evaluating the noaa coastal ecology
TRANSCRIPT
-
7/28/2019 Evaluating the NOAA coastal ecology
1/18
Evaluating the NOAA Coastal and Marine EcologicalClassification Standard in estuarine systems: A Columbia River
Estuary case study
Matthew L. Keefer a,*, Christopher A. Peery a, Nancy Wright a, William R. Daigle a,Christopher C. Caudill a, Tami S. Clabough a, David W. Griffith a, Mark A. Zacharias b
a Fish Ecology Research Laboratory, Department of Fish and Wildlife Resources, University of Idaho, Moscow, ID 83844-1141, USAb Department of Geography, University of Victoria, P.O. Box 3050 STN CSC, Victoria, BC V8W 3P5, Canada
Received 17 January 2007; accepted 19 November 2007
Available online 26 December 2007
Abstract
A common first step in conservation planning and resource management is to identify and classify habitat types, and this has led to a pro-
liferation of habitat classification systems. Ideally, classifications should be scientifically and conceptually rigorous, with broad applicability
across spatial and temporal scales. Successful systems will also be flexible and adaptable, with a framework and supporting lexicon accessible
to users from a variety of disciplines and locations. A new, continental-scale classification system for coastal and marine habitats dthe Coastal
and Marine Ecological Classification Standard (CMECS)dis currently being developed for North America by NatureServe and the National
Oceanic and Atmospheric Administration (NOAA). CMECS is a nested, hierarchical framework that applies a uniform set of rules and
terminology across multiple habitat scales using a combination of oceanographic (e.g. salinity, temperature), physiographic (e.g. depth, substra-
tum), and biological (e.g. community type) criteria. Estuaries are arguably the most difficult marine environments to classify due to large spatio-
temporal variability resulting in rapidly shifting benthic and water column conditions. We simultaneously collected data at eleven subtidal sitesin the Columbia River Estuary (CRE) in fall 2004 to evaluate whether the estuarine component of CMECS could adequately classify habitats
across several scales for representative sites within the estuary spanning a range of conditions. Using outputs from an acoustic Doppler current
profiler (ADCP), CTD (conductivity, temperature, depth) sensor, and PONAR (benthic dredge) we concluded that the CMECS hierarchy
provided a spatially explicit framework in which to integrate multiple parameters to define macro-habitats at the 100 m2 to >1000 m2 scales,
or across several tiers of the CMECS system. The classifications strengths lie in its nested, hierarchical structure and in the development of
a standardized, yet flexible classification lexicon. The application of the CMECS to other estuaries in North America should therefore identify
similar habitat types at similar scales as we identified in the CRE. We also suggest that the CMECS could be improved by refining classification
thresholds to better reflect ecological processes, by direct integration of temporal variability, and by more explicitly linking physical and
biological processes with habitat patterns.
2007 Elsevier Ltd. All rights reserved.
Keywords: classification systems; current measurement; acoustic Doppler current profiler (ADCP); estuaries; United States; Columbia River Estuary
1. Introduction
Estuaries are spatially and temporally dynamic transition
zones spanning multiple spatial and temporal scales. The
boundaries of estuarine habitats are structured over time and
space primarily as a result of daily tidal cycles, seasonal and
inter-annual variations in river discharge and temperature,
* Corresponding author.
E-mail addresses: [email protected] (M.L. Keefer), nwright@uidaho.
edu (N. Wright), [email protected] (W.R. Daigle), [email protected] (C.C.
Caudill), [email protected] (T.S. Clabough), [email protected].
mil (D.W. Griffith), [email protected] (M.A. Zacharias).
0272-7714/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ecss.2007.11.020
Available online at www.sciencedirect.com
Estuarine, Coastal and Shelf Science 78 (2008) 89e106www.elsevier.com/locate/ecss
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.elsevier.com/locate/ecsshttp://www.elsevier.com/locate/ecssmailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected] -
7/28/2019 Evaluating the NOAA coastal ecology
2/18
and long-term shifts in bedform and sediment distribution
(Sherwood and Creager, 1990; Dyer, 1997; Reed et al.,
2004). Classification of estuarine habitats is therefore both
scale dependent (Attrill and Rundle, 2002) and strongly
context driven (Elliott and McLusky, 2002), meaning the scale
of habitat partitions must appropriately correspond with the
physical or biological processes being investigated and thepartitions may vary depending on scale. For example,
ecological boundaries in estuaries may be much more spatially
restricted for benthic plants and animals than are those for
more mobile organisms that can migrate as physical habitats
change (Bottom and Jones, 1990; Colloty et al., 2002). For
the most motile and euryhaline species, estuaries might be
considered relatively uniform mixed freshwater-seawater
habitats, versus a complex continuum of interconnected habi-
tats or physical gradients for more sedentary, stenohaline, or
otherwise ecologically-restricted organisms (Neira et al.,
1992; Harris et al., 2001). Importantly, the complex interac-
tions over time and space between abiotic and biotic processes
often result in ecological boundaries that do not necessarilycorrespond with boundaries derived using physical or
chemical surrogates.
The challenges of scale and context in estuaries have long
been recognized and highlight the need for spatially and
temporally explicit classification systems (Hansen and Rattray,
1966; Cowardin et al., 1979; Jay et al., 2000; Elliott and
McLusky, 2002). Many estuarine, coastal and marine classifi-
cations have been developed in recent decades, using a variety
of conceptual and methodological approaches (e.g., Boyd
et al., 1992; Iba~nez et al., 1997; Buzzelli et al., 1999;
Whitfield, 1999; Edgar et al., 2000; Bricker et al., 2003;
Kenny et al., 2003; Roff et al., 2003). These systems havebeen generated to address an assortment of management and
conservation objectives, and as such have arisen from multiple
disciplines and were developed mostly at local or regional
scales. A number of these efforts have demonstrated both
recurring and persistent associations of marine biological
communities with oceanographic and physiographic structures
to a degree that the spatial locations of communities can be
mapped with measures of certainty (e.g. Doyle et al., 1993;
Auster et al., 2001). However, few have attempted to compre-
hensively reflect underlying ecological pattern and process at
multiple scales, which would be broadly useful for facilitating
local, regional and global comparisons. One attempt, the
Coastal and Marine Ecological Classification Standard
(CMECS), is being developed as a potential single classifica-
tion standard for North America (Madden et al., 2005). The
intent of CMECS is to provide a framework for systematically
inventorying, describing, characterizing, and predicting
habitats and, where applicable, their constituent species and
communities. The framework consists of six, nested levels
linking broad marine environment types (e.g. intertidal,
oceanic) downward to biotopes, which are discrete habitat
units that support a unique and recognizable community
type. The overall objective is that CMECS will be a flexible
and evolving classification framework, designed to be widely
applicable across all coastal and marine habitat types from
freshwater-marine and marine-terrestrial interfaces to the
open ocean (Madden et al., 2005).
While the conceptual framework has been defined and
successfully applied (mapped) at the higher levels, it is as
yet unknown whether many components of the classification
correspond with real-world habitats that can be spatially
mapped using existing inventory methods. In this paper, weattempted to apply the CMECS estuarine definitions at various
levels using several commonly-collected data types acquired
at representative sites spanning a range of ecotypes occurring
in the Columbia River Estuary (CRE, Washington-Oregon).
We focused on parameters describing energy inputs (e.g.,
current velocity), physicochemical characteristics of the water
column, and benthic substrate rather than biotic distributions
because the CMECS approach is structured by physicochemi-
cal processes at larger spatial scales and because many
management uses of classifications aim to identify suitable
habitat for species currently depressed or absent. Specific
study objectives included: (1) testing whether existing
available sampling approaches (e.g., acoustic Doppler currentprofilers [ADCP] and benthic and water column sampling
devices) can provide the underlying physical and chemical
data necessary to classify estuarine habitats at one or more
scales; (2) evaluating the sensitivity of CMECS to spatial
and temporal variability in both vertical (e.g., water column)
and horizontal gradients in environmental conditions (e.g.,
across zones of freshwater influence); (3) examining the
relationship between CMECS classification levels and the pro-
cesses underlying habitat development and persistence; and
(4) evaluating the potential for CRE habitat mapping at the
various CMECS classification levels for hypothetical mobile
(e.g., salmon) and sessile (e.g., benthic infauna) organisms.Our primary aim was to evaluate CMECS as a classification
tool and it was beyond the scope of the study to produce
a definitive classification of the CRE.
2. Materials and methods
2.1. Columbia River Estuary: Background
The Columbia River drains approximately 660,000 km2 of
the U.S. Pacific Northwest and British Columbia, and has
among the highest annual discharge of all North American
rivers (mean annual flow w7000 m3 s1) (Simenstad et al.,
1990; Naik and Jay, 2005). The estuary is characterized by
strong tidal currents and highly variable freshwater inputs
(Hamilton, 1990; Jay and Smith, 1990; Baptista et al.,
2005). Within-year river discharge near the mouth can vary
by an order of magnitude, from low flows near 2000 m3 s1
in fall to levels exceeding 20,000 m3 s1 during summer
runoff peaks (Bottom et al., 2005).
