Sediment organic carbon burial in agriculturally eutrophic
impoundments over the last century
J. A. Downing,
1,2
J. J. Cole,
3
J. J. Middelburg,
4
R. G. Striegl,
5
C. M. Duarte,
6
P. Kortelainen,
7
Y. T. Prairie,
8
and K. A. Laube
9
Received 3 October 2006; revised 13 June 2007; accepted 14 September 2007; published 15 February 2008.
[1] We estimated organic carbon (OC) burial over the past century in 40 impoundments
in one of the most intensively agricultural regions of the world. The volume of sediment
deposited per unit time varied as a function of lake and watershed size, but smaller
impoundments had greater deposition and accumulation rates per unit area. Annual
water storage losses varied from 0.1–20% and were negatively correlated with
impoundment size. Estimated sediment OC content was greatest in lakes with low ratios
of watershed to impoundment area. Sediment OC burial rates were higher than those
assumed for fertile impoundments by previous studies and were much higher than those
measured in natural lakes. OC burial ranged from a high of 17,000 g C m
2
a
1
to a low
of 148 g C m
2
a
1
and was significantly greater in small impoundments than large ones.
The OC buried in these lakes originates in both autochthonous and allochthonous
production. These analyses suggest that OC sequestration in moderate to large
impoundments may be double the rate assumed in previous analyses. Extrapolation
suggests that they may bury 4 times as much carbon (C) as the world’s oceans. The
world’s farm ponds alone may bury more OC than the oceans and 33% as much as the
world’s rivers deliver to the sea.
Citation: Downing, J. A., J. J. Cole, J. J. Middelburg, R. G. Striegl, C. M. Duarte, P. Kortelainen, Y. T. Prairie, and K. A. Laube
(2008), Sediment organic carbon burial in agriculturally eutrophic impoundments over the last century, Global Biogeochem. Cycles,
22, GB1018, doi:10.1029/2006GB002854.
1. Introduction
[2] Lakes and impoundments are depositional environ-
ments. The material that accumulates in the sediments of
these water bodies can be airborne deposits imported from
outside the watershed, air- plus water-borne deposits
imported from inside the watershed (allochthonous inputs),
or created by biological or chemical processes that occur
within the water body (autochthonous inputs). For organic
carbon (OC) the ultimate source is past and present primary
production, either in the watershed or in the water body. For
either source the OC buried in the sediments of lakes or
reservoirs represents short- to long-term-scale sequestration
of atmospheric CO
2
. Globally, this burial of OC in the
sediments of natural lakes has been estimated in the range of
30 to 70 Tg C a
1
[Dean and Gorham, 1998; Einsele et al.,
2001; Mulholland and Elwood, 1982; Stallard, 1998]. The
burial in impoundments is thought to be much larger (150 to
220 Tg C a
1
)[Mulholland and Elwood, 1982; Stallard,
1998]. While these rates are modest in comparison with the
current total storage of organic carbon (C) terrestrially
(1000 to 4000 Tg C a
1
)[Pacala et al., 2001; Randerson
et al., 2002; Schimel et al., 2001], they are comparable to
the storage of OC in the sediments of the global ocean
(about 120–240 Tg C a
1
)[Duarte et al., 2004; Meybeck,
1993; Sundqu ist, 2003]. Further, the annual amount of
organic C stored in lakes and reservoirs is of comparable
magnitude to the delivery of OC by rivers to the ocean,
about 400 Tg C a
1
[Meybeck, 1993; Probst, 2005].
[
3] The OC burial in impoundments is of particular
interest because they are manmade structures. Whatever
the ultimate source, the OC now stored in them met a
different fate prior to impoundment. Further, the global
estimates of impoundment OC s torage are based on a
limited suite of impoundment types. Missing from these
estimates are major classes of very abundant, small
impoundments that are also agriculturally impacted. These
systems receive agricultural fertilizers that enhance autoch-
thonous primary production. Further, agricultural lands tend
to have carbon-rich soils that, when tilled, are very suscep-
GLOBAL BIOGEOCHEMICAL CYCLES, VOL. 22, GB1018, doi:10.1029/2006GB002854, 2008
1
Department of Ecology, Evolution, and Organismal Biology, Iowa
State University, Ames, Iowa, USA.
