Content uploaded by Mark B. Edlund
Author content
All content in this area was uploaded by Mark B. Edlund on Mar 28, 2015
Content may be subject to copyright.
ORIGINAL PAPER
A whole-basin stratigraphic record of sediment
and phosphorus loading to the St. Croix River, USA
Laura D. Triplett ÆDaniel R. Engstrom Æ
Mark B. Edlund
Received: 4 May 2007 / Accepted: 3 March 2008
ÓSpringer Science+Business Media B.V. 2009
Abstract Lake St. Croix is a natural impoundment
of the lowermost 37 km of the St. Croix River in
Minnesota and Wisconsin, making this one of a few
large river systems in the world possessing a long-
term depositional basin at its terminus. The river’s
relatively pristine condition led to its designation as a
National Scenic Riverway in 1968, but increasing
urbanization in its lower reaches has raised concerns
about impacts on water quality. This study was
initiated to reconstruct historical loadings of sus-
pended sediment and phosphorus (P) from the
sediment record in Lake St. Croix. Twenty-four
piston cores, with an average length of 2 m, were
collected along eight transects of the lake. Dated
chronologies from
210
Pb,
137
Cs and
14
C were used to
calculate the rate of sediment accumulation in the
lake over the past 100?years. Diatom microfossil
analysis was used to reconstruct historical lakewater
P concentrations over the same time period, and
sediment P analysis quantified the amount of P
trapped in lake sediments. Using a whole-lake mass
balance approach, the loading of sediment and P to
Lake St. Croix over the last 100?years was
calculated. Beginning in 1850, sediment accumula-
tion increased dramatically to a peak in 1950–1960
of eight times background rates prior to European
settlement. The peak is driven largely by sediment
contributions from small side-valley catchment
tributaries to the downstream half of the lake. The
total P load to the lake increased sharply after 1940
and remains high, at around four times the level of
pre-European settlement conditions. The timing of
peak sediment and P loading to the lake shows that
early settlement activities, such as logging and the
conversion of forest and prairie to agricultural land
between 1850 and 1890, had only modest impacts
on the lake. By contrast, the mid-1900s brought
major increases in sediment and P loading to the
lake, suggesting that relatively recent activities on
the landscape and changes to nutrient balances in
the watershed have caused the current eutrophic
condition of this important recreational and natural
resource.
This is one of eight papers dedicated to the ‘‘Recent
Environmental History of the Upper Mississippi River’’
published in this special issue of the Journal of
Paleolimnology. D. R. Engstrom served as guest editor
of the special issue.
L. D. Triplett (&)
Department of Geology and Geophysics, University
of Minnesota, Minneapolis, MN 55455, USA
e-mail: ltriplet@gustavus.edu
D. R. Engstrom M. B. Edlund
St. Croix Watershed Research Station, Science Museum
of Minnesota, 16910 152nd St. N., Marine on St. Croix,
MN 55047, USA
Present Address:
L. D. Triplett
Department of Geology, Gustavus Adolphus College,
800 W. College Avenue, St. Peter, MN 56082, USA
123
J Paleolimnol
DOI 10.1007/s10933-008-9290-7
Keywords St. Croix River Sediment loading
Phosphorus loading Paleolimnology
Human impact Mississippi River
Introduction
Large temperate rivers around the world have been
significantly impacted by recent human activities
(Bennett et al. 2001; Bernhardt 1995; Meybeck et al.
1990; Meybeck and Helmer 1989; Saeijs and Van
Berkel 1995; Smith 2003; Zhang et al. 1999), but
because of relatively short periods of water quality
monitoring there is not a good understanding of the
magnitude of change in these systems. To protect
these resources with sound management strategies,
their ‘natural’, or pre-disturbance, condition must be
understood. In most places, regular and reliable
testing of environmental variables did not begin until
1950 or later, after major industrial and social
changes had already affected waterways and land-
scapes. Furthermore, sediment transport in most
riverine systems is too episodic and complex to
allow accumulation of continuous sedimentary
sequences, so historical water quality cannot be
reconstructed from sedimentary records, as is more
commonly done for lakes. However, several studies
of the Mississippi River’s natural impoundment at
Lake Pepin (Balogh et al. 1999; Engstrom et al. this
issue; Maurer et al. 1995) demonstrated that these
types of deposits can be interpreted as effectively as
those from more typical (non-riverine) lakes.
The St. Croix River, a major tributary to the upper
Mississippi River, is one of a very few large rivers in
the world that has a natural lake, or long-term
depositional basin, at its terminus. Fine-grained sed-
iments have accumulated on the lake bottom in a
conformable, continuous manner since the lake’s
formation around 9,000 years ago (Eyster-Smith et al.
1991), thus, paleolimnological techniques using strati-
graphic analysis of sediment cores can be applied to
evaluate the historical conditions of the river. Specif-
ically, the changes in sediment and phosphorus flux to
the lake can be reconstructed from the sediment
record, thereby inferring how changes on the land-
scape have affected the aquatic environment.
To fully elucidate the history of phosphorus loading
to Lake St. Croix, it is necessary to account for spatial
variability in the depositional basin. By analyzing
sedimentary phosphorus (total-P
sed
) in multiple cores
throughout the lake basin, a whole-lake flux of
phosphorus (P) to the sediment can be calculated
(Anderson and Rippey 1994; Evans and Rigler 1980;
Moss 1980; Schelske et al. 2006). When combined
with historical lakewater phosphorus (total-P
lake
)
reconstructed from diatoms using multivariate transfer
function techniques (Anderson et al. 1994; Bennion
et al. 1996; Hall and Smol 1992), a quantitative whole-
basin P mass balance can be established. This approach
was described by Rippey and Anderson (1996) and has
been used and refined by others including Hall and
Smol (1992) and Jordan et al. (2002).
Site description
The St. Croix River drains an area of 22,200 km
2
in
Minnesota and Wisconsin (Fig. 1) before it joins the
Mississippi River at Prescott, Wisconsin. The north-
ern part of the watershed consists of northern boreal
forest, bogs and peatlands (Curtis 1959). The south-
ern sub-watersheds were originally vegetated by
prairie and northern mixed hardwoods but have come
to be dominated by agricultural land uses (Curtis
1959; Troelstrup et al. 1993).
The lowermost 37 km of the St. Croix River
function as a lake (Lake St. Croix) due to its natural
impoundment by a delta of the Mississippi River
(Blumentritt et al. this issue) around 9500 B.P.
Today, Lake St. Croix occupies a narrow (0.5–2 km
wide) riverine basin that is divided into four sub-
basins by the deltas of side-valley tributaries along its
course: the Willow River, Valley Branch Creek and
the Kinnickinnic River (Fig. 1). The maximum depth
in these sub-basins is between 10 and 22 m and the
water residence time in the lake as a whole is on the
order of 20–50 days depending on season and
precipitation. The average annual total phosphorus
(TP) concentration in the lake is around 50 lgPO
4
–
Pl
-1
, alkalinity is near 90 mg l
-1
, and average total
suspended solids concentration is on the order of
4mgl
-1
. Dissolved organic carbon is high due to the
pine forests and peatlands in the upper watershed, so
that light attenuation is rapid and Secchi depths are
small (0.21–1.8 m) (Robertson and Lenz 2002).
Lake St. Croix is considered to be eutrophic by most
measures.
J Paleolimnol
123
Before European settlement, Native Americans
lived in the St. Croix watershed and influenced
vegetation by intentional and unintentional burning
(Curtis 1959). However, their activities probably
had negligible impacts on the river and lake water
quality. Europeans first settled in the St. Croix
valley in the 1830s (Anderson et al. 1996; Dunn
1979; McMahon and Karamanski 2002); logging
and milling were well underway by the late 1840s
and ‘‘there was a nearly complete occupation of the
prairies’’ by 1880 (Curtis 1959). Agricultural utili-
zation continued unabated through the 1900s,
although principal products changed from wheat
to dairy to corn through time. Human population in
the watershed increased to 250,000 by 1920, then
remained static until the 1970s when it began to
increase again; by 1992 there were 400,000 people
living in the St. Croix watershed (Mulla and Sekely
this issue). Counties bordering the lower St. Croix
continue to experience strong urban- and suburban-
ization pressure because of their proximity to the
rapidly-growing Minneapolis/St. Paul metropoli-
tan area. The lower St. Croix River is particu-
larly accessible to metro-area residents, and over
1 million visitors boat, swim, fish or camp along
the river each year (Lenz et al. 2003).
