Methylmercury (MeHg) is a bioaccumulative pollutant produced
in and exported from ooded soils, including those used for
rice (Oriza sativa L.) production. Using unltered aqueous MeHg
data from MeHg monitoring programs in the Sacramento River
watershed from 1996 to 2007, we assessed the MeHg contribution
from rice systems to the Sacramento River. Using a mixed-eects
regression analysis, we compared MeHg concentrations in
agricultural drainage water from rice-dominated regions (AgDrain)
to MeHg concentrations in the Sacramento and Feather Rivers, both
upstream and downstream of AgDrain inputs. We also calculated
MeHg loads from AgDrains and the Sacramento and Feather Rivers.
Seasonally, MeHg concentrations were higher during November
through May than during June through October, but the dierences
varied by location. Relative to upstream, November through May
AgDrain least-squares mean MeHg concentration (0.18 ng L−1,
range 0.15–0.23 ng L−1) was 2.3-fold higher, while June through
October AgDrain mean concentration (0.097 ng L−1, range 0.6–1.6
ng L−1) was not signicantly dierent from upstream. June through
October AgDrain MeHg loads contributed 10.7 to 14.8% of the
total Sacramento River MeHg load. Missing ow data prevented
calculation of the percent contribution of AgDrains in November
through May. At sites where calculation was possible, November
through May loads made up 70 to 90% of the total annual load.
Elevated ow and MeHg concentration in November through May
both contribute to the majority of the AgDrain MeHg load occurring
during this period. Methylmercury reduction eorts should
target elevated November through May MeHg concentrations in
AgDrains. However, our ndings suggest that the contribution and
environmental impact of rice is an order of magnitude lower than
previous studies in the California Yolo Bypass.
The Contribution of Rice Agriculture to Methylmercury in Surface
Waters: A Review of Data from the Sacramento Valley, California
K. Christy Tanner,* Lisamarie Windham-Myers, Jacob A. Fleck, Kenneth W. Tate, Stephen A. McCord,
and Bruce A. Linquist
I anoxic soils, a portion of the inorganic mercury (Hg)
pool can be methylated, predominantly by sulfate- and iron-
reducing bacteria (Compeau and Bartha, 1985; Gilmour et
al., 1992; Fleming et al., 2006; Kerin et al., 2006), forming meth-
ylmercury (MeHg). Methylmercury binds strongly to thiols in
proteins (Ballatori, 2002) and is extremely toxic to organisms,
causing neurological problems and decreased reproductive suc-
cess (Crump and Trudeau, 2009). Environmental levels of MeHg
as low as 0.1 ng L−1 in freshwater ecosystems can have negative
eects on high trophic-level organisms via bioaccumulation and
biomagnication (Rudd, 1995; Watras et al., 1998; Chan et al.,
2003). Humans are exposed to MeHg primarily through the con-
sumption of sh and other wildlife from Hg-contaminated envi-
ronments, resulting in negative health eects (Chan et al., 2003).
Mercury and MeHg contamination of surface water is wide-
spread. In the United States, sh consumption advisories have
been issued in all 50 states and one US territory, including 1.8
million km of river and 6.6 million ha of lake (USEPA, 2011).
Wetlands provide anoxic soil conditions in which
Hg-methylating microbes thrive, resulting in elevated MeHg
production and bioaccumulation in a variety of wetland types
(Marvin-DiPasquale et al., 2003; Hall et al., 2008). Unlike many
crops, rice (Oriza sativa L.) is grown in ooded elds that are
eectively agricultural wetlands. Rice is grown on approximately
150 million ha globally (Czech and Parsons, 2002), comprising a
substantial portion of the world’s estimated 1.2 billion ha of wet-
lands (Finlayson et al., 1999) and serving as important wildlife
habitat (Czech and Parsons, 2002). Seasonal wet–dry cycles and
inputs of labile organic carbon from root exudates and rice straw
promote MeHg production and bioaccumulation in rice elds
(Windham-Myers et al., 2009, 2014a; Ackerman and Eagles-
Smith, 2010; Ackerman et al., 2010; Rothenberg and Feng,
2012). Many studies have found that MeHg can accumulate in
rice grain and that rice can be the primary route of MeHg expo-
sure for people living in Hg-contaminated inland rice-growing
Abbreviations: AgDrain, agricultural drainage; AgDrain-E, drainage from
agricultural areas east of the Sacramento River, also known as Sacramento Slough;
AgDrain-W, drainage from agricultural areas west of the Sacramento River, also
known as the Colusa Basin Drain; AgDrain-Woutfall, ows from AgDrain-W that enter
the Sacramento River via the Knight’s Landing Outfall Gates; AgDrain-Wdiversion,
ows diverted from AgDrain-W to the Yolo Bypass via Ridge Cut Slough; LSM,
least-squares mean; MeHg, methylmercury.
K.C. Tanner, K.W. Tate, and B.A. Linquist, Dep. of Plant Sciences, Univ. of California,
One Shields Ave., Davis, CA 95616; L. Windham-Myers, USGS, Western Region
Bureau of Regional Research, 345 Middleeld Rd., MS 480, Menlo Park, CA
94025; J.A. Fleck, USGS, California Water Science Center, 6000 J St., Placer Hall,
Sacramento, CA 95819; S.A. McCord, McCord Environmental, Inc., 759 Bianco Ct.,
Davis, CA 95616. Assigned to Associate Editor Karen Bradham.
Copyright © American Society of Agronomy, Crop Science Society of America, and
Soil Science Society of America. 5585 Guilford Rd., Madison, WI 53711 USA.
All rights reserved.
J. Environ. Qual.
This is an open access article distributed under the terms of the CC BY-NC-ND
Supplemental material is available online for this article.
Received 11 July 2016.
Accepted 16 Oct. 2016.
*Corresponding Author (firstname.lastname@example.org).
Journal of Environmental Quality
SURFACE WATER QUALITY
• We studied the impact of rice production on MeHg at the wa-
• MeHg concentration was elevated in agricultural drainage wa-
ter during November through May.
• Watershed-scale MeHg loads were lower than expected based
on eld studies.
Published November 23, 2016
Journal of Environmental Quality
areas (e.g., Feng et al., 2008; Meng et al., 2010, 2011; Zhu et
al., 2015). In contrast, few studies have investigated export of
MeHg in rice-eld drainage water and its impact on downstream
ecosystems, with the exception of studies in the California Yolo
Bypass (Windham-Myers et al., 2014a). While it is clear that rice
elds produce and export MeHg (Alpers et al., 2014), their eect
has not been evaluated at a watershed scale.
MeHg concentration in stream channels and exposure to biota
in those channels result from the interaction of sources, sinks, and
transport processes in the watershed (Bradley et al., 2011). ough
oodplains may also produce MeHg when temporarily ooded
(Singer et al., 2016), at a watershed scale, the best predictor of MeHg
concentration in a water body is the percent wetland cover upstream
(Krabbenho et al., 1999). Methylmercury production may be
decoupled from export (Bachand et al., 2014), and processes such
as photodemethylation in open water (Seller et al., 1996) and par-
ticle settling can remove MeHg from the water column. As a result,
MeHg concentrations can be lower than would be predicted from
upstream MeHg inputs (Bradley et al., 2011).
e Sacramento River watershed (1.7 million ha) in
California provides a good opportunity to study how rice con-
tributes to watershed-scale MeHg concentrations and loads.