2
On sabbatical at Instituto Mediterraneo de Estudios Avanzados,
Esporles, Spain.
3
Institute of Ecosystem Studies, Millbrook, New York, USA.
4
Centre for Estuarine and Marine Ecology, Netherlands Institute of
Ecology, Yerseke, Netherlands.
5
United States Geological Survey, Denver, Colorado, USA.
6
Institut Mediterrani d’Estudis Avanc¸ats (CSIC-UIB), Esporles, Spain.
7
Finnish Environment Institute, Helsinki, Finland.
8
De´partement des Sciences biologiques, Universite´ du Que´bec a`
Montre´al, Montreal, Quebec, Canada.
9
Laube Engineering LLC, Greeley, Colorado, USA.
Copyright 2008 by the American Geophysical Union.
0886-6236/08/2006GB002854
GB1018 1of10
tible to rapid erosion. Thus these impoundments might be
expected to have much higher rates of sediment OC
accumulation than those in regions where agriculture does
not dominate.
[
4] There are several other unknowns concerning current-
ly accepted estimates of global sediment organic carbon
burial in reservoirs and impoundments. First, recent reviews
[e.g., Dean and Gorham, 1998; Stallard, 1998] have used
the same cited sources [Mulholland and Elwood, 1982] that
are strongly influenced by data on sediment deposition in
impoundments in certain regions of the United States
[Dendy and Champion, 1978]. These data were collected
using a variety of unstandardized met hods of unknown
accuracy collected by diverse government agencies for
divergent purposes, were assembled to analyze changes in
impoundment water storage capacity, not necessarily mass
accumulation, and, following common practice [e.g.,
Ritchie, 1989], inferred sediment dry bulk density (dry mass
per unit volume) as average values from other studies
[Dendy, 1968; Dendy and Champion, 1978] or simply as
estimates of unspecified provenance. Estimates of volume
and mass burial may thus be inaccurate. Further, many
current published studies have not made direct or indirect
estimates of sediment OC content or organic matter in the
systems analyzed but have instead computed OC burial
assuming an unweighted average sediment OC composition
found in other studies [e.g., Ritchie , 1989].
[
5] The objective of this study was to estimate sediment
OC deposition in a variety of impoundments in an agricul-
tural region using estimates of sediment volume accumula-
tion, dry bulk density, and sediment organic matter content
estimated within those same ecosystems. This approach
allows approximation of sediment OC burial over different
intervals during the past century.
2. Methods
[6] We calculated sediment burial in 40 water bodies in
the U.S. state of Iowa. The state of Iowa is one of the most
productive agricultural areas in the world [Downing et al.,
1999], in which 90% of the total land area is under some
form of agricultural use [Arbuckle and Downing, 2001]. The
impoundments we studied varied in area from 0.008 to 42
km
2
(Table 1). This size of impoundment is extreme ly
abundant in the world, covering >75,000 km
2
globally
[Downing et al., 2006], or abo ut 30% of t he Earth’s
0.26 million km
2
of impoundments [ Downing et al.,
2006]. There a re about 80,000 small impoundments
(<0.02 km
2
) in the state of Iowa alone.
[
7] We calculated the sediment accumulation rate in these
impoundments directly by comparing series of sequential,
repeated bathymetric surveys performed in different years.
Thelossofwaterstoragevolumebetweenbathymetric
surveys made in different years was considered to be the
net sediment accrual over that time period. This method is the
most accurate way of determining whole-lake sedimentation
rates [Holeman, 1975; Morris and Fan, 1998], because it
does not require upscaling from small-area sediment depo-
sition records [e.g., Dendy and Champion, 1978] that can be
biased by sediment focusing or nonrepresentative sampling.
