Ecology, 92(2), 2011, pp. 373–385
Ó 2011 by the Ecological Society of America
Marine-derived nutrients, bioturbation, and ecosystem metabolism:
reconsidering the role of salmon in streams
GORDON W. HOLTGRIEVE1,2,3
AND
DANIEL E. SCHINDLER1
1
School of Aquatic and Fishery Sciences, University of Washington, Seattle, Washington 98195 USA
2
Department of Biology, University of Washington, Seattle, Washington 98195 USA
Abstract. In coastal areas of the North Pacific Ocean, annual returns of spawning salmon
provide a substantial influx of nutrients and organic matter to streams and are generally
believed to enhance the productivity of recipient ecosystems. Loss of this subsidy from areas
with diminished salmon runs has been hypothesized to limit ecosystem productivity in juvenile
salmon rearing habitats (lakes and streams), thereby reinforcing population declines. Using
five to seven years of data from an Alaskan stream supporting moderate salmon densities, we
show that salmon predictably increased stream water nutrient concentrations, which were on
average 190% (nitrogen) and 390% (phosphorus) pre-salmon values, and that primary
producers incorporated some of these nutrients into tissues. However, benthic algal biomass
declined by an order of magnitude despite increased nutrients. We also measured changes in
stream ecosystem metabolic properties, including gross primary productivity (GPP) and
ecosystem respiration (ER), from three salmon streams by analyzing diel measurements of
oxygen concentrations and stable isotopic ratios (d18O-O2) within a Bayesian statistical model
of oxygen dynamics. Our results do not support a shift toward higher primary productivity
with the return of salmon, as is expected from a nutrient fertilization mechanism. Rather, net
ecosystem metabolism switched from approximately net autotrophic (GPP ER) to a
strongly net heterotrophic state (GPP ER) in response to bioturbation of benthic habitats
by salmon. Following the seasonal arrival of salmon, GPP declined to ,12% of pre-salmon
rates, while ER increased by over threefold. Metabolism by live salmon could not account for
the observed increase in ER early in the salmon run, suggesting salmon nutrients and
disturbance enhanced in situ heterotrophic respiration. Salmon also changed the physical
properties of the stream, increasing air–water gas exchange by nearly 10-fold during peak
spawning. We suggest that management efforts to restore salmon ecosystems should consider
effects on ecosystem metabolic properties and how salmon disturbance affects the
incorporation of marine-derived nutrients into food webs.
Key words: Alaska, USA; autotrophy; Bayesian Metabolic Model; ecosystem engineering; ecosystem
metabolism; heterotrophy; Oncorhynchus spp.; oxygen-18; Pacific salmon; primary production; resource
subsidies.
INTRODUCTION
There is a growing appreciation that landscapes are
heterogeneous mosaics of connected ecosystems (Nakano and Murakami 2001). Understanding how nutrients
and energy flow across landscapes from source to sink
ecosystems has captivated the interest of ecologists for
decades and has become increasingly recognized as
important for resource management. Subsidies of
nutrients and energy often have overwhelming importance for supporting the productivity of recipient
ecosystems (Polis et al. 1997). For example, terrestrially
derived organic matter often supports the bulk of
secondary production in small streams and lakes (Fisher
Manuscript received 30 September 2009; revised 28 May
2010; accepted 16 June 2010. Corresponding Editor: A. S.
Flecker.
3 Present address: Box 355020, Seattle, Washington 981955020 USA. E-mail: [email protected]
and Likens 1973, Vannote et al. 1980, Pace et al. 2004).
In many cases, these nutrient and energy subsidies are
represented by foraging movements and migrations of
mobile organisms that move subsidies up potential
energy gradients (i.e., gravity and wind [Polis et al.
1997]). The dynamics of lotic ecosystems are especially
influenced by subsidies both because of the intimate
connections to riparian habitats (Nakano and Murakami 2001, Polis et al. 2004) and because they provide
migration corridors for mobile aquatic species (both
native and invasive) that can have important ecosystem
effects at local scales (Baxter et al. 2004, Taylor et al.
2006).
Changing the nature and extent of subsidies by
altering the interface of adjacent ecosystems, or by
altering the population dynamics or movement patterns
of mobile organisms that produce biotic subsidies, has
profound implications for basic ecosystem processes and
the spatial patterning of these processes across landscapes. For example, dams blocking migrations of
373
374
GORDON W. HOLTGRIEVE AND DANIEL E. SCHINDLER
shrimp and fishes to headwater tropical streams
increased standing stocks of benthic resources (organic
matter, nutrients, invertebrate biomass) with potentially
large indirect effects that permeated throughout local
food webs (Pringle et al. 1993, Greathouse et al. 2006).
Similarly, Taylor et al. (2006) showed that removal of a
migratory detritivorus fish (Prochilodus mariae) from a
tropical piedmont river greatly reduced both organic
carbon export and carbon spiraling length, but increased
both gross primary productivity and ecosystem respiration. This result was somewhat surprising given the
prediction of McIntyre et al. (2007) that removing
Prochilodus would dramatically reduce nutrient recycling and subsequently primary productivity, thus
highlighting the difficulty in forecasting ecosystem
responses to the loss of biodiversity currently occurring
worldwide.
Pacific salmon (Oncorhynchus spp.) represent one of
the most widely acknowledged examples of biologically
mediated spatial subsidies. Upon their migration from
the ocean to freshwater natal habitats where they
eventually spawn and die, salmon can provide substantial subsidies of nutrients, organic matter, and energy to
coastal freshwater and riparian ecosystems throughout
the coastal North Pacific (Naiman et al. 2002, Schindler
et al. 2003). In areas with sizable salmon runs, these
subsidies are transmitted across all trophic levels, and
many species have adapted their life-history strategies to
capitalize on this consistent resource (Gende et al. 2002).
Salmon have also experienced significant population
declines throughout much of their range due to a variety
of impacts including overharvesting, habitat loss, and
pollution (Ruckelshaus et al. 2002). Loss of annual
salmon nutrient pulses to natal freshwater habitats has
been hypothesized to limit primary and secondary
productivity, including important prey for juvenile
salmon (Stockner and MacIsaac 1996, Finney et al.
2000, Stockner 2003). While the dependence of a wide
variety of consumers on salmon has been well documented (Willson and Halupka 1995, Gende et al. 2002),
there is no consensus in the literature regarding the
effect of salmon on aquatic primary productivity or
heterotrophic production (Janetski et al. 2009).
Three main mechanisms have been proposed to
describe how salmon alter freshwater ecosystem functioning: (1) increased primary production and subsequent secondary production as a result of limiting
nutrients (N and P) released from live and dead salmon,
(2) increased secondary production via heterotrophic
pathways associated with postspawning carcass decomposition (e.g., Walter et al. 2006), and (3) physical
habitat modification through bioturbation (Schindler et
al. 2003, Moore 2006). A number of studies have
attempted to link nutrients from spawning salmon to
ecosystem-level processes by focusing on communitylevel changes in benthic primary producer abundance.
