evaluating the status of puget sound’s nearshore pelagic...
TRANSCRIPT
Evaluating the status of Puget Sound’s nearshore pelagic foodweb Final report: November 28, 2012
Correigh Greene, Casimir Rice, Linda Rhodes, Brian Beckman, Joshua Chamberlin, Jason Hall, Anne Baxter, Jeff Cordell, and Sean Naman
NOAA Fisheries
Northwest Fisheries Science Center
2
TABLE OF CONTENTS
Acknowledgements ........................................................................................................................3
Executive Summary .......................................................................................................................4
Background ....................................................................................................................................5
Methods ...........................................................................................................................................7
Site selection and land use calculations ......................................................................................7
Sample and data collection ..........................................................................................................8
Stable isotopes of fish and jellyfish tissue ................................................................................14
Statistical analyses ......................................................................................................................14
Results & Discussion ....................................................................................................................15
Physical oceanographic measurements .....................................................................................15
Dissolved inorganic nutrients and microbial abundance ...........................................................18
Bacterial community characterization .......................................................................................20
Multivariate analyses using abiotic, microbial, & land use variables .......................................21
Zooplankton ..............................................................................................................................23
Fish and jellyfish ........................................................................................................................26
Stable isotopes of fish and jellyfish ...........................................................................................36
Synthesis........................................................................................................................................41
Basin differences .......................................................................................................................41
Sensitivity of indicators to land use ..........................................................................................44
Evidence for the bifurcated foodweb hypothesis ......................................................................45
Implications ...............................................................................................................................46
References .....................................................................................................................................47
Appendix .......................................................................................................................................50
3
ACKNOWLEDGEMENTS
We would like to thank the many groups and individuals who made this research effort possible.
Funding was provided by the US EPA’s National Estuary Program, with additional support from
the Washington Department of Natural Resources and the Department of Ecology. We were
greatly assisted by three tribal partners – the Squaxin Tribe, the Port Gamble S’Kallam Tribe,
and the Skagit River System Cooperative – and within these tribes we particularly thank Scott
Steltzner, Hans Daubenberger, Paul McCollum, and Eric Beamer for their support and efforts to
help staff cruises. In addition, we thank the numerous volunteers who helped with data collection
(it would take a page to list in their entirety), but especially recognize Craig Peters, Lauren
Madaras, Lindsay Anderson, Paulmer Brown, Aissa Yazzie, and Craig Wollum for consistent
support across multiple cruises. We also benefited from the help of NOAA Hollings Scholars
Michael Hoban, Megan Hess, Toby Matthews, and Sariha McIntyre, who assisted with both data
collection and analysis. University of Washington students Jessica Randall and David Berman
helped process data collected from the cruise. In addition, University of Washington researchers
Julie Keister, Dave Beauchamp, and George Hunt provided helpful design advice and support in
the field. Evelyn and Barry Sherr at Oregon State University provided invaluable advice and
assistance on the microbial analyses. Emily Runnells selflessly recorded bird observations on the
bow of the Coral Sea, and Tish Conway-Cranos was a great help on our cruises to Samish Bay
and Hood Canal. She, Kurt Fresh, and Phil Roni provided helpful reviews of the report. Our paid
contractors Dana Rudy, Alicia Godersky, Heather Cid, and Bruce Brown went above and beyond
their duties numerous times. Hiroo Imaki provided invaluable GIS support during the design
phase and even got out in the field. Anna Kagley and Kurt Fresh provided helpful input during
the experimental design phase and field expertise during cruises. Photo credits for five of the six
cover photos (and a great short movie as well) go to Alan Lovewell. Dan Lomax, Jen King, and
Skip Bold committed much time and attention to safely and effectively operate the RV Harold
W. Streeter and Coral Sea. And speaking of boats, we thank the solid support of the Streeter for
putting in a huge swan song field season without a hitch before being retired. Likewise, the Coral
Sea has moved on from her tow-netting days back to the business of fishing, and we will miss
her old-school charm.
4
EXECUTIVE SUMMARY
The pelagic zone is a large and important component of Puget Sound’s ecosystem, but basic
information is lacking on differences among oceanographic basins, linkages between abiotic
features, water quality, and pelagic biota, and the effects of anthropogenic activities. This dearth
of information complicates our ability to identify useful metrics to measure the pelagic zone’s
key characteristics determining ecological health. To address these issues, we conducted a multi-
trophic level assessment in six oceanographic basins within Puget Sound using a sampling
scheme designed to detect both basin-wide differences and relationships between pelagic
ecosystem attributes and land use in catchments surrounding sites. We measured over 20
potential indicators of nearshore pelagic ecosystem health at 79 sites in six oceanographic basins
of Puget Sound. These metrics included measurements of abiotic conditions and nutrient
availability, and abundance and diversity of phytoplankton, bacteria, zooplankton, jellyfish, and
pelagic fish species. In many taxa from lower to middle trophic levels, and for a comprehensive
suite of abiotic attributes, we observed strong seasonal and spatial structure. South Sound and
Hood Canal had the most reduced dissolved oxygen and pH, highest relative abundance of
jellyfish, and lowest abundance of forage fish and fish species richness. In contrast, Rosario
(north of Fidalgo Island) and Whidbey Basins were characterized by relatively few abiotic or
nutrient problems, few deviations in the abundance of different groups of microbes and
phytoplankton, relatively high densities of non-gelatinous (i.e., not jellyfish) zooplankton, and
high fish species richness and relatively high forage fish abundance. Admiralty Inlet and the
Central Basin scored in between this range, although they too exhibited high jellyfish abundance
and reduced forage fish abundance and fish species richness relative to Rosario and Whidbey
Basins. Furthermore, many of the potential indicators we measured were sensitive to land use,
with a general pattern that abiotic and lower trophic patterns were most sensitive, and patterns in
fish abundance and diversity were the least sensitive. We found positive relationships between
land use and jellyfish abundance, as well as shifts of jellyfish diets to lower trophic levels in sites
with greater land use. These findings provide empirical support for the bifurcated foodweb
hypothesis, which predicts that stressors from development simplifies foodweb structure, leading
to cascading effects on middle trophic levels like planktivorous salmon and forage fish, and
favoring jellyfish and other consumers of microplankton. Despite these patterns, land use rarely
explained more than 5% of the variation in observed data, indicating a dominant marine
influence and the potential for resilience of Puget Sound’s pelagic waters to anthropogenic
influence. The strong spatial structure observed in our results indicates that different pelagic food
webs exist across the system. Consequently, target conditions, current health status, or both,
cannot be uniform across greater Puget Sound. These are critical considerations for management
of the Puget Sound ecosystem, and we expect that further analysis of our results in the context of
other studies will improve our understanding of the underlying causes of the patterns we
observed across Puget Sound.
5
BACKGROUND
The pelagic ecosystem (the water column extending from the surface to just above the
benthos) is a biological and economic focal point of Puget Sound, yet it remains one of the most
poorly understood environments in the Pacific Northwest. The Sound’s deep bathymetry make
the pelagic zone the largest component of marine habitat. Not surprisingly then, the pelagic
ecosystem is at the center of the Sound’s complex marine foodweb. Within the pelagic
ecosystem, marine nutrients mix with riverine inputs at estuaries to create high primary and
secondary productivity that fuel forage fish and salmon populations (hence “The Fertile Fjord,”
Strickland 1983), which in turn are consumed by large predators such as seabirds and orcas. As a
consequence, the pelagic ecosystem is highly valued for its ecosystem services for recreational
and commercial fishing, shellfish aquaculture, boating and diving, and its tribal cultural heritage.
It also serves as the recipient of sewage treatment plant effluent and terrestrial run-off, , and may
be affected by extensive physical alteration at the land-water interface.
Recently, a number of observations have raised concern for the ecological health of Puget
Sound’s pelagic ecosystem. Herring and smelt, the dominant forage fish of Puget Sound, may be
declining in some regions of Puget Sound. Many populations of Pacific salmon that use Puget
Sound are listed as Threatened under the Endangered Species Act (Chinook and chum salmon,
and steelhead), or Species of Concern (coho salmon). Because of demand for salmon, resource
managers have supplemented declining populations with hatcheries, which often have negative
effects on the native populations due to competition, disease transmission, and genetic
introgression (Myers et al. 2004). Sea bird populations, which depend upon forage fish and
juvenile salmon living in the pelagic zone, show evidence of declines (PSP 2010). In addition,
incidents such as high abundances of jellyfish in various parts of Puget Sound (Rice et al. 2012),
harmful algal blooms, and hypoxia, all of which have been interpreted as ecological warning
signs (Anderson et al. 2002, Richardson et al. 2009, Cope and Roberts 2012), have focused our
concern of the current ecological health of Puget Sound’s pelagic habitats.
These observations beg the question: how do people affect the pelagic ecosystem? This is a
fundamentally important question for Puget Sound’s scientific community to address. While the
pathways by which people impact the pelagic ecosystem are likely complex, and may be difficult
to assess due to Puget Sound’s marine-driven hydrodynamics (Kim and Khangaonkar 2012) and
the mobility of its aquatic biota (e.g., Hay et al. 2001), this question has not been rigorously
examined. In fact, many existing field programs have not incorporated human influences such as
land use into sampling designs (Rice 2007).
The dearth of information is not due to lack of ideas. Parsons and Lalli (2002) postulated that
simple autotrophs (cyanobacteria, flagellates, and dinoflagellates) may be favored when water
quality parameters worsen, leading to predominance by jellyfish over fish at middle trophic levels,
and consequently resulting in a trophic “dead end” where little energy is transferred to upper trophic
levels such as predatory fishes, mammals, and birds. A “bifurcated foodweb” may result because
simple autotrophs constitute prey for smaller types of zooplankton and early stages of jellyfish, both
6
of which are not preferred prey for fish compared to the larger zooplankton that consume larger
diatoms. While these patterns have experimental support (Parsons et al. 1981), the bifurcated
foodweb hypothesis has not been tested in the field.
In 2011 we conducted a field study in multiple oceanographic basins within Puget Sound to
simultaneously examine different components of the pelagic foodweb, using a sampling design
that stratified for major natural environmental influences (month, oceanographic sub-basin,
shoreform) but also incorporated degree of urban and agricultural land use. This effort – the first
of its kind – sampled water quality, microbes, primary producers, zooplankton (including
jellyfish), fish, and birds and marine mammals. Because of our sampling design, we were able to
explicitly test for effects of natural influences and land use on our measurement endpoints.
The primary goals of the study were to:
1. identify how foodweb structure differs among the oceanographic basins of Puget
Sound,
2. determine whether particular measurement endpoints of the pelagic ecosystem are
sensitive to gradients of land use
3. identify a number of potential biological metrics for monitoring ecosystem health.
7
METHODS
Site selection and land use calculations
To achieve the three study goals, we used an index site-based regression design that
incorporated gradients of shoreline or catchment land use in four shoreline geomorphic types, in
each of six basins of greater Puget Sound (Figure 1). Given this goal, the number of possible
sites that can be selected is limited, and random selection can induce added variation that might
obscure underlying patterns. We therefore used the following procedure to determine a set of
index sites:
1) Determine bathymetry of Puget Sound. Some sites are inaccessible to our boats, so
bathymetry bounds the “population” of sites that can be sampled. We used PSDEM2000 (School
of Oceanography, University of Washington 2000, downloaded from
http://www.ocean.washington.edu/data/pugetsound/) and The Estuarine Bathymetry data, P290
(The National Ocean Service 2011, downloaded from http://estuarinebathymetry.noaa.gov/) to
draw a 10 m bathymetric contour of all of Puget Sound. Bathymetry was ground-truthed from
known locations where we previously sampled.
2) Determine shoreline units. We used PSNERPs drift cell framework (PSNERP Geodatabase
Version 3.0 Change File, 2010) to select shoreline segments. Because trawls are longer than
some drift cells and because land use patterns are sometimes larger or smaller than the spatial
extent of certain drift cells, we sometimes combined or divided contiguous drift cells to
determine units. In the end, units could be directly linked to shoreline segments and catchments
for which the percentage of area developed by area could be estimated. Development classes
were based on C-CAP 2006 land cover classes, and are at a 30 m resolution (Figure 2). We
further developed a list of land use metrics to include total shoreline length, total catchment area,
and amount of agriculture and development (m2) percentage of each land use type within 200 m
of shore and within the entire catchment.
Table 1. Number and sites sampled in each oceanographic basin of Puget Sound.
Geomorphic type
Basin Tidal
delta
Large
bay
Small
bay
Exposed Total
Rosario Basin 3 2 4 5 14
Whidbey Basin 6 3 3 5 17
Admiralty Inlet 3 1 4 8
Hood Canal 5 3 2 3 13
Central Basin 5 3 2 3 13
South Sound 3 5 3 3 14
Total 22 19 15 23 79
8
1
2 3
4
5
6
7
3) Determine habitat types. We used four habitat types to stratify potential sites into units of
similar geomorphic structure: embayments associated with large river deltas, large embayments
(> 2.5 km of shoreline) lacking large river deltas, small embayments (≤ 2.5 km of shoreline), and
exposed shorelines (not in embayments, Shipman 2008). We used SSHIAP embayment units
(SSHIAP 2011) to define the units.
4) Choose sites. We were limited by the number of sites we could sample in a day (maximum –
10 sites/day) and the temporal window of each sampling event. For each basin, we examined all
sample-able sites within each habitat type, and selected sites providing the most representative
gradation of development at the catchment level. Within each basin, we chose 3-6 sites of each
habitat type, depending upon the number of possible sites. The exception was Admiralty Inlet,
which has few embayments of any kind and no large river deltas. Here we chose sites to
maximize the range of land use within three of four habitat types. Because large river deltas are
an important landscape feature, we chose up to two “replicate” sites at each large river delta.
5) Once an initial site list was determined, we estimated travel times between sites, and further
reduced the number so that all sites could be visited once per month over the course of a cruise.
We tested an initial design of 94 sites in April, and found that we could not complete all
operations over 11 days of sampling per month. Therefore, we scaled back to 79 sites (Table 1,
Figure 3), primarily by removing exposed sites.
Figure 1. Project map indicating land use in the
Puget Sound Basin, ranging from high intensity
development (black) to forest, wetland, and water or
ice (lighter shadings). Lines indicate boundaries of
major oceanographic basins, which are numbered in
white: (1) South Puget Sound, (2) Central Basin, (3)
Hood Canal, (4) Admiralty Inlet, (5) Whidbey
Basin, (6) San Juan Islands, Bellingham Bay, and
Padilla Bay (“Rosario Basin”), and (7) East Strait of
Juan de Fuca. All but the last of these regions was
included in this study.
Sample and data collection
A total of 549 tows at 79 sites were sampled
across six basins between April and October
2011. Several thousand individual samples were
collected for various metrics (Table 2) at each
sampling location over the sampling period
(Table 3). All samples were collected and
processed in accordance with the methods
outlined in the Quality Assurance Project Plan (QAPP).
Sampling at each site was organized around a surface trawl for fish and jellyfish collection.
Surface trawls were conducted using a Kodiak surface trawl, or “townet” (6.1 m x 3.1 m, 6 cm
9
mesh cod-end) towed between two boats. Each trawl occurred as close to the shoreline as
feasible (within a bathymetric range of 6 to 40 m), into the current at a fixed engine RPM for 10
minutes (most trawls were approximately 0.5 km long). In a limited number of cases, trawls were
reduced to 5 minutes when jellyfish were observed at high densities, or if the site had a high
jellyfish density in previous months.
Figure 2. Graphs of each site as function of oceanographic basin (colors) and percent of land cover in
developed classes in entire catchments (x-axis) or in 200 m buffers along the shoreline (y-axis). A. land
use in the large river deltas. B. Land use in large (diamonds) and small (triangles) embayments. C. Land
use at exposed sites.
10
Figure 3. Locations of index sites
(green and black circles) sampled
throughout the course of the study.
Salmon coloration indicates
developed areas, yellow indicates
agriculture, green area is mixed
forest.
11
Upon approach to the site, the trawl was deployed approximately 0.3 km behind the site’s
midpoint. Current meters were deployed to determine amount of water swept. After 10 minutes,
the net was closed and the fish and large invertebrates were brought on board for sorting and
measuring (see below). Recorded latitude and longitudes for each sites specified the midpoint of
each trawl line.
Water column measurements were sampled at the midpoint of each tow line during each
sampling event using a SeaBird® SEACAT Profiler (SBE 19plusV2). At each site the unit was
lowered into the water at 0.3 m/sec to within 1m of the sea floor and retrieved at the same rate.