Rice is grown on more than 240,000 ha and is the main crop in
the low-lying central drainage area of the valley (USDA–NASS,
2016). Naturally enriched Hg conditions, plus a legacy of Hg
and gold mining (which used elemental mercury to amalgamate
the gold) in the mountains surrounding the valley, have resulted
in elevated Hg concentrations in river sediments, water, and sh
(Domagalski, 1998, 2001; Davis et al., 2008; Springborn et al.,
2011; Singer et al., 2013; Domagalski et al., 2016; Donovan et al.,
2016a, 2016b). Elevated MeHg concentrations in sh have led to
sh consumption advisories and enactment of a Total Maximum
Daily Load to reduce MeHg loads into the Sacramento–San
Joaquin Delta (Delta Mercury Control Program, 2010).
e overall objective of this study was to evaluate the con-
tribution from rice-producing areas to MeHg loads in the lower
Sacramento River by compiling and analyzing historic surface
water MeHg concentration and ow data. We evaluated MeHg
concentrations and loads in major agricultural drains from
rice-dominated areas and at mainstem sampling points in the
Sacramento and Feather Rivers, both upstream and downstream
of agricultural drain inputs. Furthermore, we evaluated seasonal
trends to identify the time of year in which MeHg loads exported
from rice elds may be of greatest concern, and long-term trends
to see if changes in postharvest rice straw management practices
have inuenced MeHg concentrations.
Materials and Methods
e Sacramento Valley covers 1.7 million ha and is situated in
the low-lying area between the Coast Ranges and Sierra Nevada
of California. It is 240 km long from north to south and ranges
from 32 km wide in the north to 72 km wide in the south (Bennett
et al., 2011). e Sacramento Valley has a Mediterranean climate,
with typically hot, dry conditions during April to September and
a cool, rainy season from October through March
e Sacramento River watershed has both natural and anthro-
pogenic Hg sources. Mineral springs (Youngs, 1994) and erodible
surcial soils (US Bureau of Mines, 1965) in the Coast Range to
the west and Cascade Mountains to the northeast are naturally
Hg enriched. Mercury mining occurred in the Coast Range to the
west (Jasinksi, 1995). Between 1846 and 1981, an estimated 34.5
million kg of Hg was released into the environment, partially as
Hg vapor during ore processing (Churchill, 2000). Most of the
Hg mining occurred in the Cache Creek watershed and farther
south; thus, runo from mine wastes in these areas enters the Yolo
Bypass downstream of the current study area (Domagalski, 1998).
Beginning in 1848, elemental Hg was used by the gold mining
industry in the Klamath and Sierra Nevada mountain ranges,
drained by the upper Sacramento River and Feather River, respec-
tively (Domagalski, 1998; Churchill, 2000). Combined Hg losses
from gold mining totaled 5.8 million kg , with 97% of losses occur-
ring before 1935 and 80 to 90% of losses occurring in the Sierra
Nevada (Churchill, 2000). While Hg losses were substantial, it is
unclear to what degree rice-growing areas in the central Sacramento
Valley have been contaminated with Hg. Reservoirs limit down-
stream transport of MeHg and total Hg (Slotton et al., 1995).
Domagalski (1998) measured total Hg in riverbed sediments: sites
in the central Sacramento River had 40 to 70 ng Hg g−1, while sites
within and downstream of the Feather River had 140 to 370 ng Hg
g−1. To our knowledge, there are no published Hg concentration
data for Sacramento Valley rice soils. Atmospheric deposition has
not been measured directly for the Sacramento Valley. However,
the San Francisco Estuary to the south received 19 mg Hg m−2 yr−1
as dry deposition and 4.2 mg Hg m−2 yr−1 as wet deposition in 1999
to 2000 (Tsai and Hoenicke, 2001).
Within the Sacramento Valley, there are two main waterways:
e Sacramento River on the western side of the valley and the
Feather River on the east. Additional watershed inows occur
along the eastern and western edge of the valley, including the
Yuba and Bear rivers. Both of the main rivers ow from north to
south and converge near the southern end of the Valley’s primary
rice-growing region, north of the city of Sacramento (Fig. 1).
Aer conuence with the Feather River, the Sacramento is joined
by the American River and continues south to the Sacramento–
San Joaquin Delta. Surface water hydrology is highly manipu-
lated, with a number of constructed waterways used to direct
irrigation water (supply and drainage) and to mitigate ooding.
Drainage from rice agriculture (hereaer, “rice”), and to a lesser
extent nonagricultural wetlands, makes up the majority of water
in agricultural drainage canals. Rice is planted during late April
through May. Most elds are water-seeded, whereby pregermi-
nated rice seed is dropped from an airplane onto ooded elds
(Linquist et al., 2015). Water management during the rst month
of the growing season is highly variable, with the water level and
outow adjusted for seedling establishment and herbicide applica-
tion. Later in the growing season, maintenance ow is established
where irrigation water is applied to rice in excess of evapotrans-
piration demand in part to limit salinity build up (Grattan et al.,
2002; Scardaci et al., 2002). Excess water is exported from rice
elds as drainage water. Fields are drained in August or September,
3 to 4 wk before harvest. Irrigation of other crops in the water-
shed is managed to match evapotranspiration demand closely so
that little or no drainage water is produced. Drainage water from
managed wetlands is also present; however, wetland cover within
the drainage area is approximately one fourth that of rice (Fig. 1)
(US Fish and Wildlife Service, 2016). Water-export data from
wetlands managed for wildlife is limited, but rates of outow
Journal of Environmental Quality
are also managed to prevent salinity build up (this management
objective is similar to those of rice growers). Periods of high water
export from managed wetlands are the drawdown of seasonal wet-
lands (March–May) and areas ooded during the summer (July–
August) (S. Emmons, personal communication, 2016).
Drainage water from agricultural elds and other wetlands is
collected by two major agricultural drains, hereaer referred to
as “AgDrain-E” and “AgDrain-W.” AgDrain-E collects drainage
water from the eastern valley area (e.g., Sutter Bypass) between
the Sacramento and Feather Rivers and discharges into the
Sacramento River immediately upstream of the conuence of
the Sacramento and Feather Rivers (Fig. 1). Low ows during
summer in AgDrain-E are made up primarily of rice drainage
water; however, during high winter ows, large volumes of ood-
water from the Sutter Bypass oodplain are mixed with rice drain-
age water. AgDrain-W collects drainage water from the area west
of the Sacramento River. Flows from AgDrain-W are divided
immediately downstream of the sampling location, with some
water directed into the Sacramento River (AgDrain-Woutfall) and
the remainder diverted to the Yolo Bypass oodplain (AgDrain-
Wdiversion) (Fig. 1). Drainage from AgDrain-E and the Feather River
enter the Sacramento River 25 km downstream of AgDrain-Woutfall.
We categorized sampling sites into “site groups” based on
their location with respect to the two main AgDrains (Fig.
1): Sampling sites located at the mouths of AgDrain-E and
AgDrain-W are referred to as “AgDrain” sites. “Upstream-
Sacramento” and “Upstream-Feather” sites were located on the
Sacramento and Feather Rivers, respectively, upstream of where
AgDrains empty into the Sacramento River. Upstream sites serve
as control sites because they represent river water that has been
minimally inuenced by rice. All Upstream-Sacramento sites
were upstream of rice drainage water inputs, while Upstream-
Feather site 5 may have been inuenced by rice drainage water
from rice grown east of the Feather River and north of the Bear
River. Upstream sites also represent irrigation source water: sites 1
and 3 on the Sacramento River were located close to the Tehama–
Colusa and the Glenn–Colusa Canals, respectively, which are
major agricultural diversions that provide irrigation water to the
western side of the Sacramento Valley. “Downstream” sites were
located downstream of the conuence of the Sacramento River,
Feather Riverm and AgDrains, but upstream of the Sacramento
urban area and the conuence with the American River (Fig. 1).