The St. Croix River is valued highly as a
recreational and environmental resource and is
commonly regarded as nearly pristine. Parts of
the river were designated as National Scenic
Riverway in 1968 and 1972, and more than 60
endangered and threatened species live in the
Riverway (Lenz et al. 2003). However, the eutro-
phic condition of Lake St. Croix and continuing
urbanization of the lower watershed have prompted
concern among government officials and citizen
groups determined to protect the St. Croix’s unique
character. In response, the St. Croix Basin Water
Resources Planning Team, comprised of regional,
state and federal agencies, began a program to
monitor and model the river/lake system to help
develop future water quality goals and management
policies. Several studies have used water quality
monitoring data from 1970–2000 to examine recent
nutrient dynamics and trends in the St. Croix River.
Relative to the Minnesota and upper Mississippi
rivers, the St. Croix has the lowest total-P
lake
concentration (Kroenig and Andrews 1997) and
Wisconsin
Minnesota
N
Lake
St.
Croix
map
8
1
2
3
4
5
7
02468
Kilometers
Mississippi River
Valley Creek
Trout Brook
Willow River
Kinnickinnic River
Wisconsin
6
Sub-basin C
Sub-basin D
Sub-basin B
Sub-basin A
48° N
95° W
Lake St.
Croix
0 20 40
Kilometers
Minnesota
Fig. 1 The St. Croix River
begins in northwestern
Wisconsin and flows south
and southwest along the
border of Wisconsin and
Minnesota to its confluence
with the Mississippi River.
The last 37 km of the river
function as a lake, which is
shown in detail in the right
panel. Twenty-four
sediment cores were taken
along eight transects of the
lake. Note that the lake is
divided into four sub-basins
by the deltas of side-valley
tributaries
J Paleolimnol
123
relatively good water quality compared to other
rivers in the region (Troelstrup et al. 1993).
Methods
Coring
Sediment cores were collected from the depositional
area of the lake, which was defined as those regions
of the lake bottom where fine-grained sediment
accumulates conformably. It excludes shallow-water
areas with sand and cobble substrates as well as
deltas at river mouths and submerged paleo-islands.
The delineation of depositional versus non-deposi-
tional areas of the lake was determined primarily
from GIS-based bathymetric maps (Minnesota
Department of Natural Resources) but was ‘‘ground-
truthed’’ by gravity-core sediment surveys.
Twenty-four sediment cores were collected from
1999 to 2001 along eight east-west transects of the
lake (i.e. aligned perpendicular to the flow direction)
(Wright 1991; Wright et al. 1983). Two transects
were located in each of the four sub-basins and are
numbered from 1 to 8 going upstream (Fig. 1). Cores
were wrapped in polyethylene film (food wrap) and
aluminum foil. All cores were stored at 4°C.
Core lithology
Whole-core magnetic susceptibility measurements
were made on all cores using a Geotek LTD multi-
sensor core logger with a Bartington MS2 loop sensor
(10-cm diameter). The logger had an automated track
feeder that accommodated 1.6-m sections of core.
Cores were brought to room temperature prior to
measurement and susceptibility readings were taken
at 1 cm intervals. Following susceptibility analysis,
down-core smearing was removed from the core
surface, and cores were sectioned into 2 cm intervals.
Sediment samples were placed in 120-ml screw-top
polypropylene jars for storage at 4°C.
Standard loss-on-ignition (LOI) techniques (Dean
1974) were used to determine dry density and the
weight-percent of water, organic matter and carbon-
ate of the sediments. All depth intervals in the
primary cores were homogenized then sub-sampled
for LOI analysis, while every fourth sample in the
secondary cores was analyzed. Samples of 2–4 g
(wet mass) were dried overnight at 100°C, then
ignited sequentially at 550 and 1,000°C for 1 h each
with mass measured between each step. Dry density
was calculated from each sample’s water content
and fixed density values for organic, carbonate and
inorganic matter.
Sediment dating
The primary core from each transect was dated by
210
Pb to establish chronology and calculate sediment
accumulation rates. Lead-210 was measured by the
activity of its daughter product,
210
Po, considered to
be in secular equilibrium with its parent, in 18–25
samples per core. As an internal yield tracer,
209
Po
was added. Freeze-dried samples were treated with
HCl to remove carbonate and the Po isotopes were
distilled at 550°C and plated onto silver planchets
from a 0.5 M HCl solution (adapted from Eakins and
Morrison 1978). Activity was measured for 0.8–3 9
10
5
s using an Ortec alpha spectrometry system.
Supported
210
Pb was determined from the asymptotic
activity of the lowermost samples in the core and then
subtracted from measured total activity in the upper
samples to obtain the unsupported activity at each
interval. Dates and sedimentation rates were calcu-
lated using the constant rate of supply (CRS) model
(Appleby and Oldfield 1978).
Determination of the supported
210
Pb activity is
crucial to this dating method. In three of the primary
cores (2B, 3B and 4B), the total activity as determined
by alpha spectrometry did not decline monotonically
with depth but instead co-varied with changes in
magnetic susceptibility. This covariance suggested that
supported
210
Pb activities were affected by changes in
sediment lithology at these locations in the lake. Gamma
spectrometry was thus used to directly and indepen-
dently measure supported (
214
Pb) and total
210
Pb
activities in these three cores. Freeze-dried sediment
samples weresealed with epoxy resin and ingrown for a
minimum of 30 days to achieve secular equilibrium
between the native
226
Ra and its decay products.
Isotopic activities were measured for 0.6–3 910
5
s
using an EG&G Ortec high-resolution germanium
well detector and multichannel analyzer. Supported
210
Pb was measured as
214
Pb (295.2 and 351.99 kev),
a short-lived intermediary in the radioactive decay
sequence from
226
Ra to
210
Pb. Unsupported
210
Pb was
J Paleolimnol
123
calculated as the difference between total
210
Pb (mea-
sured directly as
210
Pb at 46.52 kev) and supported
210
Pb. An efficiency curve for the detector was gener-
ated using a sediment matrix spiked with known
activities of
210
Pb,
137
Cs,
7
Be,
54
Mn, and
109
Co, and
was corrected for the small amount of native
210
Pb.
In addition, freeze-dried sediment samples from
the eight primary cores were analyzed for
137
Cs
to identify the 1963–1964 peak in atmospheric
nuclear bomb testing. Samples were measured at
661.62 kev using a high-resolution germanium well
detector multichannel analyzer, as described above.
Four to eight samples were analyzed from each
core with selection based on initial results from
210
Pb dating.
Finally, terrestrial (woody) organic matter sam-
ples were collected from near the bottom of four
cores for radiocarbon dating. Sample depths were
chosen to correspond with a distinctive feature in
the magnetic susceptibility profile of the core so that
the
14
C dates could be correlated to other cores. The
woody pieces were converted to graphite targets at
the Limnological Research Center, University of
Minnesota, and analyzed by accelerator mass spec-
trometry (AMS) at the University of Arizona in
Tucson. The CALIB program v. 4.3 (Stuiver and
Reimer 1993) and the atmospheric sample dataset
from Stuiver et al. (1998) were used to convert
14
C
dates to calendar years.
Sediment phosphorus
Freeze-dried sediments were extracted for phosphorus
determinations according to fractionation procedures
adapted from Hieltjes and Lijklema (1980) and
Engstrom and Wright (1984). Extracts were analyzed
with a Lachat QuikChem 8000 flow-injection auto-
analyzer using the ascorbic acid method. Extraction
temperatures for total-P were maintained in a hot-
water bath, and extracts were separated from sediment
residue by centrifugation. All dilutions were done
according to weight on an electronic balance. Repli-
cated extractions of 20 samples had an average
relative difference of 1.8% for total-P, 1.2% for iron-
and aluminum-bound phosphorus (NaOH–P) and
1.4% for calcium-bound phosphorus (HCl–P).
Diatom analysis and lakewater P reconstruction
Two sediment cores (1B and 6B) were analyzed for
diatom microfossils to reconstruct a detailed history
of total-P
lake
concentrations as described in Edlund
et al. (this issue-a). Four hundred diatom microfos-
sils were classified in each sediment sample, then
historical water column total P (total-P
lake
) was
reconstructed using weighted averaging calibration.
Although Lake St. Croix is a riverine system, most of
the diatom flora is lacustrine, and a diatom training
set based on 55 Minnesota lakes was used. These data
were analyzed using weighted average regression
software (CALIBRATE; Juggins 1998) to determine
total-P
lake
optima for 108 diatom taxa, providing
potential reconstructions of total-P
lake
values from
0.009 to 0.105 mg l
-1
.