The upstream extent of the salt-fresh mixing zone and the
presence or absence of a salt wedge in the CRE is strongly
linked to hydrologic conditions and spring-neap tidal cycles
(Hamilton, 1990; Jay and Smith, 1990). In contrast to many
large estuaries, high river discharge through the CRE results
in generally low salinity and rapid flushing times, and the
90 M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106
-
7/28/2019 Evaluating the NOAA coastal ecology
3/18
degree of vertical stratification varies with freshwater input
(Giese and Jay, 1989). The spatial extent of saltwater intru-
sions is also topographically controlled (Fox et al., 1984).
Two major channels, a series of minor channels, mid-estuary
sandbanks, and peripheral tidal flats, marshes, and bays shape
circulation patterns, with the greatest salinity intrusion in the
deep channels (Hughes and Rattray, 1980; Hamilton, 1990;Jay and Smith, 1990; Johnson et al., 2003).
Tidal and wave energy, combined with river currents and
large-scale circulation patterns have shaped the bedforms
and sedimentary geology of the CRE. Because the size and
distribution of benthic sediments are important determinants
of habitat type and stability, considerable effort has been
invested in CRE sediment mapping and modeling. In
a multi-year study, Sherwood and Creager (1990) analyzed
more than 2000 sediment samples and concluded that the
CRE is dominated by a narrow range of sands and finer
sediments. The sampling study and an associated energetics
study (Jay et al., 1990) found that CRE sediments were largely
riverine in origin, tended to decrease in size closer to theocean, and were mobile on daily and seasonal time scales,
particularly in areas with greater energy inputs. Local current
velocities and physical energy inputs appear to be critical fac-
tors shaping sediment distribution processes and consequently
the complex mosaic of benthic habitats in the CRE. This
suggests that spatial and temporal variability in energy inputs
will be particularly important to any habitat classification
effort in this system.
The broad-based understanding of the physical processes at
work in the CRE has been complimented by many ecological
studies, driven in part by the major decline in anadromous
Pacific salmonids (Oncorhynchus spp.). The Columbia Riverwas historically one of the worlds most productive salmon
rivers (Chapman, 1986; Augerot, 2005), but overfishing, water
withdrawal, and habitat loss and alteration decimated runs
(National Research Council, 1996; Ruckelshaus et al., 2002;
McClure et al., 2003). The construction of dozens of upriver
hydroelectric and water storage dams, combined with
estuarine diking and dredging, significantly altered hydrologic
regimes, sediment and carbon inputs, and subsequently CRE
habitats and food webs (Bottom et al., 2005). Salmonid
declines have precipitated a massive recovery effort (National
Marine Fisheries Service, 2000) and as a result, patterns of sal-
monid habitat use and mortality in the CRE have been studied
for decades (e.g., Sims and Johnsen, 1974; Ledgerwood et al.,
1991; Collis et al., 2002; Bottom et al., 2005; Schreck et al.,
2006). Non-salmonid flora and fauna in the CRE have also
been well-studied, including distribution, abundance, and
ecological relationships for fish (Haertel and Osterberg,
1967; Misitano, 1977; McCabe et al., 1983), benthic inverte-
brates (Simenstad et al., 1981; Emmett and Durkin, 1985;
Bottom and Jones, 1990; Jones et al., 1990), and phytoplank-
ton and zooplankton (Haertel et al., 1969; Frey et al., 1981;
Amspoker and McIntire, 1986; Morgan et al., 1997). A large
multidisciplinary effort (Columbia River Estuary Task Force)
has integrated much of the CRE research in a series of publi-
cations and habitat atlases (many available at http://www.
orednet.org/wcrest/index.html), providing a good backdrop
for evaluating classifications derived using CMECS. However,
these data sets were not used in the test of CMECS in the CRE
because we could not efficiently resolve temporal and spatial
scale incompatibilities between existing datasets. Specifically,
the CMECS evaluation required data collected simultaneously
across multiple scales at each location while much of thehistoric data could not be merged for single locations across
multiple scales at adequate resolutions.
2.2. Study sampling protocols
We selected 11 subtidal CRE study sites along the w25 km
reach between Astoria and Woody Island, WA (Fig. 1, Table 1).
In site selection, we aimed to capture a wide range of physical
and hydrological variability. The sampled sites included large
natural and dredged channels, sand flats, the relic channels of
Cathlemet Bay, and sloughs surrounding the islands south of
the dredged main channel. Sites were also selected so that avariety of substrates were sampled, ranging from the dominant
sand and fine sediment habitats of the CRE, to sites with
potential large woody debris or boulder fields. The sampling
design was further intended to capture gradients in salinity,
tidal influence, and water velocity.
Sites at Fort Columbia and Cliff Point (Fig. 1, sites 1 and 2)
were in the natural (non-dredged) North Channel, east and
west of the deep trough that runs from Ellice Point on the
Washington State coast to Desdemona Sands in mid-estuary
(south of sites 1 and 2). These sites include deep holes and
troughs (>15 m), subtidal (5e15 m), and shallow subtidal
(2e
5 m) habitats; neither the artificial shoreline (rock rein-forcement) nor the intertidal ribbon at the edge of Desdemona
Sands was sampled. Two other study sites, Grays Point and
Portuguese Point (Fig. 1, sites 3 and 4) were influenced by
fresh water, sediment, and nutrients from Grays River and
Deep River. These inputs move westward across the mud flats
of Grays Bay and add to the continual flux in the North
Channel; these combine to influence physical and biological
processes at these survey sites. Greys Point and Portuguese
Point sites also included the full range of CRE estuarine depth
zones. With the exception of the Tongue Point site (Fig. 1, site
6) which included a portion of the dredged shipping lane, the
remaining study sites (Fig. 1, sites 5, 7e11) were located in
the southern sloughs and natural channels from Cathlemet
Bay eastward to Woody Island. These sites were primarily
subtidal or shallow subtidal habitats with reduced flow. Shore-
line and intertidal habitats adjacent to the study sites included
a mix of freshwater and tidal marshes, mud and sandflats, tidal
drainage networks, and diked areas.
All sampling took place between 24 September and 20
October, 2004 to coincide with fall salmonid migrations and
an associated salmonid tagging study. During this time, tides
measured at Astoria ranged from approximately 1.7e3.1 m
relative to mean chart datum (NOAA station #9439040,
4612.50 N, 12346.00 W) and Columbia River discharge
measured upstream near Quincy, OR was variable but near
91M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106
http://www.orednet.org/%223C%3Bcrest/index.htmlhttp://www.orednet.org/%223C%3Bcrest/index.htmlhttp://www.orednet.org/%223C%3Bcrest/index.htmlhttp://www.orednet.org/%223C%3Bcrest/index.htmlhttp://www.orednet.org/%223C%3Bcrest/index.htmlhttp://www.orednet.org/%223C%3Bcrest/index.html -
7/28/2019 Evaluating the NOAA coastal ecology
4/18
the annual low (mean 3,850 m3 s1, SD 575 m3 s1;
USGS station #14246900, 4610.90 N, 12310.80 W).
All sampling took place from a 6.1 m inboard jetboat. Aninstrument platform for this boat included a pair of pivoting
booms to deploy an ADCP, sidescan sonar and acoustic and
radio telemetry devices for fish tracking. Other sampling
equipment included a conductivity-temperature-depth (CTD)
probe run from a transom-mounted downrigger, and a benthic
dredge (PONAR) operated with a hand line. Results of the
salmon telemetry study were reported in (Griffith, 2007).
2.3. Current velocity surveys
Velocity profiles were collected using a boat-mounted1200 kHz ADCP (Teledyne RD Instruments Inc., San Diego,
CA). The instrument, a Workhorse Rio Grande model, was
mounted to a boom and when lowered was approximately
0.3 m below the surface. Velocity accuracy for this ADCP
was 2.5 cm s1 and the near-bed region missed was typically
about 1 m. In addition to automatically filtering ambiguous
velocity data near the bottom, the ADCP checked velocity
Fig. 1. Map of Columbia River Estuary (CRE) showing depths of intertidal (high tide to 2 m), shallow subtidal (2 e5 m), subtidal (6e15 m) and deep (>15 m). The
locations show 11 sites where ADCP, CTD, and/or benthic dredge samples were collected during SeptembereOctober, 2004. Sites: 1 Fort Columbia; 2 Cliff
Point; 3 Grays Point; 4 Portuguese Point; 5 Mott Island West; 6 Tongue Point Channel; 7 Cathlamet Bay North; 8 McGregor Island; 9 South
Channel; 10 Prairie Channel; 11 Woody Island Channel.
Table 1
Names, locations and lengths of sampling sites where benthic sediment (PONAR), water velocity (ADCP), and physiochemical (CTD) data were collected. Dashes
indicate data were not collected
Site Number & Name Latitude Longitude Transect length PONAR ADCP CTD(m) Ebb Flood Ebb Flood
1. Fort Columbia 4613.90 12354.20 2,080 x x x x x
2. Cliff Point 4615.70 12350.20 450 x x x x x
3. Grays Point 4616.10 12347.00 900 x x x x x
4. Portuguese Point 4616.60 12345.30 270 x x x x x
5. Mott Island West 4611.80 12345.00 800 x x x x x
6. Tongue Point Channel 4613.2 12345.10 1,530 e x x x e
7. Cathlamet Bay Northa 4612.60 12343.60 560 x x x x x
8. McGregor Islanda 4612.20 12341.80 490 x x x x x
9. South Channel 4610.30 12342.40 180 x x x x e
10. Prairie Channelb 4610.70 12339.40 270 x x x e e
11. Woody Is. Channel 4614.60 12333.80 730 x x x x x
a Located in the North Channel of Cathlamet Bay.
b Only 2 of 6 ebb-tide and 4 of 6 flood-tide ADCP transects collected.