Downstream sites represent Sacramento River water that has
been inuenced by Sacramento Valley rice drainage.
We used data from programs that monitored aqueous, unl-
tered MeHg concentrations in the Sacramento River water-
shed from 1996 to 2007. Programs include the USGS National
Water uality Assessment Program (Domagalski et al., 2000)
dataset from 1996 to 1998 (I); three programs conducted by the
Sacramento River Watershed Program (2005, 2008), including
data from 2000 to 2003 (II), 2004 (IV), and 2006 to 2007 (V);
and one dataset from the CALFED Bay-Delta Program (Foe et
al., 2008), including data from 2003 to 2006 (III) (Supplemental
Table S1). Samples were collected either as grab samples or as
depth-integrated samples at the midpoint of the channel cross-
section. Detection limits ranged from 0.0114 to 0.0234 ng L−1
(Supplemental Table S1). Methylmercury concentrations were
below the detection limit in 1.6% of samples. Nondetects were
equally distributed among Upstream-Sacramento, Upstream-
Feather, AgDrain, and downstream samples (n = 4, 2, 3, and
2, respectively). If the lab value was reported, it was used in the
analysis; otherwise undetected MeHg concentrations were treated
as 0.5 ´ detection limit in data analysis. Sampling sites varied
among programs, and sampling frequency ranged from monthly
to quarterly. A number of sites were omitted from the analysis due
to lack of relevance to the study questions or lack of data. ese
sites included small, seasonal creeks, sites on the Sacramento River
distantly upstream or downstream of rice, and sites with fewer
than 10 samples. For sites included in this analysis, the number of
samples available by site and program is shown in Table 1.
Both precipitation and burned rice area were tested as possible
explanatory variables. Precipitation from storm events may inu-
ence MeHg concentration by dilution or ushing MeHg from
areas where it is produced (Balogh et al., 2006). Precipitation
data were obtained from the California Irrigation Management
Information System (California Department of Water Resources,
2014b). Data from two weather stations that represented the
study area within the watershed were averaged for analysis (sta-
tion identications 12 and 3). Precipitation totals from 3, 5, and
Fig. 1. Map of the study area. Site labels correspond to site identi-
cation eld in Table 1. Rice area was obtained from the California
Department of Water Resources (2013). Wetland area is from the
National Wetland Inventory (US Fish and Wildlife Service, 2016). Water
ways are depicted according to the National Hydrography Dataset
(USDA–NRCS et al., 2013). AgDrain, agricultural drainage.
Journal of Environmental Quality
7 d before MeHg sample collection were tested as predictors
of MeHg concentration. Burning of rice straw removes carbon
from elds, possibly inuencing microbial activity (Windham-
Myers et al., 2014b; Zhu et al., 2015). Burned rice area data were
obtained from the California Rice Commission (2016).
Flow data for load estimations were obtained from the
California Data Exchange Center, the California Water Data
Library (California Department of Water Resources, 2014a,
2016), and the National Water Information System (USGS, 2016)
databases (Supplemental Table S2). Mean daily ow was used
when available; otherwise, hourly or quarter-hourly ow mea-
surements were averaged to obtain mean daily ow. Missing ow
data prevented load calculation at key sites, including AgDrain-
Wdiversion, AgDrain-Woutfall during the 2005 water year (October
2004–September 2005), and AgDrain-E during winter.
Season was included in the analysis as a categorical variable
because seasonal variation occurs in rice management as well as
climate. We wanted to identify times of the year when concentra-
tions were elevated to inform future studies. We plotted MeHg
concentration data from all years by day and month and dened
two seasons based on periods with relatively higher or lower
Linear mixed-eects regression analysis was used to assess
dierences in MeHg concentration, allowing us to account
for the fact that sites were measured repeatedly but not always
concurrently. Site and year were used as random eects in the
model. Fixed eects tested included season (June–October or
November–May), site group (Upstream-Sacramento, Upstream-
Feather, AgDrain, or downstream), time (as a continuous vari-
able), fraction of rice area burned, and precipitation, as well as
site group ´ season, site group ´ time, and site group ´ fraction
of rice area burned interaction terms. Models were tted using
the lme4 package (Bates et al., 2015) in R (R Core Team, 2014).
We selected the nal model using backward stepwise regression.
Beginning with the fullest model, rst random eects, then xed
eects, were dropped stepwise if nonsignicant (p > 0.05). e
p-values for random eects were calculated using likelihood
ratio tests, whereas signicance of xed eects was determined
using F tests with denominator degrees of freedom, calculated
by Satterthwaite’s approximation. Post hoc Tukey-corrected
dierences of least-squares means (LSM) were used to identify
signicant dierences among categorical xed eects; dier-
ences signicantly dierent from zero (p < 0.05) were considered
signicant. Model selection and tests for dierences between
xed eects were done using the lmerTest package (Kuznetsova
et al., 2016). Assumptions of normalcy and homogeneity of vari-
ance were assessed using standard diagnostic plots. A natural log
transformation was used on MeHg, but not on other variables.
is transformation successfully corrected for heteroscedasticity
and normalized the residuals of the model.
Beale’s ratio estimator was used to calculate MeHg loads
(Beale, 1962; Richards, 1998) because it was found to be unbi-
ased and accurate in a number of studies comparing load calcu-
lation methods (Dolan et al., 1981; Beirman et al., 1988). e
mean daily load (concentration ´ ow) for days on which con-
centration was measured was multiplied by the ratio of average
ow during the period of interest to the average ow on days
when concentration was measured, then multiplied by a bias cor-
rection factor (Supplemental Eq. S1 and S2).
At each site, loads were calculated separately for monitor-
ing programs I, II, and V. Programs III and IV were combined
because samples were collected concurrently and program
IV did not have enough data for a separate load calculation.
Additionally, loads were calculated using data from all programs
combined. Separate calculations were done for the November
through May and June through October seasons.
Table 1. The number of methylmercury (MeHg) samples collected by each sampling program at sites used in this analysis. AgDrain, agricultural
drainage water from rice-dominated regions.
Site Site ID† Samples in each program‡ Total number
I II III IV V
Sacramento River above Bend Bridge 1 – 17 30 4 18 69
Sacramento River at Woodson Bridge 2 – – 31 – – 31
Sacramento River at Hamilton City 3 – 17 31 4 18 70
Sacramento River at Ord Ferry Bridge 4 – – 22 – – 22
Sacramento River at Butte City 5 – – 30 – – 30
Sacramento River at Colusa 6 29 17 30 4 18 98
Feather River at Gridley 7 – – 31 – – 31
Yuba River at Marysville 8 – 18 31 4 18 71
Bear River below Wheatland 9 – – 30 – – 30
Feather River at Nicolaus 10 – 18 31 4 18 71
Colusa Basin Drain at Knights Landing W 25 16 31 4 18 94
Sacramento Slough at Karnack E 23 16 28 4 18 89
Sacramento River at Verona 11 27 – –– – 27
Sacramento River at Veteran’s Bridge 12 – – –– 17 17
† Site locations are denoted in Fig. 1 using site identication (Site ID).
‡ See Supplemental Table S1 for more information about the sampling programs.
Journal of Environmental Quality
Upstream-Sacramento and Upstream-Feather loads were calcu-
lated at sites 6 and 10, respectively. ese sites were located imme-
diately upstream of AgDrain inputs but downstream of other
known MeHg inputs, including the Bear and Yuba Rivers (Fig. 1).
Additionally, these sites had the most complete datasets (Table 1).
