Ecological Applications, 18(8) Supplement, 2008, pp. A196–A212
? 2008 by the Ecological Society of America
MERCURY TROPHIC TRANSFER IN A EUTROPHIC LAKE:
THE IMPORTANCE OF HABITAT-SPECIFIC FORAGING
COLLIN A. EAGLES-SMITH,1,4THOMAS H. SUCHANEK,1,2ARTHUR E. COLWELL,3AND NORMAN L. ANDERSON3
1Department of Wildlife, Fish and Conservation Biology, University of California, Davis, California 95616 USA
2U.S. Geological Survey, Western Ecological Research Center, 3020 State University Drive East, Sacramento, California 95819 USA
3Lake County Vector Control District, P.O. Box 310, Lakeport, California 95453 USA
impacted, eutrophic lake were examined in relation to foraging habitat, trophic position, and
size. Diet analysis indicated that there were clear ontogenetic shifts in foraging habitats and
trophic position. Pelagic diet decreased and benthic diet increased with increasing fish length
in bluegill, black crappie, inland silverside, and largemouth bass, whereas there was no shift
for prickly sculpin or threadfin shad. Stable carbon isotope values (d13C) were inversely related
to the proportion of pelagic prey items in the diet, but there was no clear relationship with
benthic foraging. There were distinct differences between pelagic and benthic prey basal d13C
values, with a range of approximately ?28ø in pelagic zooplankton to approximately ?20ø
in benthic caddisflies. Profundal prey such as chironomid larvae had intermediate d13C values
of approximately ?24ø, reflecting the influence of pelagic detrital subsidies and suppressing
the propagation of the benthic carbon isotope signal up the food chain. Fish total mercury
(TotHg) concentrations varied with habitat-specific foraging, trophic position, and size;
however, the relationships differed among species and ages. When controlling for the effects of
species, length, and trophic position, TotHg and d13C were positively correlated, indicating
that Hg trophic transfer is linked to benthic foraging. When examined on a species-specific
basis, TotHg was positively correlated with d13C only for bluegill, largemouth bass, and
threadfin shad. However, diet-based multiple regression analyses suggested that TotHg also
increased with benthic foraging for inland silverside and black crappie. In both species,
benthic prey items were dominated by chironomid larvae, explaining the discrepancy with
d13C. These results illustrate the importance of foraging habitat to Hg bioaccumulation and
indicate that pelagic carbon can strongly subsidize the basal energy sources of benthic
Mercury (Hg) trophic transfer and bioaccumulation in fish from a mine-
habitat; mercury; stable isotopes; Sulphur Bank Mercury Mine; trophic transfer.
bioaccumulation; Clear Lake, California, USA; diet analysis; fish; food webs; foraging
Over the last century there have been widespread
increases in environmental mercury (Hg) contamination
(Evans et al. 1972, Appelquist et al. 1985, Hakanson et
al. 1988, UNEP 2003). In addition, recent studies have
shown elevated Hg concentrations in fish from areas
with low background levels and source inputs (Watras et
al. 1995, Bodaly and Fudge 1999). These findings have
generated significant concern regarding Hg risks to
humans and wildlife (U.S. EPA 1997). In response, a
relatively large body of literature has developed
examining environmental, physiological, and ecological
factors that drive Hg accumulation through food webs
(Wiener and Spry 1996, Gorski et al. 1999, Allen et al.
2005). Methylmercury (MeHg) is generally acknowl-
edged as the form of Hg that is of most concern due to
its toxicity and propensity to biomagnify in food webs
(Kidd et al. 1999). Thus there has been a strong focus on
the limnological factors such as pH (Cope et al. 1990),
hydroperiod (Bodaly and Fudge 1999, Snodgrass et al.
2000), dissolved organic carbon concentrations (Watras
et al. 1998), degree of wetlands (Greenfield et al. 2001),
and primary productivity rate (Pickhardt et al. 2002)
that regulate MeHg production and entry into the food
web. By controlling the input of MeHg to food webs,
these variables play a strong role in determining Hg
concentrations in fish; however, attributes of fish such as
growth rate (Swanson et al. 2006) and age (Evans et al.
2005) are also influencing factors. Moreover, ecological
properties such as trophic position of fish species and
food web structure can dictate the degree of contami-
nation (Cabana and Rasmussen 1994).
Once Hg enters a food web concentrations generally
increase with trophic position, with top predators and
older fish having the highest concentrations. Thus, the
number of trophic levels in a system can be a strong
Manuscript received 9 September 2006; revised 7 May 2007;
accepted 24 May 2007; final version received 21 June 2007.
Corresponding Editor (ad hoc): A. Fairbrother. For reprints of
this Special Issue, see footnote 1, p. A1.
4Present address: U.S. Geological Survey, Western Eco-
logical Research Center, Davis Field Station, One Shields
Avenue, Davis, California 95616 USA.
indicator of bioaccumulation potential (Cabana and
Rasmussen 1994). However, both within and among
systems there can be considerable variation in the
transfer of Hg from one trophic level to another. This
suggests that there are inherent properties of food webs
that affect Hg transfer. Food webs are extremely
complex with extensive omnivory and a great deal of
spatial and temporal variability (Winemiller 1990, Polis
1991, Lawler and Morin 1993, Dunne et al. 2002). They
consist of a network of weak and strong trophic
interactions (linkages) that vary both temporally and
spatially (Dunne et al. 2002). The strength, density, and
habitat specificity of linkages are factors likely to
confound interpretations of Hg bioaccumulation be-
cause the magnitude of trophic transfer is in part
dictated by source Hg concentrations and energetic
strength of the trophic interaction. For example,
habitat-specific foraging can expose various consumers
to differential levels of contamination, particularly if one
taxon has a strong linkage with prey that are more
highly contaminated, resulting in a disparity in Hg
concentrations in consumers occupying similar trophic
positions (Power et al. 2002). This has been demon-
strated with other contaminants (Stewart et al. 2004)
and is likely an important mechanism for determining
species- and habitat-specific accumulation risk.
The advent of stable carbon (d13C) and nitrogen
(d15N) isotopes as tracers of energy flow in food webs
has significantly improved our ability to understand
ecosystem trophic dynamics (Petersen and Fry 1987,
Cabana and Rasmussen 1996, Vander Zanden and
Rasmussen 1999). Classic approaches using gut contents
are valuable for providing taxonomic diet information
or insight regarding diel feeding cycles. However, these
approaches are limited in their applicability to under-
standing energy flow. Gut content analyses are biased
towards recent prey, and unless a population is sampled
extensively, they may reflect transient trends in prey
abundance. They also provide only an analysis of prey
consumed, but not necessarily assimilated. Moreover,
there is substantial variability in the integrity of prey
items in digestive tracts, and as a result, there can be a
strong bias towards prey that decompose slowly or
contain hard parts.
Carbon and nitrogen stable isotopes overcome some
of these limitations and have been used as a robust way
to model complex trophic dynamics (Jepesen and
Winemiller 2002). Stable isotopes, however, rarely
provide taxon-specific information about prey items.
In addition, their proper use requires some general
assumptions about tissue-specific and diet–consumer
fractionation factors, which may not be as generally
applicable as originally proposed (Goedkoop et al.
2006). Moreover, to detect differences in habitat-specific
foraging, there must be a consistent disparity in the
baseline carbon isotope signatures of the habitats
(Vander Zanden and Rasmussen 1999). Because of
weaknesses inherent in both diet and stable-isotope
studies, the two methods complement one another and
when used together can provide a robust assessment of
foraging relationships, trophic dynamics, and contami-
This study examines Hg contamination in the food
web of a eutrophic California lake, Clear Lake. The diet
and isotope signatures of six abundant fish species were
quantified, and their Hg concentrations were assessed
relative to trophic position, foraging habitat, and size.
Diet data were compared with isotope ratios to assess
comparability between methods, with the a priori
hypothesis that diet would better describe foraging
habitat because benthic energy pathways are confounded
by the Clear Lake’s trophic status. Mercury concentra-
tions were compared among fish species and size classes,
and the following questions were asked: (1) Does Hg
vary among fish species? (2) Are Hg concentrations
related to foraging habitats? (3) How important is
foraging habitat relative to fish size and trophic position
in determining Hg concentrations in Clear Lake?