92 M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106
-
7/28/2019 Evaluating the NOAA coastal ecology
5/18
data at each bin (see below), and only bins containing valid
data were retained (Gordon, 1996). The ADCP was integrated
with DGPS technology (Trimble Navigation Ltd., Sunnyvale,
CA; model DSM 132) to provide navigational positions and
facilitate transect replication. An integrated sonar feature
allowed bottom-tracking, which adjusted the ADCP velocity
measurements for boat speed relative to the ground. Real-time data presentation during the surveys was via an onboard
computer, using the Windows-based software WinRiver (v.
1.06, RD Instruments).
Twelve ADCP profiles were collected for most transects,
which included two replicates of six different tidal periods
(early ebb and flood, middle ebb and flood, and late ebb and
flood). All transects were perpendicular to the prevailing
current and transect length varied from 180e2080 m (mean
750 m) depending on site configuration. Because tidal timing
differs by location within the estuary, transects were run based
on estimated timing from the Astoria tidal gage, and the mid-
dle periods approximated maximum predicted currents.
Transect replicates were spread across 2 to 3 days at eachsite and therefore provided velocity snapshots rather than
continuous profiles. Transects were collected only during
fall, and results may or may not represent conditions present
during other seasons or discharge/tidal conditions.
The mechanics of ADCP velocity calculations have been
described in detail in Wewetzer et al. (1999), Simpson
(2001), and Dinehart and Burau (2005). In essence, ADCP
uses the Doppler Effect to determine water velocity in three
dimensions (east-west, north-south, and vertical) across the
entire water column. The ADCP collects all velocity measure-
ments, or vectors, simultaneously in a vertical profile (depth
bins 0.5 m) using a 4-beam transducer approximately onceper second. Each vertical profile is called a velocity ensemble,
and in our study the ADCP automatically averaged 5 ensem-
bles internally to compress data. This resulted in 200e800
ensembles per transect. Because velocities in adjacent ensem-
bles were typically highly correlated, we further averaged
ADCP data at 10-ensemble intervals using Matlab (v. 7.0.0)
to simplify some data presentation. Within tidal stage, transect
replicates were generally very similar.
2.4. CTD surveys
Eight CTD profiles, four each during ebb and flood tides,
were collected at the deepest point along each sampled
transect using a Hydrolab Minisonde (Hach Co., Loveland,
Colorado) (Table 1). In addition to conductivity (salinity),
depth, and temperature, the Hydrolab also collected dissolved
oxygen data for each profile. Dissolved oxygen data were
strongly correlated with temperature and salinity across sites,
but we considered the data quality for oxygen to be lower than
for other variables because of potential calibration errors.
Therefore, we present oxygen results primarily to demonstrate
oxygens utility as a classification variable. CTD data were
collected at 5 s intervals, at approximately 0.5 m increments
starting at the bottom of the water column. Within tidal cycle,
the four profiles were collected sequentially, with time gaps
between replicates typically
-
7/28/2019 Evaluating the NOAA coastal ecology
6/18
The second CMECS level (Formation) is based on major
geomorphic or hydrographic features. Examples of formations
include islands, peninsulas, trenches, seamounts, reefs,
lagoons, embayments, oceanic currents, and coastal fronts.
At the time of writing, CMECS had identified 28 estuarine
formations (http://www.natureserve.org/getData/CMECS/).
Many types of CMECS estuarine regime formations are found
in the CRE study.
The third level (Zone) is a vertical classification ranging
from the supratidal (e.g., the surf or splash zone) to the deepocean bottom. Relevant CMECS classification zones across
the littoral, water column and bottom occur in the CRE with
spatial scales ranging from 100 m2 to 10,000 km2. Because
our sampling sites were all subtidal (i.e., not in the littoral
zone), we only classified water column and bottom zones.
Nested within zones are macrohabitats (the 4th level),
which structure the distributions of ecological communities
along physical or chemical gradients. Macrohabitats occur at
scales of 100 m2 to several thousand square meters and are
repeatable and persistent physical environments with relatively
homogenous geomorphology, hydrology and vegetative struc-
tures but multiple distinct biological associations (Madden
et al., 2005). Level five (Habitat) is at the scale directly
used by biota (e.g., for food, shelter, spawning, or refuge)
and is typically modified or shaped by several specific environ-
mental variables. Habitats occur at smaller spatial scales
within formations and macrohabitats, and may be similar to
or smaller than macrohabitats in spatial extent. Often, multiple
habitats are contained within a macrohabitat, and therefore
both macrohabitat and habitat scales potentially overlappedthe scale of our sampling. Level six (Biotope) is defined
by specific faunal and/or floral relationships, where species
are physically linked to the habitat (e.g., plants or organisms
with very high site fidelity) and therefore help define it (e.g.,
seagrass beds). Classification at the biotope level was beyond
the study scope.
A critical aspect of the CMECS classification approach is
the use of attribute-based descriptors, which can be physico-
chemical, physical, geomorphologic, biological, anthropo-
genic, or biogeographic (Table 2). Descriptors can be
Regime
Littoral Water column Benthic
Zone
Macrohabitat MacrohabitatMacrohabitat
HabitatHabitatHabitat
Biotope Biotope Biotope
Geoform/Hydroform
L6: Biotope
L5: Habitat
L4: Macrohabitat
L3: Zone
L2: Formation
L1: Regime
10 km2 - >1,000 km2
10,000 m2 - 100 km2
10 m2 10,000 km2
100 m2 1000s m2
1 m2 - 100 m2
1 m2 - 100 m2
Fig. 2. Example structure of the Coastal and Marine Ecological Classification Standard (CMECS) classification, modified from Madden et al. (2005). In the full
hierarchy, Levels 2e
5 are nested within five regimes: Estuarine, Freshwater-influenced, Nearshore Marine, Neritic, and Oceanic. Descriptors and Classifiers areused to further describe and categorize ecosystem and environmental conditions at each level (see Table 2).
Table 2Descriptors, defined in Madden et al. (2005) as environmental characteristics that provide insight to a classification unit. At the higher levels of the classification
descriptors are used to differentiate broad physical and hydrologic regimes and formations. At lower levels of the classification, finer subsets of each descriptor are
used as classifiers for the identification of discrete habitats. Items in italics were not evaluated
Physico-Chemical Physical Geomorphologic Biological
Salinity Energy Type Slope Cover Type
Oxygen Energy Intensity Large Scale Relief Cover Class
Temperature Energy Direction Substrate Type Trophic Status
Turbidity Class Primary Water Source Substrate Size
Turbidity Type Enclosure Substrate Composition Anthropogenic
Turbidity Provenance Depth Class Relief Anthropogenic Impact
Tide Range Profile
Spatial Patterns Temporal Persistence Biogeographic
Photic Quality Ecological Region
Subzone
94 M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106
http://www.natureserve.org/getData/CMECS/http://www.natureserve.org/getData/CMECS/ -
7/28/2019 Evaluating the NOAA coastal ecology
7/18
applied to all six CMECS levels, if the scale and characteris-
tics of the classification unit are appropriate. Examples include
salinity, temperature, depth, oxygen and turbidity levels,
energy type and intensity, tide range, water source, slope,
substrate composition, trophic status, temporal persistence,
and degree of human impact. The final classification compo-
nents are classifiers, which are used to more precisely defineand categorize descriptors. For example, the descriptor salinity
can be classified using practical salinity units (psu) as fresh
(0 psu), oligohaline (0e5 psu), mesohaline (5e18 psu), poly-
haline (18e30 psu), euhaline (30e40 psu), marine (35 psu),
hyperaline (>40 psu), or freshwater-influenced (< 30 psu
for at least 2 months) (Madden et al., 2005). Quantification
and precise definitions of the CMECS descriptors and
classifiers are an evolving part of the classification.
3. Results
3.1. Velocity profiles
Mean mid-tide water velocities at all study sites were
between 0.16 and 0.80 m s1 (Table 3). However, velocities
varied considerably within individual transects in response to
tidal stage, stratification, and underlying bathymetry. At sites
with substantial saltwater intrusion (e.g., Fort Columbia and
Cliff Point), velocity profiles were often vertically and/or
horizontally stratified, with the magnitude of the velocity gra-
dient related to tidal stage. During the mid-ebb tide at the Fort
Columbia site, for example, velocities were
-
7/28/2019 Evaluating the NOAA coastal ecology
8/18
3.2. CTD profiles
Vertical salinity, temperature, and dissolved oxygen profiles
largely delineated the range of saltwater intrusion. The Fort
Columbia and Cliff Point sites had salt wedge characteristics
during both ebb and flood tides, with clearly stratified salinity
and temperature profiles (Table 3, Fig. 6). Salinity levels in
surface waters at these sites were mostly 4e6 psu, while salt
wedge values ranged from 16e30 psu. The Grays Point and
Portuguese Point sites were weakly stratified, with higher
salinity during flood tides and maximum salinity about
5 psu. Temperature profiles at these four sites showed patterns
that paralleled the salinity measures. In contrast, the remaining
sites were all vertically well-mixed, with no physiochemical
stratification. These sites were tidally influenced, but
maximum salinity levels did not exceed 1 psu (Table 3,
Fig. 6) indicating local tidal exchange of freshwater.