Parameters collected at each site included temperature, conductivity, density, dissolved oxygen,
depth, PAR, fluorescence, turbidity, and pH. Profile data were corrected for depth and binned by
0.5 m increments. Inorganic nutrient samples were extracted from seawater grabs taken at each
site using a General Oceanics® 5L Niskin water sampler lowered to a depth of 6m. Nutrient
samples were collected by filtering 50 cc of water through a filter into a polyethylene bottle and
stored on ice before being transferred to the lab for processing (see QAPP).
Microbial samples were extracted from seawater grabs (see above) collected at the midpoint
of each site/tow during each sampling event. Water grabs were poured into two 2 L polyethylene
bottles rinsed with water from the sample location. Production samples were collected by filling
a 15 mL polystyrene conical tube with water and stored on ice for transport to the lab. For
abundance samples, 3 mL of water was added to a 4 mL cryovial containing 60 µl 10%
paraformaldehyde. Samples were inverted and placed on ice for 10 minutes before being
transferred to liquid nitrogen for storage. Diversity samples were taken by concurrently filtering
500 mL of water through 2 µm and 0.5 µm filter using a vacuum pump. Filters were folded,
placed into a cryovial and transferred immediately to liquid nitrogen for transport. Samples
collected for chlorophyll-a consisted of filtering 50mL of water through a glass fiber filter.
Filters were folded and placed into glass tubes before being wrapped in aluminum foil and stored
on ice until storage at -20° C. Replicates were taken for each sample type and all samples were
transported to the lab within 12 hours for further processing (see QAPP).
Surface (horizontal) and water column (vertical) plankton samples were taken from each site
during each sampling event. Surface samples were collected using a 1.0 m diameter x 3.0 m long
net with 500 µm mesh. The net was deployed after the surface trawl (fish sample) and was towed
along the surface for 3 minutes at a speed of 2 knots through the water in an arc to avoid
sampling water disturbed by vessel movement. A General Oceanics® model 2030 flowmeter was
attached at the center of the net opening to quantify the amount of water swept during each tow.
Water column plankton samples were collected using a 0.5 m diameter x 2.0 m long net with 250
µm mesh size. The net was lowered to within 1 m of the sea floor and then retrieved at
approximately 0.3m/sec. Samples from both nets were washed down with water from each
sample location and filtered through a 500 µm and 250 µm mesh sieve, respectively and coarse
debris removed. Once filtered, plankton samples were preserved in a 10% neutral buffered
formalin solution and sealed for transport to the lab.
12
At the conclusion of each tow, fish and jellyfish were immediately removed from the net and
placed into live wells with a constant flow of water from the sample location. Individual of each
species were counted and up to 25 individuals of each species were measured to the nearest
millimeter (fish: fork length or total length when no fork present; jellyfish: bell diameter). All
individuals of a given species were weighed for a total species biomass. All salmonids were
checked for adipose fin clips and/or the presence of coded-wire tags.
Herring, surf smelt, Chinook, and chum were targeted for further analysis of individual life
history characteristics (otoliths), diet composition, stable isotopes, and growth (plasma IGF-1). A
subset of up to six individuals of each target species were randomly selected, sacrificed, and
processed in the onboard lab. Individual lengths and weights were recorded and otoliths and guts
were removed and placed in to ethanol- and formalin-filled vials, respectively. Blood was drawn
immediately from each individual and deposited into microfuge tubes. Several times throughout
each day blood samples were spun for 5 minutes at 5000 x G in a microcentrifuge to separate
plasma from red blood cells. Plasma was removed from the sample and frozen at -20 C for
further analysis of IGF-1 in the lab (see QAPP). Carcasses were frozen for future isotope
analysis. Up to ten individuals of a given, non-target species were randomly selected and
sacrificed for genetics and stable isotope analysis. After individuals were placed into a lethal
solution of MS-222, fin clips (genetics) were taken and placed into ethanol-filled vials and
carcasses (isotopes) were frozen. All samples were processed in the lab using accepted
methodologies (see QAPP).
Table 2. Sample types and metrics measured at each study site.
Type Metric
Environmental variables Water column measurements: (Temperature,
salinity, depth, PAR, dissolved oxygen, pH,
turbidity, density, conductivity)
Inorganic nutrients (nitrate, nitrite, ammonium,
phosphate, silicic acid)
Microbes Microbial heterotrophic production
Microbial abundance
Bacterial diversity
Autotrophic productivity (Chlorophyll a)
Zooplankton Small zooplankton abundance and composition
Large zooplankton abundance and composition
Stable isotopes
Fish and jellyfish Counts and biomass by species
Individual size
Plasma IGF-1
Stable isotopes
Birds and marine mammals Abundance and composition
13
Table 3. Total samples collected by basin and month in 2011.
Basin Month
tows zooplankton1
isotopes2
igf water column microbes3
Rosario Apr 14 28 6 0 14 75
May 14 28 78 47 14 65
June 14 28 213 148 14 70
Jul 16 32 199 143 16 76
Aug 16 32 159 118 16 80
Sep 15 30 87 59 15 75
Oct 15 30 99 44 15 75
total 104 208 841 559 104 516
Whidbey Apr 14 28 3 0 14 65
May 17 34 16 16 17 84
June 17 34 236 115 17 85
Jul 17 34 286 198 17 85
Aug 17 34 192 109 17 85
Sep 17 34 163 97 17 85
Oct 17 34 63 30 17 85
total 116 232 959 565 116 574
Admiralty Apr 8 16 0 0 8 40
May 8 16 45 10 8 30
June 7 14 52 33 7 30
Jul 7 14 72 46 7 30
Aug 7 14 60 44 7 30
Sep 7 14 45 27 7 30
Oct 6 12 17 3 6 25
total 50 100 291 163 50 215
Hood Canal Apr 13 26 0 0 13 60
May 13 26 113 22 13 65
June 14 28 108 43 14 70
Jul 14 28 77 36 14 70
Aug 13 26 75 47 13 65
Sep 14 28 52 27 14 70
Oct 14 28 54 32 14 70
total 95 190 479 207 95 470
Central Apr 14 28 0 0 14 68
May 13 26 2 2 13 64
June 13 26 131 70 13 65
Jul 13 26 195 132 13 65
Aug 13 26 149 97 13 65
Sep 13 26 103 73 13 65
Oct 13 26 58 21 13 65
total 92 184 638 395 92 457
South Sound Apr 14 28 0 14 70
May 13 26 17 17 13 69
June 13 26 109 60 13 70
Jul 13 26 176 122 13 70
Aug 13 26 144 102 13 70
Sep 13 26 83 48 13 69
Oct 13 26 55 15 13 70
total 92 184 584 364 92 488
TOTAL 549 1098 0 2253 549 2720
# of samples
1Zooplankton samples include surface (horizontal) and water column (vertical) samples. 2Isotopes samples represent individual and combined samples. Combined samples represent up to 10 individuals of a given
species. 3Microbe samples are indexed as a single combined sample representing abundance, production, diversity, nutrients and
chlorophyll-a.
14
Continuous bird and marine mammal observations were conducted opportunistically before,
during, and after surface trawls as well as during transit between trawl stations. Individuals
within 200 m in the forward 180 degrees at the boat’s bow were counted and identified to species
when possible.
Stable isotopes of fish and jellyfish tissue
We collected stable isotopes from a subset of all captured fish (n = 1,096) and jellyfish (n =
728) species across the entire duration of our study. All samples were frozen upon capture and
stored at -20 ºC. For fish, a plug of dorsal muscle tissue was extracted and lyophilized for 24 h.
Whole jellyfish were dried in an oven at 60 ºC for 24 h. All samples were ground into a
homogenous powder using a glass mortar and pestle and weighed into specified amounts in tin
capsules. Individual jellyfish species from the same site and date were combined in order to
obtain enough material for analysis; therefore jellyfish data represent a composite of individuals
present at a site at a given date. Fish were processed individually. Samples were analyzed for
their isotopic compositions of 13
C and 15
N using a continuous flow mass spectrometer. Isotope
values are expressed in the δ notation where:
δ(‰) = [(Rsample – Rstandard/ Rstandard) x 1000]
The majority of these samples are currently being analyzed at the Northwest Fisheries
Science Center. For this report, we present data from June 2011. We focus our analysis on
general patterns of isotopic enrichment among basins and focus on two finfish (Chinook and
chum) and two jellyfish (water and ctenophores) in which data were available for at least ten
individuals and three basins. We also include preliminary correlation analysis between isotopic
enrichment of fish and jellyfish and several shoreline land use metrics. For chum and Chinook,
we added reference points of potential notable diet components that may have large effects on
isotope signatures. This is for exploratory purposes and should not be interpreted as explicit diet
contributions.
Statistical Analyses
We used several analyses to examine variation in metrics, focusing on large effects (p < 0.05)
for reporting most results. We examined effects of oceanographic basin and time using general
linear models, and added effects of geomorphic type, land use variables, and other metrics as
covariates. To examine the explanatory power of different land use metrics, we used stepwise
removal procedures of land use variables to determine the smallest suite of land use
characteristics that explained the most variation.
We also used several multivariate techniques to examine correspondence among multiple
metrics. Discriminant analysis was used to determine whether land use, abiotic, and biotic
metrics systematically varied among sites. Canonical correlational analysis (CCA), and other
multivariate procedures (see below) were used to examine spatial and temporal variation in taxa
abundance and composition at various trophic levels.
15
RESULTS & DISCUSSION
Physical oceanographic measurements
A wide range of physical oceanographic measurements (water column profiles of temperature,
conductivity, density, pH, dissolved oxygen, photosynthetically active radiation, fluorescence,
and turbidity) were made at each station. These measurements exhibited strong (p < 0.05)
differences among basins, and most exhibited seasonal trends as well (Table 4).
Table 4. Results of two-way ANOVAs for abiotic variables sampled at the surface. For basin results,
pairwise differences show means of basins (R=Rosario, W=Whidbey, A=Admiralty, H=Hood, C=Central,
S=South) ranked lowest (left) to highest. Lines indicate sets of basins that lacked large pairwise
differences in means, and spaces indicate sets with large pairwise differences.
Variable Month Basin Basin differences
Salinity < 0.001 < 0.001 W SHC RA
pH < 0.001 < 0.05 AWRCSH
Turbidity > 0.1 < 0.001 ACSH RW
PAR > 0.1 < 0.001 SWRCHA
Temperature < 0.001 < 0.001 WRACHS
Dissolved
oxygen < 0.001 < 0.001 R AW SHC
For example, dissolved oxygen (DO) concentrations at the maximum depth of the water
column profile generally increased from April to May and then declined through October within
all basins (Figure 4). Decreases in dissolved oxygen concentrations at maximum depth by basin
were greatest within the Hood Canal basin, in contrast to surface measurements (Table 4). Water
column profile data from another monitoring program show dissolved oxygen concentrations
were at their lowest levels in Bellingham Bay (Rosario basin) in August 2011 while dissolved
oxygen concentrations continued to decline at South Skagit Bay, Possession Sound, and Saratoga
Passage (Whidbey basin) through October 2011 (DOE 2012). DO concentrations at 0.5 and 6
meters deep followed similar seasonal and basin level trends, with median DO concentrations
falling the most in Hood Canal (see Appendix, Figures A-1, A-2).
16
Figure 4. Box and whisker plot of DO concentrations at maximum depth by month and basin showing
median concentration (dark line inside box) for each basin and month combination, 25 – 75% interquartile
range (lower and upper limit of boxes), the maximum and minimum values excluding outliers (whiskers
extending above and below each box), and outliers (open circles above or below whiskers).
Measurements in the Rosario basin (August – October) and Whidbey basin (September – October) were
not collected due to a DO sensor failure.
Hypoxia is generally defined as less than 3 mg L-1
DO, and the upper limit of biological stress
from reduced DO concentrations for higher trophic organisms is defined as 5 mg L-1
(DOE
2002). When the minimum DO for each site throughout the survey was determined, Hood Canal
had the greatest number of sites with biologically stressful conditions (Figure 5, < 5 mg L-1
).
Across months, 92% of Hood Canal sites (12 of 13) had minimum DO concentrations surpassing
stressful conditions. Although surface DO concentrations were not below the threshold within
the monitoring period, DO at 6 m depth in Hood Canal dropped below it in October.
Depressed DO concentrations were typically associated with highly stratified water columns
featuring rapid changes in near-surface waters (Figure A-3, top graph). However, stratified water
may be necessary, but not sufficient for low DO concentrations. For example, a profile at the
same site in a different month shows salinity stratification without low DO concentrations
(Figure A-3, bottom graph).
One striking bivariate relationship was observed across all temporal and spatial scales: pH and
dissolved oxygen (DO). DO concentrations were positively correlated with pH across basins and
months at the surface, at 6 meters deep, and at the maximum depth of the water column profiles
(Figure 6). Both the strength (R2) and magnitude (slope) of this relationship increased with
depth, an observation consistent with other recent nearshore marine studies (e.g., Frieder et al.
2012). DO and pH may be tightly linked through dynamic biological (e.g., local primary
production and remineralization of organic matter) and physical (e.g., upwelling and
stratification) processes (Frieder et al. 2012), although geophysical processes are expected to be
variable across oceanographic basins (Strickland 1983) .
17
Figure 5. Minimum DO concentrations observed at the maximum depth of each site for the entire
duration of the survey (April – October). See text about data on Rosario and Whidbey basins.
Figure 6. Linear regressions of pH and dissolved oxygen at the surface (cross), 6 meters (solid circle),
and maximum depth (open circle) with slope and R2 values based on all basins and months pooled.
18
Dissolved inorganic nutrients and microbial abundance
Inorganic nutrient levels (nitrate, nitrite, ammonium, phosphate, silicic acid) are expected to
vary greatly as a function activity by both microbial autotrophs and heterotrophs (Arrigo 2005).
We examined inorganic nutrients in the context of several metrics of primary and heterotrophic
production by microbes in Puget Sound. Primary production due to phytoplankton was estimated
by measuring chlorophyll a from filtered water samples, and the abundance of specific subsets of
phytoplankton (i.e., small [< ~20 µm] non-chain-forming diatoms and picophytoplankton) was
determined by flow cytometry. Primary production due to cyanobacteria was estimated by the
abundance of Synechococcus, the dominant cyanobacterium in higher nutrient, nearshore marine
waters. Bacteria and archea are the principal effectors of nutrient cycling from all trophic levels
as well as directly from water. The abundance of bacteria and archea (= total bacterial
abundance) could be differentiated into subsets based on DNA content per cell (high-DNA
bacteria, low-DNA bacteria). High-DNA bacteria exhibit greater DNA content per cell and the
abundance of high-DNA bacteria is positively correlated with heterotrophic production
(Spearman’s rho = 0.46, p < 0.0001), suggesting this subset is undergoing cell division and
metabolically active. In contrast, low-DNA bacteria exhibit lower DNA content per cell and the
abundance is uncorrelated with heterotrophic production (Spearman’s rho = 0.02, p = 0.60),
suggesting this subset is not dividing or metabolically active.
Although the range of dissolved inorganic nutrient concentrations varied little across the
entire sampling period, there were dramatic differences in temporal patterns across the basins
(Figure A-4). For nitrate, nitrite (not shown), phosphate, and silicic acid, two basins (Rosario,
Whidbey) displayed trends of steady or slightly increasing concentrations from May through
October, whereas three basins (Admiralty, Central, South) showed a convex pattern of higher
values at the beginning and end of sampling (May, October) with a “dip” toward minimal values
during the early summer months (June, July). Hood Canal differed from all other basins in two
respects. First, nitrate levels were extremely low basin-wide from May through September.
Second, both nitrate and nitrite temporal patterns were not consistent with other basins. Overall
concentrations of nitrate, nitrite, phosphate, and silicic acid exhibited robust correlations (Table
A-1), suggesting coordinate regulation of input and/or consumption of these nutrients.
The temporal pattern of ammonium concentrations within each basin diverged from the other
four nutrients, and the patterns were more varied among basins (Figure A-4). Whidbey,
Admiralty, and Central basins exhibited a unimodal pattern with a peak concentration in May
through July, while the South basin showed peak in July with a slight increase in October. The
temporal pattern in Rosario was convex, with a minimum in June. Hood Canal ammonium
concentrations were relatively low throughout the sampling period.
Chlorophyll a concentrations across the basins showed variable temporal patterns, but peak
concentrations generally occurred by June (Figure A-5). In the Admiralty and Hood Canal
basins, a secondary peak in chlorophyll a occurred after July. Phytoplankton subsets, including
picophytoplankton and small diatoms (single cells < ~20 µm), were highly abundant in Hood
Canal in September, but occurred at relatively low levels elsewhere throughout the entire survey.
19
While picophytoplankton and small diatoms were positively correlated overall (r = 0.525, p <
0.0001), the association of these subsets with chlorophyll a varied over the months (Table A-2).
