Content uploaded by John F Bruno
Author content
All content in this area was uploaded by John F Bruno on Oct 22, 2021
Content may be subject to copyright.
301
© The Ecological Society of America www.frontiersinecology.org
A
lthough coastal regions constitute less than 4% of the
Earth’s land area, coastal habitats (eg beaches and
tidal wetlands; Figure 1) rank among the most valuable
natural resources globally (MA 2005). Over one-third of
the human population lives within 100 km of a coastline
and development is increasing rapidly in most coastal
regions of the world (MA 2005). Coastal development is
vulnerable to damage and loss from erosion, including
ambient waves, flooding, storms, and sea-level rise (SLR)
(Peterson et al. 2008).
Historically, shoreline hardening has been a common
response to coastal erosion, storm risks, and SLR, particu-
larly in industrialized countries with large coastal popula-
tions, such as the US, the Netherlands, and Japan
(Dugan et al. 2011). Shoreline hardening or armoring is
defined as the construction or placement of vertical sea-
walls or bulkheads, sloped riprap (eg rocks) revetments,
groins, jetties, or breakwaters along a shoreline (Figure
1). Although humans have been armoring coastlines for
centuries, the extent and rate at which this occurs has
increased markedly since 1900, and the effects of armor-
ing on coastal ecosystem functions and services have only
recently begun to be evaluated (NRC 2007).
Hardening of shorelines by means of seawalls or bulk-
heads can steepen and shorten shallow intertidal habitat
over time (Dugan et al. 2008). The structures themselves
also provide less physically complex habitat as compared
with natural shorelines, so that hardened shorelines gen-
erally support fewer species (Figure 1; Seitz et al. 2006;
Gittman et al. in press). When constructed landward of
tidal wetlands, hard structures may also increase seaward
scour during storm events and can prevent upslope migra-
tion of tidal wetlands as sea level rises, leading to their
eventual loss (termed “coastal squeeze”; Doody 2004).
Despite its adverse ecological effects, efforts to under-
stand the underlying causes and rates of shoreline armor-
ing have been limited (Dugan et al. 2011). Filling these
data gaps could help determine where and how much
shoreline is at risk. More specifically, identifying shore-
lines with tidal wetlands that are threatened with harden-
ing will help coastal managers to adopt more stringent
regulations for shoreline armoring that could prevent
future wetland losses. The purpose of this study was to:
(1) catalog tidal shoreline hardening in all continental
US coastal counties; (2) determine which physical and
socioeconomic features of a coastal region may be more
commonly associated with armoring; (3) identify regions
of the US likely to experience continued hardening and
subsequent coastal habitat loss; and (4) identify alterna-
tive management strategies for coastal protection.
nMethods
We selected the US for our analysis because of its exten-
sive coastline, high coastal population density (39% of
the population lives in coastal counties), and data avail-
ability, as well as its vulnerability to shoreline erosion,
flooding, and property damage (MA 2005; NOAA 2013).
Furthermore, in the US, there is growing interest in
developing a national coastal protection and adaptation
RESEARCH COMMUNICATIONS RESEARCH COMMUNICATIONS
Engineering away our natural defenses: an
analysis of shoreline hardening in the US
Rachel K Gittman1*†, F Joel Fodrie1, Alyssa M Popowich2, Danielle A Keller1, John F Bruno3, Carolyn A
Currin4, Charles H Peterson1, and Michael F Piehler1
Rapid population growth and coastal development are primary drivers of marine habitat degradation. Although
shoreline hardening or armoring (the addition of concrete structures such as seawalls, jetties, and groins), a
byproduct of development, can accelerate erosion and loss of beaches and tidal wetlands, it is a common practice
globally. Here, we provide the first estimate of shoreline hardening along US Pacific, Atlantic, and Gulf of Mexico
coasts and predict where future armoring may result in tidal wetland loss if coastal management practices
remain unchanged. Our analysis indicates that 22 842 km of continental US shoreline – approximately 14% of
the total US coastline – has been armored. We also consider how socioeconomic and physical factors relate to the
pervasiveness of shoreline armoring and show that housing density, gross domestic product, storms, and wave
height are positively correlated with hardening. Over 50% of South Atlantic and Gulf of Mexico coasts are fringed
with tidal wetlands that could be threatened by future hardening, based on projected population growth, storm
frequency, and an absence of coastal development restrictions.