Results and Discussion
A total of 681 unltered water MeHg concentration mea-
surements were compiled from the various programs and used
in this study. Only three sites were sampled by all ve programs,
two of which were the main agricultural drainage canals that
discharge directly into the Sacramento River (AgDrain-E and
AgDrain-W), while the third was Upstream-Sacramento site 6.
Methylmercury concentrations ranged from below the detec-
tion limits (0.02–0.0114) to 1.97 ng L−1, and only three samples
in the dataset were above 1 ng L−1 (Fig. 2). Waterways with MeHg
concentrations above 0.1 ng L−1 are potentially negatively impacted
(Rudd, 1995), and 57% of samples in this study had concentrations
below this level. Twenty percent of samples were below 0.06 ng L−1,
the regulatory standard for the Sacramento–San Joaquin Delta.
Despite changes in rice straw management during the study
period, mixed-eects regression analysis did not detect a signicant
change in MeHg concentration over time (Fig. 2A), and there was
no interaction between time and location (e results of variable
selection for the mixed-eects model are shown in Supplemental
Table S3). During the study period, straw removal practices
changed as a result of the Connelly–Areias–Chandler Rice Straw
Burning Reduction Act of 1991 (California Environmental
Protection Agency, 1991), which mandated that rice straw burn-
ing in the Sacramento Valley be phased down starting in 1992 and
be allowed only under specied conditions for disease control by
2001. e practice of burning rice straw was replaced by incorpo-
rating rice straw into soil during winter, followed by ooding to
facilitate its decomposition (Linquist et al., 2006). e percentage
of area where rice straw was incorporated during winter increased
from <15% in 1992 to >80% by 2001 (California Air Resources
Board, 2003). Rice straw is a source of labile organic carbon that
may increase Hg methylation in rice elds (Windham-Myers et al.,
2014a; Zhu et al., 2015); thus, an increase in AgDrain MeHg con-
centration over time was expected due to this shi from burning
to incorporation and ooding during the study period. However,
the fraction of rice area burned did not signicantly aect MeHg
concentration and did not interact with site group. is result is
consistent with a recent controlled, replicated experiment test-
ing the eect of straw removal from rice elds (Eagles-Smith et
al., 2014). Other sources of variation in MeHg concentration may
obscure any eect of rice straw management on MeHg concentra-
tion at the valley scale.
We did not detect a signicant eect of precipitation on
MeHg concentrations. Elevated MeHg concentrations were
found in early 1997, 1 mo aer a rain on snow event that caused
major ooding throughout the region (samples were not col-
lected during the event) (Fig. 2A). Although Balogh et al. (2006)
reported elevated MeHg concentrations during high ows, con-
centration changes in response to storm events can be complex,
depending on the transported material, as well as watershed
characteristics (Richards and Holloway, 1987). e sampling
frequency in this dataset may be too low to detect any MeHg
concentration changes in response to precipitation.
Plotting all data by month and day revealed a seasonal pattern in
which MeHg concentrations were consistently lower from June to
October (range: less than detection limit to 0.3 ng L−1), while con-
centrations from November through May were higher and more vari-
able (less than detection limit to 1.98 ng L−1) (Fig. 2B). We hereaer
refer to these seasons as June to October (153 d) and November to
May (212 d), respectively. Mixed eects regression revealed a strong
seasonal eect, with November to May concentrations 65% higher
than June to October concentrations (F1, 644 = 62.3, p < 0.001); how-
ever, there was an interaction between site group and season (F3, 634
= 11.8, p < 0.001; Fig. 3). e nal model includes these signicant
xed eects, as well as site group because it was part of a signicant
interaction (Supplemental Table S3). Signicant random eects for
both site and year were also included.
Fig. 2. (A) Time series of methylmercury (MeHg) concentrations through-
out the study period. Bars at the top indicate when each sampling
program occurred (Supplemental Table S1). (B) MeHg concentrations
plotted by day of year. The gure shows data from the whole study
period. Bars at top show the rice-growing season and winter fallow, and
June to October and November to May seasons as dened in this study.
Journal of Environmental Quality
With respect to water management and inputs into these sys-
tems, June to October includes runo from rice elds in the form
of maintenance ow and the nal drain in preparation for har-
vest (August). November to May corresponds to the ooding of
rice and managed wetlands (October and November) and sub-
sequent runo and nal drains from rice (February) and natural
wetlands (April and May). For the years studied, 92 ± 6% of the
annual rainfall occurred in November to May. During high rain-
fall years or large storm events, regional ooding occurs, result-
ing in runo from other agricultural elds (and urban areas), and
river water may be diverted into bypasses, both resulting in com-
ingling of water sources, particularly for AgDrain-E.
Methylmercury concentrations did not dier between
Upstream-Sacramento and Upstream-Feather in November to
May (LSM = 0.079 and 0.077 ng L−1, respectively; p = 0.3) or June
to October (0.060 and 0.075 ng L−1, respectively; p = 0.9) (Fig. 3),
indicating that rice on the east and west sides of the Valley receive
similar MeHg inputs in irrigation water. Upstream-Feather MeHg
concentrations did not dier between seasons (p = 0.8) (Fig. 3),
but Upstream-Sacramento had slightly but signicantly higher
MeHg concentrations during November to May (p = 0.002).
During June to October, AgDrain MeHg concentrations
(LSM = 0.097 ng L−1) appeared elevated compared with other site
groups; however, there were no signicant dierences (p > 0.05) in
MeHg concentrations among site groups. In contrast, November
to May AgDrain MeHg concentrations (LSM = 0.18 ng L−1; p <
0.05) were signicantly higher than November to May Upstream-
Sacramento and Upstream-Feather concentrations. Previous studies
also reported elevated MeHg concentrations in rice drainage water
relative to irrigation water (Alpers et al., 2014; Zhao et al., 2016).
Similar to AgDrains, downstream MeHg concentrations
were signicantly higher than upstream during November to
May, and there was no signicant dierence between AgDrain
and downstream MeHg concentrations during either June to
October or November to May. However, the ows and loads of
AgDrain, upstream, and downstream sites must be considered
when determining the degree to which AgDrains were inuenc-
ing downstream MeHg concentrations.
In this study, Ag Drains exhibited a much stronger seasonal pat-
tern than upstream sites, suggesting that rice and wetlands inuence
concentrations. Studies in the Yolo Bypass reported a similar sea-
sonal pattern (Bachand et al., 2014; Marvin-DiPasquale et al., 2014;
Windham-Myers et al., 2014b). While MeHg is produced in rice
elds throughout the year (Marvin-DiPasquale et al., 2014), transpi-
ration during the growing season results in the downward movement
of surface water, causing MeHg to be transported into and stored in
the rootzone (Bachand et al., 2014). e absence of transpiration
during the fallow season allows MeHg to be released into surface
water via diusion (Bachand et al., 2014). Additionally, rice plants
can store a signicant amount of MeHg during the growing season
(Windham-Myers et al., 2014b). Other studies have reported that
MeHg stored in dry sediment may be quickly mobilized into the
water column on ooding (Kelly et al., 1997; Rumbold and Fink,
2006; Alpers et al., 2014) or proposed that methylation may occur
shortly aer inundation (Singer et al., 2016). us, the early grow-
ing season drainage events, in addition to the fallow season, may be
periods of MeHg export from rice. (e.g., Fig. 2B).
Using all available data, June to October MeHg loads for
AgDrain-Woutfall, AgDrain-E, Upstream-Sacramento, and
Upstream-Feather were (mean ± SD) 0.14 ± 0.03, 0.23 ± 0.03,
2.2 ± 0.3, and 0.88 ± 0.3 g d−1, respectively (Fig. 4). e down-
stream MeHg load is expected to be the sum of all upstream
Fig. 3. Least-squares mean ± SE of methylmercury (MeHg) concentra-
tions based on mixed-eects modeling for site groups and seasons.