3.3. Benthic sampling
On average, dredge samples were predominantly sand with
finer silts and muds as the secondary substrate type. The
exception was the Mott Island West site, where samples
were largely silt/mud and organic detritus (Fig. 7). Coarser
substrate (e.g., pebbles and small rocks) and organic detritus
made up small percentages of the averaged samples at most
sites, while plants were collected only at the Mott Island
West (mean 3%) and South Channel (mean 10%) sites.
Across sites, substrate composition was significantly asso-
ciated with water velocity (Fig. 8). Mean percent sand was
positively correlated with mean water velocity during the
mid-ebb tide (P 0.036, r2 0.49, linear regression) and
a corresponding negative relationship was seen for the mean
percentage of fines (P 0.049, r2 0.45). Substitution of
mean mid-flood velocity produced less significant correlations,
probably because velocities during mid-ebb were typically the
highest among the measured tidal stages. Substrate at the Mott
Island West site was the most unique (3% sand, 83% fines),
and excluding this point reduced regression fit slightly
(r2 0.43 for sand and 0.31 for fines). Excluding the Fort Co-
lumbia site, where benthic samples were not collected along
the entire ADCP transect had little effect on regression results.
Variability in benthic sample composition within most sites
was quite high, indicating more fine-scale benthic structuring
than we anticipated. Some within-site variability appeared to
be due to velocity differences between channel and adjacent
Fig. 3. ADCP velocity profiles for thew2080 m Fort Columbia study transect during mid-ebb (top) and mid-flood (bottom) tides in October, 2004. Colors bar units
are in m s1 and range from 0.0 (magenta) to 2.5 (red). ADCP bins are 0.5 m in depth. The thick black line is the estuary bottom and white areas represent censored
data due to velocity ambiguity or interference (near-bottom) or ADCP blanking distance and boat movement (near-surface). Flow direction was downstream during
the ebb tide and upstream during the flood. Fig. is oriented so reader is looking downstream.
96 M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106
-
7/28/2019 Evaluating the NOAA coastal ecology
9/18
depositional habitats. However, testing this empirically would
be difficult using the sampling scheme we adopted for this test
study. A sampling design that retained randomness but
allowed direct comparison of velocity estimates and benthic
composition would probably allow association of some of
the finer-scale variability we observed with within-transect
differences in energy/current regime, such as potential differ-
ences between channels and adjacent flats.
Bivalves, amphipods, fine shell debris, mollusks, snails,
worms and/or worm casings were found in many samples.
However, the presence and abundance of soft sediment benthic
invertebrates was highly variable among samples within and
among transects and we did not observe consistent benthic
assemblages or associations with physical habitat features.
Such associations probably did occur at finer spatial scales
than we sampled.
3.4. CMECS classification testing
We used the above data types to classify the study sites
using CMECS (Table 4). The combined ADCP, CTD, and
benthic sampling data provided classification information for
about half of the nearly 30 CMECS descriptors (Table 2)
with potential relevance to the CRE study sites. Several
descriptors could be applied across all classification levels
we considered, from regime (Level 1) to habitat (Level 5).
For example, all study sites were in the ecological region
called the Columbian Pacific, the degree of enclosure was
partial (50e75% encircled by land), and during the sampling
month the temperature class was temperate (10e20 C) and
the tide range was moderate (1e5 m). Energy type at all sites
included wind, current, and tide. The energy direction
was a mix of horizontal and seaward, substrate composition
was mixed, and large scale relief was flat (height:width
ratio w0). The depth class at most sites included all of the
subtidal classifications for this descriptor: very shallow
(0e5 m), shallow (5e15 m), and deep (>15 m).
All sites were partially-enclosed and freshwater-influenced,
placing them in the Estuarine Regime (CMECS Level 1, code
A). We found classifying the CRE using the CMECS Level 2
(Formation) more difficult. At this range of scales (10,000 m2
to 100 km2) several geoforms and hydroforms appeared
applicable. For example, study sites in the peripheral bays
(e.g., South Channel, Mott Island West, Grays Point) had char-
acteristics of embayments (Level 2, code A02) but also could
have been characterized as subsurface channel (Level 2, code
A08) formations. Subsurface channel also seemed the most
appropriate formation classification for the Fort Columbia
Fig. 4. ADCP velocity profiles for thew730 m Woody Island Channel study transect during late-ebb (top) and late-flood (bottom) tides in October, 2004. Colors
bar units are in m s1 and range from 0.0 (magenta) to 2.5 (red). ADCP bins are 0.5 m in depth. The thick black line is the estuary bottom and white areas represent
censored data due to velocity ambiguity or interference (near-bottom) or ADCP blanking distance and boat movement (near-surface). Flow direction was
downstream during the ebb tide and upstream during the flood. Fig. is oriented so reader is looking downstream.
97M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106
-
7/28/2019 Evaluating the NOAA coastal ecology
10/18
and Tongue Point Channel sites, each of which had deep
channels >1 km wide. Surface channel (Level 2, code
A25) may also have been appropriate for the more constricted
South Channel and Prairie Channel sites. At the higher end of
the formation scale (w100 km2
), however, both surface and
subsurface channels could be viewed as part of a larger
complex of channels and subtidal flats extending for 100s of
km2 throughout the CRE. To us, this suggested a repeating
channelized subtidal flat formation, an option not currently
included in CMECS. A distinction between natural and
Fig. 5. Examples of water velocity (m s1) and direction (right East) during early, middle, and late portions of ebb and flood tides at four Columbia River Estuary
study sites, generated from ADCP transect data. Colors represent velocity in the top 5 m of the water column (ebb pink; flood green), from 5e10 m deep
(ebb red; flood blue), and at >10 m (ebb maroon; flood black). Plotted points represent 10-ensemble averages at 0.5 m depth intervals from the
ADCP profiles. Concentric rings represent 0.2 m s1 intervals, with the outer ring of each plot 1.0 m s1.
98 M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106
-
7/28/2019 Evaluating the NOAA coastal ecology
11/18
dredged channels may also be appropriate, as both types were
widely distributed in the CRE. Finally, given that the CRE is
dominated by large freshwater inputs and net seaward flow,
the current (Level 2, code A13) hydroform may have been
a reasonable formation selection.
Classification of vertical zones (Level 3) and water column
macrohabitats (Level 4) nested within the zones was more
straightforward. Only the Fort Columbia and Cliff Point sites
showed consistent vertical stratification of the water column
zone. At the Fort Columbia site, the upper water column
macrohabitat (above the pyncnocline) was mesohaline, oxic
to oxygen saturated, and with low to moderate tidal and cur-
rent energy (Table 4). The upper water column macrohabitat
at the Cliff Point site was low energy, oligohaline, and oxic.
The lower water column macrohabitatdthe salt wedgedat
both Fort Columbia and Cliff Point was low energy,
polyhaline, and oxic. The remaining nine sites were largely
unstratified, suggesting a single well-mixed water column
macrohabitat. During the sampling period all nine sites had
low energy inputs. Those closer to the estuary mouth were
oligohaline with a mix of river and estuarine water exchange,
while those further upstream were largely tidally-exchangedfreshwater sites. Water column zones and their macrohabitats
were relatively stable during the studied tidal cycles, but these
classifications and their descriptors and classifiers would
Temperature (C)
10 12 14 16 18 20
Depth(m)
0
2
4
6
8
10
12
14
16
18
Salinity (psu)
0 5 10 15 20 25 30 35
0
2
4
6
8
10
12
14
16
18
Dissolved oxygen (mg l-1)
4 6 8 10 12 14 16
0
2
4
6
8
10
12
14
16
18
Fig. 6. Examples of vertical salinity (psu), temperature (C), and dissolved
oxygen (mg L1) profiles for four Columbia River Estuary study sites during
ebb (open symbols) and flood (closed symbols) tides. Circles Cliff Point
site; triangles Mott Island West site; squares McGregor Island site;
diamonds Woody Island Channel site.
Mean water velocity (m s-1)
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Percent(%)
0
20
40
60
80
100
FC
FC
Fig. 8. Relationships between mean water velocity (v) during mid-ebb tide and
mean substrate percentages (C sand; B fines [mud and silt]) from 10
ponar grabs randomly selected near each Columbia River Estuary study tran-
sect. Linear regression results: percent sand 80.8(v) 28.1 (P 0.036,
r2 0.49) and percent fines75.6(v) 60.7 (P 0.049, r2 0.45).
FC Fort Columbia site.
Fig. 7. Mean substrate composition percentages from 10 benthic dredge grabsat each study site. Gray sands; black fines (silts and muds); white peb-
ble; vertical lines rock; hash marks organics; diagonal lines plants.
99M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106
-
7/28/2019 Evaluating the NOAA coastal ecology
12/18
Table 4
Classification grid for the 11 CRE study sites based on the collected physical data. All sites were in the CMECS Estuarine Regime (level 1). The sampling scale
was most appropriate for classifying levels 2 (Formation), 3 (Zone), and 4 (Macrohabitat). Example classifications for level 5 (habitat) are included in the text.
