Water Air Soil Pollut (2019) 230:128
https://doi.org/10.1007/s11270-019-4176-z
Source Identification of Particulate Metals/Metalloids
Deposited in the San Juan River Delta of Lake Powell, USA
Logan Frederick & William P. Johnson &
Thure Cerling & Diego Fernandez & James VanDerslice
Received: 28 September 2018 / Accepted: 2 May 2019
# Springer Nature Switzerland AG 2019
Abstract Whereas mining and non-mining sourced
particulate metal/metalloids (PM) (> 0.45 μm) are
present in the tributaries of the San Juan River,
USA, the individual contributions of PM from the
San Juan River tributaries to the sediment of the San
Juan River Delta of Lake Powell were previously
unknown. Suspended PM signatures, including
Highlights
•Contributions of suspended particulate metals and metalloids to
the San Juan River from mined and unmined tributaries were
determined using tributary geochemical signatures that reflected
legacy mining impacts or the predominant underlying geology.
•Suspended particulates in tributaries were linked to layers in San
Juan River Delta sediment cores through geochemical signatures,
such as depleted Pb isotopes and elevated Pb to Al, Cd to Al, and
Zn to Al ratios reflective of mining impact, as well as enriched Pb
isotopes, elevated Mn to Al, and red color reflective of unmined
sources
•Approximately 10% of PM deposited in a sediment core was
attributed to mining sources, whereas approximately 80% of PM
deposited in a sediment core was attributed to a mixture of sources.
•Seasonal accumulation in sediment traps indicated that the 3.37-
m core was deposited over approximately 1.3 years
Electronic supplementary material The online version of this
article (https://doi.org/10.1007/s11270-019-4176-z) contains
supplementary material, which is available to authorized users.
L. Frederick : W. P. Johnson (*) : T. Cerling :
D. Fernandez
Department of Geology and Geophysics, University of Utah, Salt
Lake City, UT 84112, USA
e-mail: william.johnson@utah.edu
J. VanDerslice
Department of Family and Preventative Medicine, School of
Medicine, University of Utah, Salt Lake City, UT 84112, USA
enrichment factors (ratios of PM concentrations to
ubiquitous metals such as aluminum), lead isotopes,
color, and particle size, were used to tie layers in a
San Juan River sediment core to upstream tributary
sources. Tributary PM concentrations and loads were
compared to Lower San Juan River suspended PM
loads to estimate the relative contribution of tributary
PM (both mining and non-mining sourced) directly
upstream of Lake Powell. Results suggest elevated
enrichment factors of lead, cadmium, copper, and
zinc in deposited sediment were sourced from mined
tributaries, whereas elevated manganese in deposited
sediment was sourced from unmined tributaries. Sed-
iment Pb isotope ratios reflected the depleted signa-
ture of the mineralized vein (present in the headwa-
ters of mined tributaries), the enriched signature of
the Chinle Sandstone (predominant underlying geol-
ogy of unmined tributaries), or a mixture of these two
endmembers. These independent lines of evidence
were used to link probable tributary source and run-
off category to sediment layers, where ~ 10% and 5%
of the overall PM deposited in the sediment core was
attributed to mining versus non-mining sources, re-
spectively. Because traditional dating methods were
not possible, runoff category signatures were used to
estimate that the 3.37 m San Juan River Delta core
was deposited over ~ 1.3 years.
Keywords Particulate metals/metalloid concentration .
Sediment cores . Sediment traps . Lead isotopes .
Principal component analysis . Lake Powell
 128 Page 2 of 18
1 Introduction
Reservoirs are responsible for providing water to more
than 31 million people each year across the Western
United States (US) (US BOR 2019). The quality of
reservoir water can be greatly affected by sediment
accumulation, which occurs primarily at alluvial deltas
formed at reservoir inflows (Mau and Christensen
2000). Finer sediment, characterized by large surface
areas with active sites for ion exchange or sorption of
metals, is generally deposited in distal lower water ve-
locity locations of deltas (Vernieu 1997). Mobilization
of metals may also occur when deltaic sediment is
exposed (i.e., during periods of prolonged drought, in-
creased water use, or changes in reservoir management).
In this way, reservoir deltas act as both sinks and sec-
ondary sources of deposited contaminants (Hornewer
2014). As droughts persist across the Western US and
population continues to grow, it becomes important to
quantify the amount and source of contaminants stored
in the sediments of large reservoirs.
