USGS Fact Sheet 080-98
July 1998
Selenium in Reservoir Sediment from the Republican River Basin
By K.E. Juracek and A.C. Ziegler
Prepared in cooperation with the
BUREAU OF RECLAMATION,
U.S. DEPARTMENT OF THE INTERIOR
Reservoir sediment quality is an important environmental concern because sediment may act
as both a sink and a source of water-quality constituents to the overlying water column and
biota. Once in the food chain, sediment-derived constituents may pose an even greater concern
due to bioaccumulation. An analysis of reservoir bottom sediment can provide historical
information on sediment deposition as well as magnitudes and trends in constituents that may
be related to changes in human activity in the basin. The assessment described in this fact
sheet was initiated in 1997 by the U.S. Geological Survey (USGS), in cooperation with the
Bureau of Reclamation (BOR), U.S. Department of the Interior, to determine if irrigation
activities have affected selenium concentrations in reservoir sediment of the Republican
River Basin of Colorado, Kansas, and Nebraska.
Table of Contents
In the Republican River Basin (fig. 1), selenium is an
environmental concern due to the presence of seleniferous soils, outcrops of the Pierre
Shale, and wide-spread irrigation. These factors may combine to produce high concentrations
of selenium in surface water at some sites located immediately downstream from irrigation
return flows. Public Law 99-294 (1986) requires BOR to investigate soil characteristics that
may result in toxic or hazardous irrigation return flows (Bureau of Reclamation, 1997). Of
interest to BOR is an assessment of the magnitude and extent of selenium concentrations
within the basin and what effects, if any, Federally supported irrigation activities are
having. Specific objectives of this assessment were to determine background selenium
concentrations in reservoir sediment, changes in selenium concentrations with time, and the
potential effects of irrigation on selenium concentrations.
The Republican River has a drainage area of about 24,900 square miles that includes parts of
Colorado, Kansas, and Nebraska (fig. 1). Soils in the basin are
productive and used mostly for growing winter wheat, grain sorghum, soybeans, corn, and sugar
beets. More than 50 percent of the basin is cropland. The basin has variable climate with
mean annual precipitation ranging from about 18 inches in the west to about 30 inches in the
east (Bureau of Reclamation, 1996). Streamflow regimens of the Republican River and several
of its tributaries have been altered by the construction of several reservoirs
(fig. 1), as well as diversion dams and canals, many of which were built for the
purpose of increasing irrigation within the basin.
Irrigation in the basin dates back to at least the 1850's. By the 1940's, prior to completion
of BOR irrigation projects, an annual average of about 41,000 acres were irrigated in the
basin, mostly using unregulated streamflow (U.S. Army Corps of Engineers, 1967). In 1996,
approximately 1,900,000 (12 percent) of the 15,900,000 acres of farmland within the basin were
irrigated. Of the total acres irrigated in 1996, about 137,000 acres (7 percent) received
water from BOR irrigation projects (Bureau of Reclamation, 1997).
Figure 2 shows the history of irrigation
development in the basin from 1949 through 1992. In 1995, total diversions for irrigation in
the basin were estimated to be about 2,335,000 acre-feet (Joan Kenny, U.S. Geological Survey,
written commun., 1998). Of that total, about 255,000 acre-feet (11 percent) were used by the
BOR's six irrigation districts (Judy Catt, Bureau of Reclamation, written commun., 1998)
(fig. 1).
Selenium can be either beneficial or toxic to plants, animals, and humans depending on its
concentration. Found throughout the environment, selenium is derived mainly from rock
weathering. In the northern Great Plains, the parent material for selenium-rich soils is the
Pierre Shale (Cretaceous age). Distribution processes for selenium include volcanic activity,
combustion of fossil fuels, soil leaching, ground-water transport, metabolic uptake and
release by plants and animals, sorption and desorption, chemical or bacterial reduction and
oxidation, and mineral formation. In most oxygen-rich environments the dominant forms of
selenium are selenite and selenate. Selnate is highly mobile, easily leached from soils, and
readily taken up by plants. Although natural water tends to have low concentrations of
selenium, relatively high concentrations can occur if the water is alkaline or if it leaches
and drains seleniferous rocks and soils (McNeal and Balistrieri, 1989). Selenium
concentations equal to or greater than 4.0 mg/kg (milligrams per kilogram) in sediment are of
concern for fish and wildlife beacause of food-chain bioaccumulation (Lemly and Smith, 1987).