Gray cells indicate habitat conditions occurred at the site. Vertical lines indicate data was not collected due to depth and current conditions (site 6), and dashes
indicate missing or incomplete data
)1.giFotdnopserrocsrebmun(etisydutS
1101987654321noitacifissalCSCEMC
)s(noitamroF ,lennahCecafrusbuS,tnemyabmE: evisulcnocnIstisopeDdna
enoZ nmulocretaW:
tatibahorcaM nmulocretawreppU:
)usp0(hserf:ytinilaS
)usp50>(enilahogilo
)usp815(enilahosem
)C0201(etarepmet:erutarepmeT
gm014(cixo:negyxO l1-
)
gm2101(detarutasnegyxo l 1- )
gm21>(detarutasrepusnegyxo l 1- )
)nk20(wol:)tnerruc(ytisnetniygrenE
)nk42(etaredom
dehsretaw:ecruosretawyramirP
egnahcxeyrautselacol
tatibahorcaM nmulocretawrewoL:
)usp0381(enilahylop:ytinilaS
)C0201(etarepmet:erutarepmeT
gm014(cixo:negyxO -1)l
)nk20(wol:)tnerruc(ytisnetniygrenE
eniram/egnahcxeyrautselacol:ecruosretawyramirP
deifitartsylgnortstonnmulocretaW
enoZ mottoB:
tatibahorcaM evisulcnocnIdetadilosnocnUdna,talfladitbuS,mottobtfoS:
)50(talF:epolS
gnipolS a )035(
c/epytetartsbuS stisopeddexim,detadilosnocnu:noitisopmo |
ezisetartsbuS b )mm70.0
-
7/28/2019 Evaluating the NOAA coastal ecology
13/18
almost certainly change with fluctuations in river discharge ortidal range.
High variability among samples within site and the
limitations related to our sampling design made classifying
bottom zone macrohabitats across all scales problematic. At
the low end of the macrohabitat scale (100 m2), each random
benthic sample at a site may have been adequate to character-
ize a bottom macrohabitat, and several recognizable macroha-
bitat types may have been present at each study site given the
apparent patchiness of fine and coarse sediments within sites.
This was seen at South Channel where a primarily sandy
substrate was mixed with rock at one sample site and vegeta-
tion at another. (Tables 4 and 5, Fig. 7). At the upper end of the
macrohabitat classification level (1000 m2, approximately the
scale of study transects), an average of several or all of
the benthic samples at a site or a classification category reflect-
ing the heterogeneity in substrate types may have been more
appropriate. In contrast, macrohabitat classification was straight-
forward when substrates were relatively homogeneous. This was
especially true at McGregor, where sand alone comprised 97% of
the sample. Clearly, different spatial scales could have pro-
duced different benthic macrohabitat classifications for most
study sites. Importantly, the scale of habitat patchiness (and
sampling) may strongly affect classification outcome and care-
fully selecting appropriate macrohabitat categories (including
mixed categories) and sampling scale will be critical to
accurately capturing habitat features at this biologically im-portant scale. Of the currently defined CMECS bottom macro-
habitats, subtidal flat, softbottom, and mixed coarse
sediments appeared to most typify the study area, though
the currently undefined estuarine unconsolidated sediments
may also be appropriate. A sample classification of three study
sites using CMECS at its current level of development (Table 5)
omits specific macrohabitat types while relying on descriptors
and classifiers to describe macrohabitat characteristics for both
the bottom and the water column. The ambiguity in our clas-
sification as it moved lower in the classification hierarchy
would likely have been reduced had our a priori emphasis
been for an individual species or ecological community and
had we adjusted our sampling intensity to match the distribu-
tion of the target organisms.
As we did not have a focal species or community, specific
habitat classifications (Level 5) were not a goal of the evalua-
tion. However, some general habitat classification examples
from the CRE sites could include pyncnocline, turbidity max-
imum, salt wedge, oligohaline surface water, and oxic bottom
water (water column habitats) as well as bare sandy, organic
mud, mud-shell hash, vegetated softbottom, mixed coarse
sediments, unconsolidated sediments, and sand wave (bottom
habitats). Beyond this, our sampling resolution was insuffi-
cient to recognize the small-scale associations between biota,
substrate, and physiochemical environment.
Table 5
Sample site classifications using the CMECS structure and terminology
Site 1: Fort Columbia (23.4 km2) The Fort Columbia site has two geohydrologic formations and consistent vertical stratification:
Formation 1: A natural (non-dredged) subsurface channel formation with
Zone 1: lower water column (>15 m depth) with
Macrohabitat 1 (hydrologic): a low energy, polyhaline, and oxic environment (salt wedge); and
Zone 2: bottom (benthic) with
Macrohabitat 1 (geomorphic): a flat to sloping (0e
30
) trough bottom, no vegetative cover,and unconsolidated sediments consisting primarily of sand and fines (mud and silt); and
Formation 2: A natural (non-dredged) current formation with
Zone 1: upper water column (6e15 m depth) with
Macrohabitat 1 (hydrologic): a mesohaline, oxic to oxygen saturated environment with low to
moderate tidal and current energy in mixed flow directions; and
Zone 2: bottom (benthic) with
Macrohabitat 1 (geomorphic): flat to sloping (0e30) bottom with unconsolidated sediments
consisting primarily of sands, fines (mud and silt) with some evidence of pebbles and plants from
a narrow rock ledge and a wedge of sand from the corner of Baker Bay
Site 8. McGregor Island (4.08 km2) The McGregor Island site has a single geohydrologic formation:
Formation 1: A natural (non-dredged) current formation; with
Zone 1: non-stratified, water column (0e15 m depth) with
Macrohabitat 1 (hydrologic): a fresh-water environment receiving flow from the watershed
(as opposed to estuarine exchange), with oxic to oxygen saturated conditions and low energy intensity; andZone 2: bottom (benthic) with
Macrohabitat 1 (geomorphic): flat to sloping (0e30) bottom comprised almost
exclusively of sand with occasional pebble inclusions and no evidence of vegetation
Site 9. South Channel (8.04 km2) The South Channel site has a single geohydrologic formation:
Formation 1: A slough formation; with
Zone 1: shallow, non-stratified subtidal (0e5 m depth) with
Macrohabitat 1 (hydrologic): a fresh-water environment receiving flow from the watershed
(as opposed to estuarine exchange), with oxic conditions and low energy intensity; and
Macrohabitat 2 (geomorphic): flat to sloping (0e30) bottom with unconsolidated sediments
including sand, fines (mud and silt), rocks, organics, and patches of full, sparse or no vegetation
101M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106
-
7/28/2019 Evaluating the NOAA coastal ecology
14/18
4. Discussion
A persistent challenge in habitat classification is the tension
between defining habitats by their physical attributes versus
defining them by biological requirements, community
structure or ecological functionality (Diaz et al., 2004; Kurtz
et al., 2006). In practice, most ecological classifications blendabiotic and biotic criteria, though the relative weight given
each category varies widely (Edgar et al., 2000; Colloty
et al., 2002). Depending on the context, each approach has
clear advantages. Abiotic classifications are useful in that
they can be relatively easily quantified, data are often available
at multiple spatial and temporal scales, and abiotic variables
may be used as surrogates for biotic composition and struc-
ture. Classifications that explicitly incorporate biologically
meaningful information, such as habitat use, species relation-
ships or limiting factors, may be more likely to accurately
reflect ecological patterns (Van Dolah et al., 1999; Jay et al.,
2000). The degree to which a classification system can be
applied to address a research or management question dependson the degree of overlap between the question and the
classification components. A primarily abiotic classification,
for example, may poorly predict the distribution of individual
biota or community composition, while a more biologically-
based classification may not be easily scalable or transferable
across ecosystem boundaries. Importantly, the question of
scale may help resolve this tension because it may be that
physiochemical attributes are adequate for classification at
coarser resolution, while biotic associations and interactions
become increasingly important at finer resolution (e.g., Poff,
1997). The CMECS attempts to bridge this gap in part by
implicitly recognizing that processes affecting habitatformation occur at multiple, hierarchical scales.
In this research, we evaluated the CMECS with a set of
abiotic data types without the aim of developing a classification
for a specific ecological or management question. This testing
approach is not unlike real-world scenarios where manage-
ment or resource agencies attempt to classify habitats using
existing or routinely-collected monitoring data. Selecting
one of the CREs specific benthic or epibenthic assemblages
(e.g., Jones et al., 1990), for example, would have reduced
ambiguity surrounding the correct scale for identifying forma-
tions or macrohabitats. A much different spatial scale may
have been appropriate had our focus been on mobile fauna
like migrating salmon, for which larger-scale processes likely
have a greater effect on behavior and distribution. This con-
text-driven ambiguity regarding scale appears to be a CMECS
weakness, because classifications of the same site could
substantially diverge for users with different target species or
communities. However, a hierarchical, multi-scale approach
such as CMECS has the potential to successfully synthesize
classifications from context-specific studies conducted at
different scales. For instance, habitats recognized in surveys
for adult salmonids at larger spatial scales may be integrated
with finer scale surveys of habitats that identify biotic commu-
nities such as oyster beds. Such explicit recognition of scale
dependence among study questions, sampling designs and
underlying ecological pattern and process among surveys is
necessary to achieving an integrated and holistic classification.