The absence of a significant relationship between these phytoplankton subsets and chlorophyll a
from July through August suggests that much of the chlorophyll a signal was associated with
larger or chain-forming phytoplankton that are not enumerated by flow cytometry.
Bacterial abundance, which is subdivided into high-DNA and low-DNA bacteria, displayed
no consistent temporal patterns across the basins (Figure A-6). Peaks in both high-DNA and low-
DNA bacterial abundance occurred at different times across the basins with no latitudinal order
(i.e., north to south). Low-DNA bacterial abundance was strikingly high in September in Hood
Canal. Synechococcus, a photosynthetic bacterium, was very low across all basins and months
except in Hood Canal, with a peak occurring in September. Although heterotrophic production
was correlated with total bacterial abundance, a larger correlation was observed with high-DNA
bacteria (r = 0.450, p < 0.0001) than with low-DNA bacteria (r = 0.141, p = 0.003), suggesting
that high-DNA bacteria are more metabolically active than low-DNA bacteria (see also
Longnecker et al. 2006).
Because both autotrophic and heterotrophic microbes may consume dissolved inorganic
nutrients, the relationship between nutrients and 1) chlorophyll a concentration or 2) high-DNA
bacterial abundance was examined. Nutrients included dissolved inorganic nitrogen (DIN; sum
of nitrate, nitrite, and ammonium) and phosphate. For all basins except Hood Canal, both
microbial measures displayed significant (p < 0.0001) negative correlations with DIN
concentrations (Figure 7). Simple linear regression of both microbial measures on DIN or
phosphate revealed negative coefficients for both parameters for all basins except Hood Canal.
The R2 values of the regressions indicate substantial explanation of nutrient concentrations by
these two measurements (Table A-3). These relatively strong inverse relationships suggest that
microbes are detectably modulating inorganic dissolved nitrogen and phosphate.
Figure 7. Scatterplots of DIN concentrations against chlorophyll a concentrations (left) or high-DNA
bacterial abundance (right) by basin with significant correlation coefficient (p < 0.0001).
20
Bacterial community characterization
The abundance of certain categories of microbes reflects conditions that are favorable or
unfavorable for those organisms. However, a profile of a community contained in a sample
would allow a multifaceted comparison of samples across space or time. The underlying
assumption is that similar community profiles have similar taxonomic and (potentially)
functional profiles. One way to examine variation in bacterial communities is with automated
ribosomal intergenic spacer analysis (ARISA). The ARISA approach to bacterial community
profiling relies on the length of 16S-23S rDNA intergenic region. While ARISA cannot be used
to identify specific bacterial taxa, it can be used to characterize the assemblage of 16S-23S
rDNA intergenic regions present in a sample and subsequently compared to assemblages in other
samples.
Canonical correspondence analysis (CCA) was used to plot each sample in relation to all other
samples based on similarity of ARISA profiles, abiotic measurements, and microbial biotic
parameters (Table A-4). Subsets of samples by basin and color-coded for month of collection
reveal several consistent patterns (Figure 8). Within a basin, samples collected during the same
Figure 8. Plots of samples based on canonical correspondence analysis (CCA) of ARISA of community
DNA from water grabs, abiotic measurements, and microbial biotic parameters. Each subplot displays
only samples from the indicated basin, and the month of each sample is color-coded. The CCA1 axis
(vertical axes) explains 5.8% of total variation, and the CCA2 axis (horizontal axes) explains 3.6% total
variation Vectors indicate the strength and direction of change of environmental variables (see Table A-4
for full list) along these two axes.
21
month cluster together, and clusters of adjacent months tend to overlap. This pattern suggests
that seasonal influences are exerted broadly within a basin, and bacterial communities are
gradually, rather than abruptly, changing in composition. Except for South Sound, community
profiles in September and October were more similar across the basins than at any other month,
based on the position of those samples (light and darker purple) in the lower left quadrant. This
quadrant corresponds to greater dissolved nutrients and reduced pH. Although all basins
exhibited seasonal patterns, each basin displayed a distinctive ARISA plot, indicating that
bacterial community structure and its seasonal progression may differ widely across the basins.
The overall lower dispersion of samples for Rosario and South basins indicate that bacterial
communities in these basins are more homogenous through time.
Partial canonical correspondence analysis (pCCA) partitioned the effect of variables on
overall bacterial community structure into three categories: land use (5.5% of total variance),
habitat characteristics (12.3% of total variance), and season (12.5% of total variance; Table A-5).
Multivariate analyses using abiotic, microbial, & land use variables
We assessed thepotential effects of shoreline and catchment land use on abiotic and biotic
metrics influencing lower trophic levels, including nutrient concentrations, chlorophyll a
concentration, bacterial abundance, and turbidity. The potential role of seasonal (i.e., monthly)
and geomorphic (i.e., delta, exposed, large bay, or small bay) variations were assessed as factors,
while land use and physical oceanographic effects were assessed as covariates. Regression
coefficients were calculated to determine the effect size of factors or covariates. Due to high
collinearity among subsets of land use parameters, certain parameters were selected as proxies
for multiple parameters (Table A-6).
Beyond the temporal and geomorphic effects, land use exerted a significant influence on
dissolved inorganic nutrients. Increasing nutrient concentrations were associated with increasing
agricultural use, although the magnitude of those effects was not as large as temporal effects
(Table A-6). Agricultural land use, either as a percentage or an area measurement of associated
uplands, had an increasing effect on nutrient concentrations for all dissolved inorganic nutrients
and turbidity, while area of developed or impervious land also contributed to DIN and phosphate
concentrations and turbidity (Table 5). In contrast, human-manipulated land use had a negative
effect on bacterial abundance, and no effect on chlorophyll a concentrations or pH. Among the
influential physical variables, temperature had opposite effects on nutrient concentrations
(negative) and microbial measurements (positive). Given the opposing effect of microbial
abundances and nutrient concentrations, temperature appears to be an environmental control over
microbial metabolism, and indirectly, over nutrient concentrations.
The large number of abiotic, microbial, and land use variables emphasized the need to identify
those parameters that are most influential in characterizing each water grab sample. Although
collinearity was used to inform parameter selection, collinear variables were not necessarily
excluded. Surprisingly, five variables over three principal components axes emerged as highly
influential in describing sample variation. The nutrient variables DIN and phosphate comprised
22
axis 1, water turbidity and the percentage of buffer in agricultural use dominated axis 2, and the
abundance of low-DNA bacteria dominated axis 3. These 3 axes explained nearly 83% of the
variation among sites.
Because there are strong pattern differences among the basins for abiotic and biotic variables
(e.g., Figures A-4, A-5 and A-6), the ability of these metrics to discriminate samples by basin
was tested, particularly in the absence of temporal or land use information. The five variables
identified by PCA exhibited mediocre to poor predictive ability for all basins except Admiralty
and Rosario basins (Table A- 7). Addition of six more variables (water temperature; water
conductivity; minimum pH; heterotrophic production; abundance of Synechococcus and small
diatoms) greatly improved predicative ability (Table 6). More than 90% of the samples from the
Rosario and Admiralty basins and more than 80% of the samples from the Whidbey and Hood
Canal basins were correctly assigned. The lowest accuracy was for samples from the Central and
Table 5. Direction ( = positive correlation, = negative correlation) of effects of land use in associated
buffer or catchment (expressed as a percentage or as area), total area of buffer or catchment or total
shoreline length, and physical oceanographic covariates on nutrients, microbial abundance, and two water
quality parameters after factoring for temporal and geomorphic effects. See Table A-6 for full display of
significant variables (p < 0.05) and associated regression coefficients.
Variable Agriculture Developed
Use or
Impervious
Surface
Total Area or
Length
Physical Covariates
DIN (area) (area) (catchment)
(buffer)
(temperature,
precipitation,
turbidity)
Ammonium (percent)
Phosphate (percent) (area) (catchment) (temperature,
turbidity)
Silicic acid (percent) (temperature,
salinity)
High-DNA
bacteria
(percent) (temperature,
turbidity, salinity,
chlorophyll a)
(precipitation)
Low-DNA
bacteria
(area) (area) (temperature)
(precipitation)
Chlorophyll a (catchment)
(shoreline)
(temperature,
turbidity)
Turbidity (area) (area) (buffer)
pH (temperature)
(precipitation)
23
Table 6. Discriminant analysis for basin assignments. Quadratic discriminant analysis using ten abiotic
and biotic variables (DIN; phosphate; turbidity; abundance of low-DNA bacteria, Synechococcus, and
small diatoms; heterotrophic production; conductivity; water temperature; minimum pH).
Figure 9. NMDS plot for surface (horizontal) plankton from individual tows in August, 2011 (data
transformed to presence/ absence due to extremely high and patchy abundances of one taxon).
BasinRosario Basin
Whidbey Basin
Central Basin
Hood Canal
South Sound
Admiralty Basin
2D Stress: 0.25
24
South basins, where 75% or more were correctly assigned. This discriminant analysis
demonstrates that a suite of abiotic and microbial parameters can be used as a relatively accurate
predictor of variation among sites without temporal or geographic information. Application of
such a multivariate tool can have utility in monitoring for changes in water quality across
seasons.
Zooplankton
Analysis of zooplankton comes from both vertical and surface plankton tows. Vertical tows
provided an index of smaller taxa across the depth range, while surface tows represented larger
taxa serving as potential prey in the depth range of fish captured in the large surface trawl.
Taxonomic composition of both vertical and surface plankton samples varied by basin (Figures 9
and 10) and month, although basin differences were stronger (Table 7). Some of the spatial
differences correspond to major environmental gradients, for example the presence of the more
oceanic copepod species Acartia longiremis in the Admiralty and Rosario Basins, and the high
relative abundance of the estuarine species Acartia hudsonica in the more river-dominated
Whidbey basin (Figure 10).
Table 7. Results of two-way ANOVAsfor plankton metrics in both vertical and horizontal tows. For basin
results, pairwise differences show means of basins (R=Rosario, W=Whidbey, A=Admiralty, H=Hood,
C=Central, S=South) ranked lowest (left) to highest. Lines indicate sets of basins that lacked large
pairwise differences in means, and spaces indicate sets with large pairwise differences.
Variable Net type Month Basin Basin differences
Crustacean density Vertical < 0.01 < 0.01 RAHCSW
Non-jelly density < 0.01 < 0.05 AHRSWC
Jellyfish density > 0.1 < 0.001 ARCWH S
Total plankton density > 0.1 < 0.001 SWRCHA
% jellyfish > 0.1 < 0.001 RACWHS
Crustacean density Horizontal < 0.05 < 0.001 WRAH SC
Non-jelly density < 0.01 = 0.001 ARHSWC
Jellyfish density > 0.1 < 0.1 ARCHWS
Total plankton density < 0.05 = 0.001 WARH SC
% jellyfish = 0.001 < 0.001 RACWH S
25
ALL TAXA
COPEPODS ONLY
Figure 10. Taxonomic composition showing all taxa (top) and copepods only (bottom) of vertical
plankton samples from August, 2011.
26
Some relationships were apparent between land use and major components of the plankton
samples. Agricultural land cover was generally positively associated with densities across taxa,
whereas urbanized or impervious land cover tended to be positively associated only with
jellyfish density and relative abundance (Table 8).
While vertical and horizontal nets differed in their potential for capturing zooplankton of
different size due to mesh size and mouth diameters, overall patterns in the datasets were fairly
similar. For example, the rank order of basins in terms of density of non-jellyfish taxa, jellyfish
taxa, and relative abundance of jellyfish, was quite similar between nets (Table 7). One
exception was crustacean taxa, which were much lower in surface trawls in Whidbey Basin,
possibly as a consequence of much lower salinity levels at the surface. Data collected with both
nets exhibited declines in non-jellyfish taxa and total plankton density as a function of
development, and increases in jellyfish density as a function of agriculture. One primary
difference between datasets was that land use metrics tended to explain approximately twice as
much of the total variation for vertical compared to horizontal tows (4.6% vs 2.6%), likely
reflecting greater fluctuations in plankton communities in surface waters.
Fish and jellyfish
Statistical analyses for pelagic macrofauna in townet catches focused primarily on
influences of oceanographic basin and month on patterns of biomass (kg wet weight per tow)
composition. Marked (adipose fin clip or coded wire tag) and unmarked (majority naturally
spawned) Chinook salmon were treated as separate species in the analysis in order to evaluate
similarities and differences between hatchery and wild Chinook. Other hatchery salmonids were
either not distinguishable (no detectable marks) or were rarely caught.
Table 8. Direction ( = positive correlation, = negative correlation) of effects of metrics (area or
percentage (%) of land use in shoreline buffer or catchment (catch)) upon zooplankton variables after
factoring for temporal and basin effects. Black arrows indicate strong (p < 0.05) effects, while gray
arrows indicate weak effects (0.05 < p < 0.1).
Variable Net type Agriculture Development
Crustacean density Vertical (shore %)
Non-jelly density (shore %)
Jellyfish density (shore %) (shore area)
Total plankton density (shore %)
% jellyfish (shore %), (catch%) (shore area)
Crustacean density Horizontal (shore %)
Non-jelly density (shore %)
Jellyfish density (shore %)
Total plankton density (shore %)
% jellyfish (shore area)
27
Relationships between biomass composition and the categorical variables of basin and month
were evaluated with two-way ANOVA for single variables, and with nonparametric multivariate
analyses for whole assemblage data (Clarke 1993, Clarke and Warwick 2001) using Primer-E
software (Clarke and Gorley 2006). For multivariate analysis biomass data were square-root
transformed in the multivariate analysis to down-weight the effect of highly abundant taxa. A
resemblance matrix of all pairwise similarities between sites, based upon the taxa present and
their biomass, was calculated for each sample using the Bray-Curtis distance measure. A
nonparametric multidimensional scaling (NMDS) procedure was applied to the resemblance
matrix to evaluate separation among samples (individual tows). This procedure was repeated for
a simplified data matrix of relative abundance (% of total catch) of each taxon in summed
biomass by basin/month combinations.
A two-way analysis of similarity (ANOSIM) procedure was applied to the resemblance matrix
to evaluate differences in the composition of biomass related to the categorical variables of basin
and month. The ANOSIM procedure calculates an R statistic based upon the difference between
average within-group rank similarities and average among-group rank similarities. Values of R
usually range between 1 (all replicates within areas or months are more similar to each other than
any replicates from different areas or months) and 0 (rank similarities between and within areas
or months are the same, on average) but can be slightly negative. An exact P value was
computed using permutation (999 iterations). Next, a two-way similarity percentages procedure
(SIMPER) was applied to the original, transformed matrix to evaluate the contribution of various
taxa to similarities in the biomass composition by month and basin.
Fish and jellyfish biomass composition varied by basin and month. Jellyfish comprised 86%
of the total wet biomass for all sites and months combined, and the percentage of jellyfish
decreased with increasing latitude (Figure 11). The majority of the biomass was jellyfish in four
of the six basins: 96% in Admiralty Inlet, 96% in Hood Canal, 91% in the Central Basin, and
93% in South Sound. In contrast, jellyfish were 39% in the Rosario basin and 34% in the
Whidbey basin.
At least 35 fish species were captured, 29 of which were identified to species, and at least 10
jellyfish species were captured, 9 of which were identified to species (Table 9). Fish species
richness increased from spring into summer then decreased in late summer (Figure 12). Fish
assemblages at each site were composed of 10 or fewer species per tow, typically averaging 3-6
species in the northern Rosario and Whidbey basins, and 1-2 species in the other basins. Jellyfish
species richness had a similar summer peak, except for in Hood Canal which showed a decline in
jellyfish species richness over the study period. Jellyfish species richness in townet catches was
typically fewer than 5 species per tow, averaging 0-2 in the northern Rosario and Whidbey
Basins, and 2-4 in the other basins. Hatchery Chinook salmon, threespine stickleback, chum
salmon, Pacific herring, and surf smelt were the most common fish species encountered. Most
fish showed strong seasonal patterns in abundance and also somewhat by basin (Figure 13), but
the most obvious difference in fish assemblages among basins was the high percentage of
28
marked (known hatchery origin) Chinook salmon and chum salmon and the low percentage of
herring and other species in all basins except the Whidbey and Rosario (Figure 13).
Sea gooseberry ctenophores, water jellyfish, cross jellyfish, lion’s mane jellyfish, and moon
jellyfish were the most commonly captured, and showed species differences in seasonal
abundance. Spatial differences were also apparent (Figure 14). Examination of fish and jellyfish
biomass composition using ANOVA (Table 10) and NMDS (Figure 15) showed clear statistical
differences based on oceanographic basin and month. ANOSIM tests for differences in biomass
composition among basins (global R = 0.41; Table 11) and months (global R = 0.32, results not
shown) both produced clear differences. The greatest pairwise spatial differences generally
corresponded to geographic distance between basins (Tables 10 and 11).
