Arch Environ Contam Toxicol (2017) 72:167177
DOI 10.1007/s00244-016-0353-x
Changes in Sport Fish Mercury Concentrations from Food Web
Shifts Suggest Partial Decoupling from Atmospheric Deposition
in Two Colorado Reservoirs
Brian A. Wolff1 Brett M. Johnson1 Jesse M. Lepak2
Received: 2 September 2016 / Accepted: 21 December 2016 / Published online: 7 January 2017
Springer Science+Business Media New York 2017
Abstract Partial decoupling of mercury (Hg) loading and
observed Hg concentrations ([Hg]) in biotic and abiotic
samples has been documented in aquatic systems. We
studied two Colorado reservoirs to test whether shifts in
prey for sport fish would lead to changes in [Hg] inde-
pendent of external atmospheric Hg deposition. We com-
pared sport fish total mercury concentrations ([T-Hg]) and
macroinvertebrate (chironomids and crayfish) methylmer-
cury concentrations ([MeHg]) before and after food web
shifts occurred in both reservoirs. We also monitored wet
atmospheric Hg deposition and sediment [T-Hg] and
[MeHg] at each reservoir. We found rapid shifts in Hg
bioaccumulation in each reservoirs sport fish, and these
changes could not be attributed to atmospheric Hg depo-
sition. Our study shows that trends in atmospheric depo-
sition, environmental samples (e.g., sediments), and
samples of species at the low trophic levels (e.g., chi-
ronomids and crayfish) may not accurately reflect condi-
tions that result in fish consumption advisories for high
trophic level sport fish. We suggest that in the short-term,
monitoring fish [Hg] is necessary to adequately protect
human health because natural and anthropogenic
& Brian A. Wolff
brian.wolff@colostate.edu
Brett M. Johnson
brett.johnson@colostate.edu
Jesse M. Lepak
salvelinus2005@gmail.com
1
Department of Fish, Wildlife and Conservation Biology,
Colorado State University, 1474 Campus Delivery,
Fort Collins, CO 80523, USA
2
New York Sea Grant Extension, SUNY Oswego, Oswego,
NY 13126, USA
perturbations to aquatic food-webs that affect [Hg] in sport
fish will continue regardless of trends in atmospheric
deposition.
Mercury (Hg) contamination is a serious concern for
human and ecological health worldwide (Mergler et al.
2007). Mercury is particularly harmful to the central ner-
vous system during fetal and early child development and
nonpoint sources of Hg contaminate aquatic food webs
resulting in Hg concentrations in fish that present health
concerns for humans who consume fish (Mergler et al.
2007). Atmospheric deposition of Hg is estimated to be the
principal global driver of Hg bioaccumulation in most
freshwater lakes and reservoirs (UNEP 2002). Regulators
are trying to reduce Hg emissions; however, Hg concen-
trations in sport fish can be decoupled from bulk loading
(Suchanek et al. 2008). The decoupling of Hg loading and
observed methylmercury concentrations [MeHg] in biotic
and abiotic samples has been documented in systems, such
as Clear Lake, California (Suchanek et al. 2008), Baltimore
Harbor and Chesapeake Bay, Maryland (Mason and
Lawrence 1999), four lakes in Voyageurs National Park,
Minnesota (Brigham et al. 2014), and in six Canadian
Shield Lakes, Ontario, Canada (Bodaly et al. 1993). This
suggests that the relationship between Hg loading to
aquatic systems and Hg concentrations in biota is complex
and dependent upon local conditions.
Mercury concentrations in sport fish can often be
explained by food web dynamics (Coelho et al. 2013;
Lavoie et al. 2010; Poste et al. 2015). It has been suggested
that changes in food web structure can have large influ-
ences on Hg bioaccumulation in fish by altering food
availability, growth, and diet (Eagles-Smith et al. 2008;
Johnston et al. 2003; Lepak et al. 2009). Sometimes the
123
168
same species within the same water body may even have
systematic differences in Hg concentrations. For instance,
in a Colorado reservoir [Hg] was higher in male walleye
(Sander vitreus) than females due to differing behaviors
(affecting various metabolic factors) and diets (Lepak et al.
2012a). It also has been experimentally demonstrated that
additions of stocked rainbow trout (Oncorhynchus mykiss;
low [Hg]) can rapidly reduce [Hg] in northern pike (Esox
lucius), via growth dilution on a nearly 1:1 rate (i.e., 20%
increase in growth equaled a 20% reduction in [Hg])
(Lepak et al. 2012b).
In 2010, twenty-four lakes and reservoirs in Colorado
were listed by the Colorado Department of Public Health
and Environment (CDPHE) as impaired due to Hg Fish
Consumption Advisories (FCAs), with total mercury con-
centrations ([T-Hg]) in fish C500 ng/g. As of 2011, twelve
additional water bodies had fish Hg FCAs when Colorado
initiated advisories at 300 ng/g in accordance with EPA
recommendations (USEPA 2009). Unlike other parts of the
United States and Canada (e.g., the Northeast and Great
Lakes regions), and despite emerging awareness of wide-
spread Hg contamination in the western United States,
there have been relatively few studies of Hg cycling and
bioaccumulation in this region.
We examined how food web shifts in two reservoir
communities in Colorado affected sport fish [Hg]. The two
main sport fish targeted by anglers at Elkhead and Horse-
tooth reservoirs are northern pike and walleye, respec-
tively. These sport fish tend to have the highest [Hg] across
the state (www.colorado.gov/cdphe/wq-fish-consumption),
and are popular sport fish for anglers throughout Colorado.
Additionally, these fish are the top predators within each
reservoir and therefore are most likely to respond to
changes in prey availability. Thus, these two sport fish
were the focus of our study. In Elkhead Reservoir, the
dominant prey (stocked rainbow trout) for northern pike
was reduced when stocking was discontinued in 2011.