[
8] Bathymetric survey s were performed through ice with
sounding rods, from boats with sounding rods, and using
sonar from boats. Surveys using sounding rods determined
horizontal position using standard surveying techniques.
Sonar bathymetry was performed by sounding transects
across predetermined courses. All depth data were adjusted
to elevation benchmarks (usually the elevation of the lake
spillway) to allow direct comparison. All maps were ana-
lyzed through planimetry to determine lake volume. Sedi-
ment accumulation was determined across the entire
impoundment by calculating the differences between vol-
umes and dividing by the number of years between bathy-
metric surveys. Bathymetric surveys were performed from
1913 to 1990 although most of the averages of first and last
survey dates were between 1940 and 1960 (Table 1). Where
more than two bathymetric surveys were performed on a
given water body, sedimentation rates over the longest
possible time period were calculated.
[
9] Sediment mass accumulation in 25 of these impound-
ments was determined as the product of total sediment
volume accumulated and sediment dry bulk density (kg
dry mass m
3
). Determinations of sediment dry bulk
density were made following standard methods [Rausch
and Heinnemann, 1984] between 1978 and 1990 at many
sites across water bodies, distributed to represent principal
areas of sediment deposition. Sediment samples were
extracted by hydraulically pushing Shelby tubes (7.6 cm
diameter) into the sediment to the approximat e depth of
sediment deposition over the period of the bathymetric
surveys. Cores were checked for compaction by noting
differences between sediment core length and depth of
hydraulic penetration [Schaefer, 1981 ]. Core samples were
taken principally in the upper 60 cm of sediment at many
sites in each lake [Iowa State Planning Board (ISPB), 1935;
Lane and Koelzer, 1943; Laube, 1991]. Sediment core
samples were divided into approximately 10 cm segments.
Dimensions of segments of sediment were measured and
volumes were calculated. Samples were oven-dried to
constant mass [Julien, 1995] and dry bulk density calculated
following standard methods and definitions [Julien, 1995;
Spangler and Handy, 1982; Vanoni, 1975]. Particle sizes
were also estimated on these sediment samples using the
pipette method recommended by Vanoni [1975]. This was
done to confirm dry bulk densities because these are often
closely related to particle size distributions. The samples
were dispersed with an air jet for 5 min rather through
shaking overnight [ Hallberg, 1978; Schaefer, 1981].
[
10] Sediment organic carbon concentration per unit dry
mass was estimated by loss on ignition [Heiri et al., 2001] in
2005 in 16 of these same lakes. Sediment samples were
sampled with a box corer from the upper 50 cm from the part
of the lake representing the area of most rapid deposition
[Morris and Fan, 1998] and that deposited over the span of
surveys. Replicate samples were homogenized and sub-
sampled before a known volume of sediment was weighed,
driedtoconstantmass(60°C), reweighed, then ashed
(550°C for 2 h), cooled in a dessicator, and reweighed. Loss
on ignition (LOI) divided by 2.13 was used as an approx-
imation of sediment organic carbon content because work
performed within 50–200 km of ou r sites [Dean and
GB1018 DOWNING ET AL.: SEDIMENT ORGANIC CARBON BURIAL
2of10
GB1018
Gorham, 1976; Dean, 1974] with similar lake water cation
chemistry [Dean, 1999] has shown that this method provides
estimates of C content of lake sediments that are comparable
in precision and accuracy to other methods across a range of
loss on ignition from 6–56%.