The results of this study are integrated with a larger
ecosystem-based assessment of Hg loading, movement,
and fate through Clear Lake’s abiotic and biotic
matrices. This study provides the linkage between MeHg
entry at the base of the food web, through accumulation
in higher trophic level fishes and piscivorous birds
(Anderson et al. 2008). It also provides a foundation for
understanding the relative importance of both foraging
habitat and trophic position in the transfer of Hg
through successive trophic levels. Moreover, the simul-
taneous use of two complementary methods (diet and
stable isotopes) show that in eutrophic systems with
strong benthic–pelagic coupling, benthic prey of fishes
can be tightly linked to pelagic primary productivity.
Clear Lake (398000N, 1228450W) is a large (17670 ha),
shallow (6.5 m mean depth) lake located in the Coast
Range of northern California, USA (Fig. 1). The lake is
moderately alkaline (pH 7.5–8.5) and, despite its name,
has likely been eutrophic to hyper-eutrophic since long
before European settlers arrived (Osleger et al. 2008,
Richerson et al. 2008). The lake is composed of three
basins, each with its own distinct hydrological and
limnological characteristics (Horne and Goldman 1972).
Along the eastern shoreline of one of those basins, the
Oaks Arm, resides the Sulphur Bank Mercury Mine, a
large open-pit mine that was designated a U.S. Environ-
mental Protection Agency (U.S. EPA) Superfund Site in
1990 and represents a significant point source for
inorganic Hg input to the lake (Suchanek et al. 2008b).
Biannual (early summer and late fall) beach seines
were conducted at two sites in the Oaks Arm (Fig. 1)
from 1985 to 2004 using a 9.131.2 m beach seine with a
3.2-mm ace mesh. To increase the size range of some
December 2008A197MERCURY TROPHIC TRANSFER
species collected, additional fish were taken in the same
vicinity as the monitoring stations in 2000, 2001, and
2004 using a boat-mounted electroshocker. Standard
length (SL) was measured to the nearest 1 mm on all
captured fish. Fish collected for TotHg and stable-
isotope analyses were immediately placed in polyethyl-
ene bags and frozen at?208C until they were processed.
Fish collected for diet analysis were immediately fixed in
10% formalin and later transferred to 70% ethanol until
stomach contents were analyzed. An incision was made
in the peritoneum of fish .100 mm SL to ensure rapid
fixation of stomach contents. Phytoplankton, periphy-
ton, benthic invertebrates, and zooplankton were
collected for TotHg, MeHg, and stable-isotope analyses
using methods outlined in detail in Suchanek et al.
(2008c). Briefly, phytoplankton and zooplankton were
collected with a Van Dorn water sampler (phytoplank-
ton) and Nitex plankton net (80 lm) at 18 locations in
the Oaks Arm between 1992 and 2001. Periphyton was
haphazardly scraped from the surfaces of rocks and tule
stems (Scirpus californicus) along the shoreline in the
Oaks Arm. Periphyton collections were from depths of
,1 m and occurred on several occasions between 1997
and 2000. Littoral invertebrates were collected by hand
from six sites in the Oaks Arm in 1993, 2000, and 2004.
Chironomids were collected with a 15-cm Eckman
dredge from 18 sites in the Oaks Arm between 1992
and 2004. Previous chironomid transect monitoring in
Clear Lake indicated that chironomid densities peaked
(10980 individuals/m2) near the shore in waters 2–3 m
deep (A. E. Colwell, unpublished data). Moreover, the
taxa were dominated almost entirely by those associated
with profundal habitats, Chironomus plumosus and
Procladius sp. (Merritt and Cummins 1996). As a result,
chironomids in this study were classified as profundal
unless they were identified as littoral species.
Gut content analysis
Dissected stomachs were opened and all contents
washed into a gridded Petri dish. Each diet item was
identified to the lowest possible taxon (family, genus, or
species) using taxonomic keys (Edmondson 1959,
Pennak 1989, Thorp and Covich 1991, Moyle 2002).
Each prey item was enumerated, and for each taxon the
proportion of total stomach content was determined on
a volumetric basis. Prior to statistical analyses, all diet
proportion data were arcsine square-root transformed
(Zar 1999), and fish with empty stomachs were noted
but excluded from analysis.
Each prey item was assigned to both a broad
taxonomic classification (e.g., zooplankton, fish, benthic
invertebrate, etc.), and primary habitat type based on
published information for each taxa and/or observations
in Clear Lake (Appendix A). Pelagic taxa were those
residing in the water column, which forage on phyto-
plankton or planktivorous herbivores. Littoral taxa
included those associated with shoreline substrate
shallow enough for light penetration to support
substantial benthic primary productivity. Profundal taxa
were considered to inhabit the substrate at depths at
which light penetration was not adequate to support
primary productivity and basal energy sources are
primarily derived from sedimentary detritus. Digestive
processes often prevented identification of chironiomids
to resolutions lower than order or family. Thus, based
on their abundance throughout the lake, most uniden-
tifiable chironomids in fish guts were assumed to be
profundal. When identification to lower resolutions was
possible, chironomids were almost always profundal
taxa. To estimate fish trophic position using diet data,
each prey item was assigned a trophic position using
literature-based trophic designations (Appendix A).
FIG. 1.Map of Clear Lake, California, USA. Solid circles in inset represent monitoring collection sites.
COLLIN A. EAGLES-SMITH ET AL.A198
Trophic position of each fish was calculated using the
following equation (Winemiller 1990):
TPi¼ 1 þ
where TPiis the trophic position of the ith consumer,
DCijis the proportion of prey j in the diet of consumer i,
TPjis the trophic position of prey j, and G is the number
of diet groups for consumer i.
To facilitate interpretation of diet data and to account
for ontogenetic changes in diet, length–frequency
distributions were examined and each species was
divided into size classes based on interpretation of those
distributions (Eagles-Smith 2006). Linear regression was
used to examine changes in diet proportion within size
classes, and ANOVA was used to test differences
between size classes.
Interpretations of proportional diet data are often
heavily biased when simple means are examined (Bowen
1996), thus modified Costello diagrams (Amundsen et
al. 1996) were developed for each species and size class.
The Costello diagrams incorporate two key descriptive
diet statistics, percentage of occurrence (number of
stomachs containing diet item j 4 total number of
stomachs 3 100) and prey-specific abundance (mean
percentage of volume of diet item j for those individuals
with diet item j in stomachs). The graphical representa-
tion of these two metrics indicates degree of specializa-
tion and prey item importance.
Prior to stable-isotope analysis, skinless axial muscle
samples were dissected from all fish .100 mm SL, and
smaller fish were decapitated, eviscerated, and their skin
and scales removed. Invertebrates were separated from
detritus and sorted to the lowest identifiable taxon. All
samples were rinsed in deionized water, blotted dry, and
weighed to the nearest 0.001 g. Samples were dried at
608C for 48 h or until a constant mass was achieved.
Each sample was then reweighed, ground to a fine
powder, and stored in a dessicator until analysis.
Stable-isotope analysis was performed on a continu-
ous flow isotope ratio mass spectrometer (IRMS; dual-
inlet Europa 20/20, PDZ Europa, Crewe, UK) at the
University of California, Davis Stable Isotope Facility.
Approximately 0.80–1.60 mg of tissue were weighed and
sealed into tin capsules. Sample combustion to CO2and
N2occurred at 10008C in an inline elemental analyzer. A
Carbosieve G column separated the gas before intro-
duction into the IRMS. Standards (PeeDee Belemnite
for d13C and N2gas for d15N) were injected directly into
the IRMS before and after sample peaks. Isotope ratios
are expressed in per mil (ø) notation. Using d13C as an
example, ratios are defined by the following equation:
Evaluation of stable-isotope signatures in food webs
can be confounded by variability in signatures at the
base of the food webs (Vander Zanden and Rasmussen
1999, Post 2002), resulting in difficulty comparing values
(particularly d15N) across habitats. Because there were
no differences in primary consumer d15N values among
habitats in Clear Lake, we used the trophic position
model below, developed by Cabana and Rasmussen
(1996) to estimate trophic position with d15N:
TP ¼ ðd15Nfish? d15NpcÞ=3:4 þ 2
where TP is the trophic position, d15Nfishis the nitrogen
isotope ratio in fish, d15Npcis the nitrogen isotope ratio
in primary consumers (snails, clams, zooplankton), and
3.4 represents the stepwise trophic fractionation of d15N.