Figure 11. Percent jellyfish in summed biomass (wet weight per tow, all months combined) for all sites
by oceanographic basin.
0
20
40
60
80
100
47 47.5 48 48.5 49
% je
llyfi
sh b
y b
iom
ass
Latitude
Admiralty
Central
Hood Canal
Rosario
South Sound
Whidbey
29
Table 9. Fish (bold) and jellyfish species captured in surface trawls from greater Puget Sound from April
through October, 2011, ranked in order of frequency of occurrence.
Species % Frequency
sea gooseberry Pleurobranchia bachei 58.1
water jelly Aequorea sp. 57.0
Chinook (marked) Oncorhynchus tshawytscha 45.9
three-spined stickleback Gasterosteus aculeatus aculeatus 43.5
cross jelly Mitrocoma sp. 34.8
Chum Oncorhynchus keta 32.2
Chinook (unmarked) Oncorhynchus tshawytscha 30.1
Pacific herring Clupea pallasii pallasii 29.1
lions mane jelly Cyanea capillata 24.4
surf smelt Hypomesus pretiosus 22.2
moon jelly Aurelia sp. 13.5
coho Oncorhynchus kisutch 11.5
unidentifed jellyfish Cnidaria sp. 8.7
Clytia jelly Clytia sp. 7.3
fried egg jelly Phacellophora camschatica 7.1
bay pipefish Syngnathus leptorhynchus 6.9
river lamprey Lampetra ayresii 5.8
Pacific sand lance Ammodytes hexapterus 5.6
northern anchovy Engraulis mordax 5.1
starry flounder Platichthys stellatus 4.4
shiner perch Cymatogaster aggregata 2.6
sockeye Oncorhynchus nerka 1.8
unidentified gadoid Gadidae spp. 1.6
steelhead Oncorhynchus mykiss 1.3
Pacific sand fish Trichodon trichodon 1.3
unidentified flatfish Pleuronectid sp. 1.1
unidentified larval fish Osteichthyes sp. 0.9
spiny lumpsucker Eumicrotremus orbis 0.9
unidentified poacher Agonidae 0.7
plainfin midshipman Porichthys notatus 0.5
unidentified fish Osteichthyes sp. 0.5
soft sculpin Psychrolutes sigalutes 0.5
American shad Alosa sapidissima 0.5
tubesnout Aulorhynchus flavidus 0.5
unidentified sculpin Cottoid sp. 0.4
grunt sculpin Rhamphocottus richardsonii 0.4
beroe ctenophore Beroe sp. 0.4
pile perch Rhacochilus vacca 0.4
sailfin sculpin Nautichthys oculofasciatus 0.2
polyorchis Polyorchis sp. 0.2
snake prickleback Lumpenus sagitta 0.2
Pacific tomcod Microgadus proximus 0.2
kelp perch Brachyistius frenatus 0.2
silver spot sculpin Blepsias cirrhosus 0.2
striped surf perch Embiotoca lateralis 0.2
30
Contributions of various taxa influencing the statistical differences by month and basin were
also clear (Table 12). Biomass of jellyfish accounted for considerable statistical within-group
similarity across basins and months, but was most dominant in Admiralty Inlet, Hood Canal, the
Central Basin, and South Sound (Table 12). Hatchery Chinook and chum salmon were the
statistically dominant fish in these areas. In contrast, jellyfish were less statistically dominant in
the northern Rosario and Whidbey basins, where three-spine stickleback where the statistically
dominant species overall, and other forage fish species (surf smelt and herring were also
important (Table 12).
Table 10. Significance values of two-way ANOVA results for fish and jellyfish variables. For basin
results, pairwise differences show means of basins (R=Rosario, W=Whidbey, A=Admiralty, H=Hood,
C=Central, S=South) ranked lowest (left) to highest, with spaces and lack of connecting grouping lines
indicating large differences among basins.
Variable Month Basin Basin differences
Salmon biomass < 0.001 = 0.001 RWHACS
Forage fish biomass < 0.05 = 0.001 AHCSWR
Jellyfish biomass > 0.1 < 0.001 WRAHC S
Total biomass > 0.1 < 0.001 AWRHC S
Fish species richness < 0.001 < 0.001 SAHC RW
% jellyfish < 0.05 < 0.001 WR HASC
Figure 12. Mean number of species per tow for fish (A) and jellyfish (B) by month and basin in 2011.
A B
31
Figure 13. Seasonal biomass of seven most commonly occurring fish species captured in surface trawls in
the six oceanographic basins of Puget Sound in 2011. Note differences in scales.
32
Figure 14. Seasonal biomass of five most commonly occurring jellyfish species captured in surface
trawls in the six oceanographic basins of Puget Sound in 2011.
33
Figure 15. NMDS plots for individual tows (top) and summed by month and basin (bottom, numbers next
to points indicate month) of biomass of fish and jellyfish species.
34
Table 11. ANOSIM R statistics for comparisons between basins. Two-way tests for differences in total biomass composition (wet weight per tow)
across all month groups; 999 permutations. Differences are ranked from most to least different.
Between Group Comparison Ra = 0.41 (P = 0.001) P
South Sound vs. Rosario 0.667 0.001
Central vs. Rosario 0.640 0.001
Hood Canal vs. Rosario 0.600 0.001
South Sound vs. Whidbey 0.516 0.001
Admiralty vs. Rosario 0.502 0.001
Hood Canal vs. Central 0.496 0.001
Central vs. Whidbey 0.428 0.001
Hood Canal vs. Whidbey 0.395 0.001
Hood Canal vs. South Sound 0.360 0.001
Admiralty vs. Hood Canal 0.349 0.001
Admiralty vs. Whidbey 0.300 0.001
Admiralty vs. South Sound 0.297 0.001
Whidbey vs. Rosario 0.207 0.001
Admiralty vs. Central 0.182 0.001
Central vs. South Sound 0.147 0.001 a
Global R statistic for overall differences
35
Table 12. Average statistical similarity of biomass composition among sites within each basin, and ranked similarity percentages of species
contributing 90 % to the similarity within each basin (two-way SIMPER test adjusted for month effect). Perfect similarity is 100, no similarity is 0.
UM = unmarked; M = marked. Fish are listed in bold type.
Rosario
Average similarity = 37
Whidbey
Average similarity = 38
Admiralty
Average similarity = 34
Hood Canal
Average similarity = 38
Central
Average similarity = 31
South
Average similarity = 21
Taxon Contrib.
%
Taxon Contrib.
%
Taxon Contrib.
%
Taxon Contrib.
%
Taxon Contrib. % Taxon Contrib.
%
stickleback 36 stickleback 25 water jelly 44 water jelly 40 water jelly 33 water jelly 42
sea
gooseberry
27 surf smelt 16 sea
gooseberry
16 Chinook
M
15 sea
gooseberry
21 sea
gooseberry
20
cross jelly 9 water jelly 9 cross jelly 16 lions mane
jelly
15 cross jelly 19 Chinook
M
9
herring 8 chum 9 chum 14 sea
gooseberry
13 Chinook
M
8 lions mane
jelly
8
Chinook M 5 herring 8 Chinook
M
5 stickleback 6 chum 8 chum 7
Chinook
UM
4 Chinook M 7 moon jelly 3 moon jelly 5
surf smelt 2 sea
gooseberry
7
lions mane
jelly
6
ChinookUM 5
36
Table 13. Direction ( = positive correlation, = negative correlation) of effects of metrics (area or
percentage (%) of land use in shoreline buffer) upon zooplankton variables after factoring for temporal
and basin effects.
Variable Agriculture Development
Salmon biomass
Forage fish biomass (shore area)
Jellyfish biomass
Total biomass (shore %)
Fish species richness (shore area) (shore %)
% jellyfish
Adjacent land use was not strongly associated with major components of the townet biomass
(Table 13). Exceptions were a positive relationship between agricultural use and forage fish
biomass, negative relationships between agricultural use and fish species richness, and negative
relationships between development and both total biomass and fish species richness.
Stable isotopes of fish and jellyfish
Stable isotopes of Carbon (12/13
C) and Nitrogen (14/15
N) have been widely applied in food
web studies and provide a powerful tool to quantify trophic relationships and nutrient dynamics
in both aquatic and terrestrial ecosystems (Post 2002; Peterson and Fry 1987). Stable isotopes are
an important component to two of the primary goals of our study. Given that nutrient inputs are
a primary mechanism by which land use may affect food webs, stable isotopes are a potentially
important metric of land use signals on pelagic food webs as isotopic enrichment are a powerful
tracer of anthropogenic nutrients in aquatic ecosystems (McClelland et al. 1999). Second, stable
isotopes contribute important information to rigorously test the bifurcated food web hypothesis
as isotopic compositions of fish and jellyfish can explicitly quantify the degree of trophic
overlap, potentially providing support for distinct trophic pathways leading to dominance of
jellyfish over finfish in Puget Sound.
Additionally, in the Puget Sound, there are several aspects of pelagic trophic dynamics which
remain poorly understood, including the trophic relationships among various species of forage
fish, salmonids and jellyfish, how these relationships vary spatially across oceanographic basins,
seasonally, and across different shoreline types with varying land use. In addition to broad scale
patterns, species-specific trophic dynamics are also of interest. For example, our understanding
of the seasonal contribution of terrestrially derived Carbon sources (i.e. terrestrial invertebrates)
to diets of endangered Chinook salmon is lacking despite ample evidence that terrestrial insects
are an important diet source for juveniles (Duffy et al. 2010; Romanuk and Levings 2005).
37
Chinook showed the greatest range of isotopic enrichment and tended to be more depleted in 13
C and more enriched in 15
N than other species (Figure 16). Examining Chinook in detail
(Figure 17) shows most individuals were enriched in 13
C relative to the reference point of
synthetic fish meal (Trueman et al. 2005) and significantly enriched in 13
C and 15
N relative to
terrestrial insects (Romanuk and Levings 2005). Chum had a similar isotopic composition to the
reference point of gelatinous zooplankton (Arai et al. 2003) but were slightly more enriched in 13
C (Figure 17). Both jellyfish species showed similar isotopic composition to chum and were
more enriched in 13
C and more depleted in 15
N than Chinook (Figure 16). Strong spatial patterns
were evident for all species (Figures 18 and 19) with ctenophores showing the greatest
differences among basins (Figure 19). Spatial patterns were not consistent across species
however; for example, ctenophores were significantly more depleted in 13
C in some basins while
chum were more enriched. Overall latitudinal patterns of isotopic composition were not evident.
Isotopic enrichment for each of the four most common species was correlated with several
metrics of shoreline land use including the percent of developed, agricultural and impervious
surfaces (Table 14).
These preliminary results suggest pelagic trophic dynamics differ spatially within Puget
Sound and that land use may be an influential factor. Chinook and chum appear to occupy
distinct trophic positions. Welch and Parsons (2007) observed similar patterns in 15
N enrichment
between Chinook and chum in the Gulf of Alaska but the reverse trend in 13
C. The reason for this
contrast is unclear but greater availability of terrestrially derived Carbon sources in the Puget
Sound as well as potential hatchery influences (described below) are possible explanations.
Isotopes appear to show a strong signal of shoreline land use (Table 14). It is especially
notable that 15
N responded positively to metrics of development in fish and negatively in
jellyfish, suggesting that finfish feed at a higher trophic level and jellyfish at a lower trophic
level in more developed areas, a result that generally supports the bifurcated food web
hypothesis. However, these results need to be tempered by several considerations. First, isotopic
enrichment of Chinook salmon in June is likely strongly influenced by hatchery effects. That is,
the isotopic composition of hatchery feed which is generally more enriched in 15
N and depleted
in 13
C relative to other prey sources in the environment. Given the large hatchery influence in
South and Central basins, these artificial inflations of isotopic enrichment may skew
comparisons among basins with varying degrees of hatchery influence. Additional analysis
incorporating the origin (wild or hatchery) of each individual is required to account for this
effect. Second, intrinsic abiotic differences among basins such as water residence time and the
influence of freshwater may create shifting isotopic baselines (sensu Peterson and Fry 2007),
rendering analysis across basins problematic. Additional consideration of these factors,
identification of stable isotope values for various trophic components (e.g., microplankton,
various crustacean taxa) as well as analysis of land use patterns within basins will address this
caveat.
Overall, stable isotopes appear a promising tool to examine the effects of land use on pelagic
food webs and characterize pelagic food web structure across the Puget Sound. Further analysis
38
incorporating our full sample set over the course of the seven month time series will provide
critical information to meet the objectives of our study. Additionally, stand-alone results from
this component will contribute important information to our understanding of pelagic trophic
dynamics in the Puget Sound.
Table 14. Pearson correlation coefficients between shoreline land use metrics and isotopic enrichment the
four most abundant species in June. Strong correlations (p < 0.05) are in bold.
Shoreline land
use
% Developed % Impervious Surface % Agriculture
Chinook δ13
C -0.25 -0.28 0.07
Chinook δ15
N 0.36 0.37 -0.11
Chum δ13
C -0.12 -0.24 0.15
Chum δ15
N 0.33 0.31 0.13
Ctenophore δ13
C 0.17 0.12 -0.16
Ctenophore δ15
N -0.53 -0.50 0.04
Water jelly δ13
C 0.09 -0.15 -0.12
Water jelly δ15
N -0.59 -0.61 0.02
Figure 16. Plots of isotopic composition of the four most abundantly captured species in June 2011. Data
are combined for all basins. Each point represents an individual fish or the composite of jellyfish captured
at a given site.
39
Figure 17. Isotopic plots of Chinook salmon (top panel) and chum salmon (bottom panel) aggregated by
basin. Literature values for fish meal, a typical food source in hatcheries (Trueman et al. 2005) as well as
mean value for terrestrial insects and marine invertebrates (Romanuk and Levings 2005) and herring (this
study) are also included as references points for potential diet components. Chum are known to consume
gelatinous zooplankton (Arai et al. 2003) so a composite mean value of several species of ctenophora and
cnidaria was calculated from our data and added as a reference point. All boxed symbols indicate stable
isotope reference points.
40
Figure 18. Box plots of median δ13
C and δ15
N values for chum and Chinook for each basin in Puget
Sound.
Figure 19. Box plots of median δ13
C and δ15
N values of gelatinous zooplankton for each basin in Puget
Sound. Data for jellyfish was only available for three of the six sampled basins in June 2011
41
SYNTHESIS
Basin differences
We found that Puget Sound’s oceanographic basins were significantly different from each
other both in quantitative measures & temporal patterns for a number of abiotic and biotic
metrics. We summarize these differences in Table 15, which shows values of metrics with a red,
yellow, or green shading. These ranks reflect scoring compared to reference conditions
determined either by a relative comparison (comparison among basin means or examination of
observations greater than 95% or less than 5% of all observations) or an absolute reference
(comparison to a biologically important reference state), and the criterion for each is indicated in
the table. This list of potential indicators includes abiotic factors, primary and microbial
production, zooplankton, and fish. The matrix of potential indicators should be considered a first
step in the process of validation of good indicators of pelagic ecosystem health. Notably, we
should strive to improve reference points from relative to absolute or historical references,
thereby providing benchmarks that reflect current status of ecological function or changes from a
baseline state, respectively. Our data were collected in a single year, and therefore do not reflect
annual patterns and temporal trends. However, the broad patterns of fish and jellyfish catches
observed in 2011 were similar to 2003 (Rice et al. 2012). In addition, some potential indicators
are exploratory at this point (e.g., picophytoplankton and small diatoms), and other metrics may
be lower risk than suggested by the color coding. For example, turbidity levels that are stressful
to fish are >10 BTU (Newcombe and Jensen 1996), a threshold that no measurement in our
dataset surpassed. Nevertheless, the metrics used in Table 15 were robust predictors of
differences among basins in Puget Sound, and many also were sensitive to measures of land use
(see below). Metrics describing similar aspects of the foodweb (e.g., abiotic conditions, bacteria,
phytoplankton , zooplankton, fish) were often concordant within their grouping, but were not
necessarily predictive of other aspects of the foodweb. Hence, measuring water quality
parameters will not likely substitute for measuring zooplankton or fish species.
The overall pattern of metrics suggests that the “healthiest” basins may be Rosario and
Whidbey Basins. These were characterized by relatively few abiotic or nutrient extremes, few
deviations in the abundance of different groups of microbes and phytoplankton, relatively high
densities of non-gelatinous zooplankton, and high fish species richness and relatively high forage
fish abundance. Rosario Basin had just a few metrics at medium levels and no metrics in “poor”
condition, while Whidbey Basin had a few abiotic/nutrient metrics at medium levels and just one
metric in poor condition. As the healthiest of Puget Sound’s oceanographic basins, these places
can benefit other regions in Puget Sound by additional protection and restoration actions. The
catchments supplying water to these basins are the centers of agriculture for Puget Sound, and
efforts to reduce agricultural-related impacts (e.g., nutrient loading) and improve estuarine and
nearshore habitats (e.g., reduced diking and shoreline hardening) will likely have the biggest
benefits to these basins.