Conversely, in Horsetooth Reservoir, the dominant prey
(rainbow smelt Osmerus mordax) for walleye increased
(re-established following presumed extirpation from 2000
to 2008). As a result, we hypothesized that top predator
[Hg] would change in opposite directions: (1) in Elkhead
Reservoir, northern pike would have higher [Hg] after the
diet shift from stocked rainbow trout to prey with lower
energy and higher [Hg] (e.g., crayfish), and (2) in Horse-
tooth Reservoir, walleye would have lower [Hg] after a diet
shift from mostly crayfish to prey with higher energy and
lower [Hg] (rainbow smelt), as predicted in a previous
bioenergetics modeling exercise (Johnson et al. 2015). To
test these predictions, we gathered data on [T-Hg] in these
predators before and after food-web shifts. We also eval-
uated if Hg in the environment (reservoir sediments,
atmospheric deposition) could be responsible for the
123
Arch Environ Contam Toxicol (2017) 72:167177
observed differences in predator [Hg] by examining sedi-
ment and invertebrate [MeHg] as indicators of patterns of
Hg bioaccumulation in each reservoir.
Methods
Study Site Description
We studied two Colorado reservoirs with elevated Hg:
Elkhead Reservoir (Moffat and Routt counties) and
Horsetooth Reservoir (Larimer County) (Fig. 1). Full-pool
surface area of Horsetooth is approximately 825 ha,
approximately twice that of Elkhead Reservoir at ca.
364 ha. Horsetooth Reservoirs local watershed is about
44,030 km2. However, Horsetooth Reservoir receives
water from the Colorado-Big Thompson Project (CBT;
watershed area ca. 2,589,990 km2), which transports water
from Colorados west-slope for users on the east-slope.
Elkhead Reservoirs watershed area is relatively small
(approximately 530,948 km2). Elkhead Reservoir has a
volume of 3.056 9 107 m3, and Horsetooth Reservoir has a
volume of 1.933 9 108 m3. Surface elevation at full-pool is
1655 m (m) above sea level at Horsetooth Reservoir and
1924 m at Elkhead Reservoir. Food webs in these systems,
and many other reservoirs in the western United States, are
dynamic because prey populations are unstable and
stocking often is required to maintain prey for piscivorous
fish (Johnson and Goettl 1999; Johnson and Martinez 2000;
Wydoski and Bennett 1981). Historically, when prey fish
were rare at Horsetooth Reservoir walleye diet was domi-
nated by invertebrates (Jones et al. 1994). Less is known
about historical diets of piscivores at Elkhead Reservoir.
Atmospheric Deposition
Atmospheric Hg deposition was determined by participa-
tion with the National Atmospheric Deposition Program
Mercury Deposition Network (NADP-MDN). Deposition
at the Buffalo Pass NADP-MDN site (CO97; latitude
40.5383, longitude -106.6766)the oldest active Hg
deposition monitoring site in Colorado (active since late
September 1998)was used to estimate Hg deposition at
Elkhead Reservoir. Elkhead Reservoir is located ca. 60 km
west of the Buffalo Pass NADP-MDN monitoring station.
A new NADP-MDN station was installed in Fort Collins
(CO13; latitude 40.5923, longitude -105.1414) on June
5, 2012 specifically for this study. Horsetooth Reservoir is
located approximately 1.75 km west of the Fort Collins
NADP-MDN site on Colorado State Universitys (CSU)
Foothills Campus.
Wet T-Hg deposition was measured using a NOAH-IV
automated collector. An ETI Instrument Systems, Inc.
Arch Environ Contam Toxicol (2017) 72:167177
Fig. 1 Map of study reservoirs
and NADP-MDN stations in
Colorado, USA. Locations of
reservoir sediment sampling
locations are depicted with
white diamonds. Black dots
represent the Fort Collins
(CO13), Buffalo Pass (CO99),
and Mesa Verde (CO99)
NADP-MDN stations
precipitation gauge was connected with the NOAH-IV to
provide accurate weekly precipitation amounts for each
deposition sample. The operation of the T-Hg deposition
collector followed the protocol specified by the NADP-
MDN. Pre-cleaned deposition sampling supplies were
provided by NADP-MDN each week. Briefly, 2 L
borosilicate glass bottles with Teflon-lined, phenolic resin
caps were prepared first by a rinse with 0.2 M BrCl, then
soaked in 30% HCl for 24 h, and thoroughly rinsed with
deionized distilled water and dried. Each bottle then
received 20 0.5 mL of 0.12 M HCl (Hg \ 0.5 ng/L).
Funnels and capillary tubes were cleaned by soaking in
30% HCl for 24 h, followed by rinsing in deionized dis-
tilled water and then dried in a clean oven at 100 C for
approximately 2 h and cooled in a laminar flow hood.
Cleaned bottles, funnels, and capillary tubes were placed in
separate new polyethylene bags and packed in shipping
containers.
Sediment Mercury
Sediment samples were collected monthly for [T-Hg] and
[MeHg] from May to September 2013 at the same location
within Elkhead Reservoir (latitude 40.5589, longitude
-107.3842) and Horsetooth Reservoir (latitude 40.5894,
longitude -105.1646). Sediment samples were generally
collected at the deepest points within each reservoir, ca. 20
and 40 m at Elkhead and Horsetooth reservoirs, respec-
tively. Sediment samples were collected using a stainless
steel 230 cubic millimeter (mm3) Ekman core sampler to a
169
depth of approximately 5 cm. A subsample core using a
sharpened section of acid-washed (10% HCl) polyvinyl
chloride (PVC) was taken from within the center of Ekman
to avoid potential contamination from contact with the steel
surface. A plastic spoon was then used to scrape the sedi-
ments from the PVC core and into a clean sample bottle.