Total Hg concentrations were determined on dried
samples prepared as above via cold-vapor atomic absorp-
tion spectroscopy (CV-AAS) using either a CETAC M-
6100 analyzer (CETAC Technologies, Omaha, Nebraska,
USA) or a Milestone DMA-80 analyzer (Milestone,
Monroe, Connecticut, USA). For the CETAC analyzer,
;0.050 g of sample was weighed into a borosilicate glass
culture tube followed by the addition of a 2:1 solution of
concentrated sulfuric and nitric acids. Samples were then
digested at 958C under pressure for 1 h, followed by 2 h of
oxidationvia the addition ofa 5% KMnO4and 5% K2SO8
solution. Upon cooling, a 10% hydroxylamine hydrochlo-
ride, 10% sodium chloride solution was added and the
samples were brought to volume with Milli-Q deionized
water (Millipore, Billerica, Massachusetts, USA). The
samples were then analyzed by CV-AAS using stannous
chloride as the reductant. For the Milestone DMA-80,
;0.05 g of ground sample was weighed into a nickel boat,
and the sample was analyzed following U.S. EPA method
7470, decomposition via combustion followed by gold
amalgamation coupled with CV-AAS.
For both methods quality assurance methods included
analysis of two certified reference materials (dogfish
muscle [DORM-2, National Research Council of Cana-
da, Ottawa, Ontario, Canada], dogfish liver [DOLT-3,
National Research Council of Canada], or lobster
hepatopancreas [TORT-2, National Research Council
of Canada]), two method blanks, two duplicates, two
matrix spikes, and two matrix spike duplicates per batch
of 20 samples. Reference material recoveries averaged
99.2% 6 1.8% (mean 6 SE; N¼106), whereas recoveries
for matrix spikes averaged 101.7% 6 3.2% (N ¼ 89).
Relative percentage of deviation (RPD) averaged 3.2%
for duplicates and 5.2% for matrix spike duplicates.
Analysis of MeHg was conducted by Battelle Marine
Science Laboratory (Sequim, Washington, USA) using
gas chromatography and cold-vapor atomic fluorescence
detection as described in Suchanek et al. (2008c).
Prior to analysis, data were examined to ensure the
underlying distributions met the assumptions of the
December 2008 A199MERCURY TROPHIC TRANSFER
parametric models. Mercury data were natural-log
transformed, and diet proportions were arcsine square-
root transformed. Analysis of variance (ANOVA),
analysis of covariance (ANCOVA), and linear regres-
sion were used to examine the influence of species and
size on diet composition. The ANCOVA and multiple
regression analyses were used to examine the effects of
species, size, trophic position, and habitat-specific diet
on Hg concentrations. An information theoretic ap-
proach, Akaike’s Information Criterion (AIC), was used
to select the most parsimonious model from an a priori
candidate set for explaining Hg concentrations in Clear
Lake fish. This method seeks parsimony between the
variation explained and the complexity of the various
models such that spurious results may be minimized
(Burnham and Anderson 2002).
A systematic, stepwise AIC approach was used
whereby the first step consisted of comparing 34
candidate models. These models included all possible
main-effects combinations of species, trophic position,
habitat-specific foraging (percentage of pelagic reliance),
and standard length, as well as interactions between
species and each covariate, and a null model (intercept
and variance only; Appendix D). The AIC values were
calculated using ANCOVA and multiple regression in
JMP version 5.01 (SAS Institute 2002). The importance
of all interactions in the top model complicated
interpretations of main effects; therefore, a second step
included separate AIC analyses for each species. In this
step, only main effects were assessed since species was no
longer a factor, resulting in eight candidate models per
species containing all possible combinations of trophic
position, pelagic reliance, and length, and the null model
(Appendix E). Finally, because of the nonlinear
responses in diet with age, each species was evaluated
separately by life stage (juvenile and adult) using the
same candidate models described above (Appendix E).
The AICcwas used as a sample size corrected AIC
value, and the model with the lowest AICcvalue was
considered the most parsimonious (Burnham and
Anderson 2002). The difference between each model’s
AICcand that of the best model, DAICc, was calculated
for model ranking, and models were considered for
biological importance when DAICc? 2 (Anderson et al.
2001). Akaike weights were used to assess each model’s
probability of being the Kullback-Leibler best model in
the set of models considered (Burnham and Anderson
2002). Variable importance was assessed using variable
weights, which were determined by summing Akaike
weights across all models incorporating the same
variable (Ackerman et al. 2007).
Bluegill.—Bluegill ranged in size from 11 to 130 mm
SL (mean¼56 mm; n¼274). The importance of pelagic
organisms decreased and littoral organisms increased in
the diet with fish size, and there was no relationship
between profundal prey and fish size (Table 1). When
corrected for year, size class had a significant effect on
pelagic (ANOVA, F2,262¼ 24.78; P , 0.001), littoral
(ANOVA, F2,262¼ 33.70; P , 0.001), and profundal
(ANOVA, F2,262¼ 4.94; P ¼ 0.007) diet. Fish ,45 mm
SL consumed a roughly equal mix of pelagic and benthic
prey, with profundal taxa (primarily Chironomidae),
accounting for nearly all zoobenthos. Fish .45 mm
switched to a more benthos-dominated diet (Fig. 2). The
proportion of Daphnia sp. in bluegill diets did not
change with size, thus the reduction in pelagic diet items
was driven mainly by diminished selection of smaller
zooplankton, such as Bosmina sp., Eurycercus sp., and
Cylcopoida (Appendix B, Fig. 3). The benthic diet for
the larger bluegill size classes shifted from profundal to
littoral dominance (Fig. 2), with amphipods (Hyallela
azteca) becoming the dominant prey item (Appendix B,
Fig. 3). The Costello diagrams (Fig. 3) agreed closely
with the mean proportion data (Appendix B, Fig. 3) and
indicated a progressive increase in littoral prey as fish
grow, although prey-specific abundance changed little.
Except for the littoral prey of ,45 mm bluegill,
relationships between diet and length within size class
were fairly weak or nonsignificant (P . 0.05) for all prey
and size class combinations (Table 2), suggesting that in
general, ontogenetic diet shifts occur between and not
within size classes.
regression models examining the relationship between fish
length and habitat-specific prey for six species in Clear Lake,
Slope direction, correlation, and P value for linear
Variable (%) Slope directionR2
COLLIN A. EAGLES-SMITH ET AL.A200
Black crappie.—Black crappie ranged in size from 17
to 195 mm SL (mean ¼ 64 mm; n ¼ 72). Foraging
patterns shifted significantly with length, particularly for
pelagic and fish prey (Table 1). Black crappie ,55 mm
were almost solely pelagic foragers (Fig. 2), with
Daphnia sp. and Bosmina sp. representing the majority
of the diet (Appendix B, Fig. 3). Fish 55–85 mm
consumed nearly equal proportions of pelagic and
benthic prey (Fig. 2). For these intermediate-sized fish,
benthic prey were mostly littoral invertebrates, including
Baetidae, Corixidae, and amphipods, whereas profundal
diet items were almost exclusively Chironomidae (Ap-
pendix B, Fig. 3). Other fish, mainly inland silverside,
were important prey for crappie .85 mm (Figs. 2 and
3). There were no relationships between fish prey and
length within any size classes. However, pelagic foraging
was negatively correlated with size within the small and
intermediate size classes, and profundal and littoral
foraging reliance increased within the small and
intermediate size classes, respectively (Table 2).
Inland silverside.—Inland silversides ranged in size
from 18 to 84 mm SL (mean ¼ 49 mm; n ¼ 839).
Although linear regression models explained ,10% of
the variance in diet, there was a significant negative
correlation between length and proportion of pelagic
prey and significant positive correlations between length
bars represent 95% confidence intervals.