Front Ecol Environ 2015; 13(6): 301–307, doi:10.1890/150065
1Institute of Marine Sciences, University of North Carolina at
Chapel Hill, Morehead City, NC; †current address: Marine Science
Center, Northeastern University, Nahant, MA *(r.gittman@
neu.edu); 2US Coast Guard, Portsmouth, VA; 3Department of
Biology, University of North Carolina at Chapel Hill, Chapel Hill,
NC; 4Center for Coastal Fisheries and Habitat Research, National
Oceanographic and Atmospheric Administration, Beaufort, NC
Shoreline hardening in the US RK Gittman et al.
302
www.frontiersinecology.org © The Ecological Society of America
framework in response to changing climatic conditions
(Peterson et al. 2008; Arkema et al. 2013).
Estimation of shoreline hardening along the US
coast
We used the National Oceanic and Atmospheric
Administration (NOAA) Office of Response and
Restoration’s Environmental Sensitivity Index (ESI) geo-
databases to calculate the linear kilometers of total shore-
line and kilometers of hardened shoreline for each coastal
county within the continental US (NOAA 2005). The
ESI identifies 15 shoreline types (eg Type 1: exposed
rocky shore or seawall) that are further subdivided into
more specific shoreline types (eg Type 1A: exposed rocky
shores; Type 1B: exposed, solid, man-made structures [sea-
walls]; Figure 1; WebTable 1). We grouped all man-made
structures (seawalls, bulkheads, riprap structures [revet-
ments, breakwaters, groins/jetties], and hybrid
seawall/bulkheads with riprap) to calculate the cumula-
tive lengths of hardened shoreline (Figure 1; WebTable 1).
We divided each state’s ESI shoreline by coastal county
and, for the Pacific and Atlantic coasts, by whether the
shoreline was “open” (ie directly exposed to the ocean) or
“sheltered” (ie located in a bay, sound, lagoon, or tidally
influenced river). We did not divide the Gulf of Mexico
coast into “open” or “sheltered” categories because much
of the Gulf coastline (eg the Louisiana coastline, the Big
Bend region of Florida) consists of reticulated wetlands
that cannot be easily classified in this way. We then calcu-
lated the length (in kilometers) of tidal shoreline (total
and armored), as well as the percentage of hardened shore
for each county. Additionally, we calculated the length of
tidal wetland shoreline (total and armored).
Regression tree analyses
To evaluate the relationship between potential drivers and
the percentage of hardened shoreline in each US coastal
county, we considered the following factors: 2010 housing
density (units per square kilometer), 2010 gross domestic
product (GDP, expressed in US dollars), coastal slope (%),
accretion/erosion rates (meters per year [m yr–1]), geomor-
phology, mean tidal range (m), mean wave height (m),
relative SLR (millimeters per year [mm yr–1]), storm fre-
quency between the years 1970–2010, relative county
shoreline position (north to south or west to east along
the coast), and years since a ban on shoreline hardening
was passed (sources included: the US Census Bureau, the
National Ocean Economics Program, the US Geological
Survey Coastal Vulnerability Index [USGS CVI] [for a
description of the variables, see Hammer-Klose and
Thiehler 2001], the Federal Emergency Management
Agency’s US Presidential Major Disaster and/or
Emergency Declarations [FEMA 2014], and federal and
state legislation and permitting procedures).