Comparisons of standing biomass and stable isotope
Ecology, Vol. 92, No. 2
ratios in algae among streams of varying salmon density
have shown positive effects of marine-derived nutrients
(MDN) on stream periphyton (Wipfli et al. 1998,
Johnston et al. 2004, Chaloner et al. 2007). However,
other similarly designed studies have yielded inconclusive results or negative effects (Ambrose et al. 2004,
Chaloner et al. 2004, Mitchell and Lamberti 2005,
Moore and Schindler 2008), highlighting the complexity
of relating nutrient subsidies and community responses
to ecosystem function.
Despite this general scientific uncertainty, current
conventional wisdom is dominated by a bottom-up
model of how salmon affect stream ecosystem metabolism, in which MDN subsidies of nitrogen (N) and
phosphorus (P) increase in situ primary productivity
and, in turn, subsidize upper trophic levels including
prey that support recruitment of juvenile salmon
(Stockner 2003, Claeson et al. 2006). This hypothesis
has heavily influenced salmon management strategies
and provided the impetus for large-scale restoration
efforts to replace missing nutrients (i.e., carcass additions) currently underway in Oregon, Washington, and
British Columbia (Claeson et al. 2006, Compton et al.
2006). However, to date, no study has addressed the
extent to which autotrophic vs. heterotrophic pathways
dominate organic matter processing in response to
salmon at the scale of whole ecosystems.
Here we present five to seven years of stream nutrient
and epilithon data combined with the first ecosystemlevel measurements of stream metabolism in salmon
streams to test the hypothesis that anadromous salmon
increases stream primary productivity. We used a novel
approach by fitting a dynamic ecosystem metabolism
model to diel variation in oxygen concentrations and
isotopic composition data within a Bayesian statistical
framework. We estimate changes in key ecosystem
characteristics including gross primary production
(GPP), ecosystem respiration (ER), and photosynthesis–irradiance relationships, and their uncertainties,
through time during a typical salmon run to determine
the ecosystem-level responses to this seasonally predictable biotic subsidy.
METHODS
Study sites and salmon surveys
We focused our study during 2002–2009 in Pick
Creek, located within the Wood River system of
southwest Alaska, supplemented with additional metabolism studies in nearby Berm Creek and 7th Creek in
2009 (Fig. 1; Appendix A). Pick Creek is a third-order
stream 6–9 m wide (wetted width) with summertime
discharge ranging from 0.2 to 0.7 m3/s. The stream is
low gradient (0.2%) and dominated by gravel and pebble
substrates. Substrate size (width) at the 50th and 90th
percentiles of the distribution was 17 and 35 mm,
respectively. The stream is composed of pool (36%),
riffle (17%), and glide (46%) reach types and is only
February 2011
SALMON AND ECOSYSTEM METABOLISM
375
FIG. 1. Map of study sites on Lake Nerka within the Wood River system of Bristol Bay, southwestern Alaska, USA. Streams
surveyed for salmon density and ecosystem metabolism measurements are identified by the thickened line segments indicating the
course of the streams in the right panel. We focused our study during 2002–2009 in Pick Creek, located within the Wood River
system, supplemented with additional metabolism studies in nearby Berm Creek and 7th Creek in 2009.
376
GORDON W. HOLTGRIEVE AND DANIEL E. SCHINDLER
minimally shaded by riparian vegetation (Pess 2009; see
Plate 1 and also Fig. A1 in Appendix A). Pick Creek
supports a substantial run of spawning salmon, ranging
from ;1000 to .30 000 returning adults per year (59year mean ¼ 10 200 adults/yr [Rogers and Schindler
2008]). The number of spawning salmon per square
meter of stream area was determined by visual survey
approximately every 7–14 days from early June to early
September from 2002 to 2008 (less often in 2009).
Stream water nutrients and periphyton
Stream water total nitrogen and phosphorus concentrations were measured approximately every 7–14 days
during the summer open water season from 2002 to
2005. Duplicate 60-mL unfiltered water samples were
collected, frozen, and analyzed at the University of
Washington Marine Chemistry Lab by persulfate
digestion and colorimetric analysis. Mass of epilithic
chlorophyll a as an index of algal biomass was
determined at similar time intervals throughout the
season for the years 2002–2008 by methods given in
Moore and Schindler (2008; also Appendix A). For the
years 2005 and 2007, a subsample of periphyton was
analyzed for carbon (C) and N isotopic composition at
either the University of California Stable Isotope
Facility or the University of Washington IsoLab. Data
within each year of sampling were grouped by week and
presented as the mean 6 SD.
Dissolved oxygen concentrations and isotopes
Seasonal sampling for stream metabolic conditions in
Pick Creek was conducted from 28 June to 9 September
2008. Dissolved oxygen concentration ([O2]) and water
temperature were monitored at a downstream station
near the stream outlet at 10-minute intervals using a YSI
6600 V2 sonde equipped with an optical dissolved
oxygen sensor (Yellow Springs Instruments (YSI),
Yellow Springs, Ohio, USA). The stream received .12
hours of daylight throughout the study. Photosynthetically active radiation (PAR) at the surface was
measured at five-minute intervals at a field station 2.4
km from Pick Creek using a LI-COR, LI-192 quantum
sensor (Fig. 1; LI-COR Biosciences, Lincoln, Nebraska,
USA).
We conducted 10 detailed analyses of diel changes in
dissolved oxygen pools, and used these patterns to
estimate in situ metabolic properties in Pick Creek
throughout the 2008 season. During 5 of the 10 diel
sampling periods, we also collected samples for dissolved oxygen isotope ratios (d18O-O2), O2:Ar (measurement of dissolved oxygen concentration), and
13
C:12C of dissolved inorganic carbon (d13C-DIC).
Samples were collected approximately hourly day and
night, with 1–3 hour breaks timed to when oxygen pools
were expected to be near atmospheric equilibrium
conditions. A detailed description of the methods for
dissolved gas ratio collections is in Appendix A.
Ecology, Vol. 92, No. 2
The 2008 seasonal sampling in Pick Creek was
designed to assess patterns in ecosystem metabolism as
a function of salmon density. In 2009, Pick Creek and
two other nearby salmon streams (Berm Creek and 7th
Creek) were also sampled before salmon and during
peak spawning to determine if the dominant patterns
observed in Pick Creek were repeatable and similar
among salmon streams. Berm Creek was also sampled a
third time in 2009 after all of the salmon had finished
spawning and died.
Bayesian estimation of ecosystem metabolic parameters
Diel irradiance, water temperature, [O2], and, when
available, d18O-O2 data were incorporated into a
dynamic aquatic ecosystem metabolism model organized within a Bayesian statistical framework (Holtgrieve et al. 2010a; see Appendix B for model
description). This model simulates stream bulk oxygen
concentrations and isotopic composition as determined
by light-dependent oxygen production via photosynthesis, temperature-dependent oxygen consumption via
respiration, and oxygen exchange between the stream
and the atmosphere dependent on the gas transfer
velocity and concentration gradient. Time periods for
individual diel analysis were chosen based on when
complete data were available and to capture both the
time immediately before salmon entered the stream and
during the period of peak spawning. Results of the
analysis for the key ecosystem metabolism parameters
described in Table 1 are presented as the mean, 2.5%,
and 97.5% credible limits of the posterior distribution.