Double-bagged samples were handled with two individu-
als, where the outside bag was handled by one individual
with gloved dirty hands, and the inside bag and sample
bottle (precleaned 250-mL I-CHEM 200 series wide-mouth
glass bottles with Teflon-line caps) were handled by
another individual with gloved clean hands. Sample
bottles were triple rinsed with reservoir water to remove
any potential contaminate residues before final collection.
Sediment samples were immediately placed on ice until
delivered to an analytical laboratory.
Mercury in Benthic Invertebrates and Prey Fish
Chironomid (Class Insecta; Order Diptera; Family Chi-
ronomidae) larvae were collected monthly MaySeptember
2013 from the same sites where sediment Hg samples were
taken. Chironomids were gently rinsed with reservoir water
and transferred into clean polyethylene microcentrifuge
tubes (partially filled with distilled water to avoid freezer
burn) and immediately placed on ice and delivered to our
laboratory at CSU for further sample preparation. All
chironomids were analyzed as a composite for each
reservoirs monthly sampling event. Chironomid MeHg
samples were analyzed as dried tissue because high
123
170
surface-area to volume ratio made accurate wet weights
difficult to obtain. Dry weight [MeHg] was then converted
to [MeHg] wet weight (Wet [MeHg] = dry
[MeHg] 9 0.148) assuming an average of 85.2% water
content (James et al. 2012).
Crayfish (Orconectes spp.) were collected using traps on
a monthly basis from June to September 2013 from each
reservoir. Following collection, all individuals were blotted
dry with a clean Kimwipe, measured, and weighed.
Composite samples of three individuals were made for
each sampling time period. All individuals included in a
composite sample were \50 mm carapace length, because
fish predation is inversely related to crayfish size (Stein
1977). Each composite was made by homogenizing equal
weights of tail muscle tissue from each individual. During
the first sampling period at Elkhead Reservoir only two
crayfish were captured, so these two samples were ana-
lyzed individually. All crayfish samples were analyzed for
[MeHg] using frozen tissue (not dried).
Rainbow smelt were collected on October 3, 2013 using
suspended gill nets in the pelagic zone of Horsetooth
Reservoir. Composite samples of three individuals were
made for four size classes: (1) 80100 mm, (2)
100120 mm, (3) 120140 mm, and (4) 140150 mm. Each
composite was made from homogenizing equal weights of
wet muscle tissue. These samples were analyzed for [MeHg].
Composite samples of rainbow smelt analyzed for [T-Hg]
were collected and compiled following the Environmental
Protection Agency guidelines (USEPA 2000). Hatchery
stocked rainbow trout [T-Hg] were obtained from a previ-
ously published study (Lepak et al. 2012b).
Sport Fish Mercury
Northern pike and walleye were collected opportunistically
during routine fish surveys with Colorado Parks and
Wildlife biologists throughout 2013. Fish total lengths (TL;
mm) were obtained on site, and samples were then placed
on ice and transported to Colorado State Universitys
Fisheries Ecology Laboratory for preparation. Fish stom-
ach contents were removed and examined for the presence
of crayfish or fish; fish remains were identified to the
lowest taxonomic level possible based on structures resis-
tant to digestion. Fish muscle tissue was frozen and shipped
to Quicksilver Scientific (Lafayette, CO) for [T-Hg] anal-
ysis (wet mass). Total mercury was analyzed as a surrogate
for MeHg, because MeHg typically comprises ca. 95%
T-Hg in fish (Bloom 1992).
Historical Mercury
To determine if food web shifts at each reservoir were
consistent with the changes in fish [T-Hg], we gathered
123
Arch Environ Contam Toxicol (2017) 72:167177
historical fish [Hg] data (CDPHE 2015) on northern pike
from Elkhead Reservoir (2005) and Horsetooth Reservoir
walleye (2006 and 2008). To determine if baseline [Hg]
changed over time at Horsetooth Reservoir, crayfish and
chironomid [MeHg] data from 2008 were used as a refer-
ence from a previously published report (Lepak and
Johnson 2010).
Analytical Analyses
All MeHg samples were analyzed by Quicksilver Scien-
tific, Lafayette, CO using ion chromatographic separation
of cationic Hg-thiourea complexes, followed by sequential
oxidation of CH3Hg? to HgII, stannous chloride reduction
of HgII to Hg0, evaporation of Hg0 into an Ar carrier,
drying of the sample gas, and finally atomic fluorescence
detection (Shade 2008). Detection limits using this method
are typically 0.001 ng/g for sediment samples. All T-Hg
samples were analyzed by ACZ Laboratories, Steamboat
Springs, CO using EPA Method 1631ECold Vapor
Atomic Fluorescence Spectrometry (CVAFS). Sediment
[T-Hg] was quantified using EPA Method M7473. Limits
of detection of Method M7473 are typically 4 ng/g. Wet
T-Hg deposition samples were analyzed by Frontier Global
Sciences, Bothell, WA, using a modification of EPA
Method 1631. The method detection limit for a 100 mL
sample is approximately 0.1 ng/L.
To determine precision of Hg samples, duplicate sam-
ples were collected at a rate C10% for field sediment and
biota. Each laboratory also implemented their own QA/QC
criteria using standard reference materials, matrix spikes,
and duplicates. Quality control results from Quicksilver
and ACZ laboratories are summarized in Table 1. The
majority of quality assurance measurements fell within
acceptable limits specified by respective laboratories, with
two exceptions: one chironomid duplicate sample had a
relative percent difference (RPD) of [MeHg] at 23.4%
(acceptance range 015%), and one sediment matrix spike
was slightly below MeHg recovery acceptance criteria
(85115%) at 84.5%.
Statistical Analyses
All statistical analyses were performed using SAS software
(SAS Institute Inc., Cary, NC). All fish [T-Hg], fish TL,
sediment [MeHg], and invertebrate [MeHg] values were
Log10(x ? 1) transformed to satisfy assumptions of
homogeneity of variance and normality. Level of statistical
significance was set to a 0.05 alpha (a) value.