3. Results
[11] Sediment burial in these impoundments was sub-
stantial but in line with water storage losses seen in similar
impoundments in other agricultural areas. Annual water
storage volume loss or its equivalent sediment deposition
(S; m
3
a
1
) increased with the size of the watershed
(Figure 1). Multiple regression, using values log trans-
formed to linearize relationships, showed that storage loss
and sediment deposition rate varied as
S ¼ 5682 A
0:3609
W
0:4155
ð1Þ
(R
2
= 0.87, n = 40, p < 0.001) where A is the lake area (km
2
),
W is the watershed area (km
2
), and partial effects of A and W
were both statistically significant (p < 0.01). Annual water
storage loss, expressed as a percentage of the volume of the
water body, was as low as 0.1– 3% in the largest
impoundments but >2% in the smallest ones (Figure 2).
The relationship between percent water storage loss and
impoundment size was statistically significant ( p = 0.009)
but weak (r
2
= 0.17, n = 40).
[
12] Sediment accumulation rates (w;cma
1
), calculated
as sediment deposition (S) divided by the area of
the impoundment, varied from 0.4 to 39.4 cm a
1
(mean
5.9 cm a
1
). Sediment accumulation varied significantly as
w ¼ 0:594 A
0:6768
W
0:3491
AW
0:1035
ð2Þ
(r
2
= 0.60, n = 40, p < 0.001), where the partial effects of A,
W, and AW were statistically significant (p < 0.05) and are
Table 1. Data Used to Analyze Organic Carbon (OC) Deposition
Lake
a
Earliest
Survey
Latest
Survey
Lake Area,
km
2
Watershed
Area, km
2
Watershed
Area:
Lake Area
Sediment
Deposition,
m
3
a
1
Annual
Water
Volume
Loss, % a
1
Sediment
Bulk
Density
Dry,
kg m
3
Sediment
Loss on
Ignition,
b
Fraction of
Dry Mass
OC Burial,
gm
2
a
1
Allerton Reservoir 1913 1939 0.414 12.5 30.1 5320 0.8 0.097
Backbone Lake 1934 1949 0.506 316.7 625.8 11400 1.5 1203 0.086 1089
Barney Mundt 1944 1952 0.016 0.9 55.2 1950 3.8 884 0.093 4864
Beeds Lake 1935 1979 0.429 81.8 191.0 8130 0.6
0.101
Black Hawk Lake 1916 1935 3.230 48.7 15.1 41400 0.8 713
0.098 419
Bloomfield 1937 1951 0.311 5.5 17.7 5720 0.5 0.096
C. A. Stiles 1940 1953 0.054 1.5 27.2 1040 1.1 839 0.094 702
Centerville Reservoir 2 1926 1937 0.207 6.8 32.8 6850 0.5 0.095
Charles Fienhold 1945 1949 0.008 1.1 141.2 3070 19.8 1011 0.093 17392
Clear Lake 1935 1971 14.743 35.5 2.4 55500 0.1
0.235
CM ST P&P RR Reservoir 1903 1918 0.052 6.5 126.0 1480 2.8 0.093
Coralville Reservoir 1958 1975 19.830 7900.0 398.4 992000 1.5 681
0.009 148
Don Williams Lake 1974 1980 0.615 83.7 136.0 17900 0.6 554
0.100 755
Fairfield 3 1927 1953 0.162 7.6 46.7 3420 1.3 827 0.094 773
Farmer’s Ditch 1941 1945 3.885 55.4 14.3 130000 15.6 1094 0.130 2227
Five Island Lake 1935 1970 4.011 34.1 8.5 26400 0.5
0.169
Fred Hollrah 1944 1949 0.010 0.6 53.1 1670 7.1 934 0.093 6537
Honey Creek F-1 1955 1961 0.034 3.1 92.3 1190 0.5 1057 0.093 1640
Jones Creek Reservoir 1942 1953 0.078 5.8 74.3 6230 2.0 878 0.094 3093
Lake Icaria 1976 1986 2.833 72.8 25.7 212000 2.3
0.090
Lake Iowa 1978 1988 0.340 5.4 15.9 5670 0.5 0.096