Results and discussion
Magnetic susceptibility
Magnetic susceptibility in all the cores shows a
common pattern: relatively constant and low values in
the oldest sediments, followed by a sharp increase to
peak values and a subsequent decrease to the present
(Fig. 2). The oldest sediment in each core ranges from
25 to 35 SI units, although cores from transect 8 have
slightly higher background levels (30–40 SI). There
are two notable exceptions to this pattern. First, the
lower half of core 4C records pronounced fluctuations
between 30 and 60 SI that are possibly due to post-
depositional sediment disturbance such as scouring or
slumping; this core was not used for detailed analysis
and correlation. Second, the three cores in transect 1
follow the general pattern described above but they
also record two large magnetic susceptibility peaks
well before the lake-wide peak described above.
These earlier peaks at the downstream end of the lake
may correspond to sediment back-washed into the St.
Croix basin by major pre-European flood events of the
Mississippi River.
Cores within a given transect have highly similar
magnetic susceptibility profiles indicating that the
pattern of sediment deposition across the width of
the lake is quite uniform and thereby confirming the
original delineation of depositional areas in the lake.
J Paleolimnol
123
This similarity also confirms that sediments are not
significantly remixed after deposition, because mix-
ing would likely cause irregular fluctuations in the
magnetic profiles from core to core. Consequently,
chronologies constructed for the primary core in each
transect can be translated with confidence to the other
cores in that transect. Likewise, because the general
pattern holds throughout the lake, some correlations
can be made among transects as well.
In the upstream half of the lake, the magnetic
susceptibility peaks are stronger in transects 8 and 7
(45–60 SI) and are more muted in transects 6 and 5
(*33 SI). This down-stream trend likely results from
the progressive settling of sediment from the main-
stem of the river as water moves through the lake
basin. However, beginning with transect 4 there is a
pronounced increase in peak intensity (70 SI) with
transects 3 and 2 showing the highest peak intensities
in the entire lake (135 and 118 SI, respectively).
Differences in magnetic susceptibility are due to
differences in sediment mineralogy, grain size, and/or
concentration of magnetic grains. Therefore, the
stronger peaks in the downstream cores suggest a
different sediment provenance than the muted
upstream peaks. This is the first indication that side-
valley tributaries flowing directly into the down-
stream half of the lake have at times contributed large
amounts of sediment to the basin. In addition, the
deltas of those tributaries were occasionally dredged
during the twentieth century to maintain an adequate
navigation channel up the river. The dredging likely
remobilized sediment that had been trapped in the
delta, and this may have contributed to the higher
magnetic susceptibility of downstream sediments.
Loss-on-ignition
Lake St. Croix sediments are consistently between
80 and 90% inorganic matter by weight as deter-
mined from LOI analysis. Carbonate comprises the
smallest proportion of the sediment, ranging from 3
to 8% by weight. All cores show a peak in inorganic
content of 82–91% at the time of highest magnetic
susceptibility, confirming that a change in sediment
character—and likely in sediment provenance—
happened at that time.
Organic matter generally ranges from 7 to 15%
with the highest values found in the upper 30–50 cm
of the cores. This up-core increase may be due in part
to incomplete diagenesis (compared to older strata in
the core), but more likely is due to increased
productivity in recent times. The depth at which
organic matter began to increase is deepest at the
upstream end of the lake (52 cm, core 8C) and
shallowest at the downstream end of the lake (30 cm,
core 1B). Regardless of lake position, the increase
0
100
200
0
100
200
0
100
200
0
100
200
300
0 30 0 30 0 30 60 0 30 60 90 0 30 60 90 0 30 60 90 120
0
100
200
0
100
200
0
100
200
0
100
200
300
7
6
8
5
4
3
2
1
Magnetic susceptibility (SI) Magnetic susceptibility (SI)
Core depth (cm)
Fig. 2 Magnetic
susceptibility for the
24 cores, which were
taken along eight transects
of Lake St. Croix
J Paleolimnol
123
always dates to about 1960, suggesting that an
increase in productivity occurred throughout the lake
at that time.
Cesium-137
Cesium-137 activities in all eight primary cores show
distinct peaks (Fig. 3) allowing for precise dating of
those depth intervals to 1963–1964. The burial depth
for these
137
Cs maxima range from 24 to 26 cm in core
1B to 62–66 cm in core 8C. Peak activities range from
1.18 pCi g
-1
(core 3B) to 2.70 pCi g
-1
(core 5B).
Lead-210 dating by alpha spectrometry
The
210
Pb activity profiles for five of the eight
primary cores (8C, 7B, 6B, 5B, 1B) show roughly
monotonic declines down-core, while the other three
(4B, 3B, 2B) exhibit more complicated activity
profiles (Fig. 4). Surface activities in all cores are
relatively low, ranging from 5.5 to around 14.0 pCi
g
-1
, with no apparent spatial pattern to the variation.
In the five cores modeled successfully with alpha
spectrometry data, the background (supported)
210
Pb
activities are between 1.2 and 2.1 pCi g
-1
, as defined
100
50
25
0
75
100
50
25
0
75
100
50
25
0
75
100
50
25
0
75
100
50
25
0
75
100
50
25
0
75
100
50
25
0
75
100
50
25
0
75
Core Depth (cm)
137Cs activity (pCi g-1)
01 23
01 23
0123
01 23
1B
4B
3B
2B
5B
8C
7B
6B
Fig. 3 Cesium-137 activity
profiles for the eight
primary cores
J Paleolimnol
123
by 4–9 intervals with near-constant
210
Pb values at
depth.
The inventory of supported
210
Pb in these five
cores ranges from 51 pCi cm
-2
(1B) to 103 pCi
cm
-2
(8C) which is equivalent to a
210
Pb flux of 1.7–
3.3 pCi cm
-2
year
-1
. The lake-wide average
210
Pb
flux to Lake St. Croix sediments (2.3 pCi
cm
-2
year
-1
) is significantly larger than the atmo-
spheric
210
Pb flux directly to the lake surface
(0.45 pCi cm
-2
year
-1
) (Urban et al. 1990), indicat-
ing that much of the lake’s
210
Pb inventory is
delivered by the St. Croix River. However, the
riverine
210
Pb input represents only a small fraction
(about 1%) of the total
210
Pb mass delivered to the
watershed from atmospheric deposition. This indi-
cates that most of the
210
Pb deposited on the
landscape is trapped in the watershed soils and
sediments where it decays away.
All the dated cores have
210
Pb activity profiles
showing changes in slope due to changes in sediment
flux to those core sites. These types of profiles are
best interpreted using the constant rate of supply
(c.r.s.) model, which allows for changes in sediment
accumulation while assuming constant flux of
210
Pb
to the core site. This assumption of constant
210
Pb
flux may be suspect in Lake St. Croix because much
02468101214 02468101214
200
150
100
50
0
250
200
150
100
50
0
250
200
150
100
50
0
250
200
150
100
50
0
250
200
150
100
50
0
250
200
150
100
50
0
250
200
150
100
50
0
250
200
150
100
50
0
250
Total 210Pb
Unsupported
210Pb
Supported
210Pb
210Pb activity (pCi g-1)
02468101214 0 2 4 6 8 10 12 14
1B
4B
3B
2B
5B
8C
7B
6B
Core Depth (cm)
Fig. 4 Lead-210 activity
profiles for the eight
primary cores. Cores 8C,
7B, 6B, 5B and 1B were
dated using alpha
spectrometry and total
210
Pb
activity is shown. Cores 4B,
3B and 2B were dated by
gamma spectrometry, and
supported
210
Pb activity
(measured) and unsupported
210
Pb activity (calculated)
are shown
J Paleolimnol
123
of the
210
Pb load comes, not from direct atmospheric
deposition, but instead is delivered by the St. Croix
River. Furthermore, unsupported activities are very
low in Lake St. Croix sediments because of dilution
by high sediment influx. Therefore, while the c.r.s.
model can be used to date most of these cores,
additional dating markers must also be used to
independently verify the chronology.
The
137
Cs peaks can be used to check the
210
Pb
dating results. The
210
Pb dates at the depths of the
137
Cs peaks ranged from 1962 ±3 year (5B) to
1975 ±8 year (4B) (in the six cores dated by
210
Pb).
The mean of those six dates is 1966, two or three
years younger than the expected date of the peak,
1963–1964. Excluding core 4B, which has a larger
error because it was counted by gamma assay, the
average date is 1965. In either case, the
210
Pb dates
are reasonably close to the
137
Cs date, confirming the
reliability of the
210
Pb chronology from mid-century
onward.
