1 University of Washington School of Aquatic and Fishery Sciences 2 PC Trask and Associates Landscape Planning Framework Fish Habitat Catena Geodatabase Methodology Mary Ramirez 1 Charles Simenstad 1 Phil Trask 2 Allan Whiting 2 Alex McManus 2 Funding provided by the Bonneville Power Administration
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1 University of Washington School of Aquatic and Fishery Sciences
2 PC Trask and Associates
Landscape Planning Framework
Fish Habitat Catena Geodatabase Methodology
Mary Ramirez 1
Charles Simenstad1
Phil Trask2
Allan Whiting2
Alex McManus2
Funding provided by the Bonneville Power Administration
Data Availability ....................................................................................................................................... 8
Data Development ........................................................................................................................................ 8
Direct Habitat ............................................................................................................................................ 8
Fish Habitat Catena ............................................................................................................................... 8
Head of Tide ....................................................................................................................................... 19
Isolated Lake ....................................................................................................................................... 20
Landscape Unit ................................................................................................................................... 21
Analysis and Application ............................................................................................................................ 22
Reach and Landscape Unit Statistics .................................................................................................. 22
Site and Landscape Unit Statistics ...................................................................................................... 26
User Manual Case Study ......................................................................................................................... 28
How To: Planning Case Study- Brix Bay | Deep River Confluence Restoration ............................... 28
Quantifying the Site and Landscape ................................................................................................... 30
Site Comparison .................................................................................................................................. 33
The length of FHC with contiguous wetland. Wetlands provide a number of
services to adjacent aquatic features (e.g. prey resource input, temperature
regulation, temper and filter floodplain drainage); knowing the proportion
of length with wetland coverage can indicate information about the quality
of the FHC feature.
Adjacent Wetland Class The composition of wetland adjacent to the FHC, which may provide
information about water temperature regulation or prey resource input.
24
Preliminary results from the LPF provide some insight into the frameworkβs utility for conservation and
restoration planning in estuarine settings. For example, LPF allows the user to compare the relative gain
in the opportunity and capacity of direct fish habitat and confluence density that would accrue with
restoring tidal-fluvial flooding to existing altered FHC features among the eight reaches (Figure 12).
Initial analyses indicate that proportional increases in direct FHC would be greatest in the mid- to upper
reaches E, F, and G. Similarly, proportional increases in confluence density would be greatest in reach A,
as well as E through G. Surveys to sample and identify the genetic stock composition of juvenile Chinook
salmon in the estuary found stock diversity was greatest in reaches A and E through G (Teel et al. 2014).
These results imply the need for multiple conservation strategies that would provide different benefits to
different stocks.
Analyses of fish habitat among landscape units are highly variable and demonstrate the complexity and
patchiness of accessible ecosystems as juvenile salmon move through the estuarine gradient (Figure 13).
There are a number of landscapes between reaches D and F that have a high proportion of altered habitat
(seen in Figure 13 as a high percent change in FHC area and confluence density with full restoration).
This would suggest that this stretch of the estuary may represent a deficiency, or gap, in sufficient habitat
for fish as they migrate downriver.
Figure 12. (A) Total area in acres of open FHC (blue) and altered FHC (yellow) by reach. Percent change (dashed line) in
FHC by reach that would accrue if all altered habitat were restored to natural tidal-fluvial flooding is shown on the
second axis. (B) Count of all open confluences (blue) and altered confluences (yellow) by reach. Percent change (dashed
line) in confluence density by reach that would accrue if all altered confluences were restored to natural tidal-fluvial
flooding is shown on the second axis.