Lake Powell is the second largest reservoir in the US,
impounding the Colorado, San Juan, and Escalante Riv-
ers at the border of Utah and Arizona (Fig. 1). These three
rivers annually supply an average of 60 million metric
tons of sediment to Lake Powell (Potter and Drake 1989),
with a majority of the sediment and water (~ 96%)
sourced from the Colorado and San Juan Rivers (Irons
et al. 1965; Evans and Paulson 1983). Increased water
withdrawals combined with increased drought have re-
sulted in decreased water level in Lake Powell (US BOR
2019) and an increase in exposed sediment. For example,
prolonged drought from 1988 to 1993 resulted in the
exposure of over 100 channel kilometer of previously
submerged deltaic sediment (Vernieu 1997).
Sediment cores collected in the San Juan River Delta
of Lake Powell in 2010 documented changes in particle-
bound metal/metalloid (PM) concentrations over time
with visually distinct layers associated with distinct
elemental concentrations (Hornewer 2014). The repeat-
ed distinct layers in the sediment core likely reflect
changes in source with changes in season/flow regime.
Whereas elevated metals concentrations are well docu-
mented in the San Juan River (Thompson and Mundorff
1982; Abell 1994; Bliesner and Lamarra 2000; Clow
and Stephens 2005), their attribution among natural
versus mining origin was previously unknown.
Hardrock mining accelerates exposure and
weathering of sulfide-bearing minerals thereby
Water Air Soil Pollut (2019) 230:128
producing acidic water and acid mine drainage
(AMD), with co-occurring heavy metals/metalloids
such as zinc (Zn), lead (Pb), cadmium (Cd), and arsenic
(As) (Church et al. 1997; Dologopolova et al. 2006).
Whereas metals/metalloids mobilized via AMD are
partitioned between dissolved, suspended particulate,
and bed sediment phases through processes such as
sorption, photoreduction, mineral precipitation, and bi-
ological interaction (Church et al. 1997), PM represent
the largest contribution to sediment for many elements
(Czuba et al. 2011).
In 2015, an estimated 500,000 kg of AMD was
inadvertently released from the Gold King Mine (EPA
2016; BOR 2015), located in the headwaters of the
Animas River, largest tributary of the San Juan River
(Fig. 1). Although this visually striking spill attained
national and international attention, the total metal mass
in the Gold King Mine plume was comparable to the
metal load carried in a single spring day of high runoff in
the Animas River (EPA 2016). However, the concentra-
tions of total metals (unfiltered) during peak plume
passage were up to four times greater than historic
spring runoff concentrations (Sullivan et al. 2017). Fol-
lowing the Gold King Mine Spill, the US Environmen-
tal Protection Agency (US EPA) began remediating
mining-sourced PM in the Bonita Peak Mining District,
CO, under Superfund (US Comprehensive Environ-
mental Response, Compensation, and Liability Act) to
improve water quality in the Animas River tributary
(EPA 2018).
Because remediation of mining and non-mining
sourced metals can be costly for large reservoirs like
Lake Powell, water managers need to understand poten-
tial sources of funds based on the balance of naturally
versus industrially derived material. The objective of
this paper is to establish geochemical signatures for
PM sourced from mined and unmined tributaries and
to use those signatures to quantify the overall contribu-
tions in sediment deposited in the San Juan River Delta
of Lake Powell. Because sedimentation rates in the San
Juan River Delta are high and fluctuate annually as well
as inter-annually (Stanford and Ward 1991; Potter and
Drake 1989; Vernieu 1997), traditional sediment dating
methods (e.g., cesium-137 or lead-210) are not reliable.
In the absence of these dating procedures, we herein use
geochemical signatures representing seasonal categories
of runoff, where the record is punctuated by annual
events such as snowmelt, as an alternative dating
method.
Water Air Soil Pollut (2019) 230:128
UT CO
AZ NM
Core 3
Core 2
Core 1 Lake
Up
Down Mid Powell
0 5 10 km
Sample Locaons Lower
San Juan
Water Samples
USGS gauges
Sediment cores
Sediment traps
Fig. 1 Map of the San Juan Watershed located in Four Corners,
USA. Tributary watersheds are outlined and labeled in black.
