, Roni P, Pollock MM. Process-
based principles for restoring river ecosystems. BioSci-
ence 2010, 60:209–222.
46. Rathbun J, Graber B, Pelto K, Turek J, Wildman L. A
sediment quality assessment and management frame-
work for dam removal projects. In: Moglen GE,
ed. Managing Watersheds for Human and Natural
Impacts: Engineering, Ecological, and Economic Chal-
lenges. Reston, VA: American Society of Civil Engi-
neers; 2005, 1–12. doi:10.1061/40763(178)19.
47. Bushaw-Newton KL, Hart DD, Pizzuto JE,
Thomson JR, Egan J, Ashley JT, Johnson TE,
Horwitz RJ, Keeley M, Lawrence J, et al. An integrative
approach towards understanding ecological responses
to dam removal: the Manatawny Creek Study. JAm
Water Resour Assoc 2007, 38:1581–1599.
doi:10.1111/j.1752-1688.2002.tb04366.x.
48. Doyle MW, Stanley EH, Orr CH, Selle AR, Sethi SA,
Harbor JM. Stream ecosystem response to small dam
removal: lessons from the heartland. Geomorphology
2005, 71:227–244.
49. Fuller BM, Sklar LS, Compson ZG, Adams KJ,
Marks JC, Wilcox AC. Ecogeomorphic feedbacks in
regrowth of travertine step-pool morphology after dam
decommissioning, Fossil Creek, Arizona. Geomorphol-
ogy 2011, 126:314–332.
50. Doyle MW, Stanley EH, Harbor JM. Channel adjust-
ments following two dam removals in Wisconsin.
Water Resour Res 2003, 39:1011.
51. Stanley EH, Luebke MA, Doyle MW, Marshall DW.
Short-term changes in channel form and macroinverte-
brate communities following low-head dam removal.
J N Am Benthol Soc 2002, 21:172–187.
52. Childers D, Kresch DL, Gustafson SA, Randle T,
Melena J, Cluer B. Hydrologic data collected during
the 1994 Lake Mills Drawdown Experiment, Elwha
River, Washington. US Geological Survey Water-
Resources Investigations Report 99-4215; 2000.
53. Konrad CP. Simulating the recovery of suspended sedi-
ment transport and river-bed stability in response to
dam removal on the Elwha River, Washington. Ecol
Eng 2009, 35:1104–1115.
54. Bromley C, Randle T, Grant G, Thorne C. Physical
modeling of the removal of Glines Canyon Dam and
Lake Mills from the Elwha River, Washington. In:
Papanicolaou AN, Barkdoll BD, eds. Sediment
Dynamics Upon Dam Removal. Reston, VA: Ameri-
can Society of Civil Engineers; 2011, 97–114.
55. Warrick JA, Bountry JA, East AE, Magirl CS,
Randle TJ, Gelfenbaum G, Ritchie AC, Pess GR,
Leung V, Duda JJ. Large-scale dam removal on the
Elwha River, Washington, USA: source-to-sink sedi-
ment budget and synthesis. Geomorphology 2015,
246:729–750. doi:10.1016/j.geomorph.2015.01.010.
56. Randle TJ, Bountry JA, Ritchie A, Wille K. Large-scale
dam removal on the Elwha River, Washington, USA:
erosion of reservoir sediment. Geomorphology 2015,
246:709–728. doi:10.1016/j.geomorph.2014.12.045.
57. Foster D, Swanson F, Aber J, Burke I, Brokaw N,
Tilman D, Knapp A. The importance of land-use lega-
cies to ecology and conservation. BioScience 2003,
53:77–88. doi:10.1641/0006-3568(2003)053[0077:
TIOLUL]2.0.CO;2.
58. Quiñones RM, Grantham TE, Harvey BN,
Kiernan JD, Klasson M, Wintzer AP, Moyle PB. Dam
removal and anadromous salmonid (Oncorhynchus
spp.) conservation in California. Rev Fish Biol Fish
2014, 25:195–215.
59. Allen JD, Castillo MM. Stream Ecology: Structure and
Function of Running Waters. 2nd ed. Dordrecht:
Springer; 2007.
60. Tullos DD, Finn DS, Walter C. Geomorphic and eco-
logical disturbance and recovery from two small dams
and their removal. PLoS One 2014, 9:e108091.
61. Doyle MW. Dam removal in the United States: emer-
ging needs for science and policy. Eos 2003, 84:29–36.
62. Brown PH, Tullos D, Tilt B, Magee D, Wolf AT. Mod-
eling the costs and benefits of dam construction from a
Advanced Review wires.wiley.com/water
Published 2016. This article is a U.S. Government work and is in the public domain in the USA.

multidisciplinary perspective. J Environ Manag 2009,
90:S303–S311.
63. Wohl E, Lane SN, Wilcox AC. The science and prac-
tice of river restoration. Water Resour Res 2015,
51:5974–5997.
64. Palmer M, Allen JD, Meyer J, Bernhardt ES. River res-
toration in the twenty-first century: data and experien-
tial knowledge to inform future efforts. Restor Ecol
2007, 15:472–481.
65. Tullos D, Collins MJ, Bellmore JR, Bountry JA,
Connolly PJ, Shafroth PB, Wilcox AC. Synthesis of
common management concerns associated with dam
removal. J Am Water Resour Assoc In Press.
66. Pohl MM. American dam removal census: available
data and data needs. In: Graf WL, ed. Dam Removal
Research: Status and Prospects. Washington, DC: The
H. John Heinz III Center for Science, Economics, and
the Environment; 2003, 29–39.
WIREs Water Status and trends of dam removal research in the U.S.
Published 2016. This article is a U.S. Government work and is in the public domain in the USA.