To determine if atmospheric deposition was different
between reservoirs, we used Analysis of Variance
(ANOVA; PROC MIXED) to test if weekly wet T-Hg
deposition samples from each NADP-MDN station
 Table 1 Quality control results from Quicksilver Scientific and ACZ laboratories
Matrix spike Duplicates Duplicates Duplicates Lab control

Laboratory Quicksilver Quicksilver Quicksilver ACZ Quicksilver

Arch Environ Contam Toxicol (2017) 72:167177

Hg type MeHg MeHg T-Hg T-Hg MeHg

Acceptance % 85115 015 015 015 85115
MEAN % 95.50 6.47 3.32 7.28 97.64
SD % 7.54 7.34 2.38 1.09 5.67
N 12 8 8 4 5
Reference materials
SQC-1238 DOLT BCR-463 HG130530 HG130710 HG130807 HG130904 HG131011 HG131025 PCN41378 PCN41382

Laboratory Quicksilver Quicksilver Quicksilver ACZ ACZ ACZ ACZ ACZ ACZ ACZ ACZ
Hg type MeHg T-Hg T-Hg T-Hg T-Hg T-Hg T-Hg T-Hg T-Hg T-Hg T-Hg
Acceptance % 85115 85115 85115 90110 90110 90110 90110 90110 90110 80120 80120
MEAN % 109.00 102.00 100.48 102.43 101.68 96.95 105.25 102.43 107.00 109.75 97.98
SD % 1.87 N/A 2.20 2.98 57.09 3.31 2.87 4.68 1.15 0.89 9.37
N 5 1 22 4 3 4 4 8 4 8 6
171

123
172 Arch Environ Contam Toxicol (2017) 72:167177

differed in 2013 (i.e., reservoir effect). To assess if

atmospheric deposition changed over the period of time of
expected food web shifts at Elkhead Reservoir, we used
ANOVA to test if monthly wet T-Hg deposition from
Buffalo Pass changed from 2005 to 2013 (i.e., year
effect). An NADP-MDN deposition site was not estab-
lished near Horsetooth Reservoir until mid 2012. There-
fore, we used ANOVA to test if monthly wet T-Hg
deposition had change from 2005 to 2013 at the Mesa
Verde NAPD-MDN station (CO99; latitude 42.5299,
longitude -108.7200)the only other NAPD-MDN sta-
tion in Colorado operating during that time. If neither
NADP-MDN sites deposition had changed, then we
assumed that deposition did not change at Horsetooth
Reservoir.
Analysis of variance (PROC MIXED) was used for all
statistical comparisons between Elkhead and Horsetooth
reservoirs of sediments, chironomids, and crayfish (2013
data only). We only had chironomid and crayfish [MeHg]
data from Horsetooth Reservoir before 2013, so we tested
the effect of year on Horsetooth Reservoir invertebrate
samples and not Elkhead Reservoir samples (i.e., no
reservoir effect).
Analysis of covariance (ANCOVA; PROC MIXED)
was used for all statistical comparisons of fish [T-Hg].
Since northern pike and walleye were only found at Elk-
head Reservoir and Horsetooth Reservoir, respectively, and Fig. 2 Weekly and yearly wet total mercury (T-Hg) atmospheric
[Hg] often increase with body size, we tested the effects of deposition from NADP-MDN sites, Buffalo Pass (CO97) and Fort
Collins (CO13), in 2013 (a), and monthly wet T-Hg atmospheric
year, TL (total length), and year 9 TL interactions
deposition from NADP-MDN Buffalo Pass site (CO97) and Mesa
separately for each reservoir. Given the potential influence Verde (CO99), 20052013 (b). A upper case letters refer to statistical
of examining fish from different length ranges obtained difference (a = 0.05)
from field sampling in 2013 and in previous years, we
included data that were over the same TL range between Sediment Mercury
years at both reservoirs.
We found that sediment [T-Hg] was significantly greater
(F value = 11.63; p = 0.0092) at Elkhead Reservoir than
Results at Horsetooth Reservoir (Fig. 3). There was a general
pattern of higher sediment [MeHg] at Elkhead Reservoir,
Atmospheric Deposition but it was not statistically significant (F value = 2.39;
p = 0.1608) between reservoirs (Fig. 3).
We found that wet T-Hg atmospheric deposition (ng/m2/
week) was significantly higher at Buffalo Pass NADP- Mercury in Benthic Invertebrates and Prey Fish
MDN site for Elkhead Reservoir (F value = 11.35;
p = 0.0011) than deposition at the Fort Collins NADP- Chironomid [MeHg] was significantly higher
MDN site for Horsetooth Reservoir in 2013 (Fig. 2a). The (F value = 6.39; p = 0.0354) in Elkhead Reservoir than in
sum of deposition was also higher at Elkhead in 2013 Horsetooth Reservoir (Fig. 4); however, crayfish [MeHg]
(Fig. 2a). The NAPD-MDN data also show that monthly was not significantly different (F value = 0.83;
T-Hg deposition had not changed significantly from 2005 p = 0.3986) between reservoirs (Fig. 4).