Mean habitat-specific diet by size class in six fish species from Clear Lake derived from stomach content analysis. Error
December 2008A201 MERCURY TROPHIC TRANSFER
codes: am, amphipod; bg, bluegill; bo, Bosmina; ca, Chaoborus; ch, Chydoridae; ci, chironomids; cp, common carp; cy, cyclopoid;
da, Daphnia; de, detritus; di, Diacyclops; dp, Diaptomus; ep, Ephemeroptera; eu, Eurycercus; ga, gastropods; ha, Harpacticoida; he,
Hemiptera; hy, Hydracarina; is, inland silverside; lb, largemouth bass; le, Leptodora; od, Odonata; os, ostracod; ps, prickly sculpin;
py, phytoplankton; te, terrestrial; ts, threadfin shad; tr, Trichoptera; uf, unidentified fish; un, unknown.
Modified Costello diagrams for six fish species from Clear Lake. Specific diet items are represented with the following
TABLE 2.Regression of relationship between standard length and habitat-specific prey for four species of fish, within size classes.
BluegillBlack crappie Inland silversideLargemouth bass
Notes: The abbreviation ‘‘N/A’’ means that data were not available. Values in boldface correspond to slopes that are significantly
different from zero.
COLLIN A. EAGLES-SMITH ET AL.A202
and proportions of both littoral and profundal diet
items (Table 1). Pelagic prey items made up the majority
of silverside diets regardless of size class. However, each
successive size class consumed less pelagic prey and
more profundal prey, mainly chironomid larvae (Fig. 2).
As with bluegill, the proportion of Daphnia sp. in
silverside diets was not related to size, and the primary
reason reliance on pelagic prey declined was a reduction
of small-bodied zooplankton prey (Fig. 3). There were
no strong relationships between diet and length within
size classes (Table 2).
Largemouth bass.—Largemouth bass ranged in size
from 17 to 303 mm SL (mean¼80 mm; n¼411). Pelagic
and littoral prey decreased and fish prey increased with
bass size; however, length explained ,10% of the
variability in diet (Appendix B, Fig. 3). Diet of bass
,45 mm contained a roughly equal proportion of
pelagic and benthic prey items, with most of the benthic
prey littoral in origin (Fig. 2, Appendix B). Substantial
piscivory (mean ; 40%) first occurred in the 45–115 mm
size class; however, even fish ,45 mm consumed some
fish (Fig. 2). Regardless of size class, fish exhibiting
piscivory generally did so to the exclusion of all other
prey types (Fig. 3). Correlations within size class
between SL and diet were substantial for the young-of-
year (YOY) fish (,115 mm). Bass ,45 mm rapidly
increased their littoral forage while decreasing pelagic
diet, and 45–115 mm bass substituted fish for littoral
prey (Table 2).
Prickly sculpin.—Prickly sculpin ranged in size from
19 to 48 mm SL (mean¼30 mm; n¼148). Diet was not
correlated with length for any habitat-specific prey
classification (Table 1), and sculpin diets were dominat-
ed by benthic prey, composed of nearly equal propor-
tions littoral and profundal taxa (Fig. 2). Profundal diet
was dominated by chironomids and ostracods, and
littoral diet was almost completely composed of
amphipods (Appendix B, Fig. 3).
Threadfin shad.—Threadfin shad ranged in size from
21 to 105 mm SL (mean ¼ 44 mm; n ¼ 162). Shad were
primarily pelagic foragers on small zooplankton (Bos-
mina sp. and Cyclops sp.). Phytoplankton also contrib-
uted a considerable proportion to the total diet (Fig. 3).
The d13C ratios of basal energy sources in the Clear
Lake food web diverged significantly between pelagic
and littoral habitats, with a mean value of ?22ø in
periphyton and ?28ø in phytoplankton (Fig. 4).
Representative primary consumers from each habitat
type (zooplankton, pelagic; amphipods, snails, and
caddisflies, littoral) reflected the d13C signature of their
assumed diet; however, profundal invertebrates (chiron-
omids) occupied an intermediate position between
pelagic and benthic d13C values (Fig. 4). The d13C
values of secondary and tertiary consumers were not
well differentiated from one another, indicating either a
fairly high degree of overlap in habitat-specific prey or a
heavy reliance on profundal resources. The d15N values
ranged from a mean of 0.5ø in Gleotrichia sp. to nearly
12.0ø in piscivores, and baseline values (habitat-specific
primary consumers) did not differ among habitats
(ANOVA, F2,63¼ 0.82; P ¼ 0.44).
The stable-isotope ratios in fish agreed relatively well
with the diet data for pelagic foragers. The d13C values
decreased with increasing percentage of pelagic diet (R2
FIG. 4. Stable-isotope biplot of Clear Lake food web. Error bars represent 95% confidence intervals.
December 2008A203 MERCURY TROPHIC TRANSFER
¼ 0.47, Fig. 5), indicating a greater reliance on benthic
diet items associated with enriched carbon isotope
values. However, the relationship between d13C and
benthic diet was equivocal. Visually there appeared to be
a trend toward increasing d13C values with increasing
littoral and profundal diet; however, linear regressions
showed no significant relationships (P . 0.05) between
diet-based benthic foraging and stable carbon isotope
ratios (littoral, R2¼ 0.05; profundal, R2¼ 0.08; total
benthic, R2¼ 0.04; Fig. 5). The d15N and diet-based
trophic position models had a strong linear correlation
with one another (R2¼ 0.93). The slope of the
relationship between the two (0.95) suggests that the
assumed trophic enrichment value (3.4ø) adequately
described the fractionation of d15N with trophic
position. However, the regression intercept (?0.44)
indicates that the isotope-based index was nearly half
a trophic level less than the diet-based estimates.
Mercury concentrations varied dramatically across
taxa, differing by three orders of magnitude from primary
producers to adult largemouth bass (Fig. 6, Appendix C).
Trophic position as indicated by d15N ratios was a strong
predictor of Hg concentrations for all taxa combined (R2
¼0.89, P , 0.001; Fig. 7); however, there was substantial
variability within species. Mercury concentration also
varied by foraging habitat, as indicated by both diet data
and d13C. Mercury concentrations increased with degree
of littoral, profundal, and total benthic foraging and
decreased with pelagic diet (Fig. 8). This pattern is
Benthic diet represents littoral and profundal items combined.
Relationship between d13C values and habitat-specific diet resources (mean 6 SE) in six species of fish in Clear Lake.
COLLIN A. EAGLES-SMITH ET AL. A204
corroborated by d13C, showing a significant positive
correlation with Hg concentrations (Fig. 9).
Using an information theoretic approach (AIC), the
most parsimonious model explaining Hg concentrations
in Clear Lake fishes included species, trophic position,
pelagic reliance, and length as main effects. This model
clearly provided the best fit relative to the others in the
candidate set as indicated by the next lowest DAIC value
of 12.00 (Appendix D). However, the top model also
included interaction terms between species and all three
covariates, suggesting that each covariate differed
among species with respect to its effect on fish Hg
concentrations (Appendix D). Post hoc AIC analysis
conducted for each species showed that the structure of
the most parsimonious models varied among species
(Appendix E), and length was the only factor that was
consistently included in each model. Because ontogenet-
ic changes in diet often were not linear and dynamics
and retention time of Hg likely differ among different-
aged fish, separate AIC analyses were conducted on
young and adult fish for each species.
Black crappie.—For all black crappie the top model
explaining Hg concentrations contained pelagic reliance
and length as factors and, based on evidence ratios, was
.2.8 times as likely as the next best model, which
included only length and had a DAIC value of 1.99
(Appendix E: Table E1). Variable weights were used to
assess the importance of each factor, and the results
indicate that across their entire size range, length (99%)
was clearly the most important factor, followed by
pelagic reliance (69%) and trophic position (28%). The
top model for young-of-year fish (,85 mm) included
both trophic position and length. The next best model,
which included only length, had a DAIC value of 0.71.
The addition of trophic position to length in the multiple
regression model increased the R2from 0.21 to 0.30. In
adult crappie (.85 mm) length alone accounted for 59%
of the variability in Hg concentrations, and neither
trophic position nor pelagic reliance substantially
improved model fit. No other candidate models had
DAIC values less than 2.00, and the next best model,
which added pelagic reliance to length, had a DAIC of
2.50 and was 3.5 times less likely than the top model.