The objectives of the assessment were accomplished by collecting and analyzing bottom sediment
from three reservoirs in the Republican River Basin-Swanson Lake and Harlan County Lake in
Nebraska and Milford Lake in Kansas (fig. 1). Swanson Lake, completed in 1953,
was selected to define background sediment concentrations of selenium as well as trends in
selenium concentrations immediately upstream from the Frenchman-Cambridge Irrigation District
(fig. 1). Harlan County Lake, completed in 1952, was selected to define background
sediment concentrations of selenium immediately upstream from the Nebraska and Kansas
Bostwick Irrigation Districts (fig. 1), and trends in selenium
concentrations at the midpoint of the BOR-irrigated reach of the Republican River Basin. In
this assessment, background is defined as the sediment concentrations of selenium that are
representative of conditions that predate the period of major irrigation development in the
basin, which began in the mid-1960's (fig. 2).
Milford Lake, completed in 1967, was selected to assess trends in selenium concentrations
resulting from cumulative effects throughout the entire basin.
In May and June 1997, bottom-sediment cores were collected near the dam at two sites for each
lake. The cores were collected near the dam because the sediment is least likely to be
disturbed (for example, by biological activity, human activity, or wind-induced currents).
Also, sediment near the dam will have the smaller grain sizes, which are most likely to
contain selenium. The cores penetrated through the entire thickness of reservoir sediment and
into the original land-surface (or channel-bed) material. At each site, four cores were
collected using a gravity corer. Multiple cores were required to provide sufficient sediment
material for laboratory analyses. Each core was divided into either 5 or 10 subsamples
depending on the total sediment thickness.
For each site, separate cores were sampled and analyzed for selenium, sediment characteristics
(that is, bulk density, percent moisture, and grain size), and age dating. The total
recoverable analyses for selenium were performed at the BOR laboratory in Bismarck, North
Dakota. The analyses of sediment characteristics were performed at the USGS laboratory in
Lawrence, Kansas. To define sampling variability, two sites were used at each reservoir
(figs. 3, 4, and 5).
An assessment of analytical variability for selenium was attempted through analysis of
duplicate samples at one site (fig.
5). Variability of the duplicate-analysis results was large, possibly due in part to the
sampling technique used to obtain the duplicate samples.
Age dating of the bottom sediment was accomplished by determining the concentration of
cesium-137 (137Cs). 137Cs is a radioactive isotope that is a by-product
of nuclear weapons testing. Measureable concentrations of this isotope first appeared in the
atmosphere in about 1952, peaked during 1963-64, and have since declined. 137Cs is
an effective marker for age dating of reservoir bottom sediment (Van Metre and others, 1997).
It also can be used to demonstrate that the sediment is undisturbed if a relatively uniform
decrease in 137Cs concentrations follows the 1963-64 peak. The age-dating analyses
were performed by Quanterra Laboratories, Denver, Colorado.
Statistical analyses were performed to characterize the magnitudes of selenium concentrations
in sediment cores from each lake and to test for trends. Trends in selenium concentration were
examined by computing a Spearman's correlation coefficient and performing a t test (alpha =
0.05) for each sediment core from each lake. Also, a median test (alpha = 0.05) was performed
to compare selenium concentrations among the reservoirs.
Background concentrations for selenium in reservoir bottom sediment were represented by
sediment-core samples from Swanson and Harlan County Lakes, which were constructed prior to
the major period of irrigation development in the Republican River Basin from the mid-1960's
through the early 1980's (fig. 2).
The 137Cs results (figs.