There is a signicant interaction between site group and season.
Site groups that were not signicantly dierent during November
to May have the same letter. Site groups did not dier signicantly
during June to October. Asterisks (*) indicate a signicant dierence
between seasons at each site group. See Supplemental Table S4 for
p-values of pairwise comparisons. AgDrain, agricultural drainage.
Fig. 4. June to October methylmercury (MeHg) loads by sampling pro-
gram (indicated by x-axis groups and shading) and site (indicated by
color in legend). Loads from tributary sources are shown as a stacked
bar and are expected to equal downstream load. “All” represents
loads calculated using data from all sampling programs. Error bars
represent the standard deviation of sampling program estimates.
Data was not available for Upstream-Feather, so a placeholder (gray)
was used with the value from “All.” Color and shading is consistent
with Fig. 5. AgDrain, agricultural drainage.
Journal of Environmental Quality
tributary loads: Upstream-Sacramento, Upstream-Feather,
AgDrain-Woutfall, and AgDrain-E. e downstream load (2.5 ± 0.6
g d−1) was 73% of the total June to October tributary load (3.4 ±
0.4 g d−1). Although the downstream load was within the range
of variability of the tributary load, the downstream load was less
than the tributary load in both programs where it was measured (I
and V), possibly suggesting MeHg loss. June to October load esti-
mates diered among datasets by less than a factor of two (Fig. 4).
AgDrain loads accounted for 10.7% of the total June to October
upstream tributary load, or 14.8% of the downstream load.
November to May loads made up 70 to 90% of the total annual
load (within sites), and estimates diered among programs by
approximately a factor of four (Fig. 5). November to May MeHg
loads for AgDrain-Woutfall, Upstream-Sacramento, Upstream-
Feather, and downstream were 0.22 ± 0.1, 7.3 ± 4, 1.7 ± 0.8, and
24 ± 17 g d−1, respectively. November to May loads for AgDrain-E
were not possible to estimate due to gaps in ow data and oodwa-
ter being diverted into and out of the river at a number of locations.
Sampling frequency in this study was lower than recommended
for accurate load calculation (Dolan et al., 1981). Low sampling
frequency is problematic because it reduces the likelihood of accu-
rately capturing loads during storm events. is is particularly
evident for November to May downstream loads, where removal
of one elevated sample concentration (1.98 ng L−1, 13 Feb. 1997,
see Fig. 2A) decreases the downstream load estimate by 40%. is
sample was taken more than a month aer a major ood event,
during an extended period of high ow, and elevated MeHg con-
centrations were also observed at other sites (quality control data
did not suggest this was a result of contamination [Domagalski,
1998]). It is possible that higher concentrations occurred during
the peak of this or other storm events, but samples were not col-
lected then. erefore, uncertainty surrounding storm event loads
limits our condence in November to May loads. ere are typi-
cally no storm events and MeHg concentrations are less variable
during June to October (Fig. 2B), so the variation in MeHg con-
centration can be captured using fewer samples; thus, the sampling
frequency is likely adequate for June to October loads.
e true MeHg load at a site should be expected to show
considerable interannual variation, with larger loads occurring
in years with high ow (Hill, 1986). It is important to consider
whether the years measured in this study are representative of
historical ow regimes. Hill (1986) recommends using data from
6 to 7 yr to obtain robust estimates of average annual loads. With
the exception of downstream, loads presented here were based
on 7 to 9 yr of data, suggesting that this study adequately cap-
tures the interannual load variation at these sites. Flow during the
study period was greater than averages of historical average ows
at all sites except AgDrain-W in June to October (Supplemental
Table S5 and Supplemental Fig. S1), suggesting that loads during
this study may have been higher than historical average loads.
Full accounting of rice eld loads requires the total MeHg load
from AgDrain-W, which is the sum of AgDrain-Woutfall (reported
above) and AgDrain-Wdiversion (not calculated because of missing
ow data). However, AgDrain-Wdiversion ow data from 2007 and
2012 shows that AgDrain-Woutfall carried 36 ± 9% of November to
May ows and 73 ± 7% of June to October ows from AgDrain-W.
rough discussions with irrigation district managers, we deter-
mined that management of AgDrain-W ows have not changed
since the beginning of the study period (Bair, personal communi-
cation, 2016). Based on the fraction of ows carried by AgDrain-
Woutfall, the total AgDrain-W load was 0.61 ± 0.17 g d−1 in November
to May and 0.19 ± 0.02 g d−1 in June to October.
Windham-Myers et al. (2014a) reported MeHg loads exported
from rice elds in the Yolo Bypass. If these loads are extrapolated to
other rice-growing regions and multiplied by the total area of rice
in the Sacramento Valley (240,000 ha), the predicted MeHg loads
are 12± 7 and 26 ± 24 g d−1 in June to October and November to
May, respectively—an order of magnitude higher than the results
of this study. is substantial dierence may result from two
potential sources. First, there are known site dierences between
the eld-scale study site in the Yolo Bypass and typical Sacramento
Valley rice-growing areas. For example, MeHg concentrations in
irrigation source water were an order of magnitude higher in the
Yolo Bypass than in irrigation source water (Upstream-Sacramento
and Upstream-Feather) in this study. e Yolo Bypass is known to
accumulate Hg-laden sediment during storm events (Springborn
et al., 2011), while much of the rice land is not in the path of storm
ows that would deliver Hg-laden sediment. Secondly, dierences
may result from scale-dependent factors. Bradley et al. (2011)
found that MeHg loads exported from wetland sources overesti-
mated the watershed-scale load because MeHg was lost through
sink processes (including photo- and microbial demethylation and
particle settling) during transport. Windham-Myers et al. (2014a)
showed that rice elds are net MeHg sources; however, aque-
ous MeHg exported from rice elds in Sacramento Valley must
be transported through a network of canals before it reaches the
Sacramento River. Future research should seek to determine the
degree to which MeHg is lost during canal transport and to quan-
tify MeHg budgets for Sacramento Valley rice elds.
Rice Field Drainage Contribution
Elevated November to May downstream concentrations and
the large November to May downstream load relative to the
Fig. 5. November to May methylmercury (MeHg) loads by site (indi-
cated by color and x-axis groups) and sampling program (indicated
by shading in legend). Note the log scale y-axis. “All” represents loads
calculated using data from all sampling programs. Error bars repre-
sent standard deviation of sampling program estimates. Color and
shading is consistent with Fig. 4. AgDrain, agricultural drainage.
Journal of Environmental Quality
upstream tributary loads suggest that the uncalculated November
to May AgDrain-E was considerably larger than AgDrain-W.
However, AgDrain-E carries a mixture of oodwater and rice
drainage water, as well as drainage from wetlands. Determining
the MeHg load from rice requires further research on the eect of
oodplains on MeHg and the contribution of managed wetlands.
It is misleading to consider only MeHg loads exported from
rice because elds receive MeHg in irrigation water and atmo-
spheric sources. Irrigation source water (upstream) had a MeHg
concentration of 60 to 70% of that of AgDrain water (Fig. 3).