Bluegill.—For all ages of bluegill, the most parsimo-
nious model included trophic position and length and
had an R2value of 0.51 (Table 3). The only other
competing model in the candidate set added pelagic
reliance as a factor and had a DAIC value of 1.41
(Appendix E: Table E2). The difference in log-likelihood
values of these two models was only 0.45, and there was
and 75th percentiles, and the center box line represents median values. Whiskers represent 10th and 90th percentiles. Open boxes
for periphyton, phytoplankton, caddisfly, amphipod, chironomid, snails, and Daphnia are methylmercury (MeHg) concentrations,
and gray boxes are total mercury (TotHg) concentrations. All fish values are TotHg concentrations (DM, dry mass). Species
abbreviations are: BGS, bluegill; BCR, black crappie; ISS, inland silverside; LMB, largemouth bass; SCP, prickly sculpin; TFS,
Boxplots of mercury concentrations in the Clear Lake food web. Upper and lower box boundaries represent the 25th
December 2008 A205MERCURY TROPHIC TRANSFER
no difference in model R2values, indicating that the
addition of pelagic reliance did not substantially help
model fit. Both length and trophic position appear to be
equally important as evidenced by their Akaike weights
of 1.00 and 0.97, respectively.
When examined by age, Hg in YOY bluegill was also
best predicted by a model that included trophic position
and length and had an R2value of 0.29 (Table 3). Yet
the best predictor of Hg concentrations in adult bluegill
(.85 mm) was length alone (Table 3). However, this
model provides a poor fit overall, only explaining 10% of
the variability in Hg values. The addition of trophic
position to the model only increased the R2to 0.12, and
Akaike weights indicate that neither trophic position
(0.22) nor pelagic reliance (0.24) are tremendously
important to determining adult bluegill Hg concentra-
tions in Clear Lake.
Inland silverside.—The top model for inland silver-
sides across all sizes included trophic position, pelagic
reliance, and length as factors and had an R2value of
0.13 (Table 3). Akaike weights of all three variables were
nearly 1.00, indicating that each variable is equally
important as the others. No other candidate models had
DAIC values below 10, suggesting that the top model
was substantially more likely than any others in the set
(Appendix E: Table E3).
For young silversides (,60 mm), length alone was the
best predictor of Hg concentrations, based on DAIC
values. The only other competing model contained both
length and pelagic reliance and had a DAIC value of
0.92 (Appendix E: Table E3). Akaike weights indicate
that the top model was 1.6 times more likely than its
lone competitor in the candidate set, and the addition of
pelagic reliance increased the R2value from 0.14 to
,0.16. However, pelagic reliance was an important
factor for large silversides (.60 mm), for which Hg
concentrations were described best using pelagic reliance
and length (Table 3). Length was still a more important
variable than pelagic reliance, as suggested by the
competing model, which contained only length and
had a DAIC value of 0.19 (Appendix E: Table E3).
Variable weights also indicate this order of relative
importance, with length having the highest value (0.83),
followed by pelagic reliance (0.55) and trophic position
Largemouth bass.—The top model for the entire size
range of largemouth bass explained 60% of the
variability in Hg concentrations and contained trophic
position, pelagic reliance, and length as independent
variables (Table 3). No other candidate models had
DAIC values ,2.00, and variable weights (all ;0.98–
1.00) indicated that all three variables were equally
In YOY bass (,115 mm), Hg concentrations were
best predicted by a combination of pelagic reliance and
length (R2¼ 0.59; Table 3). The only competing model,
geometric (for TotHg) or arithmetic (for d15N) means. Error bars represent 6SE.
Total mercury (TotHg) concentration (DM, dry mass) vs. d15N in the Clear Lake food web. Each symbol represents
COLLIN A. EAGLES-SMITH ET AL.A206
which had a DAIC value of 0.76, also included trophic
position. However, the addition of trophic position only
improved the R2value to 0.60. Akaike weights indicate
that the top model was 1.5 times more likely than its lone
competitor. In adult bass (.115 mm) however, the top
model included both length and trophic position (Table
3), which together explained 60% of the variability in Hg
concentrations. The only other candidate model with a
DAIC value of ,2.00 contained only length but was
nearly three times less likely than the top model.
Prickly sculpin.—Prickly sculpin Hg concentrations
were also best described with a model composed of
length and trophic position and had an R2value of 0.52
(Table 3). The model containing length alone was the
only competitor in the candidate set, with a DAIC value
of 1.17 (Appendix E: Table E5). Pelagic reliance did not
appear to be of much importance in sculpin, with a
variable weight of only 0.30, relative to trophic position
(0.59) or length (0.99), which clearly carried the most
Threadfin shad.—Pelagic reliance and length together
best predicted Hg concentrations in threadfin shad,
explaining 46% of the variability in the data (Table 3).
The model that also included trophic position had a
DAIC value of 0.76 and only improved the R2value to
0.48. Log-likelihood values of the two models differed
by less than 1.00, suggesting that the addition of trophic
position did not add much value relative to the other
variables (Appendix E: Table E6).
Diet was highly variable among all fish species and
differed significantly among size classes for bluegill,
black crappie, inland silverside, and largemouth bass,
indicating clear ontogenetic diet shifts. Total mercury
concentrations also varied significantly among species
determined by diet analysis. Each symbol represents geometric (for TotHg) or arithmetic (percentage of diet) means 6SE.
Relationship between total mercury (TotHg; DM, dry mass) and habitat-specific foraging in six fish species, as
December 2008A207 MERCURY TROPHIC TRANSFER
and size classes of fish. Across all taxa, stable isotopes
and diet data indicated that Hg bioaccumulation in
Clear Lake fish was influenced by both foraging habitat
and trophic position, increasing with degree of benthic
foraging and trophic position. Baseline d15N ratios did
not differ between benthic and pelagic prey. This
suggests that the relationship between fish Hg and
benthic diet was not a result of elevated trophic position,
position, d13C is correlated with Hg (R2¼0.64, F1,485¼47.52, P , 0.001). On a species-specific basis, TotHg is correlated with d13C
for bluegill (R2¼0.47, P¼0.04, N¼73), largemouth bass (R2¼0.65, P , 0.001, N¼206), and threadfin shad (R2¼0.47, P , 0.001,
N¼37), but not black crappie (P¼0.15, N¼42), inland silverside (P¼0.71, N¼111), or prickly sculpin (P¼0.48, N¼25). Each
symbol represents geometric (for TotHg) or arithmetic (d13C) means 6SE.
Total mercury (TotHg; DM, dry mass) vs. d13C in six fish species. When corrected for length, species, and trophic
full list of candidate models).
Top multiple regression models (DAIC¼0) explaining mercury concentrations in Clear Lake fishes (see Appendix E for
ln(Hg) ¼ ?1.69 – 0.61 3 PR þ 0.01 3 SL
ln(Hg) ¼ ?1.28 – 0.50 3 TP þ 0.014 3 SL
ln(Hg) ¼ ?1.57 þ 0.008 3 SL
ln(Hg) ¼ ?0.32 – 0.55 3 TP þ 0.01 3 SL
ln(Hg) ¼ ?0.15 – 0.62 3 TP þ 0.01 3 SL
ln(Hg) ¼ ?1.08 þ 0.006 3 SL
ln(Hg) ¼ ?1.85 – 0.024 3 TP – 0.058 3 PR þ 0.017 3 SL 0.13
ln(Hg) ¼ ?2.10 – 0.25 3 PR þ 0.025 3 SL
ln(Hg) ¼ ?3.38 þ 0.43 3 PR þ 0.03 3 SL
ln(Hg) ¼ ?1.56 – 0.305 3 TP – 0.59 3 PR þ 0.006 3 SL
ln(Hg) ¼ ?1.69 – 0.69 3 PR þ 0.02 3 SL
ln(Hg) ¼ ?1.75 þ 0.39 3 TP þ 0.004 3 SL
All ln(Hg) ¼ ?0.95 – 0.24 3 TP þ 0.016 3 SL 0.52
All ln(Hg) ¼ ?2.33 – 1.03 3 PR þ 0.02 3 SL
? The length thresholds for juvenile fishes were 85 mm for black crappie and bluegill, 60 mm for silversides, and 125 mm for
? Top multiple regression model for ln(Hg) as determined by Akaike Information Criteria (AIC). Abbreviations are: PR,
percentage of pelagic reliance (arcsine square-root transformed); TP, trophic position; SL, standard length (mm).