to 2013 at any of the stations where data are available; Crayfish [MeHg] from Horsetooth reservoir were not
Buffalo Pass (F value = 0.74; p = 0.6547), and Mesa significantly different between 2008 and 2013
Verde (F value = 0.74; p = 0.6527; Fig. 2b). Thus, our (F value = 0.25; p = 0.6384), nor was chironomid
assumption is that T-Hg deposition had not changed at [MeHg] (F value = 1.66; p = 0.2146). Thus, [MeHg] at
Horsetooth Reservoir. the base of the food web from Horsetooth Reservoir did not

123
Arch Environ Contam Toxicol (2017) 72:167177
Fig. 3 Mean (95% CI) sediment total mercury concentrations ([T-
Hg]) and methylmercury concentrations ([MeHg]) at Elkhead and
Horsetooth reservoirs. Sediment samples were collected once per
month from May to September 2013 in each reservoir. Sediment [T-
Hg] and [MeHg] are represented on a wet weight basis. Uppercase
letters refer [MeHg] comparisons and lowercase letters refer to [T-
Hg]; the same letters mean no statistical difference (a = 0.05)
Fig. 4 Mean (95% CI) methylmercury concentrations ([MeHg]) of
crayfish and chironomids) from Elkhead and Horsetooth reservoirs
(collected once per month from May to September 2013), rainbow
smelt from Horsetooth Reservoir (collected October 2013), and
rainbow trout from Colorado Parks and Wildlife Bellevue Fish
Hatchery (collected August 2008). All [MeHg] are presented on a wet
weight basis. Fish total mercury concentrations served as a surrogate
for [MeHg]. Uppercase letters refer crayfish comparisons and
lowercase letters refer to chironomid comparisons, separately; the
same letters mean no statistical difference (a = 0.05)
change between 2008 and 2013, suggesting that internal
loading (e.g., methylation rates and bioavailability) in
Horsetooth Reservoir did not change.
Rainbow smelt are only found in Horsetooth Reservoir,
and rainbow trout samples were sampled from Colorado
Parks and Wildlifes Bellevue Fish Hatchery, so there was
no statistical comparison between reservoirs. However, it
was clear that rainbow smelt and rainbow trout [T-Hg]
173
(surrogate for MeHg) was lower than crayfish [MeHg]
(Fig. 4).
Sport Fish Mercury
In Elkhead Reservoir, there was a limited size range of
northern pike available from 2005 (392625 mm TL;
n = 5)when trout stocking was occurringand the size
range in 2013 was substantially greater (178965 mm TL;
n = 28). Thus, we truncated the 2013 data to only include
fish [T-Hg] between 392 and 625 mm TL. Following this
adjustment, we tested the effects of year between 2005
(stocked) and 2013 (not stocked), TL, and year 9 TL
interaction. The year 9 TL interaction was not significant
(F value = 0.07; p = 0.7966). After removing the
year 9 TL interaction, TL was no longer significant
(F value = 1.09; p = 0.3109), but northern pike [T-Hg]
remained significantly higher (F value = 21.79;
p = 0.003) in 2013 than in 2005 (Fig. 5a). A T test of
combined northern pike [T-Hg] between 2005 (stocked)
versus 2013 (not stocked) showed significantly higher [T-
Hg] in 2013 than in 2005 (T value = 22.48; p = 0.0002),
with 0 of 5 fish over the Colorado FCA advisory level
(300 ng/g T-Hg) while stocked and 10 of 14 fish over the
FCA when not stocked (Fig. 5b). Changes in Hg were
consistent with observed diets before and after stocking
ceased. Northern pike collected during a period of stocking
had rainbow trout in 7 of 18 (*39%) stomachs, and
crayfish in 2 of 18 (*11%) stomachs (B.M. Johnson,
unpublished data 2011). However, post-stocking in 2013,
rainbow trout were in 0 of 36 (0%) northern pike stomachs,
and crayfish were in 10 of 36 (28%) stomachs.
In Horsetooth Reservoir, walleye collections resulted in
uneven fish size comparisons before the rainbow smelt
reestablishment (20062008; 250425 mm TL; n = 48)
and after smelt returned (2013; 184565 mm TL; n = 30).
Therefore, we truncated the data to only include [T-Hg]
data from fish between 250 and 425 mm TL, and subse-
quently tested the effects of year between 2006 and 2008
(pre-smelt) and 2013 (post-smelt), TL, and year 9 TL
interaction. The year 9 TL interaction for walleye [T-Hg]
was not significant (F value = 0.10; p = 0.7579). After
removing the interaction effect, walleye [T-Hg] increased
significantly with TL (F value = 61.79; p \ 0.0001), and
was significantly lower (F value = 21.79; p = 0.003) in
2013 following the reestablishment of rainbow smelt in
Horsetooth Reservoir (Fig. 5c). A T test of combined
walleye [T-Hg] between 2006 and 2008 (pre-smelt) versus
2013 (post-smelt) showed significantly lower [T-Hg] in
2013 than in 2006 and 2008 (T value = 28.76;
p \ 0.0001), with 40 of 48 fish over the Colorado FCA
advisory level (300 ng/g T-Hg) pre-smelt reestablishment
123
174 Arch Environ Contam Toxicol (2017) 72:167177

Fig. 5 Log-transformed total

mercury concentrations,
Log10([T-Hg] ?1), versus log-
transformed total length,
Log10(TL ?1), for Elkhead
Reservoir northern pike in 2005
(stocked) and 2013 (not
stocked) (a), and Horsetooth
Reservoir walleye in 2006 and
2008 (pre-smelt) and 2013
(post-smelt) (c). Boxplots of
untransformed fish total
mercury concentrations, [T-Hg],
of Elkhead Reservoir northern
pike in 2005 (stocked) and 2013
(not stocked) (b), and
Horsetooth walleye in 2006 and
2008 (pre-rainbow smelt) and
2013 (post-rainbow smelt) (d).
The dark horizontal line (b,
d) is the Colorado T-Hg Fish
Consumption Advisory level
(FCA; 300 ng/g), and number of
fish exceeding the FCA are
shown above each boxplot

and 3 of 19 fish over the FCA post-smelt reestablishment
(Fig. 5d). We do not have diet samples from walleye col-
lected in 2006 and 2008 (pre-smelt), but biologists reported
that crayfish were the predominant prey during that period
(Johnson et al. 2015). Walleye examined in 2013 contained
zero crayfish, and the only fish observed were rainbow
smelt (5/134, *4%) and rainbow trout (1/134, \1%).
Given the very low number of any contents from our
walleye samples suggests that they regurgitated stomach
contents during capture by the gill nets.