We ran separate regression trees for the Atlantic open
and sheltered coasts, the Pacific open and sheltered coasts,
and the Gulf of Mexico coast, to describe differences
among county-level shoreline hardening patterns, vis-à-
vis repeated partitioning of county armoring percentages
into increasingly homogeneous groups based on serial,
bimodal splits among potential drivers of hardening
(De’ath and Fabricius 2000). Because USGS CVI data
were not available for the Pacific sheltered coastal coun-
Figure 1. Types of natural and artificially hardened shorelines
found in the US: (a) rocky shore; (b) beach; (c) tidal marsh;
(d) mangrove; (e) seawall; (f) riprap revetment; (g) bulkhead;
and (h) breakwater. For images of other shoreline types found in
the US, refer to the NOAA ESI shoreline types image gallery
(http://response.restoration.noaa.gov/esi-shoreline-types).
(a–h) R Gittman
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
RK Gittman et al. Shoreline hardening in the US
303
© The Ecological Society of America www.frontiersinecology.org
ties, the regression tree for the Pacific sheltered coast does
not include these factors. We developed regression trees
using the analysis of variance (ANOVA) method of recur-
sive partitioning and we pruned over-fitted trees using k-
fold cross-validation. We ran all regression tree analyses
using R version 3.1.0 and rpart (Therneau et al. 2014).
nResults
The continental US was estimated to have 160 168 km of
tidal shoreline, of which 22 842 km (14%) was hardened
(WebTable 2). Brackish and salt marsh were the domi-
nant types of tidal wetland found along US coasts, mak-
ing up 48% of the total US shoreline (WebTable 3).
Approximately 1% (886 km) of existing tidal marsh
shoreline has been armored (eg a hard structure has been
built along and typically landward of the marsh)
(WebTable 3), although this does not account for marsh
already lost on the seaward side of man-made structures.
Shoreline hardening along US open coasts
Along the open coasts of the Atlantic and Pacific, 846 km
(9%) of the shoreline has been
hardened (Figures 2a and 3a).
Atlantic coastal counties that
experienced 17 or more storm
events between 1970 and 2010
(in Massachusetts, Maine, and
New Hampshire) had a higher
percentage of hardened shoreline
(µ = 30.7 ± 9.6%) than counties
that experienced fewer than 17
storms (µ = 8.3 ± 1.4%, R2=
0.25; Figure 2a). The percentage
of armoring along the Pacific
open coast was higher (µ = 24.1
± 2.8%) in counties where the
mean wave height was <1.3 m (ie
counties in southern California)
(Figure 3a). Counties with mean
wave heights ≥1.3 m and located
south of San Francisco County,
California, had more hardening
(µ = 11.0 ± 2.7%) than counties
north of San Mateo County,
California (µ = 2.4 ± 0.9%).
Mean wave height accounted for
70% of the variance in hardened
shoreline along the open Pacific
coast (R2= 0.80, full tree).
Shoreline hardening along US
sheltered coasts
Despite a considerable amount
of hardening along open coast-
Figure 2. The percentage of total tidal shoreline hardened by county and the regression
tree results for (a) the Atlantic open coast and (b) the Atlantic sheltered coast. “Housing
density” is the number of individual housing units per square kilometer (as defined by the
US Census Bureau), “Storms” are the total number of storms that resulted in a US
Presidential Major Disaster and/or Emergency Declaration from 1970 to 2010 (FEMA
2014), “Tide” is the mean tide range (m), “State” is the state in which the shoreline is
found, and “n” is the number of counties split into each node and used to calculate the
percentage of hardened shoreline. Error bars represent one standard error.
(a) Open coast
Hardened shoreline (%)
0.00–9.99
10.00–24.99
25.00–49.99
50.00–74.99
75.00–100.00
(b) Sheltered coast
0.00–9.99
10.00–24.99
25.00–49.99
50.00–74.99
75.00–100.00
Hardened shoreline (%)
Housing
density
Housing
density
Housing
density
Shoreline hardening in the US RK Gittman et al.