The parameters aP–I (the initial slope of the photosynthesis–irradiance curve) and R20 (the respiration rate
at standard 208C) are particularly useful in comparing
ecosystem change through time, and in response to
spawning salmon, as they standardize the rates of
photosynthesis and respiration for seasonal changes in
light and water temperature co-occurring with the
presence of salmon. In addition to the specific parameters in Table 1, the Bayesian Metabolic Model
(BaMM) also produces estimates of in situ rates of
GPP, ER, and total O2 mass flux by gas exchange (G)
integrated over a 24-hour period.
Bioenergetics model of salmon metabolism
We used the bioenergetics model of Trudel et al.
(2004) to estimate O2 consumption by live salmon in
Pick Creek during the 2008 season to determine the
proportion of ecosystem metabolism attributable to
respiration by fish in the stream (see Appendix A for the
model description). The model predicts total oxygen
consumption for an individual sockeye salmon as a
function of water temperature, mass, and swimming
speed. Individual estimates of oxygen consumption were
multiplied by the total number of spawning salmon and
converted to moles of O2 consumed by all the salmon in
the stream on a given day.
February 2011
SALMON AND ECOSYSTEM METABOLISM
377
PLATE 1. Pick Creek in southwestern Alaska, USA. The right panel shows post-spawning salmon carcasses. Color versions of
these photographs are available in Appendix A. Photo credits: G. W. Holtgrieve.
RESULTS
Sockeye salmon residence, stream nutrients,
and periphyton
The date on which adult sockeye salmon entered Pick
Creek ranged from 17 to 20 July from 2002 to 2008. Live
salmon were resident in the stream for 7.4 6 0.4 weeks,
with .95% having finished spawning and died by 10
September.
Multiple years of data showed a consistent pattern of
increased total phosphorus (TP) and nitrogen (TN) in
Pick Creek stream water coincident with seasonal
residence of spawning salmon (Fig. 2a, b). Samples
included both particulate and dissolved nutrients and
thus reflect both nutrient excretion and mineralization
directly from salmon, as well as mobilization of
nutrients in sediments through the effect of nest digging
(Moore et al. 2007). Nutrient concentrations were
generally consistent across years and among the weeks
leading up to salmon spawning, averaging 15 6 2 lg/L
and 387 6 74 lg/L (mean 6 SD) for TP and TN,
respectively. The seasonal maximum TP concentration
coincided with peak salmon density during the second
week of August, and averaged 59 6 35 lg/L, roughly
TABLE 1. Bayesian Metabolic Model (BaMM) data and estimated parameters.
Measured diel data
Local time (h)
Model estimated parameters
aP–I (mg O2/mol of photons)
Parameter description
Slope of the system photosynthesis–irradiance
relationship at low light levels.
Maximum rate of photosynthesis at light saturating
Water temperature (8C)
Pmax (mg O2m2h1)
conditions.
Community respiration standardized to a water
Irradiance (lmol of photonss1m2) R20 (mg O2m2h1)
temperature of 208C.
k20 (m/h)
Gas transfer velocity standardized to a water
O2 concentration (mg/L)
temperature of 208C.
Initial O2 concentration (mg/L)
Starting O2 concentration for model predictions.
d18O-O2 (% vs. VSMOW) Initial d18O-O2 (% vs. VSMOW) Starting d18O-O2 concentration for model
predictions.
Standard deviation of the normal likelihood for
r O2 concentration
model fits to the data.
Standard deviation of the normal likelihood for
r d18O-O2
model fits to the data.
VSMOW means Vienna Standard Mean Ocean Water.
378
GORDON W. HOLTGRIEVE AND DANIEL E. SCHINDLER
Ecology, Vol. 92, No. 2
salmon, reflecting the inputs of low N:P nutrients from
salmon tissues (Moore and Schindler 2004).
Salmon are enriched in 15N because of their high
trophic position in the ocean, allowing 15N to be used as
a tracer for marine-derived nutrients from salmon
(Kline et al. 1993). d15N of stream epilithon increased
two- to fourfold after the arrival of salmon with peak
d15N values delayed from peak spawning by ;2 weeks
(Fig. 2c). Based on two-source mixing-model estimates,
23–57% of the N in periphyton was derived from salmon
during the peak spawning period (Fig. 2c; Appendix A).
However, despite the presence of elevated stream water
nutrients, benthic algal biomass consistently declined in
Pick Creek after the seasonal arrival of salmon (Fig. 2d).
On average, algal biomass (milligrams of chlorophyll a
per square meter) was ;10-fold lower during peak
spawning compared to pre-salmon conditions. Soon
after salmon spawning had ceased, algal biomass
rebounded to approximately the levels observed before
salmon spawning (Fig. 2d).
Seasonal dissolved oxygen dynamics
FIG. 2. Seasonal trends (mean 6 SD) in stream water
nutrients, periphyton d15N, and benthic chlorophyll a density in
relation to spawning salmon. Data are grouped by week
starting on 1 June. The shaded area is mean salmon density
from 2002 to 2008 (the dashed lines bound 6 SD). Panels are:
(a) mean total phosphorus (TP), 2002–2005; (b) mean total
nitrogen (TN), 2002–2005. (c) Mean periphyton d15N, 2005 and
2007. Periphyton d15N data (inner axis) were converted to the
percentage of marine-derived nitrogen (outer axis) using a twosource mixing model. (d) Mean chlorophyll a, 2002–2008.
(Appendix A).
four times higher than average pre-salmon concentrations. TP declined to near background levels soon after
salmon spawning (Fig. 2a). TN also peaked with salmon
at 726 6 409 lg/L, but remained elevated late into the
season, possibly the result of a combined effect of both
mineralization of salmon carcasses and late season
storms mobilizing watershed N (Fig. 2b). Year-to-year
variation in TP and TN was greatest during the salmon
spawning period and was correlated with salmon density
(Moore et al. 2007). The amount of N relative to P,
expressed as the molar N:P ratio, was on average 11.2 6
1.2 prior to salmon and declined to 6.3 6 0.6 after
Salmon abundance and dissolved oxygen concentration in Pick Creek were monitored over the majority of
the 2008 open water season and during the period in
which salmon entered and completed their spawning
(Fig. 3). Salmon first entered the stream in large
numbers on 23 July 2008. The maximum number of
live spawning salmon observed was 11 802 on 15 August
2008, corresponding to a density of 0.64 live fish/m2. The
total run for Pick Creek in 2008 was 14 698 fish (0.81
fish/m2).
Seasonal dissolved oxygen concentration was heavily
influenced by the presence of spawning salmon (Fig. 3).
Early in the season, dissolved oxygen changes indicated
substantial metabolic activity with large diel swings,
strong supersaturation during the day, and undersaturation at night. Smoothing across the diel swings in
oxygen concentration showed average conditions near
atmospheric equilibrium, suggesting a roughly balanced
metabolic state (GPP ’ ER; confirmed by diel studies
below). This condition persisted until salmon entered the
stream, at which point average conditions became
uniformly undersaturated. The undersaturated state
persisted for the remainder of the season, with daily
minimum dissolved oxygen values approaching 60% of
saturation, indicating strong oxygen consumption coinciding with peak salmon abundance.