Locations for surface water samples (black squares), USGS stream
gauges (red triangles), USGS sediment cores (black circles), and
USGS sediment traps (black diamonds) are shown. USGS sedi-
ment cores were collected at USGS sites 0364200, 0382600, and
1.1 Source Identification of Metals/Metalloids
in Surface Water
Source identification of metals/metalloids from natural
and anthropogenic sources has been explored in surface
water systems utilizing parameters such as metal loads,
elemental ratios, and isotopic ratios (e.g., Axtmann and
Luoma 1991; Church et al. 1997; Coulthard and
Macklin 2003; Arditsoglou and Samara 2005; Borrok
et al. 2009; Kimball et al. 2009; Runkel et al. 2018).
Metal/metalloid loads in fluvial systems have been
analyzed via transport modeling (Paschke et al. 2005;
Runkel et al. 2012), statistical regression (Leib et al.
2003), and mass balance (Church et al. 1997; Kimball
et al. 2007) in order to determine tributary inputs to
watershed-wide loads. Elemental ratios have been used
to examine trace element enrichment relative to ambient
metals such as aluminum (Al) and iron (Fe), which are
abundant crustal elements and exist in relatively-
consistent proportions in rocks, soils, and sediments
(Covelli and Fontolan 1997; Dologopolova et al.
2006; Sullivan et al. 2017). Stable isotopes of Pb, Zn,
and iron (Fe) have been utilized in mineralized settings
for source identification and metal cycling, where Pb
and Zn isotopes do not fractionate during transport/
deposition (Church et al. 1997; Dologopolova et al.
2006; Borrok et al. 2009) and Fe isotopes fractionate
Page 3 of 18 128
Animas
McElmo
La Plata Navajo
Mancos Dam
Upper
Chinle San Juan
Chaco
0410700, for cores 1, 2, and 3, respectively. USGS sediment traps
were collected downstream of Mexican Hat, UT at upstream, mid-
stream, and downstream locations. A map with watershed mining
activity and geology is given in the Supporting Information (Fig.
SI1)
during oxidation-reduction processes (Borrok et al.
2009).
Source identification of metals/metalloids in the San
Juan River has been limited to the headwaters of the
Animas (Sullivan et al. 2017; Church et al. 1997;
Kimball et al. 2002; Leib et al. 2003; Borrok et al.
2009; Paschke et al. 2005) despite observed
exceedances of aquatic and human health standards in
many tributaries and the main channel of the San Juan
River (Abell 1994; Church et al. 1997; Clow and Ste-
phens 2005; Larrick 2010; Larrick and Ashmore 2012;
Dyer et al. 2016; Rodriguez-Ferire et al. 2016). Mass
accounting via ordinary least-square regression (Leib
et al. 2003) and mixing calculations (Church et al.
1997) indicate that the composition and source of metal
loads mobilized in the upper Animas River change
substantially with season, where PM loads mobilized
during snowmelt runoff generally exceed dissolved
loads, and the source of PM loads differs between
Mineral Creek, Cement Creek, and the Animas River
(headwater tributaries of the Animas River) (Leib et al.
2003). Fe, Zn, and Pb isotopes were used to differentiate
metal loading sources and attenuation mechanisms in
the upper Animas Watershed, where AMD contributed
enriched 56 Fe/ 54 Fe, depleted 66 Zn/ 64 Zn, and a
208
Pb/204Pb ratio representative of the mined mineral-
ized Eureka graben vein (Borrok et al. 2009; Church
 128 Page 4 of 18
et al. 1997). In the San Juan River, the ratio of total Pb to
total Al was used to track the passage of a mining waste
plume (2015 Gold King Mine Spill) beyond the distance
at which it was not visually apparent (Sullivan et al.
2017).
1.2 Source Identification in Lacustrine Sediment
In order to differentiate natural, industrial, and/or
mining-sourced PM, sediment source identification in
lacustrine systems has been performed using elemental
concentrations in principal component analysis (PCA)
(Christensen and Juracek 2001; Palau Estevan et al.
2012; Wang et al. 2015), lead (Pb) isotopes (Graney
et al. 1995), and enrichment factors relative to major
lithogenic elements that are non-reactive during trans-
port (conservative) (Riba et al. 2002; Wang et al. 2015).