304
www.frontiersinecology.org © The Ecological Society of America
lines, hardening was more prevalent on sheltered
Atlantic and Pacific coasts (14% or 14 607 km;
Figures 2b and 3b; WebTable 2). Atlantic coastal
counties with housing densities ≥658 units km–2
and counties in southern Florida with housing den-
sities ≥126 units km–2 had the highest percentages
of shoreline armoring (µ = 60.9 ± 4.8% and µ =
62.3 ± 14.5%, respectively; Figure 2b). When
housing densities were <126 units km–2 and in
counties outside South Florida, fewer storms and a
smaller mean tide range were associated with less
hardening. Housing density accounted for 41% of
the variation in hardened shoreline among
Atlantic sheltered coastal counties (R2= 0.72, full
tree). New Hampshire, Rhode Island, and
Connecticut had the highest percentage of hard-
ened marsh shoreline on the Atlantic coast (7%,
6%, and 4%, respectively; WebTable 3).
California had the highest percentage of sheltered
armored shoreline on the Pacific coast (28%) and
possessed four of the five counties with the greatest
amounts of hardening (Figure 3b; WebTable 3).
California was also characterized by the highest
percentage of hardened marsh shoreline (2%) on
the Pacific Coast; in San
Francisco County, for in-
stance, 71% of marsh shore-
line had been armored (Web-
Table 3). Counties with the
highest GDP (≥$8.95 ×
1010) along the Pacific shel-
tered coast also had the high-
est percentages of hardened
sheltered shorelines (µ = 56.7
± 10.4%; Figure 3b). When
GDP was < $8.95 × 1010, and
housing density was ≥52
units km–2, counties with
more frequent storms also
had higher percentages of
armoring (µ = 43.9 ± 6.0%)
than counties with the same
housing densities but with
fewer than 12 storms (Figure
3b). GDP and housing den-
sity accounted for 53% of the
variance in hardened shore-
line percentages along the
sheltered Pacific coast (R2=
0.69, full tree).
Shoreline hardening
along the US Gulf of
Mexico coast
The Gulf of Mexico coast
had the same percentage of
Figure 3. The percentage of total tidal shoreline hardened by county and the regression tree results for
(a) the Pacific open coast and (b) the Pacific sheltered coast. “Wave height” is the mean wave height
(m); “GDP” is the 2010 US gross domestic product (US$, as defined by the US Bureau of Economic
Analysis); and “Housing density”, “Storms”, “State”, “n”, and error bars are defined as in Figure 2.
(a) Open coast
Hardened shoreline (%)
0.00–9.99
10.00–24.99
25.00–49.99
(b) Sheltered coast
0.00–9.99
10.00–24.99
25.00–49.99
50.00–74.99
75.00–100.00
Hardened shoreline (%)
Wave height
Housing
density
RK Gittman et al. Shoreline hardening in the US
hardened shoreline (16%) as the entire Pacific coast (this
percentage drops to 9% when all of the highly reticulated
marsh shoreline of Louisiana is included) (WebTable 2;
Figure 4). Counties with higher housing densities (≥91 units
km–2, eg Orleans Parish, Louisiana; Harris County, Texas)
had higher percentages of armoring (µ = 41.7 ± 4.7%) than
counties with lower housing densities (Figure 4). When
GDP was < $3.09 × 109, Gulf counties in Texas had more
hardening (µ = 16.3 ± 2.5%) than Gulf counties outside of
Texas (µ = 3.7 ± 1.1%; Figure 4). Housing density alone par-
titioned 46% of the variation in the percentage of hardened
shoreline along the Gulf coast (R2= 0.56, full tree). Texas
had more than five times as much armored marsh shoreline
(6%) than other Gulf states (0.4–1.4%) (WebTable 3).
nDiscussion
Our analysis indicates that 14% of the contiguous US shore-
line is hardened and that 64% of armoring has occurred
along Atlantic and Pacific sheltered shorelines, such as estu-
aries, lagoons, and tidally influenced rivers (WebTable 2;
Figures 2b and 3b). Thus, shoreline hardening is likely a sub-
stantial yet largely understudied means by which humans
modify and degrade coastal ecosystems in the US.