42
Table 15. Potential indicators of ecological health in the six basins of Puget Sound. Values listed as criteria represent the 5th (when % visits ≤) or
95th (when % visites ≥) percentile for the indicator. "V" indicates vertical plankton tow, "S" indicates surface plankton tow.
Indicator Reference
type Criterion Rosario Whidbey Admiralty Central South Hood Canal
Temperature Relative % visits > 14.18°C 1% 1% 0% 2% 16% 9%
Dissolved oxygen Absolute % obs. ≤ 5 mg L-1 7% 4% 0% 0% 0% 19%
pH Absolute % obs. ≤ 7.75 13% 16% 25% 20% 11% 43%
Turbidity Relative % visits ≥ 2.72 5% 17% 0% 1% 0% 3%
DIN Absolute % visits > 16 µM 33% 51% 86% 62% 50% 37%
NH4 Relative % visits ≥ 3.04 2% 3% 2% 8% 12% 1%
Si(OH)4 Relative % visits ≥ 63.44 0% 1% 2% 11% 11% 4%
SI:DIN Relative Comparison of means 0.37 0.31 0.44 0.35 0.28 0.14
Low-DNA bacteria Relative % visits ≥ 854,024 1% 0% 0% 2% 2% 24%
High-DNA bacteria Relative % visits ≤ 528,745 7% 0% 0% 16% 2% 4%
Relative % visits ≥ 2,701,463 5% 1% 0% 2% 13% 7%
Chlorophyll a Relative % visits ≤ 0.83 2% 3% 5% 11% 3% 9%
Relative % visits ≥ 18.18 5% 1% 0% 2% 13% 7%
Synechococcus Relative % visits ≥ 77,767 0% 0% 0% 0% 0% 30%
Picophytoplankton Relative % visits ≥ 47,253 0% 3% 0% 1% 2% 23%
Small diatoms Relative % visits ≥ 7,400 1% 1% 0% 0% 6% 21%
Non-jellyfish density (V) Relative Comparison of means 1,484 2,197 2,278 1,702 1,500 1,757
Non-jellyfish density (S) Relative Comparison of means 118 154 321 257 341 64
% jellyfish (V) Relative Comparison of means 4% 7% 1% 0% 9% 3%
% jellyfish (S) Relative Comparison of means 3% 21% 2% 6% 41% 13%
% jellies (trawl) Relative Comparison of means 38% 30% 78% 84% 83% 73%
Forage fish biomass Relative Comparison of means 1,851 1,304 23 85 602 79
Fish species richness Relative % visits ≤ 2 species 49% 46% 77% 63% 74% 76%
43
In contrast, Hood Canal and South Sound were rated the lowest in our system. Hood Canal is
the most different from the other five basins. Dissolved oxygen and pH were lowest most
frequently & displayed the highest variance among Hood Canal sites. Likewise, nitrate and
nitrite concentrations were most frequently the lowest in Puget Sound. Possibly related to these
phenomena, abundances of specific phytoplankton (picophytoplankton, small diatoms), and
cyanobacteria (Synechococcus), and low-DNA bacteria were 5-10 times higher in later months
(August through October) than in other basins. The combination of these features suggests that
Hood Canal experiences a late-season rise in primary production from phytoplankton &
cyanobacteria. The strong presence of low-DNA bacteria suggests differences in bacterial taxa
or a different metabolic state of bacteria in Hood Canal. Hood Canal was also characterized by
relatively high relative abundances of jellyfish, and corresponding low abundance of
nongelatinous zooplankton, as well as low species richness and forage fish abundance. As has
been summarized recently by EPA and the Department of Ecology, Hood Canal is naturally
challenged by its unique geography and oceanography, and a recent report determined that it is
premature to assign all these problems to anthropogenic activities (Kope and Roberts 2012).
South Sound also rated relatively poorly, and was distinguished by several unique features.
Water temperatures were most frequently highest at South Sound sites, and exhibited the largest
variance among all of the basins. Ammonium & silicic acid concentrations were also most
frequently highest at South Sound sites. The highest concentrations of high-DNA bacteria and
chlorophyll a were found most frequently at South Sound sites. These features suggest that South
Sound sites experience elevated primary and heterotrophic production, possibly fueled by
ammonium and enhanced by warmer temperatures. In turn, relative abundances of jellyfish were
the highest found in Puget Sound, and fish species richness was dominated by hatchery salmon.
Forage fish were also relatively low in abundance.
Admiralty Inlet and the Central Basin appear to be at medium levels of ecosystem health.
Admiralty Inlet, the source of water for most of Puget Sound, was characterized by lowest and
relatively homogeneous water temperatures, relatively small variation in pH and dissolved
oxygen, relatively low turbidity, and low nutrient & microbial levels. Nevertheless, adult
jellyfish were at relatively high levels, and fish species richness and forage fish abundance were
low compared to other basins in Puget Sound. Admiralty Inlet’s water directly feeds the Central
Basin. Although water temperatures here did not tend toward higher values (i.e., > 14.2°C), the
variance in temperatures was second only to those for South Sound. DIN was the highest most
frequently at Central basin sites, and silicic acid levels were frequently the highest (second to
South basin). The lowest levels of chlorophyll a and high-DNA bacteria occurred most
frequently among Central basin sites. Hence, although nutrient levels do not appear limiting,
primary and heterotrophic production appear to be lower than expected. Like Admiralty Inlet, the
Central Basin features a high relative abundance of adult jellyfish, even though planktonic
jellyfish were measured at low levels. Similarly, forage fish abundance and fish species richness
were quite low, indicative of a dominant hatchery salmon component to the fish catch.
44
Sensitivity of indicators to land use
Another way to examine the utility of potential metrics is to ask which metrics are most
sensitive to land use. Of the metrics we evaluated, 12 of the 21 potential indicators exhibited
strong (p < 0.05) patterns with land use, and another seven metrics exhibited weak patterns (0.1
< p < 0.05). Even among the strong patterns, the explanatory power of land use was relatively
low. For example, land use effects on abiotic and microbial features, after accounting for basin &
temporal effects, were not strong (i.e., the proportion of the total model sum of squares due to
land use was < 5%). This finding suggests that habitat quality is strongly influenced by
oceanographic influences (Kim and Khangaonkar 2012) that can buffer Puget Sound’s pelagic
ecosystem from land use impacts.
Even so, several land use patterns warrant particular attention. Increased agricultural land use
significantly increased turbidity, the abundance of high-DNA bacteria, the abundance of
picophytoplankton, and silicic acid concentrations, while decreasing dissolved oxygen minima
Table 16. Effects of agriculture and development on various indicators. Upward pointing arrows indicate
positive effects after controlling for basin and month patterns, while downward pointing arrows indicate
negative correlations. Black arrows indicate strong effects (p < 0.05), while gray arrows indicate weak
effects (0.1 < p < 0.05).
Indicators Agriculture Development
Temperature
Dissolved oxygen
pH
Turbidity
DIN
NH4
Si(OH)4
Si:DIN
Low-DNA bacteria
High-DNA bacteria
Chlorophyll a
Synechococcus
Picophytoplankton
Small diatoms
Non-jellyfish density (V)
Non-jellyfish density (S)
% jellyfish (V)
% jellyfish (S)
% jellyfish (trawl)
Forage fish biomass Fish species richness
45
and pH. This suggests that agricultural uses may contribute to greater heterotrophic microbial
activity without necessarily increasing primary production. Likewise, increased developed land
use and its associated greater impervious surface significantly increased chlorophyll a
concentrations, silicic acid concentrations, and water temperature, while decreasing dissolved
inorganic nitrogen, abundance of both high and low-DNA bacteria, and abundance of
Synechococcus. This suggests that development and impervious surfaces may contribute to
primary production without boosting heterotrophic microbial activity. In addition, non-gelatinous
plankton taxa decreased as a function of development while the percentage of planktonic
jellyfish increased, a pattern consistent with the bifurcated foodweb hypothesis (Parsons and
Lalli 2002).
Several metrics stand out as being particular useful for the examining both differences among
basins and land use effects. We found that pH, dissolved oxygen, and high-DNA bacteria all
exhibited variation across Puget Sound’s basins, and were linked to agriculture. Likewise, we
found Si:DIN, chorophyll a as well as non-gelatinous zooplankton density and relative
abundance of planktonic jellies in the vertical net were both variable across basins and sensitive
to local development. Further monitoring of these metrics stands to improve our understanding
of how different types of land use influences pelagic foodwebs and results in differences across
the basins of Puget Sound.
We found strong differences among basins in abundance and composition of zooplankton and
fish. These taxa are consumed by salmon and other recreationally and commercially important
fish as well as marine birds and mammals. Given their importance in foodweb dynamics (Smith
et al. 2011), we believe these taxa will be critical for determining the long-term health of Puget
Sound’s pelagic zone. Therefore, we strongly support continued efforts to monitor these taxa as
indicators of ecological health at the basin level. Because fish are potentially highly mobile,
effects of land use are likely to be difficult to directly measure, and this prediction was supported
by relatively low correlations of data with land use metrics.
Evidence for the bifurcated foodweb hypothesis
Parsons and Lalli (2002) predicted that land use influences on nutrients and water quality
simplify foodweb pathways, resulting in foodweb shifts from diatom-driven primary production
to production by cyanobacteria and dinoflagellates. These shifts would be expected to affect their
consumers, resulting in shifts from crustacean zooplankton to microplankton and jellyfish taxa.
The consequences of these changes across Puget Sound is a bifurcated foodweb: jellyfish-
dominated systems in areas of high land use and fish-dominated systems in more pristine areas.
We found several lines of evidence to support the bifurcated foodweb hypothesis. Higher levels
of land use, particularly agriculture, were correlated with abiotic conditions (temperature, pH,
and dissolved oxygen) favoring simpler pathways. Likewise, the relative abundance of jellyfish
captured in vertical hauls increased as a function of land use. Furthermore, nitrogen enrichment
(a measure of trophic level position) measured in jellyfish strongly declined as a function of land
use to more basal levels, suggesting that jellyfish diets trend toward lower-trophic organisms in
46
areas of high land use. We also found several lines of evidence that did not support the
hypothesis. For example, nutrients and simpler autrotrophs did not seem strongly affected by
land use metrics. However, it is possible these patterns are a consequence of foodweb
interactions, and that a reverse pattern would be detected once removals by other organisms were
taken into account. Future modeling efforts will be used to evaluate these possibilities.
Implications
Our broad sampling of Puget Sound’s nearshore pelagic foodweb revealed 1) strong
differences in numerous attributes among the oceanographic basins, 2) small but detectable
effects of land use on these attributes, and 3) suggestions of cumulative effects of land use
leading to bifurcation of foodweb pathways that influence secondary consumers, such as fish and
jellyfish. Our study suggests that activity by microbial primary producers and heterotrophs may
influence dissolved nutrient levels, in addition to being regulated by those nutrients. Microbes
may be able to buffer against nutrient loads from anthropogenic activities, but the capacity limits
in Puget Sound are unknown. The strong correlation observed between pH and dissolved
oxygen (Fig. 6) suggests an influence from heterotrophic respiration, and that increased
microbial respiration could depress both pH and dissolved oxygen.
From an applied perspective, our findings indicate that we should expect different ecological
baselines, environmental drivers, and susceptibilities of the pelagic foodweb among Puget
Sound’s oceanographic basins. This suggests that informative indicators of ecological status are
different among the basins. Consequently, a long-term monitoring scheme that incorporates
multiple foodweb attributes across all the main basins will provide a much more informed basis
for assessing changes in ecological health than a focus on a few variables measured in a single
basin. A sampling scheme that leverages existing broad-scale efforts (e.g., Department of
Ecology’s water quality monitoring program), and involves multiple partners (e.g., tribes, USGS,
NOAA Fisheries) measuring similar attributes could make long-term assessment of multiple
trophic levels feasible.
47
REFERENCES
Arai, M. N., D.W.Welch, A.L.Dunsmuir, M. C. Jacobs, and A.R. Ladouceur. 2003. Digestion of
pelagic ctenophora and cnidaria by fish. Canadian Journal of Fisheries and Aquatic
Sciences 60: 825-835.
Arrigo, K.R. 2005. Marine microorganisms and global nutrient cycles. Nature 437:349-355.
Anderson, D.M., P.M. Glibert, and J.M. Burkholder. 2002. Harmful algal blooms and
eutrophication: nutrient sources, composition, and consequences. Estuaries 25: 704-726.
Burns, D.A. and C. Kendall. 2002. Analysis of delta N-15 and delta O-18 to differentiate NO3-
sources in runoff at two watersheds in the Catskill Mountains of New York. Water
Resources Research 38: 1051.
Casciotti, K., D. Sigman, M. Galanter-Hastings, J.Bohlke, and A. Hilkbert. 2002. A bacterial
method for the measurement of the oxygen isotope composition in marine and fresh
waters. Analytical Chemistry 74: 4905-4912.
Chang, C.C.Y., C. Kendall, and S.R. Silva. 2002. Nitrate stable isotopes: tools for determining
nitrate sources among different land uses in the Mississippi River Basin. Canadian
Journal of Fisheries and Aquatic Science 59: 1874–1885.
Clarke, K. R. 1993. Non-parametric multivariate analysis of changes in community structure.
Aust J Ecol 18:117-143.
Clarke, K. R., and R. N. Gorley. 2006. Primer v6: user manual/tutorial. PRIMER-E, Plymouth,
UK.
Clarke, K. R., and R. M. Warwick. 2001. Change in marine communities: an approach to
statistical analysis and interpretation, 2nd edition. PRIMER-E, Plymouth, UK.
Cope B. and M. Roberts. 2012. Review and synthesis of available information to estimate human
impacts to dissolved oxygen in Hood Canal. Draft Report by US EPA and WA
Department of Ecology. 74 pp.
Dahnke K., E. Bahlmann, and K. Emeis. 2008. A nitrate sink in estuaries? An assessment by
means of stable nitrate isotopes in the Elbe estuary. Limnology and Oceanography 53:
1504-1511.
Duffy, E.J., D.A. Beauchamp, R.M. Sweeting, R.J. Beamish, and J.S. Brennan. 2010.
Ontogenetic diet shifts of juvenile Chinook salmon in nearshore and offshore habitats of
Puget Sound. Transactions of the American Fisheries Society 139: 803-823.
48
Hay, D.E., P.B. McCarter, and K.S. Daniel. 2001. Tagging of Pacific herring Clupea pallasi
from 1936–1992: a review with comments on homing, geographic fidelity, and straying.
Canadian Journal of Fisheries and Aquatic Science 58: 1356–1370.
Kendall, C., E.M. Elliott, and S.D. Wankel. 2007. Tracing anthropogenic inputs of nitrogen to
ecosystems, Chapter 12, In: R.H. Michener and K. Lajtha (Eds.), Stable Isotopes in
Ecology and Environmental Science, 2nd edition, Blackwell Publishing, p. 375-449.
Kim, T., and T. Khangaonkar. 2012. An offline unstructured biogeochemical model (UBM) for
complex estuarine and coastal environments. Environmental Modelling and Software 31:
47-63.
Longnecker, K., B.F. Sherr, and E.B. Sherr. 2006. Variation in cell-specific rates of leucine and
thymidine incorporation by high and low nucleic acid content marine bacteria off the
Oregon coast. Aquatic Microbial Ecology 43: 113-125.
McClelland, J.W., I. Valiela, and R.H. Michener. 1997. Nitrogen-stable isotope signatures in
estuarine food webs: a record of increasing urbanization in coastal watersheds.
Limnology and Oceanography 42: 930–937.
Myers, R.A., S.A. Levin, R. Lande, F.C. James, W.W. Murdoch, and R.T. Paine. 2004.
Hatcheries and endangered salmon. Science 303: 1980.
Newcombe, C. P., and J. O. T. Jensen. 1996. Channel suspended sediment and fisheries: a
synthesis for quantitative assessment of risk and impact. North American Journal of
Fisheries Management 16: 693-727.
Parsons, T.R., and C.M. Lalli. 2002. Jellyfish population explosions: Revisiting a hypothesis of
possible causes. La Mer 40: 111-121.
Parsons, T.R., L.J. Albright, F.Whitney, P.J. Leb Williams, and C.S. Wang. 1981. The effect of
glucose on the productivity of seawater: An experimental approach using controlled
aquatic ecosystems. Marine Environmental Research 4: 229-242.
Peterson, B. and B. Fry. 1987. Stable isotopes in ecosystem studies. Annual Review of Ecology
and Systematics 18: 293–320.