Lead-210 dating by gamma spectrometry
Three of the eight primary cores from the lower lake
transects (2B, 3B, and 4B) had
210
Pb profiles (from
alpha spectrometry) that were uninterpretable using
any available dating model. Lead-210 activities are
generally low throughout these cores, indicating
dilution by extremely high sedimentation rates. More
importantly, the lowest total
210
Pb values are found
mid-core, rather than in the deeper (pre-settlement)
strata that are normally used to estimate supported
210
Pb. These low mid-core activities coincide with
peak values in magnetic susceptibility, implying that
supported
210
Pb is variable in these cores and that the
variation is related to major changes in sediment
lithology and provenance. Therefore, supported
210
Pb
could not be estimated from the asymptote of total
210
Pb at depth, but was instead measured directly
for each sediment interval by gamma assay of
214
Pb. Unsupported
210
Pb was estimated by differ-
ence between total
210
Pb (also measured by gamma
spectrometry) and
214
Pb.
The results from gamma assay of cores 2B, 3B,
and 4B confirm that supported
210
Pb (
214
Pb) is lower
in the mid-core strata where total
210
Pb activities are
at a minimum (Fig. 4). In core 4B, the point-
transformed estimates of unsupported
210
Pb range
from 1.8 to 12.2 pCi g
-1
, with a local minimum
between 52 and 78 cm where the peak in magnetic
susceptibility would indicate a high input of eroded
sediment. Core 4B had less variability in the
supported
210
Pb activity than did cores 2B and 3B,
so the 4B gamma values were used in the c.r.s.
model. Because the errors of gamma assay are
substantially higher than those of alpha counting,
the uncertainty of the c.r.s. model dates for 4B are
also larger.
In cores 2B and 3B, the estimates of unsupported
210
Pb (by difference) are so low—relative to analyt-
ical uncertainty—that a reliable
210
Pb chronology
cannot be modeled. For these cores, an alternative
chronology was derived by correlation of magnetic
susceptibility profiles with the
210
Pb-dated cores from
adjacent transects. Core 2B dates were derived by
correlation with 1B, and 3B dates by correlation with
4B. In both cases, the
137
Cs peak also provided a
reliable dating marker (1963–1964). Core 1B is
located downstream of 2B within the same deposi-
tional basin, and because there are no significant
tributaries between the two sites, core 1B is a muted
replica of 2B.
Core 3B was treated in a similar manner by
correlation with core 4B. In this case, 3B is down-
stream of 4B and appears from the magnetic
susceptibility profiles to be strongly impacted by
Trout Brook, which enters the lake between these two
transects (Fig. 2). Despite the differences in magnetic
character (and sediment provenance), the general
pattern of land-use change and sediment delivery
should be broadly contemporaneous between these
two depositional areas. Therefore,
210
Pb dates for the
main magnetic features in 4B can provide dates for
those features in 3B. Specifically, the base of the 4B
magnetic susceptibility rise (1872), the top of that rise
(1946), and a secondary peak (1955) were applied to
the 3B magnetic profile.
For both 2B and 3B, dates between magnetic
markers were interpolated assuming a constant
dry-mass accumulation rate (g cm
-2
year
-1
) for the
given section of core.
Carbon-14
Carbon-14 dates were obtained from cores 8C, 5B,
3A and 1B. The date from core 8C was the most
problematic, with five possible calendar dates
returned by CALIB (Stuiver and Reimer 1993)
J Paleolimnol
123
(Table 1). The 1940 and 1950 dates are not likely to
be real based on
210
Pb and
137
Cs dating of the upper
core, so these were eliminated and the median date of
the remaining options (1760) was used. Three
possible dates were returned by CALIB for the
5B sample. In this case, none of the possibilities
overlapped with
210
Pb dates so the median of the
three dates (1350) was used. CALIB returned only 1
possible date for each of the 3A and 1B samples and
these were used without modification.
The calibrated
14
C dates were used to calculate a
single pre-European settlement sediment accumula-
tion rate for each core, representing the period from
the
14
C sample depth to settlement (initial magnetic
susceptibility rise, around 1850). These rates ranged
from 0.28 to 2.81 kg m
-2
year
-1
, with the highest
value from core 8C at the upstream end of the lake.
Based on the magnetic susceptibility profiles, the
calibrated
14
C dates in these four cores were applied
to the other two cores within their respective
transects, and also to cores from adjacent transects
without
14
C dating. The 5B date was applied to
transects 7, 6 and 4, and the 1B date was translated to
transect 2. The 8C date was only used to date transect
8 cores because it is the youngest
14
C sample and has
the largest error. The 3A date was only used to date
transect 3 cores, because its associated magnetic
feature was not identifiable in other transects.
Finally, the
14
C date from 1B was used to
determine a pre-settlement sedimentation rate on core
2B. The magnetic feature associated with the 1B date
was not identifiable in transect 2 cores. Instead, the
ratio of post- to pre-settlement sediment accumulation
was calculated for 1B, where post-settlement is
defined as 1838 (magnetic susceptibility rise) to
present, and presettlement as 1838 to the
14
C
calibrated date of 1413. That ratio was applied to
2B, assuming that the magnitude of change would be
the same in these two proximal transects. Although we
would have preferred to date core 2B directly, we
were unable to find a suitable
14
C sample (terrestrial
macrofossil) in any of the transect 2 cores.
Sediment accumulation
Sediment accumulation rates for each core were
defined as the accumulated dry mass between dated
sediment intervals divided by the elapsed time. All
cores show increased sediment accumulation after
1850, coincident with the magnetic susceptibility rise
and European settlement. Most cores have a peak
sediment accumulation in the 1950s and 1960s
(Fig. 5), with maximum values ranging from 2.61 kg
m
-2
year
-1
(5B) to 36 kg m
-2
year
-1
(3B). The
sediment accumulation peak in core 2B also extends
into the 1940s, but this may be due to the dating
uncertainty for that core. After the 1960s, sediment
accumulation rates in all cores declined. Present-day
sediment accumulation rates are still well above pre-
European settlement rates, ranging from a 2.6-fold
increase in transect 8 to an 11-fold increase in
transect 4.
There is spatial, as well as temporal, variability in
sediment accumulation. Rates in the upstream end of
the lake should be higher than those further down-
stream if the mainstem of the river is the dominant
source of sediment. Where the river first widens and
deepens into a lake, sediment will quickly settle out
of the decelerating water leaving less to be deposited
downstream. Pre-European settlement data support
this depositional model: before 1850, transect 8 had a
sediment accumulation of 2.3 kg m
-2
year
-1
and
all other transects had sediment accumulations
between 0.2 and 0.8 kg m
-2
year
-1
. However,
around 1940 transects 4, 3 and 2 suddenly began
recording large inputs of sediment with maxima
from 11 kg m
-2
year
-1
(4B) to 36 kg m
-2
year
-1
(3B). Because upstream cores do not reflect such a
dramatic increase in sediment load, it appears that
this sediment was being contributed by the side-
valley tributaries in the lower half of the lake, namely
Valley Creek, Trout Brook and the Kinnickin-
nic River. In addition, dredging of the Kinnickinnic
River delta in Lake St. Croix likely mobilized
trapped sediment and may have contributed to the
high sediment accumulation rates downstream of
that tributary (core 2B). Magnetic susceptibility and
Table 1 Carbon-14 ages and calibrated dates
Core-sample
depth (cm)
14
C date
(BP) ±1 s.d.
Calibrated dates (AD)
max. (intercepts) minimum
1B-124 528 ±49 1329 (1413) 1438
3A-238 670 ±45 1284 (1297) 1387
5B-142 602 ±51 1300 (1330, 1350, 1390)
1400
8C-230 167 ±61 1660 (1680, 1760, 1800,
1940, 1950) 1950
J Paleolimnol
123
preliminary geochemical analyses confirm that
the sediment pulses in these downstream transects
have a different mineralogy, and thus likely have a dif-
ferent provenance, than contemporaneous upstream
sediments.
While sediment accumulation rates vary among
transects, cores within each transect have very similar
magnetic susceptibility profiles (Fig. 2). This similar-
ity demonstrates comparable sediment accumulation
between cores in a transect, requiring that only one of
the three cores be dated in detail. There is an
exception in transect 8 where the dated core (8C)
has magnetic susceptibility trends similar to 8A and
8B, but the profile is compressed into a shorter core
length. The uneven sediment distribution across the
river at this most upstream transect is likely due to
asymmetric growth of the river’s delta at the head of
the lake. For this transect, separate sediment accumu-
lation rates were estimated for 8A and 8B as
proportions of 8C based on the magnetic susceptibility
profiles.