Potential drivers of shoreline hardening
Understanding the potential drivers of shoreline harden-
ing could help identify where and how much shoreline and
305
© The Ecological Society of America www.frontiersinecology.org
associated habitats are at risk of being lost to this process in
the near future. Our analyses revealed that housing density
and GDP, respectively, were the best predictors of armoring
on US Atlantic and Pacific sheltered coasts, as well as
along the Gulf of Mexico coast (Figures 2–4). Globally,
man-made shoreline structures are associated with densely
populated coastlines (eg around the Mediterranean), and
have been used to protect both commercial and residential
development and infrastructure for generations (Dugan et
al. 2011). Most major US coastal metropolitan areas are
located on sheltered coasts and tend to be heavily armored
(eg New York, Los Angeles, New Orleans), regardless of
other physical shoreline characteristics or processes.
Coastal metropolitan areas also support development in
neighboring coastal counties (eg for the purposes of
industry and tourism); this likely contributes to the
spread of shoreline hardening, despite relatively low
housing densities in these adjacent areas (Figures 2b, 3b,
and 4; NOAA 2013). Specifically, these regions have a
history of coastal modification that includes dredging of
canals, to support shipping traffic from major ports (eg
New York, Corpus Christi, Miami) and for flood control
(eg South Florida canal systems), that probably con-
tributed to hardening (US Census Bureau 2010).
Outside of metropolitan areas, the addition of man-
made structures is more closely related to the vulnerability
of coastal developments to damage from physical processes
(eg storms, erosion). Along the Pacific open coast, the
rocky shorelines of northern California, Oregon, and
Figure 4. The percentage of total tidal shoreline hardened by county and the regression tree results for the Gulf of Mexico coast.
“Housing density”, “GDP”, “State”, “n”, and error bars are defined as in Figures 2 and 3.
0.00–9.99
10.00–24.99
25.00–49.99
50.00–74.99
Hardened shoreline (%)
Housing density
Shoreline hardening in the US RK Gittman et al.
Washington are associated with wave heights greater than
1.3 m, and are therefore not suitable for most types of
development; this may be the reason for the lesser degree
of armoring in these regions (Figure 3a). The regression
tree selected a single variable that reduced the most vari-
ance in shoreline hardening; however, wave height was
also strongly positively correlated with both county loca-
tion (R2= 0.82) and the number of years since a ban on
shoreline hardening was implemented (R2= 0.67).
Greater storm frequency was the most important predic-
tor of armoring on the Atlantic open coast (Figure 2a). Hard
structures are often built along coastlines in response to
damage and erosion from major storm events (eg seawall
construction in Galveston, Texas, following a major hurri-
cane; Hansen 2007); areas prone to major storms would
therefore be expected to have more hardened shoreline.
However, there is evidence that natural beach dune and
marsh shorelines experience less erosion and damage than
hardened shorelines during single storm events (Thieler
and Young 1991; Gittman et al. 2014). Given the uncer-
tainty associated with the performance of bulkheads and
seawalls during these storm events, the use of hard struc-
tures in response to storms should be evaluated further.
Given the strong relationships between both housing
density and GDP and shoreline hardening, other socio-
economic factors we did not include – such as commercial
and recreational shoreline uses, coastal land ownership
(public versus private), and coastal property values – may
provide additional insights into the patterns of observed
shoreline hardening in the US. Further refinement of the
relationship between socioeconomic factors and armoring
will allow policy makers and coastal-resource managers to
develop targeted strategies (eg tax breaks for waterfront
property owners that choose not to build a bulkhead or
seawall) for reducing shoreline hardening in areas with
vulnerable coastal habitats. These approaches may
become increasingly fine-tuned as we learn more about
how construction/repair costs or the cascading effects of
neighboring shoreline hardening (Scyphers et al. 2015)
drive decisions regarding shoreline maintenance.
Predictions for future hardening and habitat loss
Although European countries have been adding man-
made structures to their shorelines for centuries, in the
US, most of this construction likely occurred after 1900
(Dugan et al. 2011), making the rate of shoreline harden-
ing in the US about 200 km yr–1. If this rate remains con-
stant and coastal populations continue to increase
according to NOAA projections (NOAA 2013), the per-
centage of hardened shoreline will double by 2100, result-
ing in nearly one-third of the contiguous US coastline
being armored. This projected rate is probably conserva-
tive but assumes that no additional restrictions are placed
on shoreline hardening (only eight states have imple-
mented total or partial bans on construction of hard
structures along the shore).