Post, D. M. 2002. Using stable isotope methods to estimate trophic position: models, methods,
and assumptions. Ecology 83: 703–718.
Puget Sound Partnership (PSP). 2010. 2009 State of the Sound Report. Puget Sound
Partnership, Olympia, WA.
Rice, C.A. (2007). Evaluating the biological condition of Puget Sound. PhD thesis, University of
Washington.
http://www.fish.washington.edu/research/publications/ms_phd/Rice_C_PhD_Su07.pdf
49
Rice, C.A. J.J. Duda, C.M. Greene,and J. R. Karr (2012): Geographic Patterns of Fishes and
Jellyfish in Puget Sound Surface Waters, Marine and Coastal Fisheries: Dynamics,
Management, and Ecosystem Science 4: 117-128.
Romanuk T.N. and C.D. Levings. 2005. Stable isotope analysis of trophic position and terrestrial
vs. marine carbon sources for juvenile Pacific salmonids in nearshore marine habitats.
Fisheries Management and Ecology 12: 113-121.
Shipman, H. 2008. A Geomorphic Classification of Puget Sound Nearshore Landforms. Puget
Sound Nearshore Partnership Report No. 2008-01. Published by Seattle District, U.S.
Army Corps of Engineers, Seattle, Washington.
Sigman, D.M., K.L. Casciotti, M. Andreani, C. Barford, M. Galanter, and J.K. Böhlke. 2001. A
bacterial method for the nitrogen isotopic analysis of nitrate in seawater and freshwater.
Analytical Chemistry 73: 4145–4153.
Smith, A.D.M., C.J. Brown, C.M. Bulman, E.A. Fulton, P. Johnson, I.C. Kaplan, H.Lozano-
Montes, S. Mackinson, M. Marzloff, L.J. Shannon, Y. Shin, and J. Tam. 2011. Impacts of
Fishing Low–Trophic Level Species on Marine Ecosystems. Science 333: 1147-1150.
Strickland, R.M. 1983. The Fertile Fjord. Seattle, University of Washington Press.
Trueman, C. N., R. A. R. Mcgill, and P. H. Guyard. 2005. The effect of growth rate on tissue-
diet isotopic spacing in rapidly growing animals. An experimental study with Atlantic
salmon (Salmo salar). Rapid Communications in Mass Spectrometry 19: 3239–3247.
Wankel, S. D., C. Kendall, C. A. Francis, and A. Paytan .2006. Nitrogen sources and cycling in
the San Francisco Bay Estuary: A nitrate dual isotopic composition approach, Limnology
and Oceanography 51: 1654–1664.
Welch, D.W. and T.R. Parsons.1993. Delta super (13) C- delta super (15) N values as indicators
of trophic position and competitive overlap for Pacific salmon (Oncorhynchus spp.).
Fisheries Oceanography. Fisheries Oceanography 2: 11-23.
50
APPENDIX List of Supplemental Figures and Tables
Figure A- 1. Box and whisker plot of DO concentrations at 0.5 meters depth by month and
basin) .............................................................................................................................................51
Figure A- 2. Box and whisker plot of DO concentrations at 6.0 meters depth by month and
basin) .............................................................................................................................................51
Figure A- 3. Water column profile at site E00A, the southernmost Hood Canal basin site, in
October (top) and April (bottom) ..................................................................................................52
Figure A- 4. Box plots of dissolved inorganic nutrients by month and basin ..............................53
Table A- 1. Pairwise correlations (Pearsons r) between dissolved inorganic nutrient
concentrations ...............................................................................................................................54
Table A- 2. Correlations between chlorophyll a concentrations and picophytoplankton or small
diatom abundance by month ..........................................................................................................54
Table A- 3. Summary of linear regressions of chlorophyll a concentrations (“chlorophyll”) and
high-DNA bacterial abundance (“bacteria”) on nutrient concentrations by basin .......................54
Figure A- 5. Box plots of chlorophyll a concentrations, picophytoplankton abundance, and small
diatom abundance by month and basin ..........................................................................................55
Figure A- 6. Box plots of high-DNA and low-DNA bacterial abundance and Synechococcus
abundance by month and basin ......................................................................................................56
Table A- 4. Variables in the canonical correspondence analysis (CCA) of bacterial communities
........................................................................................................................................................57
Figure A- 5. Box plots of chlorophyll a concentrations, picophytoplankton abundance, and small
diatom abundance by month and basin .........................................................................................58
Table A- 6. Results of analysis of variance for temporal (month) and geomorphic (site type)
factors and covariance for land use (agricultural percentage, agricultural area,
developed/impervious percentage, developed impervious area, catchment area, buffer area,
shoreline length) and physical variables (precipitation, temperature, turbidity, salinity) ............59
Table A- 7. Quadratic discriminant analysis using the five variables dominating the first three
axes of the principal components analysis (i.e., DIN, PO4, turbidity, percentage of buffer in
agricultural use, abundance of low-DNA bacteria ........................................................................60
Appendix B. Mapping Nearshore Nodal Habitat of Juvenile Salmonids within the Hood Canal
and Eastern Strait of Juan de Fuca………………………………………………………………61
51
Figure A-1. Box and whisker plot of DO concentrations at 0.5 meters depth by month and basin showing
median concentration for each basin and month combination (dark line inside box), 25 – 75% interquartile
range (lower and upper limit of boxes), the maximum and minimum values excluding outliers (whiskers
extending above and below each box), and outliers (open circles above or below whiskers).
Figure A-2. Box and whisker plot of DO concentrations at 6.0 meters depth by month and basin showing
median concentration for each basin and month combination (dark line inside box), 25 – 75% interquartile
range (lower and upper limit of boxes), the maximum and minimum values excluding outliers (whiskers
extending above and below each box), and outliers (open circles above or below whiskers).
52
Figure A-3. Water column profile at site E00A, the southernmost Hood Canal basin site, in October (top)
and April (bottom). Changes in dissolved oxygen (green dotted line), salinity (brown dashed line), pH
(blue solid line), and temperature (purple dash and double dot line) relative to depth.
53
Figure A-4. Box plots of dissolved inorganic nutrients by month and basin.
54
Table A-1. Pairwise correlations (Pearsons r) between dissolved inorganic nutrient concentrations. For all
correlations, p ≤ 0.05. (N = 546).
Table A-2. Correlations between chlorophyll a concentrations and picophytoplankton or small diatom
abundance by month. Correlations with p value ≤ 0.05 are shown in bold.
April May June July August September October
Picophytoplankton -0.424 0.493 -0.327 -0.162 -0.093 -0.114 0.527
Small diatoms -0.110 0.425 0.428 -0.165 0.006 0.127 0.406
Table A-3. Summary of linear regressions of chlorophyll a concentrations (“chlorophyll”) and high-DNA
bacterial abundance (“bacteria”) on nutrient concentrations by basin. “ns” indicates regressions or
variables that are not significant (p > 0.05).
NO3 NO2 NH4 PO4
NO2 0.693
NH4 0.086 0.300
PO4 0.485 0.452 0.104
Si(OH)4 0.695 0.513 0.122 0.407
DIN PO4
R2 chlorophyll bacteria R
2 chlorophyll bacteria
Rosario 0.391 - ns 0.310 - ns
Whidbey 0.273 - - 0.318 - ns
Admiralty 0.575 - - 0.655 - -
Hood
Canal
0.293 ns - ns ns ns
Central 0.620 - - 0.503 - -
South 0.512 - - 0.176 - -
55
Figure A-5. Box plots of chlorophyll a concentrations, picophytoplankton abundance, and small diatom
abundance by month and basin.
56
Figure A-6. Box plots of high-DNA and low-DNA bacterial abundance and Synechococcus abundance by
month and basin.
57
Table A-4. Variables included in the canonical correspondence analysis (CCA) of bacterial communities.
.
Environmental Descriptors displayed on Canonical Correspondence Analysis (CCA) plotMeasured environmental variable Abbreviation
Salinity Sal.
Temperature Temp.
pH pH
Ammonium NH4
Nitrate NO3
Nitrite NO2
Phosphate PO4
Silicate SiOH4
Julian Date J.Date
Maximum depth Max.Depth
Latitude Lat.
Synechococcus Synecho.
Total bacteria Tot.Bacteria
Picophytoplankton Phyto.
Low DNA bacteria LowDNA
Chlorophyll a Chl_a
Turbidity Turb.
Productivity Prod
Total number of binned ARISA peaks No_Peaks
Development (m^2) in catchment Develop.
Agriculture (m^2) in catchment Agri.
7 day cumulative precipitation Precip.
Thermocline strength Thermo.Str
Halocline strength Halo.Str
Pycnocline strength Pycno.Str
Redfield Ratio - silicate: total dissolved
inorganic nitrogen Si_N
58
Partial Canonical Correspondence Analysis (pCCA)
Variation of species composition explained by grouped environmental descriptors
Environmental Descriptors Marginal effect size
Proportion of
total species
variance
Proportion of total
constrained inertia
Habitat Quality 0.282 0.123 0.576
Season 0.281 0.125 0.573
Land Use 0.11 0.055 0.224
Percent of total variance explained Percent of overlaping variance
Habitat Quality 12.3% Land Use and Habitat Quality 1.30%
Season 12.5% Land Use, Habitat and Season 1.00%
Land Use 5.5% Land Use and Season 0.40%
Habitat Quality and Season 4.40%
Habitat Quality. Land Use Latitude Shoreline length
SiOH4:DIN ratio Shore area
Average Turbidity % development in shoreline buffer
Maximum Depth % agriculture in shoreline buffer
Average pH % impervious surface in shoreline buffer
Halocline strength % development in catchment
Thermocline strength % agriculture in catchment
Pycnocline strength % impervious surface in catchment
7 day cumulative precipitation Development (m^2) in shoreline buffer
Agriculture (m^2) in shoreline buffer
Season Impervious (m^2) surface in shoreline buffer
Temperature Development (m^2) in catchment
Chlorophyl a Agriculture (m^2) in catchment
Conductivity Impervious (m^2) surface in catchment
PAR % basin developed
pH Catchment Area
Salinity
High DNA bacteria
PO4
Variable Grouping
Monte Carlo test of
significance p
0.005
0.005
0.005
Table A-5. Variable groupings and results of partial canonical correspondence analysis (pCCA) of
bacterial communities.
59
Table A-6. Results of analysis of variance for temporal (month) and geomorphic (site type) factors and covariance for land use (agricultural percentage,
agricultural area, developed/impervious percentage, developed impervious area, catchment area, buffer area, shoreline length) and physical variables (precipitation,
temperature, turbidity, salinity). Chlorophyll a included in analyses only for bacterial abundances. Proxy variables were used due to high collinearity among land
use variables; proxy variables are underlined within each group: 1) agricultural percentage (percent agricultural use in buffer, percent agricultural use in
catchment); 2) agricultural area (area of agricultural use in buffer, area of agricultural use in catchment); 3) developed/impervious percentage (percent developed
in buffer, percent impervious in buffer, percent developed in catchment, percent impervious in catchment); 4) developed/impervious area (area developed in
buffer, area impervious in buffer, area developed in catchment, area impervious in catchment). Variable name and regression coefficient (in parentheses) shown
only for variables with p < 0.05.
R2 month (reference=April) site type (reference=delta) land use physical
DIN 0.510 June (0.85)
July (1.49) August (2.36)
September (2.56)
October (2.74)
exposed (0.90)
small bay (0.46)
agricultural area (0.32)
developed/impervious area (0.16) catchment area (0.16)
buffer area (-0.44)
precipitation (0.03)
temperature (-7.40) turbidity (-0.50)
Ammonium 0.061 September (-1.00) agricultural percentage (0.16)
Phosphate 0.450 June (0.32)
July (0.77)
August (1.28) September (1.43)
October (1.62)
exposed (0.71)
large bay (0.18)
small bay (0.30)
agricultural percentage (0.15)
developed/impervious area (0.04)
catchment area (0.08)
temperature (-2.85)
turbidity (-0.17)
Silicic acid 0.345 June (0.34) July (0.58)
August (0.80)
September (1.21) October (1.20)
agricultural percentage (0.04) temperature (-2.45) salinity (-1.29)
High-DNA bacteria
0.485 June (155.46)
August (-240.04) October (-156.82)
developed/impervious percentage (-29.87) precipitation (-6.37)
temperature (1095.01) turbidity (82.59)
salinity (537.84)
chlorophyll a (91.14)
Low-DNA bacteria
0.379 May (-137.86)
September (137.76)
October (66.14)
exposed (-108.86) agricultural area (-14.82)
developed/impervious area (-23.07)
precipitation (-5.46)
temperature (275.09)
Chlorophyll a 0.168 June (-0.52)
July (-0.42)
August (-0.52) September (-1.14)
October (-0.94)
catchment area (0.15)
shoreline length (-0.07)
temperature (1.55)
turbidity (0.46)
Turbidity 0.135 exposed (-0.53) large bay (-0.18)
agricultural area (0.12) developed/impervious area (0.06)
buffer area (-0.24)
pH 0.665 June (-0.20) July (-0.31)
August (-0.50)
September (-0.59) October (-0.60)
precipitation (-0.004) temperature (1.20)
turbidity (0.03)
60
Table A-7. Quadratic discriminant analysis using the five variables dominating the first three axes of the principal
components analysis (i.e., DIN, PO4, turbidity, percentage of buffer in agricultural use, abundance of low-DNA bacteria).
61
Appendix B
Mapping Nearshore Nodal Habitat of Juvenile Salmonids within the Hood Canal and Eastern Strait of
Juan de Fuca
Hans Daubenberger Hannah Barrett
Janet Aubin Julianna Sullivan
Sarah Burlingame Port Gamble S’Klallam Tribe Natural Resource Department
October 2013
The Port Gamble S’Klallam Tribe partnered with NOAA Fisheries’ Northwest Fisheries Science Center in
the spring of 2011 to evaluate the status of the Puget Sound’s nearshore pelagic foodweb. The Tribe’s
primary interest in this undertaking was to investigate the hypothesis currently guiding salmon recovery
within nearshore habitats in the Hood Canal and Eastern Strait of Juan de Fuca.
Salmon recovery within the Hood Canal and the Eastern Strait of Juan de Fuca is governed by the Hood
Canal Coordinating Council (HCCC). The HCCC is designated as the “Lead Entity” for Hood Canal
watersheds and the “Regional Recovery Organization” for the Hood Canal and Eastern Strait of Juan de
Fuca Summer Chum population. Currently the HCCC relies on a nearshore hypothesis described in the
HCCC Salmon Habitat Recovery Strategy (HCCC 2005) for prioritization and ranking of restoration and
conservation actions within the marine nearshore environment. The implementation of the nearshore
hypothesis has resulted in strong bias for funding projects which meet Priority 1 and 2 criteria as
described in the HCCC Salmon Habitat Recovery Strategy (2005). See Table 1 and HCCC 2013 Process
Guide appendix B. The Port Gamble S’Klallam Tribe is a governing member of the Hood Canal
Coordinating Council, and has a strong interest in the success of the organization’s ability to restore
Pacific salmon. Pacific salmon are culturally and economically important to the Port Gamble S’Klallam
Tribe, and the Tribe’s utilization of the Pacific salmon resource is protected by the Point No Point Treaty.
62
Table 1: Hood Canal Coordinating Council Salmon Habitat Recovery Strategy Nearshore Priority Matrix.
Priority Nearshore Habitat Areas
Priority Habitats
P-1 • Estuarine deltas associated with Tier 1 watersheds
• Tidal marsh complexes and eel grass meadows historically contiguous and within 1 mile of P-1 estuarine deltas
• Contributing processes to P-1 segments
P-2 • Estuarine deltas associated with Tier 2 watersheds
• All other tidal marsh complexes and eel grass meadows
• Kelp forests and shallow-water shorelines within 1 mile of P-1 and P-2 estuarine deltas
• Contributing processes to P-2 segments
P-3 • All other estuarine delta habitat
• Kelp forests and shallow-water shorelines farther than 1 mile from P-1 and P-2 estuarine deltas
• Contributing processes to P-3 segment
P-4 • Non vegetated sub tidal habitats
• Non shallow-water shorelines
• Contributing processes to P-4 segments Tier 1 and 2 watersheds are described as the following: Tier 1: This Tier consists of drainages that have the capacity, or potential (historically
based) capacity, to be habitat for 2 or more ESA-listed (Summer Chum and Chinook salmon and Bull Trout) or SaSi-critical species. In addition,
several watersheds are prioritized in Tier 1 to support conservation of unique habitats supporting stocks of Summer Chum Salmon. Tier 2: This
Tier consists of drainages that have the capacity, or potential (historically based) capacity, to be habitat for 1 ESA listed species. (HCCC Salmon
Habitat Recovery Strategy (2005)).