Similar to elemental ratios, enrichment factors normal-
ize concentrations of metals to conservative elements,
e.g., Fe, Al, and Mn, can be used to determine inputs of
anthropogenic and mining-sourced PM among contam-
inated and background lakes (Wang et al. 2015; Riba
et al. 2002). PCA is a statistical tool to reduce multidi-
mensional datasets through the generation of principal
components (PCs), which represent linear combinations
of input variables (e.g., concentrations, elemental, or
isotopic ratios) (Li and Zhang 2010; Wang et al.
2015). PCA performed with elemental concentrations
was used to differentiate sources (e.g., mining, industri-
al, or natural), where sample categories occupied dis-
tinct PC space (Wang et al. 2015). Additionally, PCA
probability ellipses (also referred to as convex hulls)
represent the confidence interval within PC space of a
given sample category and can be used to predict cate-
gorical association of an independent data set (Palau
Estevan et al. 2012). Similar to source identification in
surface water, stable isotopes of Pb, which do not frac-
tionate during transport/deposition, have been utilized in
both alluvial and lacustrine sediment to attribute sources
(e.g., a mineralized vein) (Graney et al. 1995; Church
et al. 1997).
1.3 Hypotheses and Assumptions
Layers in sediment deposited in the San Juan River
Delta of Lake Powell are posited to reflect distinct
sources from upstream tributaries. Because elevations
and lithologies range among tributaries, PM loads and
PM concentrations from a given tributary are posited to
Water Air Soil Pollut (2019) 230:128
change with runoff category (i.e., snowmelt, rainfall, or
baseflow), yielding distinct layers in sediment traps and
sediment cores at the downstream end of the system.
Because mining tends to occur in high elevations,
snowmelt-activated PM loads and PM concentrations
from mined tributaries are posited to have elemental
signatures, e.g., elevated Pb, Cd, As, Zn, and Pb isotope
ratio (corresponding to mineralized vein), that can be
used to link downstream sediment layers to mining
origin.
In order to link deposited sediment to upstream trib-
utaries, PM signatures of suspended sediment were
established for each tributary watershed and runoff cat-
egory (e.g., snowmelt, rainfall, and/or baseflow). PM
signatures in the deposited sediment of Lake Powell
were linked to upstream tributary sources using PM
concentrations/ratios and Pb isotopes. This strategy as-
sumes that reductive elemental mobilization following
deposition in Lake Powell does not reposition original
strata.
The percent contribution of tributary PM (both min-
ing and non-mining sourced) to suspended sediment in
the Lower San Juan River was estimated by comparing
tributary suspended PM loads to Lower San Juan River
suspended PM loads during each runoff category. This
watershed-scale approach assumes that PM concentra-
tions collected in stream samples from a given tributary
during a given runoff category are representative of that
runoff category for that given tributary; PM remains in
particulate form during transit through the San Juan
River; and high flow during large rainfall and snowmelt
events transmits a portion of PM to Lake Powell.
Whereas some material will likely be deposited along
the reach of the San Juan River, layers of sediment
deposited in Lake Powell represent a signal of mobilized
PM that can be tied to upstream sources.
Justification for these assumptions are explored in the
Supporting Information (Appendix A). Briefly, varia-
tion in average suspended PM loads and concentrations
for each runoff category (one standard deviation) were
compared to ensure the average tributary suspended PM
signatures were representative for each runoff category.
Additionally, transport modeling in the San Juan River
suggests that PM transits the majority of the system
during high flow (~ 1500 cfs) (Sullivan et al. 2017;
USGS 2018). As San Juan River flow during snowmelt
and rainfall events often surpass 1500 cfs, a signal of
mobilized PM should be deposited during these high
flow events.
Water Air Soil Pollut (2019) 230:128
By understanding the percent contribution of mining
versus non-mining sourced PM during different runoff
events in the Lower San Juan River, the percent of
deposited PM sourced from mining versus non-mining
sources was estimated. As the San Juan River and each of
its contributing tributaries represent complex systems,
this work serves as a first order estimate to compare
PM sources from mined versus non-mined tributaries.
2 Background
2.1 Tributaries of the San Juan River
The seven main tributaries of the San Juan River (Upper
San Juan River, Animas River, La Plata River, Mancos
River, McElmo Creek, Chaco Creek, and Chinle Creek)
can be characterized by prevalence of precious metal min-
ing, predominant underlying geologic unit, as well as runoff
type (i.e., snowmelt, rainfall, or baseflow) in the tributary
(Supporting Information Fig. SI1 and SI2, Table SI1).