§ The likelihood of the model relative to others in the candidate set.
COLLIN A. EAGLES-SMITH ET AL. A208
but due to greater Hg exposure associated with habitat-
specific prey Hg concentrations. Further segregation of
benthic prey into littoral and profundal taxa revealed
that benthic subhabitat was an important distinction for
interpreting fish Hg concentrations; the degree of
profundal foraging was more strongly correlated with
fish Hg concentrations than the degree of littoral
The AIC and multiple regression models indicate that
trophic position, foraging habitat (quantified as per-
centage of pelagic reliance), and length were all very
important in determining Hg concentrations in Clear
Lake fish. However, the structure of the most parsimo-
nious models and the relative importance of variables
explaining Hg concentrations differed among fish
species and between ages. Length was a component of
all top multiple regression models and was at least as
important as the other variables, particularly when
examined across the entire size range of a species. All
three variables were equally important for silverside and
bass; however, the relative importance of foraging
habitat was greater than trophic position for crappie
and shad, whereas trophic position was more important
than habitat in bluegill and sculpin. When separate AIC
analyses were conducted for juvenile and adult fish of
each species, foraging habitat had substantially higher
variable weights than trophic position in 4 of 10 analyses
(silverside ,60 mm, silverside .60 mm, bass ,115 mm,
and shad), trophic position variable weights were higher
in 4 of 10 analyses (crappie , 85 mm, bluegill ,85 mm,
bass .115 mm, and sculpin), and the variable weights
were approximately equal in 2 of 10 analyses (crappie
.85 mm and bluegill .85 mm). These results indicate
that overall foraging habitat was at least as important as
trophic position in determining fish Hg concentrations.
The slopes of the foraging habitat component of
multiple regression models indicate that Hg was
negatively correlated with pelagic reliance and thus
increased with increasing benthic diet. Interestingly, the
parameter estimates for trophic position were negative
for several analyses, indicating that Hg actually de-
creased with trophic position in some cases. This
suggests that the increases in Hg with length and benthic
foraging for many species outweighed the influence of
The Hg mining along Clear Lake’s shoreline has
resulted in some of the most Hg-contaminated sediments
ever reported (Suchanek et al. 2008b). Chironomid
TotHg concentrations were also remarkably high and
tracked the spatial trend of sediment concentrations,
with samples near the mine much higher than in the rest
of the lake (Suchanek et al. 2008a, c). However, MeHg
concentrations in chironomids were one to two orders of
magnitude lower than TotHg and showed no clear
spatial trends, indicating that the inorganic fraction
accounted for nearly all the Hg in chironomids in the
Oaks Arm. In addition, chironomid abundance did not
differ dramatically among arms of the lake (Suchanek et
al. 2008c), suggesting that their availability as prey also
did not vary at this spatial scale. As a result, all things
being equal, fish in the Oaks Arm near the mine were
exposed to significantly more Hg from chironomids than
from their other prey items.
Because of its toxicity and bioaccumulation potential,
much of the recent literature has focused on MeHg as
the species of importance for Hg bioaccumulation in
aquatic systems, suggesting that the assimilation effi-
ciency and bioaccumulation potential of inorganic Hg
are too low to be of concern. In many circumstances this
is likely the case, particularly in areas where inorganic
Hg deposition is low and bioaccumulation is driven by
MeHg production and availability. Allen et al. (2005)
found that biomagnification was inversely correlated
with chlorophyll a concentrations in a series of lakes,
suggesting that biodilution regulated transfer of MeHg
to higher trophic levels. Moreover, MeHg concentra-
tions in biota were negatively correlated with pH,
indicating that entry into the food web was controlled
by water chemistry. In Lake Murray, Papua New
Guinea, inorganic Hg concentrations actually decreased
with trophic position, whereas MeHg accumulations
were successively higher (Bowles et al. 2001). However,
in systems in which Hg inputs are driven by a significant
point source, the trophic dynamics of inorganic Hg may
be equally important. For example, sites close to a point
source in East Fork Poplar Creek, Tennessee, USA, had
inorganic Hg as the dominant species in biota, whereas
MeHg was dominant at downstream and reference areas
(Hill et al. 1996, Southworth et al. 2000). These studies
indicate that in systems dominated by inorganic point
sources benthic infauna are likely to accumulate high
inorganic Hg concentrations, some of which is accumu-
lated in their prey. Although the assimilation efficiency
of inorganic Hg is lower than that of MeHg (Wang and
Wong 2003), the substantially elevated inorganic loads
in Clear Lake chironomids may have corresponded to
higher total Hg values in chironomid feeding fish.
In lakes, bioaccumulation of contaminants such as Hg
generally follows energetic pathways (Vander Zanden
and Rasmussen 1996). Not only can benthic productiv-
ity account for a large proportion of total primary
productivity within lakes (Vadeboncoeur et al. 2003),
but it is often intricately coupled to pelagic energy
dynamics such that benthic energy supports a substan-
tial proportion of upper trophic level production
(Vadeboncoeur et al. 2002, Vander Zanden and Vade-
boncouer 2002). Thus, understanding the relative
contributions of pelagic and benthic energy to successive
trophic levels is critical to quantifying how contaminants
move through food webs. In general, Hg bioaccumula-
tion appears to be lower through benthic pathways than
pelagic ones (Power et al. 2002, Allen et al. 2005);
however, very few attempts have been made to
understand the coupling of benthic and pelagic energy
in higher trophic levels. Benthic secondary productivity
has been shown to be correlated with pelagic energy
December 2008A209MERCURY TROPHIC TRANSFER
sources (Chandra et al. 2005) and may be strongly linked
in eutrophic systems such as Clear Lake (Welch et al.
1988). Moreover, benthic Hg bioaccumulation pathways
can be significant (Wong et al. 1997, Mason et al. 2000),
suggesting that integration of both routes is required for
Clear Lake is relatively eutrophic, so energy fixation is
dominated by pelagic pathways. Chlorophyll a concen-
trations and pelagic productivity rates can exceed 100
lg/L (Tetra Tech, unpublished manuscript) and 2000 mg
C?m?2?d?1(Goldman and Wetzel 1963), respectively.
Although there has been an increase in macrophyte
development along the shoreline in recent years, water
clarity in the littoral zone is still generally too low
(summertime secchi measurements ¼ 1.55 6 0.14 m
[mean 6 SE]; n ¼ 135) to support substantial benthic
carbon fixation. Moreover, in summer months the lake
produces dense cyanobacteria blooms, effectively fur-
ther shading benthic substrate along the shoreline. Yet
Hg transfer to upper trophic levels appears to be
strongly linked to benthic pathways, with Hg contam-
ination and bioaccumulation directly related to degree
of benthic foraging. However, the benthic conduit for
Hg transfer was subsidized by pelagic energy resources.
Profundal prey items, such as chironomid larvae, often
derive much of their nutrients from detrital organic
matter of pelagic origin (Goedkoop et al. 2000). Stable
isotopes support this interpretation, with d13C ratios of
profundal prey significantly depleted relative to those of
These results highlight previous calls (Vander Zanden
and Vadeboncoeur 2002) to integrate both gut content
and stable-isotope data in studies to obtain robust
estimates of foraging ecology and energy flow within
and between systems. This is of particular importance in
systems with strong coupling of benthic–pelagic carbon
and in eutrophic systems in which the benthic d13C
signature is obscured by the substantial pelagic carbon
fixation. In this study, diet appeared to be a better
indicator of foraging habitat of fishes than d13C.
Moreover, the variability in Hg concentrations was
better explained with the diet data than carbon isotope
ratios. Interestingly there were clear distinctions in d13C
between littoral and pelagic invertebrates; however, the
signatures of nearly all fish species were intermediate of
these values, suggesting equal mixing of pelagic and
littoral resources. Diet data indicated a strong reliance
by fish on profundal invertebrates, which had d13C
signatures matching those of many higher order
consumers. Combining the two methods revealed the
circuitous route that pelagic energy traveled from water
column to benthos to fish.