To investigate the nearshore hypothesis described in the HCCC Salmon Habitat Recovery Strategy
(2005), the Port Gamble S’Klallam Tribe conducted hydroacoustic surveys in conjunction with the
surface trawls during the Puget Sound Nearshore Pelagic Foodweb Study. Hydroacoustic surveys offer a
number of unique data collection advantages over capture based sampling techniques. They are non-
invasive, can cover significantly greater spatial areas, and can collect data from multiple depths
simultaneously. When hydroacoustic surveys are paired with capture based sampling events, species
composition may be assigned to acoustic data by size class distribution. We used a side-looking
hydroacoustic method to sample within the nearshore environment. The remainder of this appendix
describes the methodology and results of the hydroacoustic surveys conducted during the summers of
2011 and 2012.
METHODS
The hydroacoustic sampling employed the use of a BioSonics DT-X Scientific Echosounder with a 200 kHz
split beam digital transducer. The transducer was mounted on a towed body with a side-looking
orientation at a downward angle of approximately 10-degrees from horizontal to avoid surface noise
(Figure 1). A side-looking orientation was chosen over a downward-looking orientation to avoid
inaccurate estimates based on low sample volume and boat avoidance (Overman and Beauchamp 2007,
Beauchamp et al. 2009). The transducer was towed at a speed of approximately 4 knots. Hydroacoustic
data was collected using BioSonics Visual Acquisition software version 6.0 and was geo-referenced using
an integrated Garmin GPS receiver. Table 2 lists the collection parameters for the BioSonics DT-X
Echosounder and on-axis calibration values for 2011 and 2012.
63
Figure 1. Hydroacoustic sampling using a side-looking DT-X split beam digital transducer.
Table 2. BioSonics DT-X Echosounder collection parameters and on-axis calibration values.
Hydroacoustic surveys were conducted along an approximate 10 meter depth contour with the acoustic
beam looking seaward (Figure 1). Hydroacoustic sampling was conducted south from Tala Point to the
Duckabush River estuary on the west side of Hood Canal, and north from Seabeck Bay to Foulweather
Bluff on the east side of Hood Canal. In 2012, hydroacoustic sampling was also conducted in Port Ludlow
Bay, Kilisut Harbor and the Port Townsend Canal. Figure 2 shows the hydroacoustic sampling course for
Transducer ID: T206P168
Transducer Type: Split Beam
Acoustic Frequency (Hz): 208000
Digital Sampling Rate (Hz): 41667
Beamwidth (deg): 6.8° X 6.8°
Equiv. Two-Way Beam Angle (dB re 1
Steradian): -20.97
Acoustic Mode:
Active
Transmission
Transmit Pulse Duration (ms): 0.4
Start Range (m): 1
End Range (m): 100
Calibration Correction (dB): 0
Data Collection Threshold Level (dB): -130
Transducer Orientation: Side-Looking
Transducer Connector Position (deg): 180°= to stern
Transducer Depth (m): 1
Temperature (° C): 13-20
Salinity (ppt): 20-27
pH: 7
Single Echo Detection: Enabled
Transmit Pulse Duration (ms): 0.4
Echo Level Threshold (dB): -70
Min Echo Length (-6 dB reference):
0.75 times
pulse length
Max Echo Length (-6 dB reference):
2 times pulse
length
Max SD of Split-Beam Angles (deg): 10
Max Beam Compensation (dB) (one-way): -12
Environment
Echo Detection
BioSonics DTX Scientific Echosounder
Transmit/Receive
Sensors/Mounting
Date: 26-Apr-11
Time: 13:50:51
Source Level (dB re uPa): 221.2
Receiver Sensitivity (dB re uPa): -51
Noise Floor (dB Counts re uPa): -150.2
Date: 11-May-12
Time: 11:04:29
Source Level (dB re uPa): 221.2
Receiver Sensitivity (dB re uPa): -51
Noise Floor (dB Counts re uPa): -150.2
On-Axis Calibration Data
64
2011 and 2012 with surface trawl sites overlaid. Hydroacoustic surveys were conducted preceding the
August Hood Canal surface trawl events. Survey dates were August 15-19, 2011, and August 14-17 and
24th, and September 5, 2012. In 2012, survey dates were extended due to poor sea state conditions.
Figure 2. 2011 and 2012 Hydroacoustic Sampling Tract and Surface Trawl Sites.
Hydroacoustic data was analyzed using Myriax Echoview software (5.1.45.20309, Dongle version 20348).
A data analysis region of 20 meters was created 10 to 30 meters from the start range. Start range was
set at 1 meter from the transducer face (see Table 2 and Figures 1 and 3). This analysis region was
created to restrict targets to the upper 10 meters of the water column. The analysis regions were then
65
split into 1,000-meter cells. Bad data, most frequently caused by bottom interference, was manually
clipped out using the bad data region analysis tool in Echoview (see pink regions in Figure 3).
Figure 3. Screenshot of Echoview showing analysis region (yellow), bad data (pink), and single targets
(red circle).
Single targets using Echoview single target detection split beam method 2 were exported from the
analysis region by cell into an Excel file. From the exported data, single target detections between
-59.5dB to -20.5dB were extracted to isolate acoustic returns correlated with juvenile salmonids’ size
class (Love 1977). The single targets were then connected with the corresponding volume and
geospatial data, which was also exported from Echoview into Excel. Single target density was then
calculated using the sample volume of the analysis region and cell. Data was converted to single targets
per 10,000 cubic meters and imported into ArcGIS resulting in a point shapefile with points located
along the approximate 10-meter depth contour at approximately 1,000-meter intervals. Data points
with a beam volume less than 5,000 cubic meters were eliminated from the data set. The target density
data was displayed geospatially using ArcMap’s graduated symbol methods; natural breaks classification
(Figure 4) and standard deviation (Figure 5). Figure 5, which uses the standard deviation classification
method, displays only data points greater than 0.5 standard deviation.
RESULTS
When plotted, the hydroacoustic data from 2011 and 2012 resulted in uneven density distributions of
single target detections as shown in Figures 4 and 5. Single target detections per 10,000 cubic meters
are shown in Figure 4 with graduated symbols corresponding to estimated fish density. In 2011, the
highest densities occurred in Tala Point, Twin Spits, Hannon Point, Port Gamble Bay, Spring Creek,
Nordstrom Creek, Unnamed Pocket Estuary (Lat. 47.776373, Long. -122.746410), Broad Spit,
Frenchmans Point, and Pleasant Harbor. In 2012, the highest densities occurred in Kilisut Harbor, Port
Townsend Canal, Port Ludlow, Tala Point, Hannon Point, Twin Spits, Port Gamble Bay, Spring Creek,
Nordstrom Creek, Cougar-Kinman Creek, Big Beef Creek, Zelatched Point, Broad Spit, Frenchmans Point,
Point Whitney, Jackson Cove, Turner Creek, and Pleasant Harbor. Port Ludlow and sites to the north
were only surveyed in 2012. In Figure 5, single target detections from 2011 and 2012 are overlaid
illustrating the distribution of estimated fish density. We identified sites where densities with a standard
deviation greater than or equal to 0.5 fell within a common geographic feature in 2011 and 2012 (see
Figure 5 and Tables 3, 4, 5, 6 and 7).
66
Figure 4. 2011 and 2012 Estimated Fish Density (measured as single target detections per 10,000 cubic meters) for Hood Canal Hydroacoustic surveys.
67
Figure 5. 2011 and 2012 Estimated Fish Density (measured as single target detections per 10,000 cubic meters) displayed as standard deviation greater than or equal to 0.5. Overlapping data grouped by geographical feature representing nodal fish habitat.
68
Table 3. Number of points with single target detections above 0.5 standard deviation grouped by geographic region. Total points for individual year and two-year totals are given. Geographical sites are in order of greatest number of points greater than or equal to 0.5 standard deviation.
Range (SD) 0.5-1.5
1.5-2.5 >2.5
0.5-1.5
1.5-2.5 >2.5
Total 2011
Total 2012 Total Site 2011 2012
Port Gamble Bay 2 2 5 3 0 3 9 6 15
Nordstrom 2 0 0 2 0 0 2 2 4
Pleasant Harbor 1 0 1 1 0 0 2 1 3
Broad Spit 0 1 1 1 0 0 2 1 3
Tala Point 0 1 1 1 0 0 2 1 3
Spring Creek 1 0 1 1 0 0 2 1 3
Frenchmans Point 1 0 0 2 0 0 1 2 3
Twin Spits 0 1 0 0 1 0 1 1 2
Table 4. Number of single target detections per point grouped by geographic regions and standard deviation. Total single target detections per geographic region by year and two-year total are listed. Geographical sites are in order of two-year totals.
Range (SD) 0.5-1.5
1.5-2.5 >2.5
0.5-1.5
1.5-2.5 >2.5
Total 2011
Total 2012 Total Site 2011 2012
Port Gamble Bay 12, 15 29, 33
34, 35, 40, 67, 59
15, 16, 17 0
48, 66, 160 324 322 646
Spring Creek 18 0 72 27 0 0 90 27 117
Broad Spit 0 31 51 16 0 0 82 16 98
Tala Point 0 24 38 23 0 0 62 23 85
Pleasant Harbor 12 0 52 15 0 0 64 15 79
Nordstrom 12, 14 0 0 17, 18 0 0 26 35 61
Twin Spits 0 21 0 0 32 0 21 32 53
Frenchmans Point 15 0 0 14, 15 0 0 15 29 44
69
Table 5. List of all points and associated data grouped by geo- graphical feature identified by overlapping standard deviation as depicted in Figure 5. This table also includes data points associated with the Duckabush and Dosewallips river deltas.
Port Gamble Bay 2011
Date Time Lat Long Beam_Vol FD_Total FD_Density
20110815 12:46:35.60 47.856022 -122.577842 33564.516 3 1
20110815 12:54:27.42 47.852371 -122.576162 27191.85677 8 3
20110815 12:58:09.84 47.848183 -122.573653 30197.24757 20 7
20110815 13:01:55.63 47.843903 -122.571658 37214.09598 56 15
20110815 13:05:41.11 47.839463 -122.571066 44413.37537 53 12
20110815 13:09:25.06 47.835012 -122.571648 44761.0703 266 59
20110815 13:13:08.70 47.830548 -122.571932 42997.25476 288 67
20110815 13:16:52.64 47.826133 -122.572165 33571.33294 115 34
20110815 13:24:43.28 47.825602 -122.576647 40833.03036 163 40
20110815 13:28:30.89 47.829018 -122.580896 43842.58048 127 29
20110815 13:32:10.26 47.832759 -122.582577 39694.80839 137 35
20110815 13:35:49.31 47.837192 -122.582304 43781.34783 144 33
20110815 13:39:27.15 47.841645 -122.5815 43597.64986 16 4
20110815 13:43:08.03 47.846113 -122.58103 43106.23583 31 7
20110815 13:47:06.95 47.850695 -122.579737 34948.7651 29 8
20110815 14:01:16.90 47.856285 -122.57784 26198.70476 6 2
Port Gamble Bay 2012
Date Time Lat Long Beam_Vol FD_Total FD_Density
20120817 16:52:36.48 47.858149 -122.577381 9255.758738 1 1
20120817 16:57:05.12 47.852667 -122.57592 19357.40267 310 160
20120817 17:00:47.59 47.848382 -122.574062 23841.25615 158 66
20120817 17:04:26.03 47.844132 -122.572003 43271.61319 207 48
20120817 17:08:16.07 47.839697 -122.57102 34326.27566 50 15
20120817 17:12:24.37 47.835421 -122.570061 26335.22734 20 8
20120817 17:16:50.62 47.831068 -122.571456 29832.83434 17 6
20120817 17:21:28.48 47.826803 -122.571962 27089.49774 13 5
20120817 17:24:20.67 47.826018 -122.575501 8802.211109 5 6
20120817 17:27:21.43 47.826985 -122.579368 40190.33676 67 17
20120817 17:31:50.52 47.830951 -122.582237 45230.40441 28 6
20120817 17:36:09.39 47.835437 -122.582579 58567.21299 36 6
20120817 17:40:25.63 47.839949 -122.582213 85680.82603 40 5
20120817 17:44:51.61 47.844399 -122.581481 71527.16932 81 11
20120817 17:49:23.17 47.848846 -122.581222 36265.17363 7 2
20120817 17:53:20.69 47.852508 -122.579228 43706.80701 25 6
20120817 17:55:56.04 47.854908 -122.578638 32076.97646 50 16
Tala Point 2011
Date Time Lat Long Beam_Vol FD_Total FD_Density
20110815 16:03:39.98 47.922508 -122.649496 43352.71925 5 1
20110815 16:07:24.54 47.925863 -122.653753 44699.83764 2 0
20110815 16:09:42.02 47.928503 -122.654698 10348.31858 25 24
20110815 16:11:19.47 47.930343 -122.654778 27064.8332 103 38
Tala Point 2012
Date Time Lat Long Beam_Vol FD_Total FD_Density
20120817 13:52:14.16 47.922255 -122.650564 52463.80227 1 0
20120817 13:56:07.61 47.926384 -122.653251 42859.75193 14 3
20120817 13:59:46.98 47.930626 -122.655067 13106.41304 2 2
20120817 14:02:42.10 47.934032 -122.655275 5281.545087 12 23
Twin Spits 2011
Date Time Lat Long Beam_Vol FD_Total FD_Density
20110815 11:05:21.11 47.931082 -122.621719 67478.38504 38 6
20110815 11:10:35.21 47.927418 -122.618682 58232.25424 121 21
20110815 11:16:29.33 47.924085 -122.61805 83460.10782 51 6
20110815 11:23:12.92 47.921481 -122.613826 78377.79751 77 10
Twin Spits 2012
Date Time Lat Long Beam_Vol FD_Total FD_Density
20120817 15:33:51.07 47.930334 -122.617455 79601.58537 21 3
20120817 15:37:45.75 47.926309 -122.618269 63455.66268 42 7
20120817 15:41:53.89 47.923366 -122.615815 17651.8115 57 32
20120817 15:45:22.74 47.921215 -122.61163 11959.93925 6 5
Spring Creek 2011
Date Time Lat Long Beam_Vol FD_Total FD_Density
20110817 10:48:57.05 47.846378 -122.622263 48557.49486 11 2
20110817 10:52:49.87 47.841961 -122.622269 44822.30295 21 5
20110817 10:56:36.89 47.838738 -122.627309 46108.18869 330 72
20110817 11:00:41.94 47.837216 -122.633254 52170.22147 94 18
20110817 11:05:05.33 47.834739 -122.638277 53333.6419 17 3
20110817 11:09:44.91 47.832242 -122.643184 58722.11548 48 8
Spring Creek 2012
Date Time Lat Long Beam_Vol FD_Total FD_Density
20120824 11:27:56.14 47.845484 -122.620003 84237.55575 70 8
20120824 11:32:28.76 47.841297 -122.621987 88400.46963 118 13
20120824 11:36:54.70 47.8375 -122.625786 63876.08593 63 10
20120824 11:41:13.97 47.835752 -122.631698 50116.45842 67 13
20120824 11:45:04.66 47.834043 -122.637027 47621.3887 128 27
20120824 11:49:02.64 47.831548 -122.64124 56828.58374 34 6
Nordstrom Creek 2011
Date Time Lat Long Beam_Vol FD_Total FD_Density
20110818 15:59:10.97 47.816708 -122.709059 46230.654 7 2
20110818 16:02:58.59 47.81994 -122.704603 45006.00091 8 2
20110818 16:06:35.20 47.823006 -122.699706 41883.13554 51 12
20110818 16:10:16.10 47.825622 -122.693857 46536.81727 65 14
20110818 16:14:30.30 47.827971 -122.688845 55354.31949 7 1
70
20110818 16:18:48.79 47.831724 -122.685049 48251.33159 12 2
Nordstrom Creek 2012
Date Time Lat Long Beam_Vol FD_Total FD_Density
20120905 13:01:09.27 47.818071 -122.709071 94913.88699 167 18
20120905 13:05:26.57 47.821388 -122.704795 101904.6291 61 6
20120905 13:09:28.71 47.824332 -122.699883 67607.34392 82 12
20120905 13:13:37.80 47.826257 -122.693708 29756.86627 4 1
20120905 13:17:12.54 47.828668 -122.688376 88429.93366 22 2
20120905 13:21:04.61 47.832597 -122.686057 99887.50928 170 17
Broad Spit 2011
Date Time Lat Long Beam_Vol FD_Total FD_Density
20110816 15:30:13.28 47.80188 -122.816922 44026.27844 12 3
20110816 15:33:55.12 47.806005 -122.814498 44883.5356 35 8
20110816 15:37:46.73 47.810447 -122.814263 47945.16832 244 51