Bayesian estimation of ecosystem metabolic properties
Because of the large diel changes in O2 pools during
the pre-salmon period, it was possible to estimate all the
necessary parameters to obtain ecosystem metabolism
posterior distributions using the Bayesian Metabolic
Model and [O2] data alone; d18O-O2 data were also
included in the model when available. Model fits to diel
[O2] and d18O-O2 from 28 to 30 June 2008 (before
salmon) and 19–21 August 2008 (peak spawning) are
February 2011
SALMON AND ECOSYSTEM METABOLISM
379
FIG. 3. Dissolved oxygen (DO) in Pick Creek prior to and during salmon spawning in 2008. The thinner gray line is DO
recorded at 10-minute intervals expressed as a percentage of saturation concentration when at equilibrium with the atmosphere.
The thick solid line is a LOESS fit to indicate the trend of declining DO in relation to live salmon density (solid circles with dashed
line). The horizontal dashed line indicates conditions at atmospheric equilibrium. Black bars at the top of the graph indicate dates
of ecosystem metabolism analyses (see Methods for details).
given in Appendix B. In general, the model generated
excellent fits to the observed data, particularly during
daytime and nighttime transitions. Before salmon the
model did not fully capture the daytime decline in d18OO2, which is a function of photosynthetic oxygen
production and the d18O of water. Estimates of
ecosystem metabolism parameters using [O2] data alone
agreed well with estimates using both oxygen budgets.
Parameters estimates therefore appear to be relatively
insensitive to small deviations in model fits. During and
after peak spawning it was not possible to simultaneously solve for all the ecosystem metabolic parameters, in particular k20, without including both the 16O
and 18O oxygen budgets due to the weak diel signal in
[O2]. As with pre-salmon conditions, predicted oxygen
trajectories for the after-salmon sampling dates were
highly constrained with small credible intervals (Appendix B), indicating that only a small combination of
parameters were able to provide reasonably likely fits to
the data.
Seasonal trends in model parameters and in situ rates
of GPP and ER showed significant changes to the
metabolic balance in Pick Creek in response to spawning
salmon (Fig. 4). The mean posterior slope of the
photosynthesis–irradiance relationship (aP–I) declined
to ,39% of its pre-salmon value after salmon; average
pre-salmon aP–I was 144 6 138 mg O2 per mole of
photons vs. 376 6 209 mg O2 per mole of photons after
salmon, and only 22 mg O2 per mole of photons on 18–
20 August 2008 with .9400 active spawning salmon in
the stream. Respiration standardized to a constant water
temperature (R20) increased by 115%, from a prespawning average of 438 6 128 mg O2m2h1 to 942
6 850 mg O2m2h1 (and 1110 mg O2m2h1 at the
peak of spawning; Fig. 4a). These parameters standard-
ize for variations in irradiance and water temperature
and thus indicate a substantial shift toward lower
primary productivity and increased heterotrophic respiration with spawning salmon independent of seasonal
changes in irradiance. The changes in metabolic
parameters were reflected in estimates of integrated in
situ metabolism (Fig. 4b). Average post-salmon GPP
was 54% of average pre-salmon rates and only 12% at
peak spawning (mean GPP was 5.1 6 1.0 gm2d1 presalmon and 2.8 6 3.1 gm2d1 post-salmon). ER
increased by 132% on average after salmon, and was
.300% higher during peak spawning (mean ER was 5.6
6 1.6 gm2d1 pre-salmon and 13.0 6 11.7 gm2d1
post-salmon). At the end of the spawning run, when
.98% of the salmon had died, both R20 and in situ ER
declined to well below pre-salmon rates (Fig. 4a, b). The
strong decline in GPP after salmon spawning relative to
before salmon entry into the stream was corroborated
by diel measurement of dissolved inorganic carbon
isotopes (d13C-DIC), which showed substantial daytime
enrichment due to photosynthetic fractionation before
salmon and diminished enrichment after salmon (Appendix C).
Results from the 2009 sampling of Pick, Berm, and
7th creeks match patterns of metabolic change in
response to salmon observed in 2008 from Pick Creek
(Fig. 5; Appendix D). All streams changed from
autotrophic to strongly heterotrophic during peak
spawning (Fig. 5). On average across streams and years,
GPP was just 28% of pre-salmon rates during peak
spawning, while ER increased by 189%. Pick Creek had
the highest salmon densities and showed the largest
response to salmon, particularly in terms of decreased
GPP (Appendix D). Effects of salmon on ER tended to
380
GORDON W. HOLTGRIEVE AND DANIEL E. SCHINDLER
Ecology, Vol. 92, No. 2
FIG. 4. Seasonal changes in ecosystem metabolism metrics in Pick Creek in relation to spawning salmon density. Panel (a)
shows ecosystem respiration standardized to 208C (R20) and the slope of the photosynthesis–irradiance relationship at low light
(aP–I). Panel (b) shows in situ rates of ecosystem metabolism. Data are the mean of the posterior distribution from the Bayesian
Metabolic Model (BaMM). Error bars indicate the 95% credible interval.
respond to the number of salmon per volume, but this
varied among systems.
Salmon also significantly increased the gas transfer
velocity of O2 across the air–water interface (k20) in Pick
Creek during the period of high live salmon densities
(Table 2). Values of k20 increased from 157% to .300%
during salmon spawning, which translated to .700%
increase in O2 mass flux (G in Table 2) because of the
strong undersaturation in dissolved oxygen after salmon. Nonlinear regression of k20 as a function stream
velocity and salmon density included salmon as a
significant predictor (based on AICc), with the best
model explaining 78% of the variation in k20 over the
2008 season (Appendix A). Both k20 and G declined at
the end of the run because the majority of salmon had
died and discharge was ;50% of pre-salmon conditions.
FIG. 5. Metabolic state of three salmon
streams in southwestern Alaska before, during,
and after salmon spawning. Pick Creek was
sampled in both 2008 and 2009, while Berm and
7th Creeks were sampled in 2009 only. Sampling
occurred once before salmon (mid-July), once
during active spawning (early- to mid-August),
and once after the majority of salmon had died
(late August to early September, Berm and Pick
Creeks only). Data are means þ 97.5% of the
credible limit of the posterior distribution from
the Bayesian Metabolic Model. The horizontal
dashed line indicates gross primary production
(GPP) equal to ecosystem respiration (ER).
February 2011
SALMON AND ECOSYSTEM METABOLISM
381
TABLE 2. Air–water gas exchange results from diel metabolism experiments in 2008.