Discussion
This study provided a unique opportunity to investigate
how food web shifts can both increase and decrease [T-Hg]
of sport fish at the whole system scale. Other studies sup-
port the effects of food web shifts on predator diets and
food availability in our study reservoirs. In Elkhead
Reservoir in 2009, when trout stocking was ongoing, 50%
of the northern pike sampled had recently consumed rain-
bow trout, and body condition [relative weight, Wr;
(Blackwell et al. 2000)] was high, indicating that prey were
abundant (Wright 2010). At Horsetooth Reservoir, rainbow
smelt abundance increased from undetectable levels in
2010 to an estimated 2.5 9 106 fish in 2013, and walleye
body condition (Wr) increased by 20% (Davies 2014).
Changes in stocking policy (Elkhead Reservoir) and a
natural resurgence of prey fish (Horsetooth Reservoir)
caused food web shifts with implications for Hg
bioaccumulation.
Consistent with our hypotheses, northern pike [T-Hg]
was significantly higher in 2013 (without trout stocking)
than in 2005 (with trout stocking), and walleye [T-Hg] was
significantly lower in 2013 (post-rainbow smelt reestab-
lishment) than in 2006 and 2008 (pre-rainbow smelt). Our
results agreed with estimates from a recent modelling study
that predicted a shift to a diet with higher calories and
lower [Hg] (e.g., rainbow smelt) could drive the [Hg] in
walleye below the EPA consumption advisory level
(0.3 ppm) (Johnson et al. 2015).
We believe that these observed changes in [T-Hg] were
not the result of reduced atmospheric deposition because
atmospheric deposition near Elkhead Reservoir did not
change during that period. The NADP-MDN site near
Horsetooth Reservoir was not established until 2013, but
data from the Mesa Verde NADP-MDN site also did not
indicate any significant changes over that time period. We
cannot determine if there was a reduction in internal
loading of Hg to the base of the food-web over the study
period; however, this seems unlikely since [MeHg] mea-
sured in chironomids and crayfish at Horsetooth Reservoir
in 2013 and 2008 were not different. Thus, we conclude
that food web shifts at both reservoirs caused reciprocal

123
Arch Environ Contam Toxicol (2017) 72:167177
changes in sport fish [Hg], which were not related to
atmospheric deposition or [MeHg] in chironomids and
crayfish over the same period.
We estimated that Elkhead Reservoir received signifi-
cantly greater atmospheric wet T-Hg deposition than at
Horsetooth Reservoir. Increased Hg deposition at Elkhead
Reservoir likely resulted in greater [T-Hg] in sediments
than found at Horsetooth Reservoir. Despite higher con-
centrations of sediment [T-Hg] at both reservoirs, the
sediment [MeHg] did not have similar patterns. This result
was not particularly surprising since correlations between
[T-Hg] and [MeHg] in sediments have been inconsistent in
other studies (Kannan et al. 1998; Mason and Lawrence
1999; Mikac et al. 1999). Although not statistically sig-
nificant, Elkhead Reservoirs sediment [MeHg] may have
been high enough to have a biological effect, which could
have resulted in greater [MeHg] in chironomids at Elkhead
Reservoir.
In contrast to chironomid [MeHg], we did not observe
any differences in crayfish [MeHg] between reservoirs.
This demonstrates the importance of dietary sources on Hg
bioaccumulation. Crayfish are omnivores that can consume
food from decaying matter from all trophic levels of the
food-chain (Momot 1995). Therefore, crayfish likely
bioaccumulate Hg from a variety of dietary sources at
different trophic levels. Thus, crayfish [MeHg] may not
reflect bioaccumulation from sediments as seen with
chironomids.
It has been shown that species, size, and trophic position
are important factors influencing sport fish [Hg] (Bodaly
et al. 1993; Johnston et al. 2003; Power et al. 2002).
Specifically, large, piscivorous sport fish with elevated
trophic position tend to have higher [Hg] than small fish
feeding at lower trophic levels (Kidd et al. 1995). It follows
that changes in food-web structure can alter sport fish [Hg]
by changing their diet and trophic position (Eagles-Smith
et al. 2008; Lepak et al. 2009). Indeed, a whole-lake
experiment manipulating prey availability to northern pike
showed that [T-Hg] can be reduced very quickly by shift-
ing from a diet of crayfish to one of stocked rainbow trout
(Lepak et al. 2012b). As we saw in Elkhead Reservoir, the
reverse also is trueif trout stocking is discontinued
predator [T-Hg] can increase. Thus, fish stocking policy
can have implications for Hg bioaccumulation and FCAs
when stocked fish are low in Hg and also become prey for
resident predators.
Our study, and others (Clements et al. 2012), emphasize
that whether or not a given Hg load to a system is harmful
(e.g., results in FCAs) depends partly on the context of the
food web. Thus, there is some level of decoupling between
mechanisms such as Hg deposition and bioaccumulation in
sport fish. This implies that undetected food web shifts may
obscure potential relationships between long-term trends in
175
Hg deposition and changes in methylation or bioaccumu-
lation and [Hg] in fish. Additionally, these shifts can hap-
pen quickly and unpredictably, adding complexity to
relationships among external loading and bioaccumulation
in top predators.
Although [Hg] in invertebrates and fish can respond
rapidly to changes in Hg deposition (Harris et al. 2007),
unexpected relationships between Hg bioaccumulation and
deposition have been observed. For example, responses in
fish [Hg] to a 32% reduction in wet Hg deposition differed
across four study lakes in Voyageurs National Park from
1998 to 2012 (Brigham et al. 2014). During the same time
period, yellow perch [Hg] increased by 80% in one system,
decreased by 35% in two others, and stayed the same in a
fourth system. The authors concluded that complex inter-
actions were occurring that obscured the effects of reduc-
tions in Hg deposition by increasing Hg bioavailability.