20110816 15:41:50.86 47.812397 -122.818575 46442.19947 143 31
20110816 15:45:57.14 47.816803 -122.820115 48741.19283 24 5
Broad Spit 2012
Date Time Lat Long Beam_Vol FD_Total FD_Density
20120815 13:32:38.05 47.799523 -122.817867 124219.9504 60 5
20120815 13:37:57.97 47.803839 -122.815905 80306.22763 77 10
20120815 13:43:42.29 47.80807 -122.814269 142105.7269 93 7
20120815 13:49:36.36 47.811113 -122.817045 129861.2452 211 16
20120815 13:54:54.28 47.814792 -122.820323 114687.006 64 6
Frenchmans Point 2011
Date Time Lat Long Beam_Vol FD_Total FD_Density
20110816 13:33:25.37 47.773963 -122.863427 44638.60499 14 3
20110816 13:37:06.29 47.778172 -122.865098 41907.15279 34 8
20110816 13:40:47.49 47.782048 -122.864315 29000.94294 43 15
20110816 13:44:28.09 47.786076 -122.86582 38634.85401 14 4
Frenchmans Point 2012
Date Time Lat Long Beam_Vol FD_Total FD_Density
20120816 17:22:50.49 47.773926 -122.86473 128355.8595 142 11
20120816 17:28:01.30 47.77829 -122.865311 90510.96064 131 14
20120816 17:33:13.50 47.782296 -122.865354 121298.3727 59 5
20120816 17:36:21.99 47.784687 -122.866663 14566.97609 22 15
Pleasant Harbor 2011
Date Time Lat Long Beam_Vol FD_Total FD_Density
20110817 15:44:06.25 47.666732 -122.903777 10544.04735 55 52
20110817 15:40:43.04 47.663657 -122.900514 46842.98054 9 2
20110817 15:44:34.97 47.667117 -122.903248 45802.02542 56 12
20110817 15:48:22.64 47.669874 -122.898144 45434.62949 6 1
Pleasant Harbor 2012
Date Time Lat Long Beam_Vol FD_Total FD_Density
20120816 13:31:59.04 47.661868 -122.90163 143872.0961 154 11
20120816 13:37:45.86 47.665339 -122.905157 47833.47328 19 4
20120816 13:43:32.77 47.663798 -122.911486 33634.22974 38 11
20120816 13:47:51.32 47.661487 -122.914816 10025.71594 15 15
20120816 14:29:04.50 47.662356 -122.915547 54898.56077 15 3
20120816 14:42:55.09 47.666907 -122.906183 38415.42564 7 2
20120816 14:47:55.98 47.670585 -122.902386 121280.9059 36 3
Duckabush Delta 11
Date Time Lat Long Beam_Vol FD_Total FD_Density
20110817 15:02:14.90 47.636254 -122.934918 44883.5356 5 1
20110817 15:05:57.37 47.637947 -122.928763 44332.44172 26 6
20110817 15:09:38.01 47.64057 -122.923543 44026.27844 19 4
20110817 15:13:21.41 47.642368 -122.917555 45557.0948 7 2
20110817 15:17:09.38 47.642148 -122.911675 45740.79277 20 4
20110817 15:20:54.91 47.644613 -122.906241 44638.60499 4 1
Duckabush Delta 2012
Date Time Lat Long Beam_Vol FD_Total FD_Density
20120816 12:35:46.80 47.635808 -122.938347 129806.5645 34 3
20120816 12:41:04.95 47.638137 -122.932893 124441.5372 68 5
20120816 12:46:24.34 47.638977 -122.927187 130815.7506 21 2
20120816 12:51:44.35 47.641661 -122.922095 132203.3816 60 5
20120816 12:57:09.33 47.6422 -122.916122 131072.1275 138 11
20120816 13:01:31.13 47.643666 -122.911347 80122.07387 37 5
20120816 13:05:49.76 47.644902 -122.906488 132012.1179 177 13
Dosewallips Delta 2011
Date Time Lat Long Beam_Vol FD_Total FD_Density
20110817 15:56:00.72 47.677809 -122.892513 46169.42135 49 11
20110817 15:59:09.28 47.68072 -122.889317 29391.67407 17 6
20110817 16:02:19.66 47.682926 -122.885113 46842.98054 18 4
20110817 16:06:13.44 47.685917 -122.880485 46842.98054 13 3
20110817 16:10:04.47 47.689628 -122.877225 43592.22579 9 2
20110817 16:13:49.69 47.693577 -122.876926 44148.74375 15 3
20110817 16:17:31.25 47.697928 -122.878473 43475.18456 15 3
20110817 16:21:09.14 47.701898 -122.878662 37712.71986 14 4
20110817 16:24:01.50 47.705403 -122.879518 25044.15561 5 2
Dosewallips Delta 2012
Date Time Lat Long Beam_Vol FD_Total FD_Density
20120816 14:58:43.99 47.67866 -122.89571 138132.35 81 6
20120816 15:04:17.49 47.68132 -122.891177 133117.9727 110 8
20120816 15:09:43.75 47.683112 -122.88576 132203.3816 93 7
20120816 15:14:59.41 47.685994 -122.880943 108084.7546 61 6
20120816 15:20:11.87 47.688457 -122.876325 127758.7371 55 4
20120816 15:24:22.59 47.691764 -122.874484 76445.8485 19 2
20120816 15:28:13.04 47.694686 -122.876418 112902.697 58 5
20120816 15:32:40.03 47.697808 -122.879687 92810.58202 130 14
20120816 15:37:00.42 47.701806 -122.878824 107541.3958 60 6
20120816 15:41:19.73 47.705065 -122.88236 105523.0235 69 7
71
Table 6. List grouped by geographical feature and year with features sorted by feature density. Table displays volume sum, single target detection sums, and total target detection feature density.
Site Name Total Feature Beam_Vol Total Feature FD
Total Feature FD_Density
Port Gamble Bay 2011 609913.87 1462 23.97
Port Gamble Bay 2012 603280.01 1065 17.65
Spring Creek 2011 694794.51 1001 14.41
Twin Spits 2012 476382.96 647 13.58
Nordstrom Creek 2012 714538.54 964 13.49
Broad Spit 2011 823218.53 963 11.70
Frenchmans Point 2012 695061.42 653 9.39
Spring Creek 2012 674342.80 630 9.34
Frenchmans Point 2011 566300.75 523 9.24
Tala Point 2011 364887.86 323 8.85
Broad Spit 2012 873717.57 752 8.61
Nordstrom Creek 2011 765762.43 656 8.57
Twin Spits 2011 553597.34 445 8.04
Tala Point 2012 544317.30 379 6.96
Pleasant Harbor 2011 598584.09 410 6.85
Dosewallips Delta 2012 1134520.74 736 6.49
Duckabush Delta 2012 1117461.58 656 5.87
Dosewallips Delta 2011 874758.54 500 5.72
Duckabush Delta 2011 1129652.30 616 5.45
Pleasant Harbor 2012 719139.16 365 5.08
Table 7. List grouped by geographical feature sorted in order of combined total densities from 2011 & 2012.
2011 & 2012
Site Name Totals Feature FD_Density
Port Gamble Bay 41.62
Spring Creek 23.75
Nordstrom Creek 22.06
Twin Spits 21.62
Broad Spit 20.30
Frenchmans Point 18.63
Tala Point 15.81
Dosewallips Delta 12.20
Pleasant Harbor 11.93
Duckabush Delta 11.32
72
To compare hydroacoustic returns to surface trawl species composition we have summarized the catch
data from August surface trawling events and hydroacoustic results (Tables 8 and 9). Hydroacoustic data
is limited in its ability to distinguish amongst species. When collecting data where the fish axis is known,
acoustic returns can be correlated directly to capture-based sampling. However, when a side-looking
transducer is used, as in this study, acoustic returns vary and an analysis range must be used
(Beauchamp et al. 2009). In our analysis, we used the decibel range -59.5dB to -20.5dB, which is
expected to exclude post-larval fish, species lacking significant gas bladders, and gelatinous zooplankton
(Overman and Beauchamp 2007). The species in boldface in Table 8 are expected to return
hydroacoustic energy within the -59.5dB to -20.5dB analysis range. We analyzed only for single target
detections and did not use echointegration, thus fish schools were excluded from the analysis. As a
result, boldfaced fish species in Table 8 that exhibit strong schooling behavior are likely
underrepresented in this analysis. These species include surf smelt, sand lance, herring, Northern
Anchovy and stickleback (Pitcher et al. 1996, Wark 2011, WDFW 2010). We expect that pink and chum
are also underrepresented in this analysis because they exhibit schooling behavior more frequently than
Chinook (behavioral observation made during surface trawls and beach seines in Hood Canal between
2009 and 2013 PGST). In order to relate our hydroacoustic results with our surface trawling catch data,
we calculated an average of number of individuals of each species captured per site in trawls. Using the
surface trawl volumes sampled, we then calculated an average number of individuals per 10,000 cubic
meters sampled (Tables 8 and 9).
73
Table 8. Surface trawl catch for sites overlapping with hydroacoustic surveys by species and year with total catch of each species and average number of individuals per site. The number of individuals per 10,000 cubic meters was estimated using an average trawl sampling volume of 6,000 cubic meters to compare with hydroacoustic data.
Table 9. Summary of 2011 and 2012 Hydroacoustic Results.
Species
Total
Counts
Average #
per site
# per 10,000
cubic
meters
Total
Counts
Average #
per site
# per 10,000
cubic
meters
Size
Range
mm
Chinook 54 4.50 7.50 59 3.93 6.56 108-177
Chum 3 0.25 0.42 67 4.47 7.44 89-148
Crescent Gunnel 0 0 0 1 0.07 0.11 56
Clytia 25 2.08 3.47 0 0.00 0.00 0
Cross Jelly 1 0.08 0.14 0 0.00 0.00 0
Fried Egg Jelly 0 0 0 2 0.13 0.22 270-280
Herring 4 0.33 0.56 157 10.47 17.44 51-104
Lions Mane Jelly 11 0.92 1.53 6 0.40 0.67 121-223
Midshipman 0 0 0 1 0.07 0.11 72
Moon Jelly 11 0.92 1.53 1 0.07 0.11 140
Northern Anchovy 0 0 0 1 0.07 0.11 114
Pink 0 0 0 30 2.00 3.33 83-119
Pipefish 3 0.25 0.42 4 0.27 0.44 114-178
Polyorchis 0 0 0 2 0.13 0.22 11-18
Saddleback Gunnel 0 0 0 1 0.07 0.11 89
Sand Lance 0 0 0 39 2.60 4.33 51-95
Shiner Perch 0 0 0 2 0.13 0.22 86-91
Slimy Sculpin 0 0 0 1 0.07 0.11 35
Snake Prickleback 0 0 0 2 0.13 0.22 87-145
Stickelback 1119 93.25 155.42 49 3.27 5.44 60-88
Surf Smelt 8 0.67 1.11 57 3.80 6.33 39-115
Unk Fish 0 0 0 9 0.60 1.00 20-41
Unk Gadid 0 0 0 2 0.13 0.22 65-84
Unk Larval Fish 24 2.00 3.33 52 3.47 5.78 18-46
Water Jelly 92 7.67 12.78 37 2.47 4.11 20-62
2011 2012
Count 283 Count 326
Minimum 0 Minimum 0
Maximum 72 Maximum 160
Sum 1700 Sum 2500
Mean 6.2 Mean 7.5
Median 3 Median 4.5
Standard Deviation 9.9 Standard Deviation 13
2011 Hydroacoustic Results 2012 Hydroacoustic Results
74
DISCUSSION
Single target detection densities recorded in the Port Gamble Bay area were substantially greater than
those found in any other geographical feature within our study. Geographical features were delineated
by overlapping fish abundances from 2011 and 2012 with standard deviations above 0.5 (see Figure 5).
These areas were evaluated for 2011 and 2012 using several different methods, in all cases Port Gamble
Bay consistently had the highest density ranking (see Tables 3-7).
Based on field observations (behavioral observation made during surface trawls and beach seines in
Hood Canal between 2009 and 2013 PGST), there appears to be a correlation between prey abundance
and salmonid densities. Our hypothesis for why Port Gamble Bay consistently had higher densities of
single target detections can be explained by the presence of high densities of larval forage fish due to
the utilization of this area as a spawning ground by herring, surf smelt, and sand lance (NOSC 2005, Stick
and Lindquist 2009). These larval fish are a high energy prey item for juvenile chinook (Duffy et al. 2010).
An additional explanation for juvenile salmonid presence within Port Gamble Bay is explained by the
geomorphology of the Hood Canal fjordal system. Within the Hood Canal there are relatively few areas
where the seafloor is within reach of the productive photic zone. Port Gamble Bay, however, is a
relatively shallow embayment within this greater system. This shallower area permits for a highly
productive aquatic environment allowing for the presence of eelgrass and attached macroalgae.
Another intriguing finding was the lack of single target detections recorded along the Dosewallips and
Duckabush river deltas (see Figure 5 and Tables 5-7). This result was unexpected as these deltas are the
terminus to natal systems for large populations of coho, chinook, fall and summer chum, steelhead and
cutthroat trout. This may be explained by the unique geomorphology of these delta systems. During
daily low tide events both river deltas are dewatered until reaching the precipitous edge of their alluvial
fans (Figure 6). During low tides we expect that juvenile salmon rapidly vacate these areas in search of
more productive habitat as well as to avoid predation in this highly exposed environment.
75
Figure 6. Dosewallips and Duckabush River delta depth profiles. Y-axis describes water depth, X-axis demarcates distance. Units are in feet.
Acoustic surveys have been completed for the summer 2013 season and are scheduled for 2014.
Continuing these surveys will increase our ability to evaluate nearshore nodal habitats for juvenile
salmonids within the Hood Canal. As we continue this effort it may be possible to define geomorphic
features that are associated with high densities of juvenile salmon. These identified features could
potentially also be applied to locations outside of the immediate study area. Within the mid-Hood Canal
systems, our research indicates that juvenile salmon may be rapidly moving away from estuaries
associated with their natal system in search of more productive and protective habitat. This data will be
utilized to re-evaluate nearshore restoration and habitat conservation efforts within the mid to outer
portions of the Hood Canal.
76
Citations
Beauchamp, DA, D Parrish, R Whaley. 2009. Salmonids/coldwater species in large standing waters. Pages
97-117 in S Bonar, D Willis, W Hubert (eds), Standard Sampling Methods for North American Freshwater
Fishes. American Fisheries Society. Bethesda, MD.
Duffey, E.J., D.A. Beauchamp, R.M. Sweeting, R.J. Beamish, and J.S. Brennan. 2010. Ontogenetic Diet
Shifts of Juvenile Chinook Salmon in Nearshore and Offshore Habitats of Puget Sound. American
Fisheries Society. 139:803-823.
Hood Canal Coordinating Council (HCCC). 2005. Salmon Habitat Recovery Strategy for the Hood Canal &
Eastern Strait of Juan de Fuca Version 09.
Hood Canal Coordinating Council (HCCC). 2013. 2013 Salmon Recovery Grant Process Guide –
Developing Salmon Habitat Recovery Projects in Hood Canal & the Eastern Strait of Juan de Fuca.
North Olympic Salmon Coalition (NOSC). 2005. Intertidal Forage Fish Spawning Site Investigation for
East Jefferson, Northwestern Kitsap, and North Mason Counties 2001-2004.
Overman, N.C., and D.A. Beauchamp. 2007. Growth, Distribution, and Abundance of Pelagic Fishes
in Lake Washington. Final Report to Seattle Public Utilities. Washington Cooperative Fish and
Wildlife Research Unit. Report # WACFWRU-07-01.
Pitcher, T. J., O. A. Misund, A. Fernö, B. Totland, and V. Melle. 1996. Adaptive behaviour of herring
schools in the Norwegian Sea as revealed by high-resolution sonar. – ICES Journal of Marine
Science, 53: 449–452.
Stick, K.C, and A. Lindquist. 2009. 2008 Washington State Herring Stock Status Report. Final Report
to Washington Department of Fish and Wildlife Fish Management Division. Report # FPA 09-05.
Love, R.H. 1977. Target strength of an individual fish from any aspect. Journal of the Acoustical
Society of America. 62:1397-1403.
Wark A.R., A.K. Greenwood, E.M. Taylor, K. Yoshida, and C.L. Peichel. 2011. Heritable Differences in
Schooling Behavior among Threespine Stickleback Populations Revealed by a Novel Assay. PLoS ONE
6(3): e18316. doi:10.1371/journal.pone.0018316.
Washington Department of Fish and Wildlife (WDFW). 2010. Washington State Surf Smelt Fact
Sheet.