Due to evapotranspiration and percolation losses, the volume of
drainage water is £40% of irrigation water applied to a rice eld
during the growing season (Linquist et al., 2015). us, growing
season increases in MeHg concentrations between irrigation and
drainage water during the growing season were of a similar mag-
nitude to that expected from evapoconcentration alone. e June
to October AgDrain MeHg load contributed to the Sacramento
River may be similar to the MeHg load diverted from the river
for irrigation. During November to May (the fallow season),
upstream MeHg concentrations were 43% of AgDrain, while
drainage water exports are expected to be a larger fraction of irri-
gation water applied (no transpiration is expected in fallow rice
elds), resulting in increased MeHg exports relative to imports.
is is consistent with Bachand et al. (2014), who found that rice
elds might store MeHg during the growing season but release it
during the winter fallow. Wet and dry atmospheric deposition
may be important sources of MeHg and total Hg (Munthe et al.,
1995, Conaway et al., 2010) and have been shown to inuence
runo uxes from catchments (Hultberg and Munthe, 1995).
However, atmospheric deposition data is not available for the
Sacramento Valley. Deposition studies would help quantify the
degree to which rice elds are MeHg sources or sinks.
Elevated November to May AgDrain MeHg concentrations
and November to May AgDrain-W loads being fourfold higher
than June to October loads both indicate that November to May
is the period of higher concern for MeHg export from rice elds.
Without knowledge of the full annual cycle of MeHg production
in and export from a system, studies risk missing important peri-
ods of MeHg export. If the environmental management objective
is to reduce annual loads of MeHg, control eorts should focus
on the November to May period, when MeHg concentrations
in AgDrains are elevated. Finally, this study indicates that care
should be taken when extrapolating the impact of rice produc-
tion on MeHg in surface water at the eld to valley scale. While
the seasonal patterns of MeHg concentrations and exports were
similar to those observed at the eld scale, there were substantial
dierences in the magnitudes of concentrations and loads.
Graduate research assistantship funding for K.C. Tanner was provided
by the California Rice Research Board and the UC Davis Department
of Plant Sciences. L. Windham-Myers and J.A. Fleck were funded
by the USGS. anks to members of the Agroecosystems lab for
encouragement and support.
Ackerman, J.T., and C.A. Eagles-Smith. 2010. Agricultural wetlands as po-
tential hotspots for mercury bioaccumulation: Experimental evidence
using caged sh. Environ. Sci. Technol. 44:1451–1457. doi:10.1021/
Ackerman, J.T., A.K. Miles, and C.A. Eagles-Smith. 2010. Invertebrate mer-
cury bioaccumulation in permanent, seasonal, and ooded rice wetlands
within California’s Central Valley. Sci. Total Environ. 408:666–671.
Alpers, C.N., J.A. Fleck, M. Marvin-DiPasquale, C.A. Stricker, M. Stephenson,
and H.E. Taylor. 2014. Mercury cycling in agricultural and managed wet-
lands, Yo lo Bypass, California: Spatial and seasonal variations in water qual-
ity. Sci. Total Environ. 484:276–287. doi:10.1016/j.scitotenv.2013.10.096
Bachand, P.A.M., S. Bachand, J.A. Fleck, C.N. Alpers, M. Stephenson, and L.
Windham-Myers. 2014. Methylmercury production in and export from
agricultural wetlands in California, USA: e need to account for physi-
cal transport processes into and out of the root zone. Sci. Total Environ.
Ballatori, N. 2002. Transport of toxic metals by molecular mimicry. Environ.
Health Perspect. 110:689–694.
Balogh, S.J., E.B. Swain, and Y.H. Nollet. 2006. Elevated methylmercury con-
centrations and loadings during ooding in Minnesota rivers. Sci. Total
Environ. 368:138–148. doi:10.1016/j.scitotenv.2005.09.045
Bates, D., M. Maechler, B. Bolker, and S. Walker. 2015. Fitting linear mixed-ef-
fects models using lme4. J. Stat. Sow. 67:1–48. doi:10.18637/jss.v067.i01
Beale, E.M.L. 1962. Some uses of computers in operational research. Industrielle
Bennett, G.L., M.S. Fram, and K. Belitz. 2011. Status of groundwater quality in
the southern, middle, and northern Sacramento Valley study units, 2005–
08: California GAMA priority basin project. Scientic Investigations Rep.
2011–5002 USGS. https://pubs.er.usgs.gov/publication/sir20115002
(accessed 1 Mar. 2016).
Bierman, V.J., Jr., S.D. Preston, and S.E. Silliman. 1988. Development of estima-
tion methods for tributary loading rates of toxic chemicals. IWRRC Tech.
Rep. 183 Water Resources Research Center, Lafayette, IN. http://docs.lib.
cessed 1 Aug. 2015).
Bradley, P.M., D.A. Burns, K. Riva-Murray, M.E. Brigham, D.T. Button, L.C.
Chasar et al. 2011. Spatial and seasonal variability of dissolved methylmer-
cury in two stream basins in the eastern United States. Environ. Sci. Tech-
nol. 45:2048–2055. doi:10.1021/es103923j
California Air Resources Board. 2003. Progress report on the phase-down of rice
straw burning in the Sacramento Valley air basin. California Air Resources
Board and the California Department of Food and Agriculture. http://
www.arb.ca.gov/research/apr/reports/l2075.pdf (accessed 20 July 2015).
California Department of Water Resources. 2014a. California Data Exchange
Center. California Department of Water Resources. http://cdec.water.
ca.gov (accessed 25 Aug. 2014).
California Department of Water Resources. 2014b. California Irrigation Man-
agement Information System. California Department of Water Resources.
http://www.cimis.water.ca.gov (accessed 30 June 2014).
California Department of Water Resources. 2013. Land use surveys. California
Department of Water Resources. http://www.water.ca.gov/landwateruse/
lusrvymain.cfm (accessed 9 Oct. 2013).
California Department of Water Resources. 2016. Water Data Library, Califor-
nia Department of Water Resources. http://www.water.ca.gov/waterda-
talibrary (accessed 8 Jan. 2016)
California Environmental Protection Agency. 1991. Connelly–Areias–Chan-
dler Rice Straw Burning Reduction Act of 1991. California health and
safety code sec. 41865. California Environmental Protection Agency, Sac-
ramento, CA. http://calrice.org/pdf/Reduction_Act_1991.pdf (accessed
21 July 2015).
California Rice Commission. 2016. Air quality. California Rice Commission.
http://calrice.org/industry/air-quality (accessed 12 Apr. 2016).
Chan, H.M., A.M. Scheuhammer, A. Ferran, C. Loupelle, J. Holloway, and S.
Weech. 2003. Impacts of mercury on freshwater sh-eating wildlife and
humans. Hum. Ecol. Risk Assess. 9:867–883. doi:10.1080/713610013
Churchill, R. 2000. Contributions of mercury to California’s environment from
mercury and gold mining activities—insights from the historical record.
In: Proceedings and Summary Report of the Workshop on Assessing and
Managing Mercury from Historic and Current Mining Activities, San
Francisco, CA. 28–30 Nov. 2000. USEPA, San Francisco, CA. p. 36–39.
Compeau, G.C., and R. Bartha. 1985. Sulfate-reducing bacteria: Principal meth-
ylators of mercury in anoxic estuarine sediment. Appl. Environ. Microbiol.
Conaway, C.H., F.J. Black, P. Weiss-Penzias, M. Gault-Ringold, and A.R.
Flegal. 2010. Mercury speciation in Pacic coastal rainwater, Mon-
terey Bay, California. Atmos. Environ. 44:1788–1797. doi:10.1016/j.
Crump, K.L., and V.L. Trudeau. 2009. Mercury-induced reproductive impair-
ment in sh. Environ. Toxicol. Chem. 28:895–907. doi:10.1897/08-151.1
Journal of Environmental Quality
Czech, H.A., and K.C. Parsons. 2002. Agricultural wetlands and waterbirds: A
review. Waterbirds 25:56–65.