Sediment phosphorus
Prior to European settlement, the total phosphorus
concentrations in the sediment (total-P
sed
) were fairly
024 681012
Sediment accumulation (kg m-2 yr -1)
Year
1800
1850
1900
1950
2000
1800
1850
1900
1950
2000
1800
1850
1900
1950
2000
1800
1850
1900
1950
2000
024 681012
02 4 6 8 1012 02 4 6 8 10 12
1800
1850
1900
1950
2000
1800
1850
1900
1950
2000
1800
1850
1900
1950
2000
1800
1850
1900
1950
2000
1B
4B
3B
2B
5B
8C
7B
6B
0 8 16 24 32 40 48
04 812162024
Fig. 5 Sediment
accumulation rates in the
eight primary cores. Core
8C is at the upstream
end of the lake and core
1B is at the downstream
end. Error bars
represent ±1 s.d.
propagated from counting
statistics
J Paleolimnol
123
uniform at all core sites (1.5 ±0.5 mg g
-1
), with
NaOH–P generally comprising half or more of the
total P concentration and Organic-P being the least
abundant fraction (Fig. 6). The concentrations of
total-P
sed
began to gradually increase between 1800
and 1950, and between 1900 and 1950 the cores (with
the exception of 2B) record a slight increase in the
relative abundance of NaOH–P and a decrease in the
other two fractions. Beginning c. 1940 the total-P
sed
concentrations decreased in all cores except 8C and
1B due to massive influxes of sediment at that time.
However, that decrease was temporary, and the
total-P
sed
increased sharply from 1950 to the present,
again with the exception of core 2B, which changed little
because of dilution from increased sediment loading.
Peak total P
sed
concentrations in the 1990s ranged
from 1.35 mg g
-1
in core 2B to 3.73 mg g
-1
in 5B.
Whole-lake fluxes
To understand the cumulative effects of human
activity on the sediment and P loads to Lake St.
Croix, the information from all cores must be
integrated. Eighteen to twenty samples from each of
Fig. 6 Phosphorus
concentrations in the eight
primary cores. NaOH–P is
iron- and aluminum-bound
phosphorus, HCl–P is
carbonate-bound
phosphorus, and Organic-P
was calculated by
difference from a total-P
extraction
J Paleolimnol
123
the eight primary cores were analyzed for total-P
sed
.
In addition, two sediment samples from each of the
sixteen secondary cores were analyzed for total-P
sed
:
the ‘‘top’’, or most recent 4 cm of sediment, and the
‘‘bottom’’, or the 4 cm immediately preceding the
increase in magnetic susceptibility marking European
settlement. The ‘‘top’’ and ‘‘bottom’’ values from the
secondary cores were weighted proportionally and
averaged into the flux calculations to account for
within-transect variation and thus better constrain the
fluxes during those two important time intervals. The
flux of each sediment fraction to a core site was
calculated as the product of the sediment accumula-
tion rate and the concentration of that fraction in
different strata. Subsequently, fluxes for the whole
lake were determined by weighting the flux of each
transect of cores by the portion of the depositional
basin represented by that transect.
The resolution of the dating models limits the
resolution at which these stratigraphic changes can be
examined. Cores 2B and 3B in particular have coarse
chronological resolution because of the failure of the
210
Pb dating method for these two profiles. Thus,
instead of reporting annual changes in lake-wide
fluxes, data are integrated as decadal averages for the
period 1930–2000, and as 20-year averages from 1850
to 1930 where dating uncertainty is greater (Oldfield
et al. 1999). Fluxes before 1850 reflect an average of
the 100–700 years prior to 1850 when the respective
14
C samples were deposited. All graphs are truncated
at 1800 to simplify visual interpretation of the data,
but the 1800–1850 values actually extend back to the
respective
14
C sample depth. This strategy harmo-
nizes the chronologies among cores with different
dating resolutions, yet provides sufficient resolution
of major trends in the data over the past 150 years.
Whole-lake sediment accumulation began to
increase shortly after European settlement from
15,000 t year
-1
to a peak in the 1950s of 130,000 t
year
-1
, eight times the pre-settlement rate (Fig. 7a).
Lake-wide sediment accumulation then declined
rapidly to a current level of 60,000 t year
-1
, around
four times the pre-settlement rate. The sharp peak in
the 1950s is driven in large part by sediment
deposited in transects 4, 3 and 2.
Total-P
sed
flux follows closely the temporal and
spatial patterns in sediment accumulation from pre-
1850 to the 1950s. Total-P
sed
flux rose from
23 t year
-1
before 1850 to a peak of 181 t year
-1
in the 1950s, an eight-fold increase. After 1950,
however, total-P
sed
did not decrease as much as the
sediment accumulation, but instead stayed near six
times the pre-1850 level at 133 t year
-1
(Fig. 7b).
That is, post-1950 sediment had more P per sediment
mass, and/or the proportion of P-rich organic matter
increased in the sediment. This abrupt switch from a
close correlation between P and sediment delivery to
a distinct offset in that relationship strongly suggests
that there was a new P source in the watershed after
1950.
Lakewater P reconstruction
Diatom microfossil analysis was performed on cores
6B and 1B, but only the downstream core (1B) is
described here. Diatom microfossil results from
both cores are presented in detail by Edlund et al. (this
issue-a). Three hundred and fifty-one diatom taxa were
Fig. 7 Sediment
accumulation rate for the
whole lake as determined
by multiplying the
accumulation rates of
eight dated primary cores
by the respective
depositional areas (a).
Phosphorus accumulation
rates for the whole lake,
determined by multiplying
each component of
sediment phosphorus by the
whole-lake sediment
accumulation rate for each
time period (b)
J Paleolimnol
123
identified in Lake St. Croix sediments and these can be
divided into benthic and planktic groups. During the
twentieth century, the relative ecological importance
of the groups in Lake St. Croix dramatically shifted.
Before 1920, benthic species were around 75% of the
diatom flora. By 1950, planktic species had become
more dominant with a relative abundance of 78% of the
diatom flora, and have remained the dominant species
group to the present day.
Thirty-three of the diatom taxa in core 1B,
representing a minimum of 55% and a maximum of
85% of the diatoms counted in each sample, were
used with the Ramstack et al. (2003) training set to
reconstruct historical concentrations of P in the water
column. Pre-settlement total-P
lake
as inferred from
diatoms varied from 20 to 25 lgl
-1
, and did not
begin to rise until about 1920. The most dramatic
increase began about 1950, and levels remained near
the maximum value 60 lgl
-1
, with some variation,
through the 1980s and 1990s.
The most recent diatom-inferred TP values closely
match TP values from direct monitoring of the lake
outlet (within 5% during the 1990s, within 20%
during the 1980s). However, as described in com-
panion papers by Lafrancois et al. (this issue) and
Edlund et al. (this issue-a) the directly measured TP
values during the 1970s were much higher than the
corresponding diatom-inferred values. As a result,
measured values indicate a decline in TP concentra-
tions from the 1970s to 2000, while the diatom-
inferred values do not. Edlund et al. (this issue-a)
attribute the discrepancy to infrequent and discontin-
uous monitoring during the 1970s, particularly before
1976, and/or potential underestimation of diatom-
inferred TP due to limitations of the diatom training
set. In summary, we used diatom-inferred TP for all
of our analyses in order to have a consistent dataset
extending back before the onset of monitoring.
Phosphorus mass balance
Phosphorus flux out of Lake St. Croix to the
Mississippi River was calculated as the product of
the St. Croix’s hydrologic discharge and the recon-
structed total-P
lake
concentration (core 1B) at the
downstream end of the lake (Fig. 8a). The first
consistent flow data for the St. Croix River were
recorded at St. Croix Falls, Wisconsin beginning in
1892. The mean of all flow data from 1892 to 2001
(4 910
9
m
3
year
-1
) was used for pre-1892 flux
calculations. Three large tributaries (the Apple,
Willow and Kinnickinnic Rivers) enter the river and
lake downstream of St. Croix Falls, increasing the flow
through Lake St. Croix by about 11%. To account for
that increase, measured and estimated flows for the
tributaries were added to the St. Croix Falls data.
Seventy-five years of flow data exist for the Apple
River, but flows for the other two tributaries (and for
unmonitored years on the Apple River) were estimated
using representative ratios of tributary flow to main-
stem flow, or tributary to Apple River flow. The Apple
River has a good correlation with mainstem flows
(R
2
=0.70), and the Willow and Kinnickinnic River
flows were estimated to be 0.30 and 0.36 of the Apple
River flow based on 1–3 years of flow data.