Lake John Deere 1979 1988 0.050 3.4 68.4 6030 3.0 0.093
Lake Panorama 1970 1980 5.666 1139.6 201.1 657000 2.8 1202 0.107 6987
Lake of Three Fires 1936 1950 0.393 15.5 39.3 33700 2.2
0.085
Lake Wapello 1937 1982 1.153 20.2 17.5 16400 0.3 521
0.146 508
Lower Pine Lake 1924 1932 0.263 39.2 149.0 28900 3.2 1256
0.089 5771
Max Miller 5 1941 1952 0.013 0.6 45.4 1410 2.2 1179 0.093 5593
Prairie Rose Lake 1971 1985 0.882 18.5 21.0 43900 1.8 1474
0.083 2862
Red Rock Reservoir 1977 1984 42.089 15695.9 372.9 2240000 2.9
0.070
Silver Lake 1935 1973 2.699 32.4 12.0 14400 0.3
0.122
Springbrook Lake 1936 1980 0.069 4.9 71.3 2010 0.9 767
0.105 1111
Swan Lake 1935 1980 0.526 3.1 6.0 3950 0.7 881
0.114 353
Theobold C 1949 1952 0.041 0.6 14.6 1420 3.1 1088 0.093 1635
Theobold Main 1949 1952 0.106 1.1 10.8 5230 4.1 1158 0.094 2519
Tracy North 1939 1953 0.018 0.6 34.0 1010 1.7 868 0.093 2122
Tuttle Lake 1935 1973 11.396 444.4 39.0 104000 0.7 0.192
Union Grove Lake 1936 1950 0.478 27.9 58.4 14800 1.4 686
0.097 958
Upper Pine Lake 1934 1990 0.239 35.7 149.6 12600 1.6 0.094
Wilbur Meyer 1944 1952 0.010 0.8 73.0 1490 2.8 900 0.093 5647
William Esbeck 1944 1949 0.010 0.5 50.9 1110 7.1 900 0.093 4201
a
Lake names either refer to known geographic place names or the names of owners of impoundments.
b
Italicized data were predicted from equation (3); underlined data are observed loss on ignition (LOI) values.
GB1018 DOWNING ET AL.: SEDIMENT ORGANIC CARBON BURIAL
3of10
GB1018
listed in order of their statistical strength. Impoundment area
had the strongest influence on w (Figure 3).
[
13] The sediment dry bulk density varied substantially
among lakes and was roughly correlated with the rate of
fractional annual water storage loss (Figure 4) as well as
with w. Dry bulk density was correlated with particle size
distributions in the expected way [Lane and Koelzer, 1943].
Dry bulk density varied from about 500 kg m
3
, which is
about 4 times the dry bulk density of peat [Schlotzhauer and
Price, 1999], or about the dry bulk density of dry corn and
wheat seeds [Trabelsi et al., 2001], to 1500 kg m
3
, which
is about the bulk density of sand or firmly packed soils
[Henderson et al., 1988; McNabb et al., 2001; Morris and
Figure 1. Relationship between the volume of sediment
deposited annually in impoundments and the size of the
watershed. Sediment deposition was measured as annual
storage loss estimated from repeated bathymetric surveys.
Figure 2. Relationship between the annual fractional loss
of lake water volume and lake area. Dashed line is the fitted
least squares regression (r
2
= 0.17, n = 40, p = 0.009).
Figure 3. Relationship between sediment accumulation
rate and the area of lakes. Dashed line is the fitted least
squares regression (r
2
= 0.30, n = 40, p = 0.003).
Figure 4. Relationship between the amount of dry matter
per unit volume of sediments (bulk density) and the annual
storage loss of lakes. Dashed line indicates the fitted least
squares regression (r
2
= 0.20, n = 25, p = 0.025).