0
5
10
15
20
25
30
35
40
45
50
0
2000
4000
6000
8000
10000
12000
14000
A B C D E F G H
FH
C %
Change
FH
C A
rea (
acre
s)
A
0
10
20
30
40
50
60
70
0
400
800
1200
1600
2000
A B C D E F G HC
onfluence D
ensity %
Change
Confluence C
ount
B
25
Figure 13. (A) Total area in acres of open FHC (blue) and altered FHC (yellow) by landscape unit. Percent change
(dashed line) in FHC by landscape unit that would accrue if all altered habitat were restored to natural tidal-fluvial
flooding is shown on the second axis. (B) Count of all open confluences (blue) and altered confluences (yellow) by
landscape unit. Percent change (dashed line) in confluence density by landscape unit that would accrue if all altered
confluences were restored to natural tidal-fluvial flooding is shown on the second axis.
0
20
40
60
80
100
0
1000
2000
3000
4000
5000
6000
Ba
ke
r B
ay
Ba
ke
r B
ay s
ha
llow
s
Pa
cific
Du
ne
s
Tre
stle B
ay s
ha
llow
Yo
un
gs B
ay
Yo
un
gs B
ay s
hallo
w
Ca
thla
me
t B
ay
Gra
ys B
ay
Gra
ys B
ay S
hallo
ws
Julia
Bu
tle
r H
anse
n
Cla
tska
nie
Fla
ts
Pu
ge
t Is
land
Rin
ea
rso
n
Co
wlit
z R
ive
r
Tro
jan
De
er
Isla
nd
Lew
is R
ive
r
Mu
ltno
mah
Ch
an
ne
l
Rid
ge
fie
ld
Sa
uvie
Isla
nd
Scap
poo
se B
ay
Will
am
ette
Riv
er
Po
rtla
nd
Va
nco
uve
r
Wa
sh
ou
ga
l R
ive
r
Bo
nn
evill
e
Go
rge W
ate
rfa
lls
Mir
ror
La
ke
A B C D E F G H
FH
C %
Change
FH
C A
rea (
acre
s)
A
1450
1500
1550B
0
20
40
60
80
100
0
50
100
150
200
250
300
Ba
ke
r B
ay
Ba
ke
r B
ay s
ha
llow
s
Pa
cific
Du
ne
s
Tre
stle B
ay s
ha
llow
Yo
un
gs B
ay
Yo
un
gs B
ay s
hallo
w
Cath
lam
et
Bay
Gra
ys B
ay
Gra
ys B
ay S
hallo
ws
Julia
Bu
tle
r H
anse
n
Cla
tska
nie
Fla
ts
Pu
ge
t Is
land
Rin
ea
rso
n
Cow
litz R
ive
r
Tro
jan
Dee
r Is
lan
d
Lew
is R
ive
r
Mu
ltno
mah
Ch
an
ne
l
Rid
ge
fie
ld
Sa
uvie
Isla
nd
Scap
poo
se B
ay
Will
am
ette
Riv
er
Po
rtla
nd
Va
nco
uve
r
Wa
sh
ou
ga
l R
ive
r
Bo
nn
evill
e
Go
rge W
ate
rfa
lls
Mir
ror
La
ke
A B C D E F G H
Confluence D
ensity %
Change
Confluence C
ount
26
Site and Landscape Unit Statistics Through the concept of landscape allometry and its application in restoration ecology, Hood describes the
correlation of landscape form and ecological processes (Hood 2002, 2007a, 2007b, 2014, 2015). Working
in Puget Sound deltas and the lower Columbia River estuary, Hood has documented patterns between
marsh surface area and various metrics of the tidal channels that drain the marshes. By accounting for
marsh size, relationships generated from a large number of active reference marshes can be used as a
standard for comparison of restoration sites, improving upon restoration design and monitoring (Hood
2007b, 2014, 2015). For example, when designing dike breaching in tidal marsh restoration, managers
can cite the number of channel outlets in reference marshes throughout the landscape to determine how
many breaches should be made at the restoration site (Hood 2015).
Following these principles, the Landscape Planning Framework was used to examine the scaling
relationship of tidal channel surface area and channel outlet count with total wetland surface area. In the
following example, individual surge plain (active and isolated) wetlands were identified in the Grays Bay
Landscape using the Ecosystem Complex designation from CREEC and dike locations (Figure 14).