As discussed above, reconstructed total-P
lake
con-
centrations increased dramatically after 1950. When
multiplied by flow to become mass flux, P export
from the St. Croix River to the Mississippi more
than doubled from 127 t year
-1
before 1850 to
285 t year
-1
in the 1990s. At all times, the majority
of the P load (between 52 and 78%) was not trapped
in lake sediments, but was instead discharged to the
Mississippi River. During 1976–2000 when both in-
lake and inflow TP were directly measured, as
reported by Lafrancois et al. (this issue), Lake
St. Croix generally fit the Vollenweider model for
P retention in reservoirs as a function of water
residence time (Vollenweider and Kerekes 1982).
Residence time in Lake St. Croix is so short, on the
order of 20–60 days, that annual or decadal variation
in flow (lake residence time) was not enough to
explain the observed changes in P retention.
The major phosphorus losses from Lake St. Croix
are thus quantified as (1) P burial in the sediments
(whole lake total-P
sed
flux), and (2) P outflow to the
Mississippi River (reconstructed total-P
lake
from core
1B). Because there are no significant losses of P
through other pathways (e.g. volatilization), the sum of
the sedimented and discharged P equals the historical P
loading to the lake system (Fig. 8b). That is,
ðP loading to the lakeÞ¼ðP trapped in sedimentÞ
þðP in outflowÞ
(Rippey and Anderson 1996; Vollenweider 1975)
Total P loads to the lake averaged 166 t year
-1
before European settlement. Phosphorus input increased
J Paleolimnol
123
slightly during 1890–1910 to 210 t year
-1
, then
increased sharply from the 1940s onward. During the
1990s the total P load to the lake was 459 t year
-1
,a
three-fold increase from the presettlement load and the
highest P load the lake has received during the period of
study (*200 years). This upward trend was driven by
increasing total-P
lake
concentrations which, in turn,
were driven by the proliferation of anthropogenic P
sources in the watershed from 1940 to 2000 (Edlund
et al. this issue-b).
In some systems, anoxic bottom water releases
significant amounts of P from the sediments to the
overlying water. While periods of anoxia have been
documented in parts of the Lake St. Croix hypolim-
nion, an analysis reported by Robertson and Lenz
(2002) suggests that the amount of P released from
Lake St. Croix sediments would be small, on the
order of 6% of the total P load to the lake. The
relatively high sediment accumulation in the lake
likely insures that most P delivered to the lake bottom
is buried before it can be remobilized. Regardless of
magnitude, P recycling does not affect the P mass
balance because all P inputs are eventually buried in
the sediment or discharged from the lake.
Synthesis
In the 170 years since European settlement of the
St. Croix watershed, significant changes have occurred
in sediment accumulation, phosphorus loading and
algal production and ecology in Lake St. Croix. With
the multiple-core, whole-basin record developed in this
study, these changes can be quantified and observed
lake effects linked to specific anthropogenic causes.
Sediment
The sediment accumulation in Lake St. Croix
increased gradually during the first century after
European settlement, then rose sharply after 1930 to
a peak of 129,000 t year
-1
in the 1950s. Today the
Water TP Load Calculation
Flow (10
9
m
3
yr
-1
) Water TP (µg l
-1
) Outflow TP (t yr
-1
)
1800
1850
1900
1950
2000
0246
Year
1800
1850
1900
1950
2000
0 25 50 75 100
1800
1850
1900
1950
2000
0 100 200 300 400
X=
P Mass Balance
1850
1900
1950
2000
1800
0
100
200
300
400
500
1800
1850
1900
1950
2000
0
100
200
300
400
500
1800
1850
1900
1950
2000
0
100
200
300
400
500
+=
Outflow TP (t yr
-1
) Sediment TP (t yr
-1
) Total TP load (t yr
-1
)
Year
a
b
Fig. 8 Reconstructed total
phosphorus (TP)
concentration in the water
column multiplied by
outflow volume equals the
TP load leaving Lake St.
Croix (a). The sum of TP
load leaving the lake via
outflow and TP load
sequestered in the lake
sediments equals the total
TP load to the lake (b)
J Paleolimnol
123
lake accumulates around 60,000 t year
-1
of sedi-
ment, about four times the pre-settlement rate. The
sediment accumulation increase is presumably due to
increased topsoil or stream-bank erosion from the
watershed. Importantly, the distribution of the sedi-
ment load among Lake St. Croix’s four sub-basins
varies through time. In its ‘‘natural’’ pre-European
state, the lake’s most upstream sub-basin (D)
received the largest proportion of the total sediment
load (51–56%), as expected where the river velocity
decreases suddenly upon entering the lake. This
pattern shifted dramatically between 1940 and 1960
when the two downstream sub-basins (A and B)
became the primary depositional centers and together
received 56–61% of the total sediment load. These
downstream spikes in sediment accumulation almost
certainly originated in the small side-valley tributar-
ies, specifically Valley Branch Creek, Trout Brook
and the Kinnickinnic River. Despite the relatively
small size of these watersheds (120, 45 and 520 km
2
,
respectively) they were able to deliver large amounts
of sediment to the lake due to their steep gradients,
proximity to the depositional basin and possibly
because of locally-high erodibility of soil and glacial
outwash (Dearing and Foster 1993; Mulla and Sekely
this issue). Apparent accumulation changes down-
stream of the Kinnickinnic River may also have been
affected by dredging of that river’s delta.
In summary, Lake St. Croix sediment accumulation
responded only modestly to initial logging of northern
forests and conversion of southern prairie and wood-
lands to agricultural uses during 1840–1880. Sediment
eroded from land distant from the lake may have been
temporarily stored in the river channel (including
sloughs, sandbars, etc.) or behind impoundments such
as the multitude of wing dams constructed along the
river from 1850 to 1890 and the hydropower dam at
St. Croix Falls built in 1904. For this reason, the
amount of sediment accumulated in the lake does not
directly equal that eroded from upstream landscapes
after the original land conversion. It is clear, however,
that the lake was strongly affected by land-use changes
in the lower watershed during the mid-1900s, possibly
including increased road-building, urbanization and
the mechanization of agriculture. After 1970 the
sediment accumulation declined, suggesting that
agricultural soils and streambanks were stabilized
and/or the local, episodic disturbances of the 1950s
and 1960s have not occurred since.
Phosphorus
In calculating the total P load to the lake, it is
apparent that total-P
lake
is at all times a larger
component than total-P
sed
. The total P load to the lake
increased after 1940 with a maximum during 1980–
2000 of three times the average pre-settlement load.
The increase began after land conversion was com-
plete and was not due simply to eroded soils entering
the lake.
Today there are over 50 permitted point source
discharges in the watershed, a few of which are large
municipal wastewater treatment plants (WWTPs).
However, even during a dry year (when non-point
inputs are normally low) the P input from point sources
represents less than 30% of the total P load to the lake
(Robertson and Lenz 2002). In addition, P discharges
from WWTPs in the nearby Upper Mississippi
watershed peaked in the 1960s and 1970s (*1,000 t
year
-1
), then decreased significantly (*800 t
year
-1
in 1992) after the ban on phosphate detergents
(Mulla and Sekely this issue). The ban applied equally
to WWTPs in the St. Croix watershed, so it can be
assumed that a similar trend occurred in these dis-
charges (Edlund et al. this issue-b).
A more significant cause of the increased P load is
likely the importation of phosphorus into the
watershed via inorganic fertilizers and livestock
feed supplements, leading to accumulation of P in
watershed soils (Mulla and Sekely this issue). Soil-P
has been increasing on a global scale (Bennett et al.
1999,2001), and specific studies have documented
deleterious effects on water quality. For instance, Foy
et al. (1995) reported an increase in soluble reactive
phosphorus (SRP) loading to an Irish lake despite
reductions in point-source P discharges and a con-
stant use of phosphate fertilizer in that watershed.
The continued rise of SRP resulted from the buildup
of P in the watershed and a decreasing ability of the
soils to bind the additional P. This phenomenon has
been observed in other locations as well (Baker and
Richards 2002; Jordan and Rippey 2003; Jordan et al.
2001; Mulla and Sekely this issue; Richards et al.
2002). Furthermore, the inverse trend has been
observed, that decreased application of agricultural
fertilizers can lead to a decline in TP in a catchment’s
streams (Moog and Whiting 2002). There is evidence
that soils in the St. Croix watershed are already high
in P (Laboski and Lamb 2003), and phosphate
J Paleolimnol
123
fertilizer applications have remained steady since the
1970s (Mulla and Sekely this issue), so agricultural
phosphorus (fertilizers and feed supplements) is
likely a major source of P to Lake St. Croix.