Mined tributaries, defined as > 50 precious metal mines,
include the Animas, Upper San Juan, Mancos, and La Plata
Rivers, whereas unmined tributaries, defined as < 50 pre-
cious metal mines, include McElmo, Chinle, and Chaco
Creeks. The watersheds of McElmo Creek, Mancos River,
and La Plata River are predominantly underlain (> 50%
area) by the Mancos Shale, a Cretaceous-age marine shale
naturally high in As and Se (US DOE 2011; Larrick and
Ashmore 2012; Dyer et al. 2016; MRDS 2017), whereas
the watersheds of Chinle Creek and Chaco Creek are
predominantly underlain (> 50% area) by the Chinle Sand-
stone, a Pennsylvanian to Cretaceous age sandstone natu-
rally high in Al, As, cobalt (Co), chromium (Cr), Fe,
vanadium (V), and Zn (Newman 1962; MRDS 2017)
(Supporting Information Fig. SI1, Table SI1).
Stream hydrographs in snowmelt-dominated water-
sheds (Upper San Juan and Animas) are characterized
by sustained high flow during spring, whereas
hydrographs from rainfall-dominated watersheds (La
Plata, McElmo, Chinle, and Chaco) are characterized
by short high intensity flow events (Supporting
Information Fig. SI2).
2.2 San Juan River Delta of Lake Powell
In order to document the presence of elevated elemental
concentrations in San Juan River deltaic sediment, the
United States Geological Survey (USGS) collected
Page 5 of 18 128
three sediment cores in August 2010 ranging in depth
from 1.48 to 4.6 m (Hornewer 2014) (Fig. 1). The three
sediment cores (USGS 371712110364200, USGS
371425110382699, and USGS 371545110410700)
showed variation in elemental concentration with depth;
however, a sediment accumulation history was not de-
termined (Hornewer 2014). Additionally, in response to
the 2015 Gold King Mine spill, the USGS deployed
sediment traps at three locations (upstream, mid-stream,
and downstream, Fig. 1) in the San Juan River Delta
during baseflow with intermittent rainfall and snowmelt
(Supporting Information Fig. SI3) (UDWQ et al. 2018).
Lower San Juan River baseflow conditions punctuated
by four rainfall events in the watershed were captured in
48 cm at the upstream location from 8/23/2015 to 11/19/
2015 (USGS MRP-15078). Snowmelt runoff in the
Lower San Juan River (5/17/2016 to 7/14/2016) was
captured with three sediment traps (upstream, mid-
stream, and downstream locations), where 12.7 cm were
deposited in each trap over 59 days (USGS MRP-
15597). Lower San Juan River baseflow conditions
punctuated by a large rainfall event and two smaller
rainfall events in the watershed (7/14/2016 to 10/26/
2016) were captured in 3.5, 6, and 2 cm from the
upstream, mid-stream, and downstream locations, re-
spectively (USGS MRP-16105).
Sediment deposition rates in Lake Powell are high
(meter per year range) and vary substantially within and
between years (Stanford and Ward 1991; Potter and Drake
1989; Vernieu 1997). For example, USGS sediment traps
collected during baseflow with intermittent rainfall show
interannual variation in deposition rates, e.g., 1.97 m/year
in 2015 and 0.21 m/year in 2016. Average sediment
accumulation rates in the San Juan River Delta were
reported to range from 0.48 to 2 m/year (Potter and
Drake 1989; Stanford and Ward 1991; Vernieu 1997).
The high observed sedimentation rates in the San Juan
River Delta lead to an expectation that the USGS sediment
cores (ranging in length from 1.48 to 4.6 m) likely repre-
sent less than a decade of sediment deposition. Whereas
cesium (Cs)-137 and Pb-210 isotopes are commonly used
to determine sediment accumulation rates in modern lake
records (post-1950) (e.g., Malher et al. 2006; Bloesch and
Evans 1982; USGS 2016), the combination of fluctuating
deposition rates and the decade (short) timescale make
traditional dating methods unfeasible. Therefore, seasonal
signatures of San Juan River tributaries may help in
estimation of deposition time in the San Juan River Delta
sediment cores.