Davis, J.A., B.K. Greeneld, G. Ichikawa, and M. Stephenson. 2008. Mercury
in sport sh from the Sacramento–San Joaquin Delta region, California,
USA. Sci. Total Environ. 391:66–75. doi:10.1016/j.scitotenv.2007.10.050
Delta Mercury Control Program. 2010. Amendments to the water quality con-
trol plan for the Sacramento River and San Joaquin River Basins for the
control of methylmercury and total mercury in the Sacramento–San Joa-
quin Delta Estuary. Resolution R5-2010-0043. California Regional Water
uality Control Board Central Valley Region, Rancho Cordova, CA.
Dolan, D.M., A.K. Yui, and R.D. Geist. 1981. Evaluation of river load esti-
mation methods for total phosphorus. J. Great Lakes Res. 7:207–214.
Domagalski, J. 1998. Occurrence and transport of total mercury and methyl
mercury in the Sacramento River Basin, California. J. Geochem. Explor.
Domagalski, J. 2001. Mercury and methylmercury in water and sediment of
the Sacramento River Basin, California. Appl. Geochem. 16:1677–1691.
Domagalski, J., M.S. Majewski, C.N. Alpers, C.S. Eckley, C.A. Eagles-Smith, L.
Schenk, and S. Wherry. 2016. Comparison of mercury mass loading in
streams to atmospheric deposition in watersheds in the western U.S.: Evi-
dence for non-atmospheric mercury sources Sci. Total Environ. 568:638–
Domagalski, J.L., P.D. Dileanis, D.L. Knifong, C.M. Munday, J.T. May, B.J. Daw-
son et al. 2000. Water-quality assessment of the Sacramento River Basin,
California: Water-quality, sediment and tissue chemistry, and biological
data, 1995–1998. Open-File Rep. 2000–391 USGS. http://ca.water.usgs.
gov/sac_nawqa/waterindex.html (accessed 7 Aug. 2013).
Donovan, P.M., J.D. Blum, M.B. Singer, M. Marvin-DiPasquale, and M.T.K.
Tsui. 2016a. Isotopic composition of inorganic mercury and methylmer-
cury downstream of a historical gold mining region. Environ. Sci. Technol.
Donovan, P.M., J.D. Blum, M.B. Singer, M. Marvin-DiPasquale, M.T.K. Tsui.
2016b. Methylmercury degradation and exposure pathways in streams and
wetlands impacted by historical mining. Sci. Total Environ. 568:1192–
Eagles-Smith, C.A., J.T. Ackerman, J. Fleck, L. Windham-Myers, H. Mcuillen,
and W. Heim. 2014. Wetland management and rice farming strategies to
decrease methylmercury bioaccumulation and loads from the Cosumnes
River Preserve, California. Open-File Rep. 2014–1172. USGS, Reston,
Feng, X.B., P. Li, G. Qiu, S. Wang , G. Li, L. Shang et al. 2008. Human exposure to
methylmercury through rice intake in mercury mining areas, Guizhou Prov-
ince, China. Environ. Sci. Technol. 42:326–332. doi:10.1021/es071948x
Finlayson, C.M., N.C. Davidson, A.G. Spiers, and N.J. Stevenson. 1999. Global
wetland inventory: Current status and future priorities. Mar. Freshwater
Res. 50:717–727. doi:10.1071/MF99098
Fleming, E.J., E.E. Mack, P.G. Green, and D.C. Nelson. 2006. Mercury methyla-
tion from unexpected sources: Molybdate-inhibited freshwater sediments
and an iron-reducing bacterium. Appl. Environ. Microbiol. 72:457–464.
Foe, C., S. Louie, and D. Bosworth. 2008. Task 2. Methyl mercury concentra-
tions and loads in the Central Valley and Freshwater Delta. Methylmercury
Central Valley Regional Water uality Control Board, Sacramento, CA,
task2mmhg_nal.pdf (accessed 1 Sept. 2013).
Gilmour, C.C., E.A. Henry, and R. Mitchell. 1992. Sulfate stimulation of mer-
cury methylation in freshwater sediments. Environ. Sci. Technol. 26:2281–
Grattan, S., L. Zeng, M. Shannon, and S. Roberts. 2002. Rice is more sensitive to
salinity than previously thought. Calif. Agric. 56:189–198. doi:10.3733/
Hall, B.D., G.R. Aiken, D.P. Krabbenho, M. Marvin-DiPasquale, and C.M.
Swarzenski. 2008. Wetlands as principal zones of methylmercury produc-
tion in southern Louisiana and the Gulf of Mexico region. Environ. Pollut.
(Oxford, U. K.) 154:124–134. doi:10.1016/j.envpol.2007.12.017
Hill, A.R. 1986. Stream nitrate-N loads in relation to variations in annual
and seasonal runo regimes. J. Am. Water Resour. Assoc. 22:829–839.
Hultberg, H., and J. Munthe. 1995. Cycling of methyl mercury and mercury—
Responses in the forest roof catchment to three years of decreased atmo-
spheric deposition. Water Air Soil Pollut. 80:415–424. doi:10.1007/
Jasinksi, S.M. 1995. e material ow of mercury in the United States. Resour.
Conserv. Recycl. 15:145–179.
Kelly, C.A., J.W.M. Rudd, R.A. Bodaly, N.P. Roulet, V.L. St. Louis, A. Heyes et
al. 1997. Increases in uxes of greenhouse gases and methylmercury fol-
lowing ooding of experimental reservoir. Environ. Sci. Technol. 31:1334–
Kerin, E.J., C.C. Gilmour, E. Roden, M.T. Suzuki, J.D. Coates, and R.P.
Mason. 2006. Mercury methylation by dissimilatory iron-reducing
bacteria. Appl. Environ. Microbiol. 72:7919–7921. doi:10.1128/
Krabbenho, D.P., J.G. Wiener, W.G. Brumbaugh, M.L. Olson, J.F. DeWild,
and T.J. Sabin. 1999. A national pilot study of mercury contamination of
aquatic ecosystems along multiple gradients . Water-Resources Investiga-
tion Rep. 99-4018B USGS, Reston, VA, http://toxics.usgs.gov/pubs/
Kuznetsova, A., P.B. Brockho, and R.H.B. Christensen. 2016. lmerTest: Tests
in linear mixed eects models. R package version 2.0-30. R Foundation
for Statistical Computing, Vienna, Austria. https://CRAN.R-project.org/
Linquist, B.A., S.M. Brouder, and J.E. Hill. 2006. Winter straw and water man-
agement eects on soil nitrogen dynamics in California rice systems.
Agron. J. 98:1050–1059. doi:10.2134/agronj2005.0350
Linquist, B.A., R. Snyder, F. Anderson, L. Espino, G. Inglese, S. Marras et al.
2015. Water balances and evapotranspiration in water- and dry-seeded rice
systems. Irrig. Sci. 33:375–385. doi:10.1007/s00271-015-0474-4
Marvin-DiPasquale, M., J. Agee, R. Bouse, and B. Jae. 2003. Microbial cy-
cling of mercury in contaminated pelagic and wetland sediments of San
Pablo Bay, California. Environ. Geol. (Heidelberg, Ger.). 43:260–267.
Marvin-DiPasquale, M., L. Windham-Myers, J.L. Agee, E. Kakouros, L.H. Kieu,
J.A. Fleck et al. 2014. Methylmercury production in sediment from agricul-
tural and non-agricultural wetlands in the Yolo Bypass, California, USA.