3, 4, and
5) indicated that the two deepest sample
intervals from the Swanson and Harlan County Lake sediment cores were deposited prior to the
mid-1960's and thus are representative of background conditions. In cores from Swanson Lake,
background selenium concentrations (four samples) ranged from 0.1 to 1.0 mg/kg (milligram per
kilogram) with a mean of 0.6 mg/kg. In cores from Harlan County Lake, background selenium
concentrations (three samples) ranged from 0.8 to 1.5 mg/kg with a mean of 1.2 mg/kg. In
comparison, post-background selenium concentrations in cores from Swanson Lake (16 samples)
ranged from 0.2 to 2.0 mg/kg with a mean of 1.0 mg/kg. Post-background selenium
concentrations in cores from Harlan County Lake (12 samples) ranged from 1.1 to 2.7 mg/kg
with a mean of 1.8 mg/kg. In cores from Milford Lake, which represent post-background
conditions, selenium concentrations (20 samples) ranged from 0.2 to 2.2 mg/kg with a mean of
1.0 mg/kg.
For Swanson Lake, both sediment cores indicated a trend of increasing selenium concentrations
with decreasing depth (fig. 3). The
respective Spearman's correlation coefficients for sites S1 and S2 were -0.92 and -0.47. The
correlation for site S1 was statistically significant (two-sided p-value = 0.00014), whereas
the correlation for site S2 was not (two-sided p-value = 0.17246). Likewise, for Harlan
County Lake, both sediment cores indicated a trend of increasing selenium concentrations with
decreasing depth (fig. 4).
The respective Spearman's correlation coefficients for sites H1 and H2 were -0.56 and -0.50,
neither of which was statistically significant (respective two-sided p-values were 0.09158
and 0.39100). However, if the most-recent (shallowest) depth interval is omitted from site
H1, the resultant correlation is a statistically significant -0.94 (two-sided p-value =
0.00019). The most-recent depth interval may be anomalous as it presumably includes a large
input of sediment resulting from the 1993 flood. Inspection of annual discharge data from
USGS streamflow gages located immediately upstream from each reservoir
(fig. 1) indicates that the 1993 flood was minor at Swanson Lake, pronounced at
Harlan County Lake, and extreme at Milford Lake.
The initial results were mixed for Milford Lake (fig. 5). Sites M1 and M2 had respective Spearman's correlation coefficients of
-0.38 and 0.18, neither of which was statistically significant (respective two-sided p-values
were 0.28141 and 0.62287). However, if the two most-recent depth intervals are omitted from
sites M1 and M2, the respective Spearman's correlation coefficients (and associated
significances) are -0.71 (two-sided p-value = 0.04807) and -0.30 (two-sided p-value =
0.47126). The two most-recent depth intervals may be anomalous as they presumably include a
large input of sediment associated with extreme effects of the 1993 flood.
The possible effect of irrigation on selenium concentrations was assessed in two ways. First,
background and post-background selenium concentrations were compared for sediment cores from
Swanson and Harlan County Lakes. Although the small number of background sediment samples
precluded a statistical analysis to determine significance, the post-background increases
previously noted for both lakes indicate that irrigation within the basin may be having an
effect on selenium concentrations.
Second, in an attempt to isolate the contributions of BOR irrigation projects, post-background
selenium concentrations in sediment from Swanson Lake were compared to those from Harlan
County and Milford Lakes using a median test. BOR irrigation projects are located in the river
valleys upstream from both Harlan County and Milford Lakes. However, no BOR irrigation
projects are located upstream from Swanson Lake (fig. 1).
Post-background selenium concentrations in sediment from Swanson Lake (16 samples) were
compared to those from Harlan County Lake (12 samples), and a significant difference was
indicated (two-sided p-value = 0.00263). However, a comparison of post-background selenium
concentrations in sediment from Swanson Lake with those from Milford Lake (20 samples)
indicated no significant difference (two-sided p-value = 0.73732). Selenium concentrations in
sediment from Harlan County Lake were also significantly different from selenium
concentrations in sediment from Milford Lake (two-sided p-value = 0.03540). The respective
median post-background selenium concentrations in sediment from Swanson, Harlan County, and
Milford Lakes were 0.9, 1.8, and 0.8 mg/kg.