The CMECSs use of hierarchical, nested spatial scales,
including the vertical component, is a conceptual strength
of the CMECS framework and largely accommodated the
spatial complexity and ambiguity of habitats within the
CRE (e.g., Johnson et al., 2003). Importantly, the CMECSclassification categories identified for the CRE study sites
aligned well with the habitats defined in the much more spa-
tially extensive studies of the Columbia River Estuary Task
Force. For example, CMECS bottom habitat classifications
were consistent with the types and distributions of sediments
described in Fox et al. (1984) and Sherwood and Creager
(1990). Similarly, salinity intrusion and energy patterns
were similar to those in Jay et al. (1990) and Hamilton
(1990) despite our temporally restricted sampling regime.
However, two types of spatial complexity could perhaps be
improved within the CMECS. First, a mechanism is
necessary to describe hybrid or transitory habitats within
a classification level, which are those where more than oneof the predefined classification options appeared appropriate
at a given scale. Second, as described above, we recognize
that the operational thresholds we used within CMECS to de-
fine the borders between habitat types were somewhat arbi-
trary. While the broad ranges within each classification
level allowed considerable flexibility in defining habitats,
clear correspondence between spatial thresholds and some
measure of ecological functionality was not necessarily
evident. Such ambiguity in drawing distinctions in scale
within and across units may frustrate efforts to map discrete
habitats or to compare habitats at regional or global scales.
One potential analytical approach to examining this issue isto perform sensitivity analyses whereby the threshold values
are varied across a range of plausible values and the affect
on the classification and resulting maps are compared. Such
an analysis was beyond the scope of this study.
In contrast to its relatively explicit handling of spatial
variability, using CMECS to capture temporal variability is
more challenging. A single physical descriptordtemporal
persistencedwas largely qualitative (e.g., permanent, vari-
able, stochastic, low persistence, high persistence; Madden
et al., 2005), which may at times be inadequate in highly
dynamic estuarine systems like the CRE. The historical inter-
disciplinary CRE studies identified predictable and quantifi-
able daily and seasonal events that clearly shape ecological
processes (e.g., Simenstad et al., 1990; Prahl et al., 1998)
and animal behaviors (Bottom and Jones, 1990; Morgan
et al., 1997), but CMECS lacks a suitable framework to
capture and classify this type of predictable variability. One
approach may be to incorporate a hierarchy of appropriate
temporal scales for averaging environmental conditions in
predictable, but varying habitats. For instance, in the CRE it
may be most appropriate to average local salinity and/or water
velocity within and across tidal cycles (hours), within seasons
(days to weeks), or across seasons or years (months to years),
depending on the ecological scale of interest. Overall, we
suspect that long-term averaging is appropriate in many cases,
102 M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106
-
7/28/2019 Evaluating the NOAA coastal ecology
15/18
but not in others where stochastic or episodic events may have
strong organizing effects (e.g., brief periods of anoxia).
Addressing temporal variability in any estuarine classifica-
tion effort is critical because temporally restricted data collec-
tion methods run the risk of misrepresenting aspects of
estuarine ecological composition, structure, and function.
The snapshot approach to data collectiond
point-grab ben-thic samples, haphazard CTD or velocity profiles, or seasonal
biological samplingdis common in habitat surveys and in
existing databases, but classifications based on these data
types are inherently incomplete. The ADCP data we collected,
for example, illustrated predictable patterns of tidal effects and
current velocity gradients, all within a fairly narrow range of
energy input. What we did not capture, however, were shifts
in tidal phase and amplitude or seasonal order-of-magnitude
changes in CRE freshwater input (Jay et al., 1990; Naik and
Jay, 2005), any of which could alter classification conclusions
from the temporally restricted data. Had our sampling effort
been during a high-discharge spring flood event, for example,
velocity, salinity, and temperature profiles could have beensubstantially different and some classification decisions would
need to be reconsidered. Strong spatial-temporal coupling
between physiochemical conditions (e.g., saltwater intrusion
distance, oxygen availability) and some biological events
(e.g., animal aggregations or migrations, primary production
cycles, etc.) suggests that temporal variability should be
more fully integrated into CMECS, perhaps as an additional
category, nested within each hierarchical level. The collection
of time-series data to populate the temporal category could be
focused on the physical and hydrological characteristics of the
classes identified at each level. Adding clear temporal resolu-
tion to a highly structured two-dimensional framework may bedifficult, but should result in much greater flexibility for many
classification scenarios.
Considerable recent attention has also been given to how
physical and biochemical processes structure estuarine
environments (Uncles, 2002; Thain et al., 2004; Deloffre
et al., 2005; Ralston and Stacey, 2005). Some processes,
such as erosion and deposition cycles (Dyers et al., 2000) or
tidal forcing of water column stratification and mixing
(Uncles, 2002), directly affect habitat creation and persistence.
Therefore, understanding mechanisms that shape and trans-
form habitats has clear value for habitat classification. In the
CRE, the combination of river, tidal and wind energy affect
habitats by their influence on large-scale morphology,
bedforms, sediment distribution, and the limits of salinity
intrusion (Giese and Jay, 1989; Jay et al., 1990). On a small
scale, our ADCP data also suggested a CRE benthic habitat
gradient structured by energy input. CMECS most explicitly
addresses this type of hydrodynamic process using the
physical descriptors energy type, energy intensity and energy
direction. Among these, we believe energy intensity should
perhaps be further classified. Although our study sites had sig-
nificantly different benthic habitats, apparently as a function of
energy inputs, all but one were classified as low energy using
the CMECS intensity scale (none, low, moderate, high),
suggesting that this scale was too coarse for differentiating
estuarine habitats that structure in response to subtle but crit-
ical energy differences. It is also possible that low frequency
but high energy inputs sort benthic sediments and that our
sampling period was not suitably scaled to capture these events.
Integrating process-based details into any habitat classifica-
tion is a conceptual challenge because process-based models
tend to be data intensive (e.g., Buzzelli et al., 1999; Monteet al., 2006; Simionato et al., 2004), while the intention of
most classifications is to simplify information and reduce
data inputs (e.g., Roy et al., 2001). Conceptually, CMECS
frequently linked classification to ecological process, referring
to reciprocal relationships between biological and ecological
processes and habitat morphology and structure at various
scales (Madden et al., 2005). However, the framework
currently uses descriptors and/or spatial scale as proxies for
ecological processes, which differs from explicitly allowing
users to incorporate specific processes into classification
decisions. Striking a balance between the rigor of mechanistic
approaches and the convenience of descriptive approaches
may be achievable by critically selecting parameters that arereliable indicators for controlling ecological processes. How-
ever, this selection process must be transparent and explicit,
with clear explanations of how classification parameters
substitute for specific processes. Continued testing and
refinement of CMECS, perhaps including procedures for users
to add and refine parameters, may help identify and develop
more precise process surrogates.
One of our goals in this evaluation was to test how well
existing (and relatively low cost) technology could be used
to help populate the various CMECS tiers. In general, the
combined ADCP, CTD and PONAR data collection types
were useful for mid-level classification units (zones andmacrohabitats), and each has obvious potential for application
at the finer habitat and biotope scales. The ADCP was a partic-
ularly versatile tool, providing physical and geomorphologic
data including basic bathymetry and relief and detailed
measures of energy direction and intensity. In other studies,
repeated ADCP measurements have been used to understand
hydrographic structure over varying time scales (Reed et al.,
2004; Dinehart and Burau, 2005), and such an approach would
likely have improved our test classification of CRE habitats. It
is also possible to use backscatter data from ADCP profiles to
make inferences about the type and volume of suspended
sediments and plankton (e.g., Holdaway et al., 1999; Gartner,
2004), which may be helpful for some classification questions.
The point grab nature of both the CTD and PONAR data
suggested to us that additional sampling intensity with these
instruments would probably be necessary to clearly delineate
boundaries between habitat types, particularly at mid-level
and low-level CMECS tiers. Co-location of ADCP, CTD and
benthic grabs would have provided the ability to better test
for associations between parameters at finer scales. We also
briefly tested a sidescan sonar during the preparation for the
CMECS evaluation, and this instrument provided high
resolution data on sand wave bedforms, rock bars, and other
estuarine bottom morphology. Sidescan sonar has been used
to map a variety of marine habitat types (e.g., Kenny et al.,
103M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106
-
7/28/2019 Evaluating the NOAA coastal ecology
16/18
2003; Allen et al., 2005) and has obvious potential for use as
a habitat classification tool. Finally, the CMECS descriptors we
could not apply in our classification test (e.g., turbidity, photic
quality, trophic status) could have been included by collecting
a handful of additional data types (e.g., Secchi depth, turbidity,
and/or chlorophyll a) that require only basic instrumentation.
Though still a work in progress, the CMECS frameworkshows promise as a classification standard. The approach is
general and therefore applicable to a wide range of coastal
and marine habitat types. The supporting terminology,
combined with the nested hierarchical design, appears to
provide a good conceptual foundation for generating classifi-
cations that address a spectrum of research and management
objectives. As the classification moves towards greater quanti-
fication and more precise definitions, however, it will be
essential for CMECS to maintain an open and adaptive stance.