 128 Page 6 of 18
3 Methods
The concentration of total and filtered metals (<
0.45 μm) and total suspended solids (TSS) for each of
the seven tributary watersheds and the main channel of
the San Juan River were obtained from the USEPA
Storage and Retrieval database (STORET) (USEPA
2017) (Fig. 1) and supplemented with opportunistic
water quality samples collected at each tributary and
main channel of the San Juan River in May and Sep-
tember 2017. Sample sites from the downstream-most
locations in each of the seven tributary watersheds were
utilized since they represent an integrated signature of
PM loads in each watershed (Fig. 1). Samples collected
in the San Juan River downstream of all tributary wa-
tersheds near Mexican Hat, UT (Lower San Juan River)
represented the integrated PM loads from all tributary
watersheds (Fig. 1). Within each tributary, suspended
PM loads and concentrations were likely impacted by
settling and resuspension. By utilizing the downstream-
most locations for each tributary in suspended sediment
characterization, we are able to capture the integrated
signature of material entering the San Juan River from
each tributary. Supplemental water quality samples were
analyzed for total and filtered elemental concentrations
and Pb isotopes using an Agilent 7500ce quadrupole
ICP-MS with collision cell and Neptune Plus
multicollector.
STORET data included samples collected by the
USEPA, United States Geological Survey (USGS), Na-
vajo Nation Environmental Protection Agency
(NNEPA), Ute Mountain Ute Environmental Protection
Agency (UMUEPA), Colorado River Watch (CRW),
New Mexico Environmental Department (NMED),
and Utah Department of Environmental Quality
(UDEQ). STORET data from 1980 to 2017 that were
measured via inductively coupled plasma atomic emis-
sion spectroscopy (ICP-AES) or inductively coupled
plasma mass spectroscopy (ICP-MS) were included.
Number of samples for each tributary watershed and
collecting agency are given in Supporting Information
(Table SI2). STORET results with a total concentration
below the reported detection limit for a given element
were not included. STORET results with a filtered con-
centration below the reported detection limit for a given
element were replaced with one half that detection limit.
All water quality samples were collected as grab
samples in the shallow water column, which can neglect
PM in the deeper water column. However, direct
Water Air Soil Pollut (2019) 230:128
comparisons between shallow grab samples and sam-
ples composited across water column depth collected
May to July 2016 and 2017 from the Lower San Juan
River, found < 10% difference in metal concentrations
between these two sampling methods, demonstrating
that grab samples adequately capture mobilized PM
under the conditions in which this comparison was
made (UDWQ et al. 2018).
Both PM loads (mgmetal/s) and PM concentrations
(mgmetal/mgsed) were determined for the analyzed
suspended sediment samples. Conversion of discrete
elemental concentrations into loads can be done by
regression of discrete concentrations to continuous or
more frequent variables (e.g., discharge) (Runkel et al.
2004; Leib et al. 2003). However, an integration ap-
proach was not possible in this study because of infre-
quent water quality measurements (Supporting
Information Fig. SI4). Daily and instantaneous (15-
min interval) discharge data was downloaded from the
USGS (USGS 2018) for all gaging stations in the trib-
utaries and main channel of the San Juan River (USGS
gauge locations given in Fig. 1). Discharge from Chinle
Creek was used as an estimated flow for Chaco Creek
because discharge values were not available for Chaco
Creek, and historic discharge data from both tributaries
suggests similar ephemeral flow regimes (Supporting
Information, Figure SI4). Discrete elemental concentra-
tions (sampling events) were associated with discharge
values according to spatial proximity (within watershed
and less than 25 km separation) and temporal proximity
(the closest 15-min measurement).
Discrete PM loads were calculated from the product
of discharge and PM. A similar method of determining
PM loads was used for the Animas River (Sullivan et al.
2017; Besser and Leib 2007). Suspended PM concen-
trations (mgmetal/mgsed) were determined for each sam-
ple by differencing total and filtered concentrations di-
vided by the TSS concentration. Both suspended PM
loads and PM concentration samples were categorized
and averaged according to runoff category (i.e., snow-
melt, rainfall, or baseflow), under the assumption that
each sample mobilized PM concentrations representa-
tive of the assigned runoff category. Student’s t tests
(unpaired, two-tailed, unequal variance) were used to
determine whether average suspended PM loads or con-
centrations were distinct (p < 0.05) between tributary or
among runoff categories within a given tributary. Aver-
age percent PM load contribution of each tributary
watershed to the Lower San Juan River was determined
Water Air Soil Pollut (2019) 230:128
for each runoff category, where conservative transport
with no deposition was assumed.