We thank many colleagues for their dedicated field and
laboratory assistance. We are especially grateful to Liz Vignola,
Matt Fine, Rebekka Woodruff, Levi Lewis, Amanda Bern,
Patrick Crain, and Katie Small. Peter Moyle, Dan Anderson,
Peter Green, and Doug Nelson graciously provided access to
analytical equipment and laboratory space. We thank Peter
Moyle, Dan Anderson, Sudeep Chandra, Josh Ackerman,
Patrick Crain, and Darell Slotton for many thoughtful
discussions. We also thank Julie Yee, Karen Phillips, and two
anonymous reviewers for comments that substantially improved
the quality of the manuscript. This work was supported by the
U.S. EPA-funded (R819658 and R825433) Center for Ecolog-
ical Health Research at UC–Davis, the U.S. EPA Region IX
Superfund Program (68-S2-9005) to T. H. Suchanek and UC
Toxic Substances Research and Teaching Program (UC–Davis
Lead Campus), UC–Davis block grant, and Jastro-Shields
fellowships to C. A. Eagles-Smith. Although portions of this
work have been funded wholly or in part by the U.S.
Environmental Protection Agency, it may not necessarily reflect
the views of the Agency, and no official endorsement should be
inferred. Any use of trade, product, or firm names in this
publication is for descriptive purposes only and does not imply
endorsement by the U.S. government.
Ackerman, J. T., C. A. Eagles-Smith, J. Y. Takekawa, S. A.
Demers, T. L. Adelsbach, J. D. Bluso, A. K. Miles, N.
Warnock, T. H. Suchanek, and S. E. Schwarzbach. 2007.
Mercury concentrations and space use of pre-breeding
American avocets and black-necked stilts in San Francisco
Bay. Science of the Total Environment 384:452–466.
Allen, E. W., E. E. Prepas, S. Gabos, W. M. J. Strachan, and
W. Zhang. 2005. Methyl mercury concentrations in macro-
invertebrates and fish from burned and undisturbed lakes on
the Boreal Plain. Canadian Journal of Fisheries and Aquatic
Amundsen, P. A., H. M. Gabler, and F. J. Staldvik. 1996. A
new approach to graphical analysis of feeding strategy from
stomach contents data—modification of the Costello (1990)
method. Journal of Fish Biology 48:607–614.
Anderson, D. R., W. A. Link, D. H. Johnson, and K. P.
Burnham. 2001. Suggestions for presenting the results of data
analyses. Journal of Wildlife Management 65:373–378.
Anderson, D. W., T. H. Suchanek, C. A. Eagles-Smith, and
T. M. Cahill, Jr. 2008. Mercury residues and productivity in
Osprey and grebes from a mine-dominated ecosystem.
Ecological Applications 18(Supplement):A227–A238.
Appelquist, H., I. Drabaek, and S. Asbirk. 1985. Variation in
mercury content of Guillemot feathers over 150 years,
Marine Pollution Bulletin 16:244–248.
Bodaly, R. A., and R. J. P. Fudge. 1999. Uptake of mercury by
fish in an experimental boreal reservoir. Archives of
Environmental Contamination and Toxicology 37:103–109.
Bowen, S. H. 1996. Quantitative description of the diet. Pages
513–529 in B. R. Murphy and D. W. Willis, editors. Fisheries
techniques. Second edition. American Fisheries Society,
Bethesda, Maryland, USA.
Bowles, K. C., S. C. Apte, W. A. Maher, M. Kawei, and R.
Smith. 2001. Bioaccumulation and biomagnification of
mercury in Lake Murray, Papua New Guinea. Canadian
Journal of Fisheries and Aquatic Sciences 58:888–897.
Burnham, K. P., and D. R. Anderson. 2002. Model selection
and inference: a practical information-theoretic approach.
Springer-Verlag, New York, New York, USA.
Cabana, G., and J. B. Rasmussen. 1994. Modeling food chain
structure and contaminant bioaccumulation using stable
nitrogen isotopes. Nature 372:255–257.
Cabana, G., and J. B. Rasmussen. 1996. Comparison of aquatic
food chains using nitrogen isotopes. Proceedings of the
National Academy of Sciences 93:10844–10847.
Chandra, S., M. J. Vander Zanden, A. C. Heyvaert, B. C.
Richards, B. C. Allen, and C. R. Goldman. 2005. The effects
of cultural eutrophication on the coupling between pelagic
primary producers and benthic consumers. Limnology and
COLLIN A. EAGLES-SMITH ET AL. A210
Cope, W. G., J. G. Wiener, and R. G. Rada. 1990. Mercury
accumulation in yellow perch in Wisconsin seepage lakes:
relation to lake characteristics. Environmental Toxicology
and Chemistry 9:931–940.
Dunne, J. A., R. A. Williams, and N. D. Martinez. 2002. Food-
web structure and network theory: the role of connectance
and size. Proceedings of the National Academy of Sciences
Eagles-Smith, C. A. 2006. Mercury in fish: food web structure,
trophic transfer, and bioaccumulation in two California
Lakes. Dissertation. University of California, Davis, Cal-
Edmondson, W. T. 1959. Fresh-water biology. Second edition.
John Wiley and Sons, New York, New York, USA.
Evans, M. S., W. L. Lockhart, L. Doetzel, G. Low, D. Muir, K.
Kidd, G. Stephens, and J. Delaronde. 2005. Elevated
mercury concentrations in fish in lakes in the Mackenzie
River Basin: the role of physical, chemical, and biological
factors. Science of the Total Environment 351–352:479–500.
Evans, R. J., J. D. Bails, and F. M. D’Itri. 1972. Mercury levels
in muscle tissues of preserved museum fish. Environmental
Science and Technology 6:901–905.
Goedkoop, W., N. Akerblom, and M. H. Demandt. 2006.
Trophic fractionation of carbon and nitrogen stable isotopes
in Chironomus riparius reared on food of aquatic and
terrestrial origin. Freshwater Biology 51:878–886.
Goedkoop, W., L. Sonesten, G. Ahlgren, and M. Boberg. 2000.
Fatty acids in profundal benthic invertebrates and their
major food resources in Lake Erken, Sweden: seasonal
variation and trophic indications. Canadian Journal of
Fisheries and Aquatic Sciences 57:2267–2279.
Goldman, C. R., and R. G. Wetzel. 1963. A study of the
primary productivity of Clear Lake, Lake County, Califor-
nia. Ecology 44:283–294.
Gorski, P. R., R. C. Lathrop, S. D. Hill, and R. T. Herrin.
1999. Temporal mercury dynamics and diet composition in
the mimic shiner. Transactions of the American Fisheries
Greenfield, B. K., T. R. Hrabik, C. J. Harvey, and S. R.
Carpenter. 2001. Predicting mercury levels in yellow perch:
use of water chemistry, trophic ecology, and spatial traits.
Canadian Journal of Fisheries and Aquatic Sciences 58:1419–
Hakanson, L., A. Nilsson, and T. Andersson. 1988. Mercury in
fish in Swedish lakes. Environmental Pollution 49:145–162.
Hill, W. R., A. J. Stewart, and G. E. Napolito. 1996. Mercury
speciation and bioaccumulation in lotic primary producers
and primary consumers. Canadian Journal of Fisheries and
Aquatic Sciences 53:812–819.
Horne, A. J., and C. R. Goldman. 1972. Nitrogen fixation in
Clear Lake, California. I. Seasonal variation and the role of
heterocysts. Limnology and Oceanography 17:678–692.
Jepesen, D. B., and K. O. Winemiller. 2002. Structure of
tropical food webs revealed by stable isotope ratios. Oikos
Kidd, K. A., M. J. Paterson, R. H. Hesslein, D. C. G. Muir,
and R. E. Hecky. 1999. Effects of northern pike (Esox lucius)
additions on pollutant accumulation and food web structure,
as determined by d13C and d15N, in a eutrophic and an
oligotrophic lake. Canadian Journal of Fisheries and Aquatic
Lawler, S. P., and P. J. Morin. 1993. Food web architecture and
population dynamics in laboratory microcosms of protists.
American Naturalist 141:675–686.