Uncertainties
Three components of this study have the most
potential for introducing uncertainty to the calcula-
tions. First, the lack of flow data before 1892 means
that earlier outflow-P calculations are estimated
based on an average of twentieth century flows.
Sediment transport is also correlated to flow. The
flow in the St. Croix has certainly changed less since
European settlement than in other upper Mississippi
tributaries such as the Minnesota River, where
extensive drainage of wetlands and installation of
agricultural drain-tile has significantly increased the
rate of water runoff to the river. However, variations
in flow due to climate are still not accounted for by
using the average twentieth century flow, so this
value is only a best estimate of nineteenth century
discharge.
Second, there is some uncertainty in dating the
sediment, due to both the dating models themselves
and to the fluctuating sediment provenance and
accumulation rates in this lake. High sediment
accumulation rates and variable supported
210
Pb
activities in transects 2, 3 and 4 cause unacceptably
large errors in the c.r.s.
210
Pb model. This problem
was overcome in transect 4 by directly measuring
supported and unsupported
210
Pb with gamma spec-
trometry. In transects 2 and 3, dates were obtained by
correlating magnetic susceptibility features from
cores 2B and 3B to dated cores in adjacent transects.
With only 4–5 known (dated) points for cores 2B
and 3B, there is more uncertainty in the timing of
specific events, such as the first increase in total-P
sed
.
Coarser temporal resolution may also mask short,
intense fluxes of sediment. The most significant
consequence for this study is that the timing and
duration of the large sediment fluxes to sub-basins A
and B cannot be constrained to better than a two- or
three-decade interval. Nevertheless, dating detail is
sufficient to distinguish very significant changes in
sediment accumulation, and to integrate these two
cores into the whole-lake flux calculations.
Third, the total-P
lake
reconstruction using a diatom
inference model entails some uncertainty. The diatom
inference model is constructed from a lacustrine
dataset, whereas it is applied here to the riverine
system of Lake St. Croix. Furthermore, while the
model itself is statistically robust, hypereutrophic
lakes are under-represented in the calibration lake set,
so diatom-inferred TP reconstructions greater than
85 lgl
-1
are not possible. This may explain some of
the discordance between diatom TP and monitoring
records of total-P
lake
from the 1960s and early 1970s
when measured total-P
lake
values in Lake St. Croix
were higher than the reconstructed TP values. Thus,
the historical TP outflow from Lake St. Croix inferred
from the diatoms and historical flows during these
time periods are likely conservative.
Conclusions
The unique geomorphic history of the Upper Missis-
sippi River which created Lake St. Croix at the
terminus of the St. Croix River allows us to success-
fully apply paleolimnological techniques within a
whole-basin context to quantitatively reconstruct pre-
historical water quality in a large riverine system. The
lake sediments describe changes occurring throughout
the St. Croix watershed and constrain the list of
historic events that had significant influence on water
quality. The chronologies developed here make it
clear that the earliest European settlement activities
caused only modest increases in sediment loading to
the river and did not affect nutrient levels appreciably.
Major changes occurred in the mid-1900s, and Lake
St. Croix is now far from its natural state in terms of
sediment and phosphorus loads and algal ecology.
This study provides historical context for citizens and
policy-makers as they decide what level of lake/
watershed management and restoration is desirable
and achievable (Davis 2004).
Acknowledgements Many colleagues were involved in the
successful completion of this project. Shawn Schottler
(St. Croix Watershed Research Station) performed the
137
Cs
analyses as well as the
210
Pb by gamma spectrometry. Scott
Schellhaass (Metropolitan Council Environmental Services,
MCES) ably navigated the lake for us during all of the core
collection work, and Angella Craft-Reardon (MCES) performed
the sediment phosphorus extractions. Brenda Moraska
Lafrancois (National Park Service) and Greg Johnson
(Minnesota Pollution Control Agency) provided water quality
monitoring data and statistical analysis thereof. In addition, we
are grateful to Kelly Thommes, Tara Bromenshenkel, Jill
J Paleolimnol
123
Coleman, Erin Mortenson, and Jerry Mliner (SCWRS) who
helped with much of the loss-on-ignition analysis.
References
Anderson NJ, Rippey B (1994) Monitoring lake recovery from
point source eutrophication: the use of diatom-inferred
epilimnetic total phosphorus and sediment chemistry.
Freshw Biol 32:625–639. doi:10.1111/j.1365-2427.1994.
tb01153.x
Anderson NJ, Korsman T, Renberg I (1994) Spatial heteroge-
neity of diatom stratigraphy in varved and non-varved
sediments of a small, boreal-forest lake. Aquat Sci 56:40–
58. doi:10.1007/BF00877434
Anderson O, Crow TR, Lietz SM, Stearns F (1996) Transfor-
mation of a landscape in the upper mid-west, USA: the
history of the lower St. Croix river valley, 1830 to present.
Landscape Urban Plan 35:247–267. doi:10.1016/S0169-
2046(96)00304-0
Appleby PG, Oldfield F (1978) The calculation of lead-210
dates assuming a constant rate of supply of unsupported
210
Pb to the sediment. Catena 5:1–8. doi:10.1016/S0341-
8162(78)80002-2
Baker DB, Richards RP (2002) Phosphorus budgets and riv-
erine phosphorus export in northwestern Ohio watersheds.
J Environ Qual 31:96–108
Balogh SJ, Engstrom DR, Almendinger JE, Meyer ML, Johnson
DK (1999) History of mercury loading in the Upper Mis-
sissippi River reconstructed from the sediments of Lake
Pepin. Environ Sci Technol 33:3297–3302. doi:10.1021/
es9903328
Bennett EM, Reed-Andersen T, Houser JN, Gabriel JR, Car-
penter SR (1999) A phosphorus budget for the Lake
Mendota watershed. Ecosystems (NY, Print) 2:69–75. doi:
10.1007/s100219900059
Bennett EM, Carpenter SR, Caraco NF (2001) Human impact
on erodable phosphorus and eutrophication: a global
perspective. Bioscience 51:227–234. doi:10.1641/0006-
3568(2001)051[0227:HIOEPA]2.0.CO;2
Bennion H, Juggins S, Anderson NJ (1996) Predicting epi-
limnetic phosphorus concentrations using an improved
diatom-based transfer function and its application to lake
eutrophication management. Environ Sci Technol 30:
2004–2007. doi:10.1021/es9508030
Bernhardt H (1995) Reservoirs and drinking-water supply—a
global perspective. Aqua 44:2–17
Blumentritt DJ, Wright HE Jr, Stefanova V (this issue) For-
mation and early history of lakes Pepin and St. Croix of
the Upper Mississippi River. J Paleolimnol. doi:10.1007/
s10933-008-9291-6
Curtis JT (1959) The vegetation of Wisconsin: an ordination of
plant communities. The University of Wisconsin Press,
Madison, WI, 655 pp
Davis PJ (2004) St. Croix Basin phosphorus-based water
quality goals. Minnesota Pollution Control Agency,
St. Paul, MN, 28 pp
Dean WE (1974) Determination of carbonate and organic
matter in calcareous sediments and sedimentary rocks by
loss on ignition: comparison with other methods. J Sedi-
ment Petrol 44:242–248
Dearing JA, Foster IDL (1993) Lake sediments and geomor-
phological processes: some thoughts. In: McManus J,
Duck RW (eds) Geomorphology and sedimentology of
lakes and reservoirs. Wiley, Chichester, pp 5–14
Dunn JT (1979) The St Croix: Midwest Border River. Min-
nesota Historical Society Press, St. Paul, 310 pp
Eakins JD, Morrison RT (1978) A new procedure for the
determination of lead-210 in lake and marine sediments.