Restored surge plain wetlands were identified where dike breaching has allowed reconnection of tidal
channels with the tributary channel; however, dikes were not fully removed. These site were historically
wetland before being leveed and converted to agriculture. The proposed Brix Bay β Deep River
Confluence restoration site was also identified for comparison with active reference wetlands (see the
following section for more information on this project). Within each wetland, the number of channel
outlets (tidal channel confluences) was counted and tidal channel surface area was summed. The wetland
surface area was also calculated, excluding the channel area. Wetland area was plotted against the
dependent metrics (channel area and channel outlet count) for all reference wetlands. All variables were
log transformed for regression analysis to fit power functions (Hood 2014). The slope of the log-linear
regression trendline is equal to the exponent of the power function and describes how the dependent
metric changes in relation to wetland area (Hood 2014). Restoration sites were then plotted to examine
deviation from the reference wetland regression relationship.
Tidal channel area and wetland area in reference surge plain habitats of the Grays Bay Landscape was
highly correlated (Figure 15A). The data indicate channel area increased at a slightly more rapid rate than
wetland area (scaling exponent equals 1.27). The channel area to wetland area relationship in restored
wetlands and the Brix Bay β Deep River restoration site was nearly identical to reference wetlands. This
suggests an appropriate amount of total channel habitat in restored wetlands compared to reference
habitats.
The number of channel outlets also scaled with wetland area in reference surge plain habitats, though
outlets increased more slowly than wetland area (scaling exponent equals 0.37; Figure 15B). A previous
study that looked at the relationship between channel outlet count and marsh area in surge plain islands of
the Columbia River Estuary found much higher densities of channel outlets than those reported here
(Hood 2015). This difference emphasizes the heterogeneous distribution of fish habitat and the
importance of examining relationships within the context of the surrounding landscape. In the Grays Bay
landscape, the number of channel outlets in restored wetlands and at the restoration site was consistently
lower than surrounding reference wetlands, with all data points falling below the reference trendline. This
agrees with results from Hoodβs (2015) study where completed and proposed tidal marsh restoration
projects had on average 5 times fewer channel outlets than reference marshes.
In addition, the average channel size per outlet was significantly greater in restored wetlands than in
reference wetlands in the Grays Bay Landscape (p<0.001). Average channel area in restored wetlands
(including the restoration site) was 2.17 acres, compared to an average channel area of 0.42 in reference
wetlands. Such discrepancies in the size of channels and the number of access points may have
consequences in the restored habitatβs ability to effectively support rearing juvenile salmon.
27
Figure 14. Map of surge plain wetlands in the Grays Bay Landscape. Active surge plain is distinguished from isolated
surge plain, as well as wetlands with restored tidal channels.
Active Wetland Restored Wetland Restoration Site
Figure 15. Scaling of tidal channel (FHC) area (A) and channel outlet count (B) with wetland size in the Grays Bay
Landscape. The trendline and equation shows the power function of active wetland data points.
y = 0.0087x1.2735 RΒ² = 0.9486
0.01
0.1
1
10
100
1 10 100 1000
FH
C A
rea (
acre
s)
Wetland Area (acres)
A y = 1.1099x0.3682 RΒ² = 0.765
1
10
1 10 100 1000
Channel O
utlet
Count
Wetland Area (acres)
B
28
User Manual Case Study
The Landscape Planning Framework allows users to evaluate the effects of restoration to juvenile salmon
habitat. Once the existing features, or proposed in the case of restoration planning, have been
characterized, their spatial attributes can be quantified and compared to a reference site, other restored
sites, other restoration scenarios, or pre-restoration conditions. Discrete project areas and their proposed
features can also be assessed for their contribution of open FHC to larger landscapes. This allows the LPF
user to quantify the change a project provides to broader landscapes (in terms of open habitat versus
potential habitat).
The following restoration planning case study will illustrate how to quantify the landscape, calculate LPF
metrics, and interpret those metrics to tell a compelling story about the effects of restoration to juvenile
salmon habitat. The following case study is just an example, and does not calculate every single LPF
metric. However, a new user should be able to replicate the processes outlined below and establish a
foundation for using the LPF.