Because the distribution of BOR-irrigated acres between Swanson and Harlan County Lakes (51
percent) and between Harlan County and Milford Lakes (49 percent) is virtually equal, the
observed difference in median post-background selenium concentrations in sediment from Harlan
County Lake and Milford Lake does not appear to be explained by BOR irrigation. However,
median post-background selenium concentrations in sediment from all three lakes do appear to
be related to total irrigated acres (that is, BOR plus non-BOR irrigated lands). The
distribution of total irrigated acres in the basin upstream from Swanson Lake (29 percent),
between Swanson and Harlan County Lakes (54 percent), and between Harlan County and Milford
Lakes (17 percent) parallels the median post-background selenium concentrations determined for
sediment from the three lakes (table 1). Also, the primary increase in selenium concentrations
(figs. 3, 4, and 5)
corresponds with the substantial increase in the number of irrigated acres in the basin
(fig. 2). It is also noteworthy that most of the
exposed Pierre Shale in the basin is located upstream from Harlan County Lake.
Table 1. Irrigated acres and median post-background selenium concentrations
[mg/kg, milligrams per kilogram]
Subbasin area |
Percentage of total irrigated acres in basin |
Percentage of Bureau of Reclamation irrigated
acres in basin |
Median post-background selenium concentration
(mg/kg) |
Upstream from Swanson Lake |
29 |
0 |
0.9 |
Between Swanson and Harlan County Lakes |
54 |
51 |
1.8 |
Between Harlan County and Milford Lakes |
17 |
49 |
0.8 |
"Overall the trend analyses indicate a general increase in selenium concentrations, through
time, in sediment from all three reservoirs."
As evidenced by a comparison of background and post-background selenium concentrations and the
results of trend tests, irrigation appears to have increased selenium concentrations in
reservoir sediment within the Republican River Basin. The changes in selenium concentrations
attributable to BOR irrigation are likely commensurate with the percentage of total irrigated
acres that receive water from BOR's six irrigation districts. Because most of the irrigated
land and exposed Pierre Shale is located upstream from Harlan County Lake and median
post-background selenium concentrations in sediment from Harlan County Lake are significantly
larger than the median concentrations in sediment from Swanson and Milford Lakes, Harlan
County Lake may be acting as a sink for selenium deposition within the basin.
- Bureau of Reclamation, 1996, Resource management assessment, Republican River
Basin, water service contract renewal: Billings, Montana, July 1996, various
pagination.--1997, Republican River Basin irrigation water quality study, draft sample
plan: Billings, Montana, April 1997, 11 p.
- Lemly, D.A., and Smith, G.J., 1987, Aquatic cycling of selenium-implications for
fish and wildlife: U.S. Fish and Wildlife Service, Fish and Wildlife Leaflet 12, 10 p.
- McNeal, J.M., and Balistrieri, L.S., 1989, Geochemistry and occurrence of
selenium-an overview, in Jacobs, L.W., ed., Selenium in agriculture and the
environment, 1989: Madison, Wisconsin, American Society of Agronomy, Inc., and Soil
Science Society of America, Inc., p. 1-13.
- U.S. Army Corps of Engineers, 1967, Republican River Basin reservoir regulation
manual, master manual: Kansas City, Missouri, March 1967, various pagination.
- U.S. Environmental Protection Agency, 1996, The National Sediment Quality Survey-a
report to Congress on the extent and severity of sediment contamination in surface
waters of the United States (draft): Washington, D.C., July 1996, EPA-823-D-96-002,
various pagination.
- Van Metre, P.C., Callender, Edward, and Fuller, C.C., 1997, Historical trends in
organochlorine compounds in river basins identified using sediment cores from
reservoirs: Environmental Science & Technology, v. 31, no. 8, p. 2339-2344.
For more information please contact:
District Chief
U.S. Geological Survey
4821 Quail Crest Place
Lawrence, Kansas 66049-3839
(785) 842-9909
email: waucott@usgs.gov