Int J Appl Radiat Isot 29:531–536. doi:10.1016/0020-
708X(78)90161-8
Edlund MB, Engstrom DR, Triplett LD, Lafrancois BM,
Leavitt PR (this issue-a) Twentieth century eutrophication
of the St. Croix River (Minnesota–Wisconsin, USA)
reconstructed from the sediments of its natural impound-
ment. J Paleolimnol. doi:10.1007/s10933-008-9296-1
Edlund MB, Triplett LD, Tomasek MD, Bartilson K (this issue-
b) From paleo to policy: partitioning the historical point and
nonpoint phosphorus loads to the St. Croix River, Minne-
sota-Wisconsin, USA. J Paleolimnol. doi:10.1007/s10933-
008-9288-1
Engstrom DR, Wright HE Jr (1984) Chemical stratigraphy of
lake sediments as a record of environmental change. In:
Haworth EY, Lund JWG (eds) Lake sediments and
environmental history. University of Minnesota Press,
Minneapolis, pp 11–67
Engstrom DR, Almendinger JE, Wolin JA (this issue) Historical
changes in sediment and phosphorus loading to the upper
Mississippi River: mass-balance reconstructions from the
sediments of Lake Pepin. J Paleolimnol. doi:10.1007/
s10933-008-9292-5
Evans DR, Rigler FH (1980) Measurement of whole lake
sediment accumulation and phosphorus retention using
lead-210 dating. Can J Fish Aquat Sci 37:817–822
Eyster-Smith NM, Wright HE Jr, Cushing EJ (1991) Pollen
studies at Lake St. Croix, a river lake on the Minnesota/
Wisconsin border, USA. Holocene 1:102–111. doi:10.1177/
095968369100100202
Foy RH, Smith RV, Jordan C, Lennox SD (1995) Upward trend
in soluble phosphorus loadings to Lough Neagh despite
phosphorus reduction at sewage treatment works. Water
Res 29:1051–1063. doi:10.1016/0043-1354(94)00259-A
Hall RI, Smol JP (1992) A weighted-averaging regression and
calibration model for inferring total phosphorus concen-
tration from diatoms in British Columbia (Canada) lakes.
Freshw Biol 27:417–434. doi:10.1111/j.1365-2427.1992.
tb00551.x
Hieltjes AHM, Lijklema L (1980) Fractionation of inorganic
phosphates in calcareous sediments. J Environ Qual 9:
405–407
Jordan P, Rippey B (2003) Lake sedimentary evidence of
phosphorus, iron and manganese mobilisation from
intensively fertilised soils. Water Res 37:1426–1432. doi:
10.1016/S0043-1354(02)00488-8
Jordan P, Rippey B, Anderson NJ (2001) Modeling diffuse
phosphorus loads from land to freshwater using the sedi-
mentary record. Environ Sci Technol 35:815–819. doi:
10.1021/es0001308
Jordan P, Rippey B, Anderson NJ (2002) The 20th century
whole-basin trophic history of an inter-drumlin lake in an
J Paleolimnol
123
agricultural catchment. Sci Total Environ 297:161–173.
doi:10.1016/S0048-9697(02)00135-3
Juggins S (1998) CALIBRATE ver. 0.82. University of New-
castle, Newcastle upon Tyne, UK
Kroenig S, Andrews WJ (1997) Water-quality assessment of part
of the upper Mississippi River Basin, Minnesota and Wis-
consin—nitrogen and phosphorus in stream, streambed
sediment and ground water, 1971–1994. U.S. Geological
Survey, Mounds View, MN
Laboski CAM, Lamb JA (2003) Changes in soil test phos-
phorus concentration after application of manure or
fertilizer. Soil Sci Soc Am J 67:544–554
Lafrancois BM, Magdalene S, Johnson DK (this issue) A
comparison of recent water quality trends (1976–2004)
with sediment-core records for two riverine lakes of the
upper Mississippi River basin: Lake St. Croix and Lake
Pepin. J Paleolimnol. doi:10.1007/s10933-008-9294-3
Lenz BN, Robertson DM, Fallon JD, Ferrin R (2003) Nutrient
and suspended-sediment concentrations and loads, and
benthic-invertebrate data for tributaries to the St. Croix
River, Wisconsin and Minnesota 1997–1999. U.S.
Geological Survey, Middleton, WI
Maurer WR, Claflin TO, Rada RG, Rogala JT (1995) Volume
loss and mass balance for selected physicochemical con-
stitutents in Lake Pepin, upper Mississippi River, USA.
Reg Rivers 11:175–184. doi:10.1002/rrr.3450110207
McMahon EM, Karamanski TJ (2002) Time and the river: a
history of the St. Croix. National Park Service, Omaha,
NE
Meybeck M, Helmer R (1989) The quality of rivers: from
pristine stage to global pollution. Palaeogeogr Palaeocli-
matol Palaeoecol 75:283–309. doi:10.1016/0031-0182
(89)90191-0
Meybeck M, Chapman DV, Helmer R (1990) Global fresh-
water quality: a first assessment. Blackwell Reference,
Cambridge, MA, 306 pp
Moog DB, Whiting PJ (2002) Climatic and agricultural con-
tributions to changing loads in two watersheds in Ohio.
J Environ Qual 31:83–89
Moss B (1980) Further studies on the palaeolimnology and
changes in the phosphorus budget of Barton Broad, Nor-
folk. Freshw Biol 10:261–279. doi:10.1111/j.1365-2427.
1980.tb01201.x
Mulla DJ, Sekely AC (this issue) Historical trends affecting the
accumulation of sediment and phosphorus in Lake Pepin,
upper Mississippi River, USA. J Paleolimnol. doi:10.1007/
s10933-9293-4
Oldfield F, Appleby PG, van der Post KD (1999) Problems of
core correlation, sediment source ascription and yield
estimation in Ponsonby Tarn, West Cumbria, UK. Earth
Surf Process Landf 24:975–992. doi:10.1002/(SICI)1096-
9837(199910)24:11\975::AID-ESP23[3.0.CO;2-W
Ramstack JM, Fritz SC, Engstrom DR, Heiskary SA (2003)
The application of a diatom-based transfer function to
evaluate regional water-quality trends in Minnesota since
1970. J Paleolimnol 29:79–94. doi:10.1023/A:1022869
205291
Richards RP, Baker DB, Eckert DJ (2002) Trends in agricul-
ture in the LEASEQ watersheds, 1975–1995. J Environ
Qual 31:17–24
Rippey B, Anderson NJ (1996) Reconstruction of lake phos-
phorus loading and dynamics using the sedimentary
record. Environ Sci Technol 30:1786–1788. doi:10.1021/
es950670i
Robertson DM, Lenz BN (2002) Response of the St. Croix River
pools, Wisconsin and Minnesota to various phosphorus-
loading scenarios. U.S. Geological Survey, Middleton, WI
Saeijs HLF, Van Berkel MJ (1995) Global water crisis: the
major issue of the 21st century, a growing and explosive
problem. Eur Water Pollut Control 5:26–40
Schelske CL, Stoermer EF, Kenney WF (2006) Historic low-
level phosphorus enrichment in the Great Lakes inferred
from biogenic silica accumulation in sediments. Limnol
Oceanogr 51:728–748
Smith VH (2003) Eutrophication of freshwater and coastal
marine ecosystems: a global problem. Environ Sci Pollut
Res 10:126–139. doi:10.1065/espr2002.12.142
Stuiver M, Reimer PJ (1993) Extended (super 14) C data base
and revised CALIB 3.0 (super 14) C age calibration
program Calibration 1993. Radiocarbon 35:215–230
Stuiver M, Reimer PJ, Bard E, Beck JW, Burr GS, Hughen KA,
Kromer B, McCormac FG, van der Plicht J, Spurk M
(1998) INTCAL98 radiocarbon age calibration 24,000-
0 cal BP. Radiocarbon 40:1041–1083
Troelstrup NHJ, Foley JL, Perry JA (1993) Changing patterns
of water quality and biology within the lower St. Croix
national scenic riverway. Legislative Commission on
Minnesota Resources, St Paul, MN
Urban NR, Eisenreich SJ, Grigal DF, Schurr KT (1990)
Mobility and diagenesis of Pb and
210
Pb in peat. Geochim
Cosmochim Acta 54:3329–3346. doi:10.1016/0016-7037
(90)90288-V
Vollenweider RA (1975) Input-out models with special refer-
ence to the phosphorus loading concept in limnology.
Swiss J Hydrol 37:53–84. doi:10.1007/BF02505178
Vollenweider RA, Kerekes J (1982) Eutrophication of waters:
monitoring, assessment and control. Environment Direc-
torate, OECD, Paris, 154 pp
Wright HE Jr (1991) Coring tips. J Paleolimnol 6:37–49. doi:
10.1007/BF00201298
Wright HE Jr, Mann DH, Glaser PH (1983) Piston corers for lake
and wetland sediments. Ecology 65:657–659. doi:10.2307/
1941430
Zhang J, Chen HT, Zhang ZF, Liu SM, Wu Y (1999) Human
impacts on the large world rivers: would the Changjiang
(Yangtze River) be an illustration? Glob Biogeochem
Cycles 13:1099–1105. doi:10.1029/1999GB900044
J Paleolimnol
123