To provide ecologically relevant information and value to
a wide range of potential end users, the classification must
be flexible enough to accommodate an array of data types at
multiple spatial and temporal scales. Classification inputs forlocalized estuary studies or species management (e.g., Harris
et al., 2001; Johnson et al., 2003) will differ greatly from those
for continental-scale studies of estuarine fish habitats (e.g.,
Harrison and Whitfield, 2006), regional studies of intertidal
diversity (e.g., Zacharias and Roff, 2001), or broad efforts to
classify and protect marine biodiversity (e.g., Ward et al.,
1999; Zacharias and Roff, 2000; Halpern, 2003). Such diverse
objectives will be accompanied by specialized, discipline-
oriented data and vocabulary. To be widely adopted, CMECS
must accommodate these existing knowledge systems while
simultaneously encouraging a disciplined, standardized
classification framework and lexicon.
Acknowledgements
The authors would like to thank P. Wilbur (NOAA National
Marine Fisheries Service) for vision, direction and funding;
K. Nielsen (NOAA National Ocean Service) for excellent da-
tabase design and GIS support; C. Simenstad (University of
Washington) for guidance in site selection; A. Baptista and
M. Wilkin (Oregon Health & Science University) for project
guidance and technical assistance; and the Fish Ecology
Research Laboratory, University of Idaho, who supported
this project from conception to completion.
References
Allen, Y.C., Wilson, C.A., Roberts, D.H., Supan, J., 2005. High resolution
mapping and classification of oyster habitats in nearshore Louisiana using
sidescan sonar. Estuaries 28, 435e446.
Amspoker, M.C., McIntire, C.D., 1986. Effects of Sedimentary Processes and
Salinity on the Diatom Flora of the Columbia River Estuary. Botanica
Marina 29, 391e399.
Attrill, M.J., Rundle, S.D., 2002. Ecotone or ecocline: ecological boundaries in
estuaries. Estuarine, Coastal and Shelf Science 55, 929e936.
Augerot, X., 2005. Atlas of Pacific Salmon: The First Map-Based Assessment
of Salmon in the North Pacific. University of California Press, Berkeley,
California.
Auster, P.J., Joy, K., Valentine, P.C., 2001. Fish species and community
distributions as proxies for seafloor habitat distributions: The Stellwagen
Bank National Marine Sanctuary example (Northwest Atlantic, Gulf of
Maine). Environmental Biology of Fishes 60 (4), 331e346.
Baptista, A.M., Zhang, Y.L., Chawla, A., Zulauf, M., Seaton, C., Myers, E.P.,
Kindle, J., Wilkin, M., Burla, M., Turner, P.J., 2005. A cross-scale model
for 3D baroclinic circulation in estuary-plume-shelf systems: II. Applica-
tion to the Columbia River. Continental Shelf Research 25, 935e972.
Bottom, D.L., Jones, K.K., 1990. Species Composition, Distribution, and
Invertebrate Prey of Fish Assemblages in the Columbia River Estuary.
Progress in Oceanography 25, 243e270.
Bottom, D.L., Simenstad, C.A., Burke, J., Baptista, A.M., Jay, D.A.,
Jones, K.K., Casillas, E., Schiewe, M.H., 2005. Salmon at rivers end:
the role of the estuary in the decline and recovery of Columbia River
salmon. U.S. Department of Commerce, p. 246.
Boyd, R., Dalrymple, R., Zaitlin, B.A., 1992. Classification of clastic coastal
depositional environments. Sedimentary Geology 80, 139e150.
Bricker, S.B., Ferreira, J.G., Simas, T., 2003. An integrated methodology for
assessment of estuarine trophic status. Ecological Modelling 169, 39e60.
Buzzelli, C.P., Wetzel, R.L., Meyers, M.B., 1999. A linked physical and
biological framework to assess biogeochemical dynamics in a shallow
estuarine ecosystem. Estuarine, Coastal and Shelf Science 49, 829e851.
Chapman, D., 1986. Salmon and steelhead abundance in the Columbia River
in the nineteenth century. Transactions of the American Fisheries Society
115, 662e670.
Collis, K., Roby, D.D., Craig, D.P., Adamany, S., Adkins, J.Y., Lyons, D.E.,
2002. Colony size and diet composition of piscivorous waterbirds on the
lower ColumbiaRiver:Implicationsfor losses of juvenilesalmonids to avian
predation. Transactions of the American Fisheries Society 131, 537e550.
Colloty, B.M., Adams, J.B., Bate, G.C., 2002. Classification of estuaries in the
Ciskei and Transkei regions based on physical and botanical characteris-
tics. South African Journal of Botany 68, 312e321.
Cowardin, L.M., Carter, V., Golet, F.C., LaRoe, E.T., 1979. Classification of
Wetlands and Deepwater Habitats of the United States. Fish and Wildlife
Service. U.S. Department of the Interior. FWS/OBS-79/31.
Deloffre, J., Lafite, R., Lesueur, P., Lesourd, S., Verney, R., Guezennec, L.,
2005. Sedimentary processes on an intertidal mudflat in the upper
macrotidal Seine estuary, France. Estuarine, Coastal and Shelf Science64, 710e720.
Diaz, R.J., Solan, M., Valente, R.M., 2004. A review of approaches for
classifying benthic habitats and evaluating habitat quality. Journal of
Environmental Management 73, 165e181.
Dinehart, R.L., Burau, J.R., 2005. Repeated surveys by acoustic Doppler
current profiler for flow and sediment dynamics in a tidal river. Journal
of Hydrology 314, 1e21.
Doyle, M.J., Morse, W.W., Kendal, A.W., 1993. A comparison of larval fish
assemblages in the temperate zone of the northeast pacific and northwest
Atlantic oceans. Bulletin of Marine Science 53 (2), 588e644.
Dyer, K.R., 1997. Estuaries: A Physical Introduction, second ed. John Wiley,
London.
Edgar, G.J., Barrett, N.S., Graddon, D.J., Last, P.R., 2000. The conservation
significance of estuaries: a classification of Tasmanian estuaries using
ecological, physical and demographic attributes as a case study. BiologicalConservation 92, 383e397.
Elliott, M., McLusky, D.S., 2002. The need for definitions in understanding
estuaries. Estuarine, Coastal and Shelf Science 55, 815e827.
Emmett, R.L., Durkin, J.T., 1985. The Columbia River Estuary: an important
nursery for Dungeness crabs, Cancer magister. Marine Fisheries Review
47, 21e25.
Fox, D.S., Bell, S., Nehlsen, W., Damron, J., 1984. The Columbia River
Estuary: atlas of physical and biological characteristics. Columbia River
Estuary Data Development Program, Seattle.
Frey, B.E., Lara, R.L., Small, L.F., 1981. Spatial and temporal patterns of
phytoplankton primary productivity in the Columbia River Estuary.
Estuaries 4, 275e275.
Gartner, J.W., 2004. Estimating suspended solids concentrations from
backscatter intensity measured by acoustic Doppler current profiler in
San Francisco Bay, California. Marine Geology 211, 169e
187.
104 M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106
-
7/28/2019 Evaluating the NOAA coastal ecology
17/18
Giese, B.S., Jay, D.A., 1989. Modeling tidal energetics of the Columbia River
Estuary. Estuarine, Coastal and Shelf Science 29, 549e571.
Gordon, L., 1996. Acoustic Doppler Current ProfilerdPrinciples of Operation:
A Practical Primer, second ed. RD Instruments, San Diego.
Griffith, D.W., 2007. Behavior of Adult Fall Chinook Salmon in the Lower
Columbia River Estuary. M.S. Thesis, University of Idaho, Moscow, Idaho.
Haertel, L., Osterberg, C., 1967. Ecology of zooplankton, benthos and fishes in
Columbia River Estuary. Ecology 48, 459e472.
Haertel, L., Osterberg, C., Curl, H., Park, P.K., 1969. Nutrient and plankton
ecology of the Columbia River Estuary. Ecology 50, 962e978.
Halpern, B.S., 2003. The impact of marine reserves: do reserves work and does
reserve size matter? Ecological Applications 13, 117e137.
Hamilton, P., 1990. Modeling salinity and circulation for the Columbia River
Estuary. Progress in Oceanography 25, 113e156.
Hansen, D.V., Rattray, M., 1966. New Dimensions in Estuary Classification.
Limnology and Oceanography 11, 319e326.
Harris, S.A., Cyrus, D.P., Beckley, L.E., 2001. Horizontal trends in larval fish
diversity and abundance along an ocean-estuarine gradient on the northern
KwaZulu-Natal coast, South Africa. Estuarine, Coastal and Shelf Science
53, 221e235.
Harrison, T.D., Whitfield, A.K., 2006. Temperature and salinity as primary
determinants influencing the biogeography of fishes in South African
estuaries. Estuarine, Coastal and Shelf Science 66, 335e345.
Holdaway, G.P., Thorne, P.D., Flatt, D., Jones, S.E., Prandle, D., 1999. Com-
parison between ADCP and transmissometer measurements of suspended
sediment concentration. Continental Shelf Research 19, 421e441.
Hughes, F.W., Rattray, M., 1980. Salt flux and mixing in the Columbia River
Estuary. Estuarine and Coastal Marine Science 10, 479e493.
Iba~nez, C., Pont, D., Prat, N., 1997. Characterization of the Ebre and Rhone
estuaries: A basis for defining and classifying salt-wedge estuaries.
Limnology and Oceanography 42, 89e101.