Thus, the magnitude and timing of responses to changes in
Hg deposition are difficult to anticipate due to inherent
complexity in Hg cycling.
The causal relationship between Hg deposition and [Hg]
in fish is well established (Hammerschmidt and Fitzgerald
2006; Sorensen et al. 1990; Wiener et al. 2006). Ultimately,
lasting reductions in Hg bioaccumulation in sport fish will
probably require large-scale reductions in Hg emissions
and associated atmospheric deposition. Many studies sug-
gest a general decline in atmospheric deposition of Hg
across large portions of the United States and southern
Canada, and concentrations of Hg in upper trophic level
fish also have been declining in those areas (summarized in
Cross et al. 2015). However, our work and others (Zillioux
2015) show that monitoring concentrations in fish will be
necessary to protect human health, because natural and
anthropogenic perturbations to aquatic food webs that can
increase and decrease [Hg] in sport fish will continue
regardless of trends in atmospheric deposition.
Acknowledgements Funding for this study came from the Colorado
Department of Public Health and Environment, Water Quality Con-
trol Division, Nonpoint Source Program through a Clean Water Act
Section 319 assistance Grant C9-99818610 from the U.S. Environ-
mental Protection Agency. The authors thank Dr. Chris Shade and
Nathan Brady of Quicksilver Scientific Laboratory for technical
assistance and field support, respectively. They also thank Colorado
Parks and Wildlife biologists, Kurt Davies and Kyle Battige, for field
assistance, Bill Pate of the Colorado State University Fisheries
Ecology Laboratory for field and laboratory support, William Cle-
ments of Colorado State University, and the anonymous reviewers
from this journal for providing insightful critiques.
References
Blackwell BG, Brown ML, Willis DW (2000) Relative weight (Wr)
status and current use in fisheries assessment and management.
Rev Fish Sci 8:144
123
176
Bloom NS (1992) On the chemical form of mercury in edible fish and
marine invertebrate tissue. Can J Fish Aquat Sci 49:10101017.
doi:10.1139/f92-113
Bodaly RA, Rudd JWM, Fudge RJP, Kelly CA (1993) Mercury
concentrations in fish related to size of remote Canadian Shield
lakes. Can J Fish Aquat Sci 50:980987
Brigham ME, Sandheinrich MB, Gay DA, Maki RP, Krabbenhoft DP,
Wiener JG (2014) Lacustrine responses to decreasing wet
mercury deposition ratesresults from a case study in northern
Minnesota. Environ Sci Technol 48:61156123. doi:10.1021/
es500301a
Clements WH, Hickey CW, Kidd KA (2012) How do aquatic
communities respond to contaminants? It depends on the
ecological context. Environ Toxicol Chem 31:19321940.
doi:10.1002/etc.1937
Coelho JP, Mieiro CL, Pereira E, Duarte AC, Pardal MA (2013)
Mercury biomagnification in a contaminated estuary food web:
effects of age and trophic position using stable isotope analyses.
Mar Pollut Bull 69:110115. doi:10.1016/j.marpolbul.2013.01.
021
Cross FA, Evans DW, Barber RT (2015) Decadal declines of mercury
in adult bluefish (19722011) from the Mid-Atlantic coast of the
USA. Environ Sci Technol 49:90649072. doi:10.1021/acs.est.
5b01953
Davies K (2014) Horsetooth Reservoir: fish survey and management
data. Colorado Parks and Wildlife Report
Eagles-Smith CA, Suchanek TH, Colwell AE, Anderson NL, Moyle
PB (2008) Changes in fish diets and food web mercury
bioaccumulation induced by an invasive planktivorous fish. Ecol
Appl 18:A213A226. doi:10.1890/06-1415.1
Hammerschmidt CR, Fitzgerald WF (2006) Methylmercury in
freshwater fish linked to atmospheric mercury deposition.
Environ Sci Technol 40:77647770. doi:10.1021/es061480i
Harris RC, Rudd JWM, Amyot M, Babiarz CL, Beaty KG,
Blanchfield PJ, Bodaly RA, Branfireun BA, Gilmour CC,
Graydon JA, Heyes A, Hintelmann H, Hurley JP, Kelly CA,
Krabbenhoft DP, Lindberg SE, Mason RP, Paterson MJ,
Podemski CL, Robinson A, Sandilands KA, Southworth GR,
Louis VLS, Tate MT (2007) Whole-ecosystem study shows
rapid fish-mercury response to changes in mercury deposition.
Proc Natl Acad Sci USA 104:1658616591. doi:10.1073/pnas.
0704186104
James DA, Csargo IJ, Von Eschen A, Thul MD, Baker JM, Hayer CA,
Howell J, Krause J, Letvin A, Chipps SR (2012) A generalized
model for estimating the energy density of invertebrates. Freshw
Sci 31:6977. doi:10.1899/11-057.1
Johnson BM, Goettl JP (1999) Food web changes over fourteen years
following introduction of rainbow smelt into a Colorado
reservoir. N Am J Fish Manage 19:629642
Johnson BM, Martinez PJ (2000) Trophic economics of lake trout
management in reservoirs of differing productivity. N Am J Fish
Manage 20:127143
Johnson BM, Lepak JM, Wolff BA (2015) Effects of prey assemblage
on mercury bioaccumulation in a piscivorous sport fish. Sci Total
Environ 506:330337. doi:10.1016/j.scitotenv.2014.10.101
Johnston TA, Leggett WC, Bodaly RA, Swanson HK (2003)
Temporal changes in mercury bioaccumulation by predatory
fishes of boreal lakes following the invasion of an exotic forage
fish. Environ Toxicol Chem 22:20572062. doi:10.1897/02-265
Jones MS, Goettl JP, Flickinger SA (1994) Changes in walleye food
habits and growth following a rainbow smelt introduction. N Am
J Fish Manage 14:409414
Kannan K, Smith RG, Lee RF, Windom HL, Heitmuller PT,
Macauley JM, Summers JK (1998) Distribution of total mercury
and methyl mercury in water, sediment, and fish from south
Florida estuaries. Arch Environ Contam Toxicol 34:109118
123
Arch Environ Contam Toxicol (2017) 72:167177
Kidd KA, Hesslein RH, Fudge RJP, Hallard KA (1995) The influence
of trophic level as measured by delta-n-15 on mercury concen-
trations in freshwater organisms. Water Air Soil Pollut
80:10111015. doi:10.1007/Bf01189756
Lavoie RA, Hebert CE, Rail JF, Braune BM, Yumvihoze E, Hill LG,
Lean DRS (2010) Trophic structure and mercury distribution in a
Gulf of St. Lawrence (Canada) food web using stable isotope
analysis. Sci Total Environ 408:55295539. doi:10.1016/j.