Some of the largest increases in population density are
predicted to occur along the South Atlantic and Gulf
coasts, where most of the US’s remaining tidal salt
marshes (> 50%) and mangrove forests (100%) are cur-
rently found (WebTable 3; Kennish 2001). As much as
50% of US salt marsh has been lost over the past century,
largely as a result of human activities (Kennish 2001).
Although only a small percentage (1%) of existing tidal
wetlands is currently hardened, this percentage does not
account for the wetlands likely lost in the past to harden-
ing (WebTable 3). On the Atlantic coast, 60% of the
land between 0–1 m above current sea level is expected
to be developed and hardened, thereby resulting in a
large-scale coastal squeeze on tidal wetlands (Titus et al.
2009). Given the prevalence and ecological conse-
quences of shoreline hardening, steps should be taken to
reduce the rate of armoring and to implement alternative
stabilization strategies (eg submerged sills, marsh plant-
ing; see Gittman et al. in press).
nConclusions
Our assessment demonstrates that much of the US shore-
line is vulnerable to future habitat loss if actions are not
taken to revise current coastal management strategies.
Our estimate – that 14% of US tidal shoreline is hard-
ened – is conservative, as all types of tidal shoreline were
included in our analysis. When shorelines that are not
likely to be armored (eg naturally rocky shores) are
excluded, the majority of the natural shoreline most vul-
nerable to future hardening (eg beaches, wetlands) is
found along the Gulf of Mexico and South Atlantic
coasts. Given the projected increases in population for
these two regions, these coasts will likely experience the
highest rates of future hardening and the associated loss
of marsh and other vegetated intertidal habitat to coastal
squeeze (Doody 2004). This, in turn, will result in the loss
of critical coastal ecosystem services such as provision of
nursery habitat for commercially and recreationally valu-
able fish and crustaceans, filtration of nutrients and pol-
lutants from terrestrial runoff, carbon burial, and erosion
protection (Peterson et al. 2008).
Further analyses of the socioeconomic drivers of shore-
line hardening are needed to determine how to prevent
or reduce further destruction of sensitive habitats such as
tidal wetlands. Although our analysis provides baseline
estimates of armoring, continued updates will be required
to calculate region-specific rates. Coastal managers could
use these rates to assess the cumulative impacts on coastal
habitats and to assess the risk of future habitat loss. Policy
makers should also use these assessments to develop
informed legislation and regulations, including a revision
of the US Army Corps of Engineers’ policy on nationwide
permits to account for the future loss of habitat as a result
of coastal squeeze, which will likely extend well beyond
the construction footprint of a hard structure. Finally, we
recommend that new management guidelines be devel-
306
www.frontiersinecology.org © The Ecological Society of America
RK Gittman et al. Shoreline hardening in the US
oped to incorporate green infrastructure and planning for
shoreline migration (eg rolling easements, bulkhead
removal) with SLR. Without substantial changes to
coastal management policies and development practices,
the US coastlines will likely lose their natural defenses.
nAcknowledgements
We thank J Grabowski, T Rodriguez, and S Scyphers for
their guidance and advice. This research was funded by a
NOAA National Estuarine Research Reserve System
graduate fellowship to RKG, a US National Science
Foundation grant to FJF (NSF OCE-1155628), and the
University of North Carolina at Chapel Hill. The scien-
tific results and conclusions, as well as any views or opin-
ions expressed herein, are those of the author(s) and do
not necessarily reflect the views of NOAA or the US
Department of Commerce.
nReferences
Arkema KK, Guannel G, Verutes G, et al. 2013. Coastal habitats
shield people and property from sea-level rise and storms.
Nature Clim Change 3: 1–6.
De’ath G and Fabricius KE. 2000. Classification and regression
trees: a powerful yet simple technique for ecological data analy-
sis. Ecology 81: 3178–92.
Doody JP. 2004. “Coastal squeeze” – an historical perspective. J
Coast Conserv 10: 129–38.