Diel sampling
dates
Salmon density
(live fish/m2)
Discharge
(m3/s)
Velocity
(m/s)
Depth
(m)
28–30 June
9–11 July
14–15 July
17–19 July
20–22 July
29–31 July
13–14 August
18–20 August
29–31 August
4–5 September
0
0
0
0
0
0.27
0.49
0.52
0.16
0.05
0.72
0.58
0.52
0.52
0.46
0.41
0.32
0.28
0.21
0.24
0.33
0.31
0.38
0.29
0.27
0.27
0.24
0.18
0.20
0.22
0.40
0.35
0.33
0.34
0.32
0.34
0.29
0.30
0.26
0.23
k20 (m/h)
0.15
0.16
0.27
0.17
0.15
0.34
0.56
0.29
0.10
0.06
(0.15,
(0.16,
(0.26,
(0.17,
(0.15,
(0.31,
(0.44,
(0.27,
(0.09,
(0.06,
G ,à
(g O2m2h1)
0.16)
0.17)
0.28)
0.18)
0.16)
0.38)
0.73)
0.31)
0.11)
0.06)
2.7
2.6
4.5
2.8
2.9
7.2
24.3
14.7
3.2
1.5
(2.6, 2.8)
(2.4, 2.7)
(4.3, 4.7)
(2.7, 2.8)
(2.8, 3.0)
(6.5, 8.0)
(19.1, 31.9)
(13.6, 15.9)
(3.0, 3.4)
(1.4, 1.5)
Data are mean, 2.5%, and 97.5% credible limits from posterior distributions of model estimates.
à G is total mass flux of O2 via air–water gas exchange.
Contribution of live salmon to ecosystem respiration
Metabolism by live salmon alone could not account
for the full increase in ER observed in Pick Creek (Fig.
6). Based on bioenergetics modeling of metabolism,
respiration by live salmon accounted for 7–86% of postsalmon ER, and the contribution of salmon metabolism
to ER increased as the run progressed. For example, at
one-quarter of the way into the run (1 August), total ER
in Pick Creek had increased by 262% over background
rates, but metabolism by the 4600 salmon in the stream
could only account for 32% of total oxygen consumption. At three-quarters through the run (25 August), ER
had fallen to 80% above pre-salmon background, and
salmon could account for 79% of total ER. This
indicates a large initial heterotrophic response to the
nutrients and benthic disturbance by spawning salmon.
As the stream cleared of sediment over the course of the
spawning season, respiration of in situ organic matter
declined and salmon respiration became a larger fraction
of total ER. d13C-DIC again corroborated these results
which showed modest enrichment at night as the run
progressed. Nighttime d13C-DIC values reflect respired
carbon sources in the absence of photosynthesis; salmon
are an enriched source of carbon compared to both
terrestrial and aquatic organic matter (Appendix C).
There have been mixed results from previous studies
examining the effects of salmon (or salmon nutrients) on
benthic chlorophyll a as a community-level metric to
infer primary productivity (e.g., Wipfli et al. 1998,
Mitchell and Lamberti 2005, Chaloner et al. 2007).
Some of this variation stems from experimental procedures, i.e., carcass addition experiments vs. studies of
natural runs, and many studies have not considered
important interactions between live salmon and the
physical environment (Moore 2006, Janetski et al. 2009).
Chlorophyll density also does not directly scale with
productivity due to factors such as shading and seasonal
variation in PAR. A recent meta-analysis of salmon
studies indicated that, in most cases, chlorophyll a on
small rocks declines in response to seasonal presence of
salmon in streams (Janetski et al. 2009). Our results
confirm this finding and elevates it to full ecosystem
functioning, showing that GPP declined by an order of
magnitude during salmon spawning. Research on
salmon in streams has also generally been biased toward
detecting effects on primary production and has ignored
potentially large effects on heterotrophic processes.
DISCUSSION
Spawning anadromous salmon are the archetypal
example of large-scale fish migrations that are common
worldwide. They are an important resource subsidy to
coastal regions, adding substantial amounts of nutrients
and organic matter to inland aquatic ecosystems (Gende
et al. 2002, Naiman et al. 2002, Janetski et al. 2009) (Fig.
2a, b). Salmon are also ecosystem engineers that can
substantially alter the physical structure of streams,
primarily through digging redds (nests), resulting in the
displacement of sediments and benthic organisms
(Minakawa and Gara 1999, Peterson and Foote 2000,
Moore 2006, Tiegs et al. 2009). These mechanisms have
contrasting effects on ecosystem metabolism, and the
relative balance will determine the ecosystem-level
response to salmon subsidies.
FIG. 6. Pick Creek whole-stream respiration (thick line) and
respiration by live salmon (thin line) based on the bioenergetics
model by Trudel et al. (2004). Live salmon contributed 7–86%
of total O2 consumption, the remainder coming from microbial
respiration.
382
GORDON W. HOLTGRIEVE AND DANIEL E. SCHINDLER
Nitrogen stable isotopes (d15N) of periphyton in Pick
Creek reflected the increase in salmon-derived nutrients
following seasonal salmon entry and the incorporation
of these nutrients into the benthic community (Fig. 2c).
However, the redds of female sockeye salmon are 1–2
m2, and with the densities of salmon found in Pick
Creek, ;50–70% of the stream bed was occupied by
redds. The continuous and repeated turnover of benthic
substrates translated to a steep decline in primary
producer abundance with spawning salmon (Fig. 2d;
Appendix D). We further show that the presence of
salmon switched ecosystem metabolism from autotrophic or near neutral metabolic conditions (GPP ER)
to strong net heterotrophy (GPP ER; Figs. 4 and 5).
This switch to net heterotrophy was caused by both
substantial decreases in GPP and coincident increases in
ER and was consistent across the three streams and two
years of the study (Fig. 5). The effect of bioturbation
from salmon to remove primary producers therefore
dominated over the potential for increased GPP from
nutrient fertilization in these ecosystems.
Despite differences in salmon density among the three
streams surveyed, all streams responded relatively similarly to salmon (Appendix D). Salmon disturb sediments
nearly continuously during spawning, and at the densities
found in the Wood River system, substrates are often
turned over multiple times during a salmon run (Moore
2006). The first pulse of salmon to enter a stream is
therefore likely to have the greatest impact on ecosystem
metabolism, in particular reducing GPP. True control
streams, streams that lack salmon but are otherwise
identical to salmon streams, do not exist in this and most
other salmon regions. In areas with strong salmon stocks,
streams without salmon are different in their geomorphology (e.g., they are usually higher gradient and
susceptible to scour) and biological composition; otherwise salmon would rapidly colonize the available habitat.
Further, the consistent pattern of partial recovery after
salmon spawning (increasing GPP and decreasing ER)
strongly indicates salmon as the primary driver of
metabolic changes observed in these systems.
Autotrophic production can be an important resource
to food webs even when the net carbon balance is neutral
or heterotrophic (Lamberti 1996). Pre-salmon sampling
in this study indicated high rates of GPP and ER. Our
results of decreased primary production and increased
heterotrophic respiration during salmon spawning are in
contrast to the model of MDN (marine-derived nutrients) fertilization, which suggests that the primary effect
of salmon subsidies is to increase primary productivity
by relieving nutrient limitation. Salmon can increase ER
by a number of mechanisms including: (1) mobilizing
sediments, allowing greater accesses and surface area for
heterotrophic microbes; (2) increasing nutrient concentrations, alleviating stochiometric constraints on organic
carbon processing; (3) importation and release of highly
labile carbon from salmon; and (4) the direct consumption of O2 by live salmon.