To perform the LPF restoration evaluation, the following software is needed:
a spreadsheet program (Microsoft Excel, Apache OpenOffice Spreadsheet, Google Sheets) for
organizing your landscape values, LPF metrics, and change percentages;
a geographic information system (GIS) (ArcGIS, QuantumGIS) for displaying, selecting (using
feature attributes and location), geoprocessing FHC features, and quantifying landscape values.
All screenshots and directions are written with the use of Microsoft Excel and ESRI ArcMap for
Windows desktop. Users performing the LPF restoration evaluation with different software should be able
to follow along, but may need to alter some steps slightly to fit within the constructs of different software
packages.
How To: Planning Case Study- Brix Bay | Deep River Confluence Restoration The Brix Bay β Deep River Confluence site is located in a transition zone for migrating juvenile
salmonids in freshwater tidal rearing habitats before transitioning to the broader Columbia River estuary
(Figure 16). The project site, directly adjacent to Deep River, Brix Bay, and Grays Bay, historically
provided important rearing habitat within a broader freshwater tidal swamp complex. The project is also
very close to the North Channel of the Columbia River estuary. North Channel is a semi-diffused
distributary channel off the mainstem that begins upriver from Rice Island and meanders closely to Gray
Bay area. Fish tagging studies completed by Pacific Northwest National Laboratory (PNNL) in 2010
show a high proportion (87%) of subyearling Chinook migrating across shallows surrounding North
Channel (McMichael et al. 2011).
The 175-acre project site was historically connected to the Deep River by three large tidal channel
systems, providing access to a complex network of tidal meanders and a diverse mosaic of Sitka spruce
surge plain wetlands. Today, the site is constrained by a road levee with three tidegates at the historical
tidal channel confluences that control minimal juvenile salmonid ingress/egress into the site. The project
goal map (Figure 17) characterizes primary restoration actions planned for the Brix Bay β Deep River
Confluence site. The goal of the project is to re-establish tidal hydrology by removing the tidegates and
replacing them with bridge structures that will allow full tidal volume exchange, reshaping and restoring
diverse and complex estuarine habitat over time.
29
Figure 16. Map of the Brix Bay - Deep River Confluence restoration site.
Figure 17. Map of the Deep River Confluence primary restoration actions.
30
Quantifying the Site and Landscape The first step in using the LPF is choosing the scales of analysis. For the Brix Bay β Deep River
Confluence restoration project, the Grays Bay Landscape and Deep River Sub-landscape were chosen to
highlight contributions of the project in the context of these larger landscapes (see Figure 16). This
approach provides a nested landscape consideration to understand the contributions of the project to the
Deep River system and the Grays Bay Landscape as a whole. These local landscapes all reside within
Reach B, a reach situated near the freshwater tidal β oligohaline transition within the estuary. Habitat
transitions in the estuary near the ocean are considered relevant to rearing and migrating juvenile
salmonids, as this transition involves dramatic shifts in prey and predators (Simenstad, Cordell 2000).
At the site scale, the Brix Bay β Deep River Confluence project site is compared to an upstream wetland
that was breached 12 years ago on the Deep River. The site was historically wetland before being leveed
and converted to agriculture. The site underwent restoration in 2004 and has had twelve years of tidal and
fluvial inundation and propagation of native wetland vegetation, primarily characterized as tidal
coniferous forest. As the restored site has responded to the reintroduction of increased hydrologic
volumes, coniferous forest die-off and shrub scrub propagation has been observed. It could be expected
that the project site will evolve on a similar trajectory. The project site is also compared to two
undeveloped surge plain sites of approximately the same size located within the Grays Bay Landscape.
Reference Site A is located along Grays River and is primarily characterized as a mixed coniferous-
deciduous forested tidal wetland; Reference Site B is a mixed tidal and non-tidal coniferous forested
wetland located along Crooked Creek. These sites serve as references for wetland conditions where
development has not impeded the geomorphic structure of the habitat and are suggestive of an endpoint
target for the project site as hydrologic processes are restored.