Jay, D.A., Smith, J.D., 1990. Circulation, density distribution and neap-spring
transitions in the Columbia River Estuary. Progress in Oceanography 25,
81e112.
Jay, D.A., Giese, B.S., Sherwood, C.R., 1990. Energetics and sedimentary pro-
cessesin theColumbiaRiver Estuary. Progress in Oceanography 25,157e174.
Jay, D.A., Geyer, W.R., Montgomery, D., 2000. An ecological perspective on
estuarine classification. In: Estuarine Science: a Synthetic Approach toResearch and Practice. Island Press, Washington, DC, pp. 149e176.
Johnson, G.E., Thom, R.M., Whiting, A.H., Sutherland, G.B., Berquam, T.,
Ebberts, B.D., Ricci, N.M., Southard, J.A., Wilcox, J.D., 2003. An
Ecosystem-Based Approach to Habitat Restoration Projects with Emphasis
on Salmonids in the Columbia River Estuary. Pacific Northwest National
Laboratory for US Department of Energy, p. 82.
Jones, K.K., Simenstad, C.A., Higley, D.L., Bottom, D.L., 1990. Community
structure,distribution, and standingstock of benthos,epibenthos, and plank-
ton in the Columbia River Estuary. Progress in Oceanography 25, 211e241.
Kenny, A.J., Cato, I., Desprez, M., Fader, G., Schuttenhelm, R.T.E., Side, J.,
2003. An overview of seabed-mapping technologies in the context of
marine habitat classification. Ices Journal of Marine Science 60, 411 e418.
Kurtz, J.C., Detenbeck, N.D., Engle, V.D., Ho, K., Smith, L.M., Jordan, S.J.,
Campbell, D., 2006. Classifying coastal waters: current necessity and
historical perspective. Estuaries and Coasts 29, 107e
123.Ledgerwood, R.D., Thrower, F.P., Dawley, E.M., 1991. Diel sampling of
migratory juvenile salmonids in the Columbia River Estuary. Fishery
Bulletin 89, 69e78.
Madden, C.J., Grossman, D.H., Goodin, K.L., 2005. Coastal and marine
systems of North America: framework for an ecological classification
standard: Version II. NatureServe. Virginia, Arlington, 48 pp.
McCabe, G.T., Muir, W.D., Emmett, R.L., Durkin, J.T., 1983. Interrelation-
ships between Juvenile Salmonids and Nonsalmonid Fish in the Columbia
River Estuary. Fishery Bulletin 81, 815e826.
McClure, M.M., Holmes, E.E., Sanderson, B.L., Jordan, C.E., 2003. A large-
scale, multispecies status, assessment: anadromous salmonids in the
Columbia River Basin. Ecological Applications 13, 964e989.
Misitano, D.A., 1977. Species composition and relative abundance of larval
and post-larval fishes in Columbia River Estuary, 1973. Fishery Bulletin
75, 218e
222.
Monte, L., Hakanson, L., Perianez, R., Laptev, G., Zheleznyak, M.,
Maderich, V., Angeli, G., Koshebutsky, V., 2006. Experiences from
a case study of multi-model application to assess the behaviour of pollut-
ants in the Dnieper-Bug Estuary. Ecological Modelling 195, 247e263.
Morgan, C.A., Cordell, J.R., Simenstad, C.A., 1997. Sink or swim? Copepod
population maintenance in the Columbia River estuarine turbidity-maxima
region. Marine Biology 129, 309e317.
Naik, P.K., Jay, D.A., 2005. Estimation of Columbia River virgin flow: 1879 to
1928. Hydrological Processes 19, 1807e1824.
National Marine Fisheries Service, 2000. Endangered Species ActdSection 7
consultation. Biological Opinion. Reinitiation of consultation on operation
of the Federal Columbia River Power System, including the juvenile fish
transportation program, and 19 Bureau of Reclamation projects in the
Columbia Basin. National Marine Fisheries Service, Seattle, Washington.
National Research Council, 1996. Upstream: Salmon and Society in the Pacific
Northwest. National Academy Press, Washington, DC.
Neira, F.J., Potter, I.C., Bradley, J.S., 1992. Seasonal and spatial changes in the
larval fish fauna within a large temperate Australian estuary. Marine
Biology 112, 1e16.
Poff, N.L., 1997. Landscape filters and species traits: towards mechanistic
understanding and prediction in stream ecology. Journal of the North
American Benthological Society 16, 391e409.
Prahl, F.G., Small, L.F., Sullivan, B.A., Cordell, J., Simenstad, C.A.,
Crump, B.C., Baross, J.A., 1998. Biogeochemical gradients in the lower
Columbia River. Hydrobiologia 361, 37e52.
Ralston, D.K., Stacey, M.T., 2005. Stratification and turbulence in subtidal
channels through intertidal mudflats. Journal of Geophysical Research-
Oceans 110.
Reed, R.E., Glasgow, H.B., Burkholder, J.M., Brownie, C., 2004. Seasonal
physical-chemical structure and acoustic Doppler current profiler flow
patterns over multiple years in a shallow, stratified estuary, with
implications for lateral variability. Estuarine, Coastal and Shelf Science
60, 549e566.
Roff, J.C., Taylor, M.E., Laughren, J., 2003. Geophysical approaches to the
classification, delineation and monitoring of marine habitats and their
communities. Aquatic Conservation: Marine and Freshwater Ecosystems
13, 77e90.
Roy, P.S., Williams, R.J., Jones, A.R., Yassani, I., Gibbs, P.J., Coates, B.,West, R.J., Scanes, P.R., Hudson, J.P., Nichol, S., 2001. Structure and
function of south-east Australian estuaries. Estuarine, Coastal and Shelf
Science 53, 351e384.
Ruckelshaus, M.H., Levin, J., Johnson, J.B., Kareiva, P.M., 2002. The Pacific
salmon wars: What science brings to the challenge of recovering species.
Annual Review of Ecology and Systematics 33, 665e706.
Schreck, C.B., Stahl, T.B., Davis, L.E., Roby, D.D., Clemens, B.J., 2006.
Mortality estimates of juvenile spring-summer Chinook salmon in the
lower Columbia River and estuary, 1992-1998: evidence for delayed
mortality? Transactions of the American Fisheries Society 135, 457e475.
Sherwood, C.R., Creager, J.S., 1990. Sedimentary geology of the Columbia
River Estuary. Progress in Oceanography 25, 15e79.
Simenstad, C., Houghton, J., Kinney, W., Cordell, J., Williams, G., 1981.
Structure and role of epibenthic invertebrate assemblages in the Columbia
River Estuary: initial conclusions. Estuaries 4, 275e
276.Simenstad, C.A., Small, L.F., McIntire, C.D., Jay, D.A., Sherwood, C., 1990.
Columbia River Estuary studies: an introduction to the estuary, a brief
history, and prior studies. Progress in Oceanography 25, 1e13.
Simionato, C.G., Dragani, W., Meccia, V., Nu~nez, M., 2004. A numerical
study of the barotropic circulation of the Rio de la Plata estuary: sensitivity
to bathymetry, the Earths rotation and low frequency wind variability.
Estuarine, Coastal and Shelf Science 61, 261e273.
Simpson, M.R., 2001. Discharge Measurements using a Broad-Band Acoustic
Doppler Current Profiler. US Geological Society, Sacramento, California,
Open-file report 01-1, 123 pp.
Sims, C.W., Johnsen, R.H., 1974. Variable-mesh beach seine for sampling
juvenile salmon in the Columbia River Estuary. Marine Fisheries Review
36, 23e26.
Thain, R.H., Priestley, A.D., Davidson, M.A., 2004. The formation of a tidal
intrusion front at the mouth of a macrotidal, partially mixed estuary: a field
105M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106
-
7/28/2019 Evaluating the NOAA coastal ecology
18/18
study of the Dart estuary, UK. Estuarine, Coastal and Shelf Science 61,
161e172.
Uncles, R.J., 2002. Estuarine physical processes research: Some recent studies
and progress. Estuarine, Coastal and Shelf Science 55, 829e856.
Van Dolah, R.F., Hyland, J.L., Holland, A.F., Rosen, J.S., Snoots, T.R., 1999.
A benthic index of biological integrity for assessing habitat quality in estuar-
ies of the southeastern USA. Marine Environmental Research 48, 269e283.
Ward, T.J., Vanderklift, M.A., Nichols, A.O., Kenchington, R.A., 1999. Select-
ing marine reserves using habitats and species assemblages as surrogates
for biological diversity. Ecological Applications 9, 691e698.
Wewetzer, S.F.K., Duck, R.W., Anderson, J.M., 1999. Acoustic Doppler
current profiler measurements in coastal and estuarine environments:
examples from the Tay Estuary, Scotland. Geomorphology 29, 21e30.
Whitfield, A.K., 1999. Ichthyofaunal assemblages in estuaries: A South
African case study. Reviews in Fish Biology and Fisheries 9,
151e186.
Zacharias, M.A., Roff, J.C., 2000. A heirarchical ecological approach to
conserving marine biodiversity. Conservation Biology 14, 1327e1334.
Zacharias, M.A., Roff, J.C., 2001. Explanations of patterns of intertidal
diversity at regional scales. Journal of Biogeography 28, 471 e483.
106 M.L. Keefer et al. / Estuarine, Coastal and Shelf Science 78 (2008) 89e106