Dugan JE, Hubbard DM, Rodil IF, et al. 2008. Ecological effects of
coastal armoring on sandy beaches. Mar Ecol-Prog Ser 29:
160–70.
Dugan JE, Airoldi L, Chapman MG, et al. 2011. Estuarine and
coastal structures: environmental effects, a focus on shore and
nearshore structures. In: Wolanski E and McLusky D (Eds).
Treatise on estuarine and coastal science. Waltham, MA:
Academic Press.
FEMA (Federal Emergency Management Agency). 2014.
Presidential and Emergency Disaster Declarations 1970–2010.
Washington, DC: FEMA. www.fema.gov/disaster-process-disas-
ter-aid-programs. Viewed 16 Aug 2014.
Gittman RK, Popowich AM, Bruno JF, and Peterson CH. 2014.
Marshes with and without sills protect estuarine shorelines
from erosion better than bulkheads during a Category 1 hurri-
cane. Ocean Coast Manage 102: 94–102.
Gittman RK, Peterson CH, Currin CA, et al. Living shorelines can
enhance the nursery role of threatened estuarine habitats. Ecol
Appl. In press; doi:10.1890/14-0716.1.
Hammer-Klose ES and Thiehler ER. 2001. Coastal vulnerability to
sea-level rise: a preliminary database for the US Atlantic,
Pacific and Gulf of Mexico coasts. Washington, DC: US
Geological Survey. Digital Data Series 68. http://pubs.usgs.gov/
dds/dds68. Viewed 14 Jan 2014.
Hansen B. 2007. History lesson – weathering the storm: the
Galveston seawall and grade raising. Civil Eng 77: 32–33.
Kennish MJ. 2001. Coastal salt marsh systems in the US: a review
of anthropogenic impacts. J Coastal Res 17: 731–48.
MA (Millennium Ecosystem Assessment). 2005. Ecosystems and
human well-being. Washington, DC: Island Press.
NOAA (National Oceanographic and Atmospheric Admini-
stration). 2013. The US population living at the coast. Silver
Spring, MD: NOAA. http://stateofthecoast.noaa.gov/
population/welcome.html. Viewed 30 Sep 2014.
NOAA (National Oceanographic and Atmospheric Admini-
stration). 2005. Environmental sensitivity index geodatabases.
Silver Spring, MD: NOAA. http://response.restoration.noaa.
gov/maps-and-spatial-data/download-esi-maps-and-gis-
data.html. Viewed 10 Oct 2011.
NRC (National Research Council). 2007. Mitigating shore erosion
along sheltered coasts. Washington, DC: National Academies
Press.
Peterson CH, Barber R, Cottingham KL, et al. 2008. National estu-
aries. In: Julius SH and West JM (Eds). Preliminary review of
adaptation options for climate-sensitive ecosystems and
resources. Washington, DC: US Climate Change Science
Program.
Scyphers SB, Picou JS, and Powers SP. 2015. Participatory conser-
vation of coastal habitats: the importance of understanding
homeowner decision making to mitigate cascading shoreline
degradation. Conserv Lett 8: 41–49.
Seitz R, Lipcius R, Olmstead N, et al. 2006. Influence of shallow-
water habitats and shoreline development on abundance, bio-
mass, and diversity of benthic prey and predators in
Chesapeake Bay. Mar Ecol-Prog Ser 326: 11–27.
Therneau T, Atkinson B, and Ripley B. 2014. rpart: recursive par-
titioning and regression trees. R package version 4.1-9.
http://CRAN.R-project.org/package=rpart. Viewed 1 Jun
2015.
Thieler ER and Young RS. 1991. Quantitative evaluation of
coastal geomorphological changes in South Carolina after
Hurricane Hugo. J Coast Res S8: 187–200.
Titus JG, Hudgens DE, Trescott DL, et al. 2009. State and local
governments plan for development of most land vulnerable to
rising sea level along the US Atlantic coast. Environ Res Lett 4:
044008.
US Census Bureau. 2010. US detailed county census data.
Washington, DC: US Census Bureau.
307
© The Ecological Society of America www.frontiersinecology.org