scitotenv.2010.07.053
Lepak JM, Johnson BM (2010) Bioaccumulation of mercury in
aquatic food webs: integrating research and management
towards remediation. Colorado Parks and Wildlife Report
Lepak JM, Robinson JM, Kraft CE, Josephson DC (2009) Changes in
mercury bioaccumulation in an apex predator in response to
removal of an introduced competitor. Ecotoxicology
18:488498. doi:10.1007/s10646-009-0306-5
Lepak JM, Hooten MB, Johnson BM (2012a) The influence of
external subsidies on diet, growth and Hg concentrations of
freshwater sport fish: implications for management and fish
consumption advisories. Ecotoxicology 21:18781888. doi:10.
1007/s10646-012-0921-4
Lepak JM, Kinzli KD, Fetherman ER, Pate WM, Hansen AG,
Gardunio EI, Cathcart CN, Stacy WL, Underwood ZE, Brandt
MM, Myrick CA, Johnson BM (2012b) Manipulation of growth
to reduce mercury concentrations in sport fish on a whole-
system scale. Can J Fish Aquat Sci 69:122135. doi:10.1139/
f2011-136
Mason RP, Lawrence AL (1999) Concentration, distribution, and
bioavailability of mercury and methylmercury in sediments of
Baltimore Harbor and Chesapeake Bay, Maryland, USA. Envi-
ron Toxicol Chem 18:24382447. doi:10.1897/1551-
5028(1999)018\2438:Cdabom[2.3.Co;2
Mergler D, Anderson HA, Chan LHM, Mahaffey KR, Murray M,
Sakamoto M, Stern AH (2007) Methylmercury exposure and
health effects in humans: a worldwide concern. Ambio 36:311
Mikac N, Niessen S, Ouddane B, Wartel M (1999) Speciation of
mercury in sediments of the Seine estuary (France). Appl
Organomet Chem 13:715725
Momot WT (1995) Redefining the role of crayfish in aquatic
ecosystems. Rev Fish Sci 3:3363
Poste AE, Muir DCG, Guildford SJ, Hecky RE (2015) Bioaccumu-
lation and biomagnification of mercury in African lakes: the
importance of trophic status. Sci Total Environ 506:126136.
doi:10.1016/j.scitotenv.2014.10.094
Power M, Klein GM, Guiguer KRRA, Kwan MKH (2002) Mercury
accumulation in the fish community of a sub-arctic lake in
relation to trophic position and carbon sources. J Appl Ecol
39:819830
Shade CW (2008) Automated simultaneous analysis of monomethyl
and mercuric Hg in biotic samples by Hg-thiourea complex
liquid chromatography following acidic thiourea leaching.
Environ Sci Technol 42:66046610. doi:10.1021/es800187y
Sorensen JA, Glass GE, Schmidt KW, Huber JK, Rapp GR (1990)
Airborne mercury deposition and watershed characteristics in
relation to mercury concentrations in water, sediments, plankton,
and fish of 80 northern Minnesota lakes. Environ Sci Technol
24:17161727. doi:10.1021/Es00081a015
Stein RA (1977) Selective predation, optimal foraging, and predator-
prey interaction between fish and crayfish. Ecology
58:12371253. doi:10.2307/1935078
Suchanek TH, Eagles-Smith CA, Harner EJ (2008) Is Clear Lake
methylmercury distribution decoupled from bulk mercury load-
ing? Ecol Appl 18:A107A127. doi:10.1890/06-1649.1
UNEP (2002) Global mercury assessment. United Nations Environ-
mental Program http://www.uneporg/gc/gc22/Document/UNEP-
GC22-INF3pdf. Accessed 16 Jan 2016
Arch Environ Contam Toxicol (2017) 72:167177
USEPA (2000) Guidance for assessing chemical contaminant data for
use in fish advisories. US Environmental Protection Agency,
Office of Water, EPA 823-B-00-007 1
USEPA (2009) Guidance for implementing the January 2001
methylmercury water quality criterion. EPA 823-R-09-002 US
Environmental Protection Agency, Office of Water, Washington
Wiener JG, Knights BC, Sandheinrich MB, Jeremiason JD, Brigham
ME, Engstrom DR, Woodruff LG, Cannon WF, Balogh SJ
(2006) Mercury in soils, lakes, and fish in Voyageurs National
Park (Minnesota): importance of atmospheric deposition and
ecosystem factors. Environ Sci Technol 40:62616268. doi:10.
1021/es060822h
177
Wright B (2010) Elkhead Reservoir: fish survey and management
information. Colorado Parks and Wildlife Report
Wydoski RS, Bennett DH (1981) Forage species in lakes and
reservoirs of the western United States. Trans Am Fish Soc
110:764771. doi:10.1577/1548-8659(1981)110\764:fsilar[2.0.
co;2
Zillioux EJ (2015) Mercury in fish: history, sources, pathways,
effects, and indicator usage. In: Environmental indicators.
Springer, Berlin, pp 743766
123