Ecology, Vol. 92, No. 2
Bioenergetics estimates of O2 consumption by live
salmon in Pick Creek (2008) showed that salmon
metabolism contributed 10–20% of ER early in the
salmon run, rose to 60–85% at peak spawning, and
declined as the salmon died postspawning (Fig. 6).
Suspended sediments follow a similar pattern to ER,
where the highest sediment loads are early in the run and
begin to decline just prior to peak spawning as the
stream is cleared of sediment by nest digging (Moore et
al. 2007). This finding indicates that the mobilization of
sediments in the presence of salmon nutrients fueled
particularly high rates of microbial heterotrophic
production that dominated whole-ecosystem metabolism
(Richey et al. 1975). The stimulation of heterotrophic
processing of organic matter with increased nutrients is
well recognized (Robinson and Gessner 2000, Benstead
et al. 2009), and, combined with disturbance of benthic
substrates, are the primary effects of salmon and MDN
in these streams. Given these results and the extensive
literature demonstrating the support of stream food
webs by allochthonous organic matter, heterotrophic
effects are likely to be a significant and important aspect
of salmon subsidies across all salmon systems.
By subsidizing heterotrophs, accelerating secondary
production, while diminishing primary production,
salmon have the potential to alter the primary pathway
through which organic matter is incorporated into
stream food webs. Rex and Petticrew (2008) demonstrated that salmon nutrients can be incorporated into
benthic organic matter where they can be stored and
possibly rereleased when salmon are no longer in the
stream. That work, performed in an artificial stream,
used salmon nutrient analogs and therefore did not
consider the important effect of salmon disturbance on
these processes.
Our multiple years of data showed that TP rapidly
returned to background levels and benthic algae began
to recover immediately after disturbance from salmon
spawning had ended (Fig. 2a, c). It is possible GPP may
have briefly increased postspawning, which is partially
indicated in the data from Berm and Pick Creeks after
salmon (Fig. 5). However, by late September GPP likely
would have declined with substantially shorter days and
cooler temperatures (Appendix E). Previous work has
also shown that there are few aquatic macroinvertebrates present after salmon spawning (Moore and
Schindler 2008), and juvenile sockeye salmon are not
present in the fall, suggesting any increase in primary
productivity probably did not substantially impact the
aquatic food web. Spring d15N of periphyton in Pick
Creek before salmon (late May–June) wasindistinguishable from nearby streams without salmon, indicating
little annual carryover of MDN among years (Holtgrieve et al. 2010b). A survey of 18 streams in the area
also showed there was no relationship between salmon
density and either periphyton d15N or algal biomass the
following spring (Holtgrieve et al. 2010b). The high rates
of GPP observed in spring across all of our study
February 2011
SALMON AND ECOSYSTEM METABOLISM
streams are therefore not directly attributable to storage
and carryover of MDN in these ecosystems.
Physical features within streams may be important in
controlling ecosystem response to salmon subsidies or
ecosystem engineering. Substrate geomorphology varies
among salmon streams in the Wood River system. Pick
Creek is a relatively simple stream with evenly distributed
salmon spawning (University of Washington Alaska
Salmon Program, unpublished data), and nearly the entire
streambed was easily disturbed through nest digging. In
this case, the physical role of salmon was greater than the
chemical role. In streams with a wider rock size
distribution it may be the case that large rocks are
invulnerable to disturbance by salmon, and that primary
productivity would increase in response to nutrients from
salmon (Tiegs et al. 2008). In a companion study
(Holtgrieve et al. 2010b), we show that rocks .65 mm
in width were relatively invulnerable to salmon disturbance and, in some cases, increased in algal abundance
after salmon. However, small, vulnerable rocks dominate
the aerial distribution of rock size in these sockeye
salmon spawning streams, and whole-stream algal
biomass declined as a function of salmon density.
Sockeye salmon densities in Pick Creek were very
similar to other intensively studied pink and chum
salmon systems in southeastern Alaska and British
Columbia (Chaloner et al. 2004, Hocking and Reimchen
2009, Tiegs et al. 2009). Bristol Bay streams and other
sockeye salmon systems are typically dominated by this
single salmon species, which has a highly concentrated
timing of activity. Other nutrient-limited systems with
more protracted run timing, multiple species, longer
growing season, or greater geomorphic diversity may see
increased GPP from MDN inputs; the prevalence and
extent of such areas has yet to be characterized.
Salmon also fundamentally changed the biogeochemical and physical dynamics of oxygen in Pick Creek.
Nest digging and competitive interactions among
spawners created substantial and continuous disturbance at the air–water interface that was associated with
increased gas transfer velocity (k20) and O2 exchange
with the atmosphere (G; Table 2). High rates of ER kept
Pick Creek undersaturated in dissolved oxygen during
spawning, despite the significant increase in k20 and
large mass flux of O2 into the stream. If the effect of
salmon increasing k20 is removed in our model, dissolved
oxygen concentrations are predicted to have ranged
from just 5.5–6.2 mg O2/L (46–56% of saturation)
during the period of peak spawning, nearing the levels
where physiological functions of salmonids become
highly impaired (;5.0 mg O2/L [Spence et al. 1996]).
383
reviewed the basis for these salmon carcass additions to
coastal streams in Oregon, USA and cited considerable
lack of scientific evidence for ecosystem benefits from
nutrient enrichments, and point to potential negative
consequences, such as the spread in disease and toxins
and deteriorating water quality. All told, a high degree
of scientific uncertainty remains as to the extent,
importance, and mechanisms through which MDN
supports productivity in stream ecosystems. Resource
management based solely on salmon-derived nutrient
dynamics misrepresents the functional roles of salmon in
aquatic ecosystems, and probably does not achieve the
stated goals of restoring the important processes that
salmon perform in ecosystems where populations
remain abundant.
While a detailed understanding of the many roles that
migratory fishes like salmon play in freshwaters has not
yet been achieved, is it apparent that they can
fundamentally change ecosystem functioning. We suggest that the energy provided by salmon to aquatic food
webs and the role they play as sources of physical
disturbance are an equal, if not more important,
external subsidy than nutrients alone. The nutrient
effects of salmon are likely to impact heterotrophic
processes equally or more than autotrophic processes,
and this may be the primary pathway of MDN entry
into food webs. A growing body of literature is
recognizing the role of animals in shaping ecosystems
(Jones and Lawton 1995, Polis et al. 2004). Our example
from sockeye salmon in streams demonstrates how a
single species can rapidly alter the fundamental metabolic properties of ecosystems through a combination of
energetic, disturbance, and nutrient effects.