The three sediment cores collected from the San Juan
River Delta of Lake Powell in August 2010 (Core 1,
Core 2, and Core 3) (Fig. 1) (Hornewer 2014) had
individual thickness of 1.48 m, 3.37 m, and 4.60 m,
respectively. The complete sampling methodology and
coring procedure are given in Hornewer (2014) and Hart
et al. (2005). Various strata were subsampled (20, 32,
and 49 subsamples, for cores 1, 2, and 3, respectively),
dried, digested via total digestion (HCl, HNO3, HClO4,
and HF at low temperature), and analyzed via inductive-
ly coupled plasma atomic emission spectrometry (ICP-
AES) (Hornewer 2014). Additional sediment samples
were sub-sectioned from the frozen cores in Fall 2017 at
the USGS laboratory in Boulder, CO (6, 19, and 15
samples from cores 1, 2, and 3, respectively).
Because the three sediment cores have similar aver-
age elemental concentrations (Student t test p > 0.05)
(Supporting Information, Fig. SI 5), Core 2 was selected
and analyzed for elemental concentration and Pb iso-
topes via ICP-MS with a 10% HCl digestion for 48 h at
22 °C. Grain size was analyzed via Sympa Tec Lixell
Helos/KF laser diffraction using two lenses, which cap-
tured particles sizes 0.25 to 87.5 μm and 0.5 to 350 μm,
respectively. Particle sphericity was analyzed via Sympa
Tec Lixell QICPIC.
Sediment color was assigned based on Munsell Soil
Color Charts (Munsell Color x-rite 2009). Water content
and bulk density of the sediment samples were deter-
mined by weighing a known volume of sample before
and after drying. Sediment elemental concentrations
were then corrected to represent the dry elemental con-
centration. A 10% HCl leaching was assumed to extract
only the metal/metalloid coating on the mineral (e.g.,
elements that precipitate upon mixing acidic mine waste
with circum-neutral river water) (extractable phase),
rather than the entire mineral matrix (total digestion).
While inclusion of an oxidizing extractant (nitric acid)
might have recovered additional trace elements via de-
struction of organic matter, increased chloride, and pro-
ton concentration (via HCl) also promotes trace element
release from organic matter. In this way, suspended PM
concentrations in the water, also digested via 10% HCl,
can be directly compared to sediment elemental
concentrations.
Sediment traps were sub-sectioned and analyzed by
the USGS (UDWQ et al. 2018), where layers deposited
during rainfall/baseflow 2015 were sub-sectioned into
Page 7 of 18 128
10 samples at 4- to 5-cm increments. Sediment deposit-
ed during peak snowmelt at the upstream location was
sub-sectioned by visual layer (total of three subsec-
tions), whereas the sediment from the other two loca-
tions was homogenized. Collection and analysis of the
sediment traps were performed by the USGS indepen-
dent from and prior to our project (MPR-15078, MPR-
15597, and MPR-16105). Sediment deposited during
rainfall/baseflow 2016 was homogenized and analyzed
separately for each trap location. All sediment trap sam-
ples were dried, digested via total digestion, and ana-
lyzed via ICP-AES-MS multi-acid (ICP-42) (USGS
method 19-42, USGS 2018).
PCA was performed in R (R Core Team 2013) to
generate linear combinations of sediment elemental con-
centrations, where PC1 accounts for the majority of
variance among elemental concentrations,and PC2 ac-
counts for the largest remaining variance orthogonal to
PC1. Because sediment elemental concentrations
spanned multiple orders of magnitude, elemental con-
centrations were log transformed to avoid
overweighting of large concentrations. PCA was per-
formed for the sediment trap elemental concentrations
in order to generate runoff-specific probability ellipses
representing 95% confidence interval. PC1 and PC2
weights were then applied to log-transformed sediment
core elemental concentrations (also measured via total
digestion).
Enrichment factors, PCA, and Pb isotopes were
used as independent lines of evidence for source iden-
tification of San Juan River Delta sediment layers.