Sci. Total Environ. 484:288–299. doi:10.1016/j.scitotenv.2013.09.098
Meng, B., X. Feng, G. Qiu, Y. Cai, D. Wang, P. Li et al. 2010. Distribution pat-
terns of inorganic mercury and methylmercury in tissues of rice (Oryza
sativa L.) plants and possible bioaccumulation pathways. J. Agric. Food
Chem. 58:4951–4958. doi:10.1021/jf904557x
Meng, B., X. Feng, G. Qiu, P. Liang, P. Li, C. Chen, and L. Shang. 2011. e
process of methylmercury accumulation in rice (Oryza sativa L.). Environ.
Sci. Technol. 45:2711–2717. doi:10.1021/es103384v
Munthe, J., H. Hultberg, and Å. Iverfeldt. 1995. Mechanisms of deposition of
methylmercury and mercury to coniferous forests. In: D.B. Porcella, J.W.
Huckabee, and B. Wheatley, editors, Proceedings of the 3rd International
Conference on Mercury as a Global Pollutant, Whistler, BC, Canada. 10–
14 July 1994. Springer Science + Business Media, Dordrecht, the Nether-
lands. p. 363–371.
R Core Team. 2014. R: A language and environment for statistical computing.
Version 3.2.4. R Foundation for Statistical Computing, Vienna, Austria.
Richards, R.P. 1998. Estimation of pollutant loads in rivers and streams: A
guidance document for NPS programs. USEPA Region VIII Grant
X99837-01-0. Water uality Laboratory, Heidelburg University, Tin,
Richards, R.P., and J. Holloway. 1987. Monte Carlo studies of sampling strat-
egies for estimating tributary loads. Water Resour. Res. 23:1939–1948.
Rothenberg, S.E., X . Feng. 2012. Mercury cycling in a ooded rice paddy. J. Geo-
phys. Res. 117(G3):G0300. doi:10.1029/2011JG001800
Rudd, J.W. 1995. Sources of methyl mercury to freshwater ecosystems: A review.
Water, Air, Soil Pollut. 80:697–713. doi:10.1007/BF01189722
Rumbold, D.G., and L.E. Fink. 2006. Extreme spatial variability and unprec-
edented methylmercury concentrations within a constructed wetland.
Environ. Monit. Assess. 112:115–135. doi:10.1007/s10661-006-0767-4
Sacramento River Watershed Program. 2005. Annual monitoring report: 2003–
2004. http://swim.sacriver.org/page.php?id=139 (accessed 22 July 2013).
Sacramento River Watershed Program. 2008. Final proposition 50 grant moni-
toring report 2005–2007. http://swim.sacriver.org/page.php?id=139 (ac-
cessed 22 July 2013).
Scardaci, S., M. Shannon, S. Grattan, A. Eke, S. Roberts, S. Goldman-Smith, and
J. Hill. 2002. Water management practices can aect salinity in rice elds.
Calif. Agric. 56:184–188. doi:10.3733/ca.v056n06p184
Seller, P., C.A. Kelly, J.W.M. Rudd, and A.R. MacHutchon. 1996. Pho-
todegredation of methylmercury in lakes. Nature 380:694–697.
Singer, M.B., R. Aalto, L.A. James, N.E. Kilham, J.L. Higson, and S. Ghoshal.
2013. Enduring legacy of a toxic fan via episodic redistribution of Cali-
fornia gold mining debris. Proc. Natl. Acad. Sci. USA 110:18436–18441.
Journal of Environmental Quality
Singer, M.B., L.R. Harrison, P.M. Donovan, J.D. Blum, and M. Marvin-Di-
Pasquale. 2016. Hydrologic indicators of hot spots and hot moments of
mercury methylation potential along river corridors. Sci. Total Environ.
Slotton, D.G., S.M. Ayers, J.E. Reuter, and C.R. Goldman. 1995. Gold mining
impacts on food chain mercury in northwestern Sierra Nevada streams.
Project W-816. University of California Water Resources Center, Oak-
Springborn, M., M.B. Singer, and T. Dunne. 2011. Sediment-adsorbed total
mercury ux through Yolo Bypass, the primary oodway and wetland in
the Sacramento Valley, California. Sci. Total Environ. 412–413:203–213.
Tsai, P., and R. Hoenicke. 2001. San Francisco Bay atmospheric deposition pilot
study part 1: Mercury. San Francisco Estuary Institute, San Francisco, CA.
1 Mar 2016).
US Bureau of Mines. 1965. Mineral potential of the United States. Information
circular 8252. US Bureau of Mines, Washington, DC.
USDA–NASS. 2016. uickstats database. USDA–NASS. http://quickstats.
nass.usda.gov/ (accessed 1 Feb. 2016).
USDA–NRCS, USGS, and USEPA. 2013. National hydrography dataset.
USGS, Reston, VA. http://nhd.usgs.gov/data.html (accessed 28 Oct.
USEPA. 2011. National listing of sh advisories: Technical fact sheet EPA-
820-F-13-058. USEPA. https://www.epa.gov/sites/production/les/2015-06/
documents/technical-factsheet-2011.pdf (accessed 1 June 2015).
US Fish and Wildlife Service. 2016. National wetlands inventory. US Depart-
ment of the Interior, US Fish and Wildlife Service, Washington, DC.
http://www.fws.gov/wetlands (accessed 19 Apr. 2016).
USGS. 2016. USGS water data for the nation. National Water Information Sys-
tem: Web interface. USGS, Reston, VA. doi:10.5066/F7P55KJN
Watras, C.J., R.C. Back, S. Halvorsen, R.J.M. Hudsonc, K.A. Morrisona, and S.P.
Wentec. 1998. Bioaccumulation of mercury in pelagic freshwater food webs.
Sci. Total Environ. 219:183–208. doi:10.1016/S0048-9697(98)00228-9
Windham-Myers, L., J.A. Fleck, J.T. Ackerman, M. Marvin-DiPasquale, C.A.
Stricker, W.A. Heim et al. 2014a. Mercury cycling in agricultural and
managed wetlands: A synthesis of methylmercury production, hydrologic
export, and bioaccumulation from an integrated eld study. Sci. Total En-
viron. 484:221–231. doi:10.1016/j.scitotenv.2014.01.033
Windham-Myers, L., M. Marvin-DiPasquale, E. Kakouros, J.L. Agee, L.H. Kieu,
C.A. Stricker et al. 2014b. Mercury cycling in agricultural and managed
wetlands of California, USA: Seasonal inuences of vegetation on mer-
cury methylation, storage, and transport. Sci. Total Environ. 484:308–318.
Windham-Myers, L., M. Marvin-Dipasquale, D.P. Krabbenho, J.L. Agee,
M.H. Cox, P. Heredia-Middleton et al. 2009. Experimental removal
of wetland emergent vegetation leads to decreased methylmercury
production in surface sediment. J. Geophys. Res. 114(G2):G00C05.
Youngs, L.E. 1994. California Low-temperature geothermal resource update:
1993. Rep. DOE/ID/13223–T10. California Division of Mines and Ge-
ology, Sacramento, CA. http://www.osti.gov/scitech/biblio/205566 (ac-
cessed 1 May 2016).
Zhao, L., C.W. Anderson, Q. Qiu, B. Meng, D. Wang, and X. Feng. 2016.
Mercury methylation in paddy soil: Source and distribution of mercury
species at a Hg mining area, Guizhou Province, China. Biogeosciences
Zhu, H., H. Zhong, D. Evans, and H. Hintelmann. 2015. Eects of rice resi-
due incorporation on the speciation, potential bioavailability and risk
of mercury in a contaminated paddy soil. J. Hazard. Mater. 293:64–71.