ACKNOWLEDGMENTS
This is a contribution of the University of Washington (UW)
Alaska Salmon Program supported by the Gordon and Betty
Moore Foundation, Alaska salmon processors, and the
National Science Foundation. G. W. Holtgrieve was generously
supported by the ARCS Foundation, U.S. Environmental
Protection Agency Science to Achieve Results (EPA-STAR)
fellowship program, and the H. Mason Keeler scholarship fund
to the UW School of Aquatic and Fishery Sciences. We thank
the staff of the Wood River State Park, Johnny Evans, and Bill
Berkhahn, for coordination of our field research. C. Boatright,
J. Richey, and A. Schauer provided substantial lab and
logistical assistance. A. Hilborn, P. Lisi, J. Mudra, A. Paulson,
and C. Ruff helped immensely with field work. J. Moore
supplied nutrient data for Pick Creek along with interesting
discussions. Dissolved oxygen isotope samples were analyzed at
the UW School of Oceanography Stable Isotope Lab with the
guidance of M. Haught and P. Quay. M. C. Horner-Devine and
three anonymous reviewers provided helpful comments on the
manuscript.
CONCLUSIONS
LITERATURE CITED
Resource managers have instituted fertilization and
MDN restoration programs in the Pacific Northwest
(Oregon, Washington, and British Columbia) in an
attempt to offset declining salmon returns (Slaney et al.
2003, Compton et al. 2006). Compton et al. (2006)
Ambrose, H. E., M. A. Wilzbach, and K. W. Cummins. 2004.
Periphyton response to increased light and salmon carcass
introduction in northern California streams. Journal of the
North American Benthological Society 23:701–712.
Baxter, C. V., K. D. Fausch, M. Murakami, and P. L.
Chapman. 2004. Fish invasion restructures stream and forest
384
GORDON W. HOLTGRIEVE AND DANIEL E. SCHINDLER
food webs by interrupting reciprocal prey subsidies. Ecology
85:2656–2663.
Benstead, J. P., A. D. Rosemond, W. F. Cross, J. B. Wallace,
S. L. Eggert, K. Suberkropp, V. Gulis, J. L. Greenwood, and
C. J. Tant. 2009. Nutrient enrichment alters storage and
fluxes of detritus in a headwater stream ecosystem. Ecology
90:2556–2566.
Chaloner, D. T., G. A. Lamberti, A. D. Cak, N. L. Blair, and
R. T. Edwards. 2007. Inter-annual variation in responses of
water chemistry and epilithon to Pacific salmon spawners in
an Alaskan stream. Freshwater Biology 52:478–490.
Chaloner, D. T., G. A. Lamberti, R. W. Merritt, N. L. Mitchell,
P. H. Ostrom, and M. S. Wipfli. 2004. Variation in responses
to spawning Pacific salmon among three south-eastern
Alaska streams. Freshwater Biology 49:587–599.
Claeson, S. M., J. L. Li, J. E. Compton, and P. A. Bisson. 2006.
Response of nutrients, biofilm, and benthic insects to salmon
carcass addition. Canadian Journal of Fisheries and Aquatic
Sciences 63:1230–1241.
Compton, J. E., C. P. Andersen, D. L. Phillips, J. R. Brooks,
M. G. Johnson, M. R. Church, W. E. Hogsett, M. A. Cairns,
P. T. Rygiewicz, B. C. McComb, and C. D. Shaff. 2006.
Ecological and water quality consequences of nutrient
addition for salmon restoration in the Pacific Northwest.
Frontiers in Ecology and the Environment 4:18–26.
Finney, B. P., I. Gregory-Eaves, J. Sweetman, M. S. V. Dougas,
and J. P. Smol. 2000. Impacts of climatic change and fishing
on Pacific salmon abundance over the past 300 years. Science
290:795–799.
Fisher, S. G., and G. E. Likens. 1973. Energy flow in Bear
Brook, New Hampshire: integrative approach to stream
ecosystem metabolism. Ecological Monographs 43:421–439.
Gende, S., R. Edwards, M. Willson, and M. Wipfli. 2002.
Pacific salmon in aquatic and terrestrial ecosystems. BioScience 52:917–928.
Greathouse, E. A., C. M. Pringle, W. H. McDowell, and J. G.
Holmquist. 2006. Indirect upstream effects of dams: consequences of migratory consumer extirpation in Puerto Rico.
Ecological Applications 16:339–352.
Hocking, M. D., and T. E. Reimchen. 2009. Salmon species,
density and watershed size predict magnitude of marine
enrichment in riparian food webs. Oikos 118:1307–1318.
Holtgrieve, G. W., D. E. Schindler, T. A. Branch, and Z. T.
A’Mar. 2010a. Simultaneous quantification of aquatic
ecosystem metabolism and re-aeration using a Bayesian
statistical model of oxygen dynamics. Limnology and
Oceanography 55:1047–1063.
Holtgrieve, G. W., D. E. Schindler, C. Gowell, C. P. Ruff, and
P. J. Lisi. 2010b. Stream geomorphology regulates the effects
of ecosystem engineering and nutrient enrichment by Pacific
salmon on periphyton. Freshwater Biology 55:2598–2611.
Janetski, D. J., D. T. Chaloner, S. D. Tiegs, and G. A.
Lamberti. 2009. Pacific salmon effects on stream ecosystems:
a quantitative synthesis. Oecologia 159:583–595.
Johnston, N. T., E. A. MacIsaac, P. J. Tschaplinski, and K. J.
Hall. 2004. Effects of the abundance of spawning sockeye
salmon (Oncorhynchus nerka) on nutrients and algal biomass
in forested streams. Canadian Journal of Fisheries and
Aquatic Sciences 61:384–403.
Jones, C. G., and J. H. Lawton, editors. 1995. Linking species
and ecosystems. Chapman and Hall, New York, New York,
USA.
Kline, T. C., J. J. Goering, O. A. Mathisen, P. H. Poe, P. L.
Parker, and R. S. Scalan. 1993. Recycling of elements
transported upstream by runs of Pacific salmon: delta N-15
and delta C-13 evidence in the Kvichak River watershed,
Bristol Bay, southwestern Alaska. Canadian Journal of
Fisheries and Aquatic Sciences 50:2350–2365.
Ecology, Vol. 92, No. 2
Lamberti, G. A. 1996. The role of periphyton in benthic food
webs. Pages 533–573 in R. J. Stevenson, M. L. Bothwell, and
R. L. Lowe, editors. Algal ecology: freshwater benthic
ecosystems. Academic Press, San Diego, California, USA.
McIntyre, P. B., L. E. Jones, A. S. Flecker, and M. J. Vanni.
2007. Fish extinctions alter nutrient recycling in tropical
freshwaters. Proceedings of the National Academy of
Sciences USA 104:4461–4466.
Minakawa, N., and R. I. Gara. 1999. Ecological effects of a
chum salmon (Oncorhynchus keta) spawning run in a small
stream of the Pacific Northwest. Journal of Freshwater
Ecology 14:327–335.
Mitchell, N. L., and G. A. Lamberti. 2005. Responses in
dissolved nutrients and epilithon abundance to spawning
salmon in southeast Alaska streams. Limnology and
Oceanography 50:217–227.
Moore, J. W. 2006. Animal ecosystem engineers in streams.
BioScience 56:237–246.
Moore, J. W., and D. E. Schindler. 2004. Nutrient export from
freshwater ecosystems by anadromous sockeye salmon
(Oncorhynchus nerka). Canadian Journal of Fisheries and
Aquatic Sciences 61:1582–1589.
Moore, J. W., and D. E. Schindler. 2008. Biotic disturbance and
benthic community dynamics in salmon-bearing streams.