Once the scales of analysis are selected, the user must define which LPF metrics to calculate (see Table 9
in the Data Summary section above). In this example, Confluence Density (CD), Direct FHC Percent
Landscape (PLAND), Direct FHC Edge Density (ED), and SHAPE Index were chosen. Confluence
Density is a measure of the number of confluence points divided by the analysis area (acres), multiplied
by 100 (to convert to confluence count per 100 acres). Confluences are important opportunities for
juvenile salmonids as entry points into discrete habitat patches. Percent Landscape is a measure of FHC
area divided by landscape area, multiplied by 100 (to convert to a percentage). Understanding the percent
of the landscape that is made up of direct FHC can inform the user of the relative amount of habitat within
a defined landscape that is regularly available to juvenile salmonids. Edge Density is a measure of the
length of FHC perimeter divided by the analysis area, and provides information about the complexity and
foraging interface of the FHC relative to the total landscape area. SHAPE Index is calculated at the site
scale for the largest tidal channel in each of the selected wetlands and equals the channel perimeter
divided by the square root of the channel area, adjusted by a constant to adjust for a square standard
(McGarigal et al. 2012). The index equals 1 when the feature is square and increases as the shape
becomes more irregular. The largest channel represents a significant portion of the available fish habitat
(ranging from 33 to 80 percent in the selected wetlands) and this index provides a representative measure
of channel irregularity, or complexity, which can be compared among sites.
The site is a marsh located in the surge plain (although it is currently isolated from tidal influence) with
floodplain channels as the main hydrologic feature. Since these features are the focus of the restoration,
they will also be the focus of the analysis. In the selection process (below), only surge plain and surge
plain isolated complexes will be included for the marsh area and only floodplain and tidal channels will
be included for the confluences, Direct FHC area, and Direct FHC edge. For this analysis, small channels
(where total channel area is less than 100 square meters; Confluence Size = 0) will also be omitted.
To populate the metrics table with values, FHC must be selected with a combination of selection tools
within GIS: select by location, select by attribute, and manual selection. Try selecting features from the
31
FHC based on their location (completely within, intersect, etc.) relative to your scale of analysis polygon
(site, sub-landscape, landscape). Be careful when using select by location with polygons that do not
originate from the FHC layers (project site or reference site polygons) because they may not share the
same boundaries as the FHC polygons. It may help to alter your project site or reference site polygon (if
possible) to better fit within the FHC polygon boundaries. This is a good time to zoom in and make sure
all the FHC features on your site have been selected, and no external features are included. If the selection
needs to be adjusted, use the manual selection tool to add or subtract features from the selection. After
using the select by location tool, isolate just open or just altered FHC using the select by attribute tool
(and selecting from the current selection). Repeat the selection process until the table has all the necessary
values. Alternatively, instead of using the selection tools within GIS, geoprocessing tools like intersect
and union also isolate features for analysis using the scale of analysis polygon (project site, sub-
landscape, landscape) and an FHC layer as inputs to the tools. Table 10 lists the queries used to isolate the
target features. With the scales of analysis and LPF metrics chosen, the user can quantify the features to
calculate metrics. For the four metrics mentioned above (CD, PLAND, ED, SHAPE) the values needed
for each scale of analysis calculation are listed in Table 11.
Table 10. Select by attribute queries used to isolate FHC features for site and landscape analysis.