GB1018 DOWNING ET AL.: SEDIMENT ORGANIC CARBON BURIAL
4of10
GB1018
Fan, 1998]. The average dry bulk density observed in these
lake sediments (942 kg m
3
) was only about 50% lower
than the dry bulk densities of normal terrestrial soils
[Calhoun et al., 2001]. The average and range of our dry
bulk densities were comparable to those found for reservoirs
and impoundments (1100 kg m
3
; range 320– 2000) by
Dendy and Champion [1978] and similar to those typifying
dry bulk densities of material s found perma nently sub-
merged in impoundments [Stall, 1981].
[
14] Sediment LOI varied substantially among the 16
impoundments on which we measured it (Table 1). The
highest L OI was found in Clear Lake, a eutrophic,
impounded natural lake with a small watershed, while the
lowest LOI was found in Coralville Lake, a large flood
control impoundment with a very large watershed. If
sediment carbon is 47% of loss on ignition, as was found
for lakes near our sites [Dean, 1999; Dean and Gorham,
1976; Dean, 1974], the accumulating sediments we ana-
lyzed were composed of from 0.5% to 11% OC. Further,
using standard least squares multiple regression analysis, we
found that measured LOI, expressed as a fraction of
sediment dry mass, varied systematically with both lake
size and watershed size as
LOI ¼ 0:093 þ 0:0096 A
ðÞ
4:2 10
5
W
þ 3:497 10
7
AW
ð3Þ
(r
2
= 0.84, n = 16, p < 0.001) where all variables are as in
equation (1) and all regression coefficients were statistically
significant (p < 0.05). Because of the joint influence of lake
and watershed area on LOI, sediment OC also varied
systematically with the ratio of the watershed area to the
area of the lake. LOI as a fraction of dry mass of sediment is
shown plotted in Figure 5 for the 16 impoundments for
which we made direct estimates. Because some lakes were
on private property or were recently drained or filled with
soil, equation (3) was used to predict loss on ignition in
impoundments where it was not estimated by direct means
(italics on Table 1). The average LOI in the impoundments
we studied was about 10.3%, which corresponds to a
sediment OC content of about 4.8%, assuming that
sediment OC is approximately 47% of LOI [Dean, 1974].
[
15] We approximated sediment OC burial using direct
measures of hydrographic change and dry bulk density from
25 impoundments and approximations of sediment OC from
direct measures of LOI on 16 impoundments and predicted
LOI (equation (3)) for nine of them. Estimated sediment OC
burial (B; g m
2
a
1
) varied from only about 150 g m
2
a
1
in the large flood control Coralville Reservoir to more than
17,000 g m
2
a
1
in the tiny (0.008 km
2
) impoundment
owned by farmer Charles Fienhold (Table 1). Organic
carbon burial was generally high in small impoundments,
declining exponentially in large ones (Figure 6). The data fit
a power function:
B ¼ 1060A
0:298
ð4Þ
(r
2
= 0.35, n = 25, p = 0.002), suggesting a rapid decline in
carbon burial in large impoundments. Multiple regression
indicated that there was no significant trend in burial rates
with the median year of bathymetric surveys (p > 0.05).
4. Discussion
[16] Sediment deposition rate increased with both im-
poundment area and watershed size (equation (1)). Water
storage loss rates seen in this study were similar to those
Figure 5. Sediment dry mass loss on ignition (LOI)
expressed as fraction of dry sediment mass related to the
ratio of watershed to lake area. Joint, independent
influences of watershed and lake area are implied by
equation (3). Dashed line is the fitted least squares, semilog
linear regression (r
2
= 0.60, n = 16, p < 0.0005).
Figure 6 . Sediment organic carbon (OC) burial in
impoundments measured in this study. Dashed line is the
least squares regression between carbon burial and lake size
(r
2
= 0.35, n = 25, p = 0.002).
GB1018 DOWNING ET AL.: SEDIMENT ORGANIC CARBON BURIAL
5of10
GB1018