Journal of Animal Ecology 77:275–284.
Moore, J. W., D. E. Schindler, J. L. Carter, J. Fox, J. Griffiths,
and G. W. Holtgrieve. 2007. Biotic control of stream fluxes:
Spawning salmon drive nutrient and matter export. Ecology
85:1278–1291.
Naiman, R. J., R. E. Bilby, D. E. Schindler, and J. M. Helfield.
2002. Pacific salmon, nutrients, and the dynamics of
freshwater and riparian ecosystems. Ecosystems 5:399–417.
Nakano, S., and M. Murakami. 2001. Reciprocal subsidies:
dynamic interdependence between terrestrial and aquatic
food webs. Proceedings of the National Academy of Sciences
USA 98:166–170.
Pace, M. L., J. J. Cole, S. R. Carpenter, J. F. Kitchell, J. R.
Hodgson, M. C. Van de Bogert, D. L. Bade, E. S. Kritzberg,
and D. Bastviken. 2004. Whole-lake carbon-13 additions
reveal terrestrial support of aquatic food webs. Nature 427:
240–243.
Pess, G. R. 2009. Patterns and processes of salmon colonization. Dissertation. University of Washington, Seattle, Washington, USA.
Peterson, D. P., and C. J. Foote. 2000. Disturbance of smallstream habitat by spawning sockeye salmon in Alaska.
Transactions of the American Fisheries Society 129:924–934.
Polis, G. A., W. B. Anderson, and R. D. Holt. 1997. Toward an
integration of landscape and food web ecology: the dynamics
of spatially subsidized food webs. Annual Review of Ecology
and Systematics 28:289–316.
Polis, G. A., M. E. Power, and G. R. Huxel. 2004. Food webs
at the landscape level. University of Chicago Press, Chicago,
Illinois, USA.
Pringle, C. M., G. A. Blake, A. P. Covich, K. M. Buzby, and A.
Finley. 1993. Effects of omnivorous shrimp in a montane
tropical stream—sediment removal, disturbance of sessile
invertebrates and enhancement of understory algal biomass.
Oecologia 93:1–11.
Rex, J. F., and E. L. Petticrew. 2008. Delivery of marinederived nutrients to streambeds by Pacific salmon. Nature
Geoscience 1:840–843.
Richey, J. E., M. A. Perkins, and C. R. Goldman. 1975. Effects
of kokanee salmon (Oncorhynchus nerka) decomposition on
ecology of a subalpine stream. Journal of the Fisheries
Research Board of Canada 32:817–820.
Robinson, C. T., and M. O. Gessner. 2000. Nutrient addition
accelerates leaf breakdown in an alpine springbrook.
Oecologia 122:258–263.
February 2011
SALMON AND ECOSYSTEM METABOLISM
Rogers, L. A., and D. E. Schindler. 2008. Asynchrony in
population dynamics of sockeye salmon in southwest Alaska.
Oikos 117:1578–1586.
Ruckelshaus, M. H., P. Levin, J. B. Johnson, and P. M.
Kareiva. 2002. The Pacific salmon wars: what science brings
to the challenge of recovering species. Annual Review of
Ecology and Systematics 33:665–706.
Schindler, D. E., M. D. Scheuerell, J. W. Moore, S. M. Gende,
T. B. Francis, and W. J. Palen. 2003. Pacific salmon and the
ecology of coastal ecosystems. Frontiers in Ecology and the
Environment 1:31–37.
Slaney, P. A., B. R. Ward, and J. C. Wightman. 2003.
Experimental nutrient addition to the Keogh River and
application to the Salmon River in coastal British Columbia.
Pages 111–126 in J. G. Stockner, editor. Nutrients in
salmonid ecosytems: sustaining production and biodiversity.
American Fisheries Society, Bethesda, Maryland, USA.
Spence, B. C., G. A. Lomnicky, R. M. Hughes, and R. P.
Novitzki. 1996. Ecosystem approach to salmonid conservation. TR-4501-96-6057. ManTech Environmental Research
Services Corporation, Corvallis, Oregon, USA. (Available
from the National Marine Fisheries Service, Portland,
Oregon, USA.)
Stockner, J. G., editor. 2003. Nutrients in salmonid ecosytems:
sustaining production and biodiversity. American Fisheries
Society, Bethesda, Maryland, USA.
Stockner, J. G., and E. A. MacIsaac. 1996. British Columbia
lake enrichment programme: Two decades of habitat
enhancement for sockeye salmon. Regulated Rivers—Research and Management 12:547–561.
385
Taylor, B. W., A. S. Flecker, and R. O. Hall. 2006. Loss of a
harvested fish species disrupts carbon flow in a diverse
tropical river. Science 313:833–836.
Tiegs, S. D., E. Y. Campbell, P. S. Levi, J. Ruegg, M. E.
Benbow, D. T. Chaloner, R. W. Merritt, J. L. Tank, and
G. A. Lamberti. 2009. Separating physical disturbance and
nutrient enrichment caused by Pacific salmon in stream
ecosystems. Freshwater Biology 54:1864–1875.
Tiegs, S. D., D. T. Chaloner, P. Levi, J. Ruegg, J. L. Tank, and
G. A. Lamberti. 2008. Timber harvest transforms ecological
roles of salmon in southeast Alaska rain forest streams.
Ecological Applications 18:4–11.
Trudel, M., D. R. Geist, and D. W. Welch. 2004. Modeling the
oxygen consumption rates in Pacific salmon and steelhead: an
assessment of current models and practices. Transactions of
the American Fisheries Society 133:326–348.
Vannote, R. L., G. W. Minshall, K. W. Cummins, J. R. Sedell,
and C. E. Cushing. 1980. River continuum concept.
Canadian Journal of Fisheries and Aquatic Sciences 37:
130–137.
Walter, J. K., R. E. Bilby, and B. R. Fransen. 2006. Effects of
Pacific salmon spawning and carcass availability on the
caddisfly Ecclisomyia conspersa (Trichoptera: Limnephilidae). Freshwater Biology 51:1211–1218.
Willson, M. F., and K. C. Halupka. 1995. Anadromous fish as
keystone species in vertebrate communities. Conservation
Biology 9:489–497.
Wipfli, M. S., J. Hudson, and J. Caouette. 1998. Influence of
salmon carcasses on stream productivity: response of biofilm
and benthic macroinvertebrates in southeastern Alaska,
USA. Canadian Journal of Fisheries and Aquatic Sciences
55:1503–1511.
APPENDIX A
Expanded materials and methods (Ecological Archives E092-033-A1).
APPENDIX B
Bayesian Metabolic Model description and example fits to data (Ecological Archives E092-033-A2).
APPENDIX C
Diel d13C-DIC data from Pick Creek before and after salmon spawning (Ecological Archives E092-033-A3).
APPENDIX D
Table of ecosystem metabolic rates for three salmon streams in southwestern Alaska before, during, and after salmon in 2008
and 2009 (Ecological Archives E092-033-A4).
APPENDIX E
The 2008 seasonal trends in PAR and water temperature at Pick Creek, Alaska (Ecological Archives E092-033-A5).