Target Feature Select by Attribute Query
Open Floodplain and Tidal
Channel Confluence
(("ChannelType_a" = 'Floodplain channel' OR
"ChannelType_a" = 'Tidal channel') OR
("ChannelType_b" = 'Floodplain channel' OR
"ChannelType_b" = 'Tidal channel')) AND
("ConfluenceSize" = 1 AND
"ConfluenceStatus" = 'Open')
Altered Floodplain and Tidal
Channel Confluence
(("ChannelType_a" = 'Floodplain channel' OR
"ChannelType_a" = 'Tidal channel') OR
("ChannelType_b" = 'Floodplain channel' OR
"ChannelType_b" = 'Tidal channel')) AND
("ConfluenceSize" = 1 AND
"ConfluenceStatus" = 'Altered')
Open Floodplain and Tidal
Channel Direct FHC
("ChannelType" = 'Floodplain channel' OR
"ChannelType" = 'Tidal channel') AND
("Complex" = 'Surge plain' OR
"Complex" = 'Isolated surge plain') AND
"FishHabitatStatus" = 'Open'
Altered Floodplain and Tidal
Channel Direct FHC
("ChannelType" = 'Floodplain channel' OR
"ChannelType" = 'Tidal channel') AND
("Complex" = 'Surge plain' OR
"Complex" = 'Isolated surge plain') AND
"FishHabitatStatus" = 'Altered'
Complex Selection "Complex" = 'Surge plain' OR
"Complex" = 'Surge plain (isolated)'
32
Table 11. Summary statistics used to quantify landscape metrics at each scale of analysis for the Deep River Confluence restoration case study.
Confluence
(Count)
Direct FHC Area
(Acres)
Direct FHC
Length
(1,000 Feet)
Largest
Channel
Area
(Acres)
Largest
Channel
Length
(1,000
Feet)
Surge
Plain
(Acres)
Isolated
Surge
Plain
(Acres)
Total
Complex
(Acres) Scale Open Altered Open Altered Open Altered
Appendix A. Datasets and descriptions included in the Fish Habitat Catena geodatabase (FHCv1_FINAL.gdb).
Dataset Description Data Sources Analysis Metric(s)
Direct_fish_habitat_ catena
Polygon; categorized as open or altered habitat that may be directly utilized by juvenile salmon
CREEC Catena (2012) CREEC Cultural Features
(2012) OR NAIP (2009) WA NAIP (2009) LiDAR/Columbia River Terrain
Model (1930-2010) T-sheets (1868-1901)
Capacity area edge
Opportunity channel type diversity tributary
confluence connectivity
Indirect_wetland Polygon; adjacent wetlands occurring within 2 meters (herbaceous), 5 meters (scrub-shrub), or 20 meters (forested) of fish habitat catena
Lower Columbia River Estuary Land Cover (2010)
Capacity wetland type area channel
connectivity
Indirect_drainage Polygon; estimate of tidally influenced and tidally impaired areas around the fish habitat catena
Lower Columbia River Estuary Land Cover Hydrologic Information (2010)
Lower Columbia River Estuary Land Cover (2010)
Capacity area
Indirect_USACE_ 2y_flood
Polygon; estimate of area inundated under the 2-year flood elevation around the fish habitat catena
USACE 50% AEP Stage for Columbia River Estuary (2011)
LiDAR/Columbia River Terrain Model (1930-2010)
Capacity area
LandscapeFeature_ confluence
Point; channel confluence point where dissimilar FHC aquatic features converge
CREEC Catena (2012) CREEC Cultural Features
(2012) OR NAIP (2009) WA NAIP (2009) LiDAR/Columbia River Terrain
Model (1930-2010) T-sheets (1868-1901)
Opportunity occurrence nearest neighbor
LandscapeFeature_ potential_beaver_ habitat
Polygon; potential locations of American beaver habitat selected from the fish habitat catena based on size, channel type, and location in a wooded ecosystem criteria
CREEC Catena (2012) Opportunity occurrence
LandscapeFeature_ head_of_tide
Point; up-valley extent of strong tidal influence, defined by the Classification
CREEC Catena (2012) Opportunity occurrence
Isolated_lake Polygon; naturally isolated lakes with no channelized connection to the Columbia River system
CREEC Catena (2012) CREEC Cultural Features
(2012) OR NAIP (2009) WA NAIP (2009) LiDAR/Columbia River Terrain
Model (1930-2010) T-sheets (1868-1901)
Landscape_unit Polygon; level of analysis between the scale of an ecosystem complex and hydrogeomorphic reach