Mason, R. P., J. M. Laporte, and S. Andres. 2000. Factors
controlling the bioaccumulation of mercury, methylmercury,
arsenic, selenium, and cadmium by freshwater invertebrates
and fish. Archives of Environmental Contamination and
Merritt, R. W., and K. W. Cummins. 1996. An introduction to
the aquatic insects of North America. Third edition.
Kendall/Hunt, Dubuque, Iowa, USA.
Moyle, P. B. 2002. Inland fishes of California. University of
California Press, Berkeley California, USA.
Osleger, D. A., R. A. Zierenberg, T. H. Suchanek, J. S. Stoner,
S. Morgan, and D. P. Adam. 2008. Clear Lake sediments:
anthropogenic changes in physical sedimentology and
magnetic response. Ecological Applications 18(Supplement):
Pennak, R. W. 1989. Fresh-water invertebrates of the United
States. Third edition. John Wiley and Son, New York, New
Petersen, B. J., and B. Fry. 1987. Stable isotopes in ecosystem
studies. Annual Review of Ecology and Systematics 18:293–
Pickhardt, P. C., C. L. Folt, C. Y. Chen, B. Klaue, and J. D.
Blum. 2002. Algal blooms reduce the uptake of toxic
methylmercury in freshwater food webs. Proceedings of the
National Academy of Sciences 99:4419–4423.
Polis, G. A. 1991. Complex trophic interactions in deserts: an
empirical critique of food web theory. American Naturalist
Post, D. M. 2002. Using stable isotopes to estimate trophic
position: models, methods, and assumptions. Ecology 83:
Power, M., G. M. Klein, K. R. R. A. Guiguer, and M. K. H.
Kwan. 2002. Mercury accumulation in the fish community of
a sub-Arctic lake in relation to trophic position and carbon
sources. Journal of Applied Ecology 39:819–830.
Richerson, P. J., T. H. Suchanek, R. A. Zierenberg, D. A.
Olseger, A. C. Heyvaert, D. G. Slotton, C. A. Eagles-Smith,
and C. E. Vaughn. 2008. Anthropogenic stressors and
changes in the Clear Lake ecosystem as recorded in sediment
cores. Ecological Applications 18(Supplement):A257–A283.
SAS Institute. 2002. JMP (the statistical discovery software).
Version 5.0. SAS Institute, Cary, North Carolina, USA.
Snodgrass, J. W., C. H. Jagoe, A. L. Bryan, Jr., H. A. Brant,
and J. Burger. 2000. Effects of trophic status and wetland
morphology, hydroperiod, and water chemistry on mercury
concentrations in fish. Canadian Journal of Fisheries and
Aquatic Sciences 57:171–180.
Southworth, G. R., R. R. Turner, M. J. Peterson, M. A. Bogle,
and M. G. Ryon. 2000. Response of mercury contamination
in fish to decreased aqueous concentrations and loading of
inorganic mercury in a small stream. Environmental Mon-
itoring and Assessment 63:481–494.
Stewart, A. R., S. N. Luoma, C. E. Schlekat, M. A. Doblin, and
K. A. Hieb. 2004. Food web pathway determines how
selenium affects aquatic ecosystems: a San Francisco Bay
case study. Environmental Science Technology 38:4519–
Suchanek, T. H., C. A. Eagles-Smith, and E. J. Harner. 2008a.
Is Clear Lake methylmercury distribution decoupled from
bulk mercury loading? Ecological Applications 18(Supple-
Suchanek, T. H., C. A. Eagles-Smith, D. G. Slotton, E. J.
Harner, and D. P. Adam. 2008b. Mercury in abiotic matrices
of Clear Lake, California: human health and ecotoxicological
implications. Ecological Applications 18(Supplement):A128–
Suchanek, T. H., C. A. Eagles-Smith, D. G. Slotton, E. J.
Harner, D. P. Adam, A. E. Colwell, N. L. Anderson, and
D. L. Woodward. 2008c. Mine-derived mercury: effects on
lower trophic species in Clear Lake, California. Ecological
Swanson, H. K., T. A. Johnston, D. W. Schindler, R. A.
Bodaly, and D. M. Whittle. 2006. Mercury bioaccumulation
in forage fish communities invaded by rainbow smelt
December 2008A211 MERCURY TROPHIC TRANSFER
(Osmerus mordax). Environmental Science and Technology Download full-text
Thorp, J. H., and A. P. Covich. 1991. Ecology and classification
of North American freshwater invertebrates. Academic Press,
New York, New York, USA.
UNEP [United Nations Environment Programme]. 2003.
Global mercury assessment. United Nations Environment
Programme, Chemicals, Geneva, Switzerland.
U.S. EPA [U.S. Environmental Protection Agency]. 1997.
Mercury report to Congress. EPA-452/R-97-008. U.S.
Environmental Protection Agency, Washington, D.C., USA.
Vadeboncoeur, Y., E. Jeppesen, M. J. Vander Zanden, H. H.
Schierup, K. Christoffersen, and D. M. Lodge. 2003. From
Greenland to green lakes: cultural eutrophication and the
loss of benthic pathways in lakes. Limnology and Oceanog-
Vadeboncoeur, Y., M. J. Vander Zanden, and D. M. Lodge.
2002. Putting the lake back together: reintegrating benthic
pathways into lake food web models. BioScience 52:44–54.
Vander Zanden, M. J., and J. B. Rasmussen. 1996. A trophic
position model of pelagic food webs: impact on contaminant
bioaccumulation in lake trout. Ecological Monographs 66:
Vander Zanden, M. J., and J. B. Rasmussen. 1999. Primary
consumer d13C and d15N and the trophic position of aquatic
consumers. Ecology 80:1395–1404.
Vander Zanden, M. J., and Y. Vadeboncoeur. 2002. Fishes as
integrators of benthic and pelagic food webs in lakes.
Wang, W., and R. S. K. Wong. 2003. Bioaccumulation kinetics
and exposure pathways of inorganic mercury and methyl-
mercury in a marine fish, the sweetlips Plectorhinchus
gibbosus. Marine Ecology Progress Series 261:257–268.
Watras, C. J., R. C. Back, S. Halvorsen, R. J. M. Hudson,
K. A. Morrison, and S. P. Wente. 1998. Bioaccumulation of
mercury in pelagic freshwater food webs. Science of the Total
Watras, C. J., K. A. Morrison, and N. S. Bloom. 1995. Mercury
in remote Rocky Mountain lakes of Glacier National Park,
Montana, in comparison with other temperate North
American regions. Canadian Journal of Fisheries and
Aquatic Sciences 52:1220–1228.
Welch, H. E., J. K. Jorgenson, and M. F. Curtis. 1988.
Emergence of Chironomidae (Diptera) in fertilized and
natural lakes at Saqvaqjuac, Northwest Territories [Canada].
Canadian Journal of Fisheries and Aquatic Sciences 45:731–
Wiener, J. G., and D. J. Spry. 1996. Toxicological significance
of mercury in freshwater fish. Pages 297–339 in W. N. Beyer,
G. H. Heinz, and A.W. Redmon-Norwood, editors. Envi-
ronmental contaminants in wildlife: interpreting tissue
concentrations. Lewis, Boca Raton, Florida, USA.
Winemiller, K. O. 1990. Spatial and temporal variation on
tropical fish trophic networks. Ecological Monographs 60:
Wong, A. H. K., D. J. McQueen, D. D. Williams, and E.
Demers. 1997. Transfer of mercury from benthic inverte-
brates to fishes in lakes with contrasting fish community
structures. Canadian Journal of Fisheries and Aquatic
Zar, J. H. 1999. Biostatistical analysis. Prentice-Hall, Upper
Saddle River, New Jersey, USA.
Habitat-specific designations for fish prey items (Ecological Archives A018-078-A1).
Mean percentage of diet of major prey items by fish species and size class (Ecological Archives A018-078-A2).
Mercury concentrations (geometric mean) in Clear Lake fishes (Ecological Archives A018-078-A3).
Akaike Information Criteria values and structure of candidate models describing mercury concentration in Clear Lake fishes
(Ecological Archives A018-078-A4).
Akaike Information Criteria values and structure of candidate models describing mercury concentrations in black crappie,
bluegill, inland silversides, largemouth bass, prickly sculpin, and threadfin shad (Ecological Archives A018-078-A5).
COLLIN A. EAGLES-SMITH ET AL.A212