Enrichment factors, Pb isotopes, sediment color, and
grain size tie sediment deposited in the San Juan River
Delta to upstream tributaries, where enrichment fac-
tors, Pb isotopes, and color will characterize the
bioavailable/redox sensitive component of sediment
(extractable phase). Enrichment factors and color of
San Juan River Delta sediment layers may record
changes in PM source, where distinct PM concentra-
tions and/or color exists between tributaries or between
runoff categories (assuming no diagenetic processes).
Pb isotopes in San Juan River Delta sediment may
reflect the mineralized vein (sourced from mined trib-
utaries), predominant underlying geologic unit
(sourced from unmined tributaries), or a combination
of sources. Sediment trap elemental concentration
PCA (total digestion) establishes runoff category
trends in deposited sediment to be used predictively
for San Juan River Delta sediment.
 128 Page 8 of 18
4 Results
4.1 Signatures in Extractable Phases in Suspended PM
and Sediment
4.1.1 Signatures in Suspended PM
Tributary PM concentrations that were uniquely elevat-
ed in a given tributary or during a given runoff category
within a given tributary were used to establish enrich-
ment factors for comparison to those from San Juan
River Delta sediment. For each runoff category, distinct-
ly elevated PM concentrations were correlated with
specific tributaries. Cd, Cu, Pb, selenium (Se), and Zn
PM concentrations were higher in the Animas River
relative to other tributaries during all runoff categories
(p < 0.05, Fig. 2; Supporting Information Fig. SI6). An-
imas River rainfall/baseflow beryllium (Be) and molyb-
denum (Mo) PM concentrations were greater than other
tributary rainfall or baseflow PM concentrations
(Fig. 2a, b). Additionally, Animas River snowmelt As,
manganese (Mn), and Mo PM concentrations were
greater than other tributary snowmelt PM concentrations
(Fig. 2b). Mancos River snowmelt Be, Cr, and nickel
(Ni) PM concentrations and baseflow Mo PM concen-
trations exceeded other tributary PM concentrations for
the respective runoff categories (Fig. 2b, c). During
rainfall, Chaco Creek Mn concentrations were greater
than other tributary watersheds (Fig. 2a).
Whereas PM concentrations were generally indistinct
between runoff categories within a given tributary
(Fig. 2; Supporting Information Fig. SI6), a few excep-
tions existed which differentiated tributary snowmelt
PM concentrations from rainfall or baseflow PM con-
centrations. These exceptions included Animas River
As PM concentrations, which were greater for snowmelt
relative to rainfall and baseflow (p < 0.05) (Fig. 3) as
well as Animas River Se PM concentrations, which
were greater for baseflow relative to snowmelt and
rainfall (p < 0.05) (Fig. 3). Additionally, Mancos River
Mo and Se PM concentrations were greater for snow-
melt relative to rainfall or baseflow (Supporting
Information, Fig. SI7), although elevated Mancos River
snowmelt Mo and Se PM concentrations were insignif-
icant compared to Animas River Mo and Se PM con-
centrations (Fig. 2).
PM concentrations that were uniquely elevated in a
given tributary or during a given runoff category in a
given tributary were used to characterize the suspended
Water Air Soil Pollut (2019) 230:128
PM of that tributary for subsequent tie to San Juan River
Delta sediment. Elevated PM concentrations of Cd, Cu,
Pb, and Zn were indicative of Animas River contribu-
tions, whereas elevated As and Se concentrations were
indicative of Animas River snowmelt contributions.
Elevated Ni PM concentrations were indicative of
Mancos River snowmelt contributions and elevated
Mn PM concentrations were indicative of Chaco Creek
rainfall contributions.
That PM Al concentrations were similar between
tributaries (Fig. 2) and between runoff categories
(Fig. 3), which justifies the use of Al as a conservative
element for enrichment factors in the San Juan River.
4.1.2 Signatures in Delta Sediment
For elements with unique tributary or runoff suspended
PM concentrations (e.g., As, Se, and Pb), enrichment
factors were calculated relative to Al for sediment core
2, where variations in these ratios with depth were
examined to indicate potential changes in PM sources
(Fig. 4b). Samples 2–03 had the greatest Pb to Al, Cu to
Al, Cd to Al, and Zn to Al ratios, suggesting Animas
River-sourced PM (Fig. 4b; Supporting Information,
Fig. SI8). Peaks in Ni to Al occurred in samples 2–11
and 2–14, suggesting Mancos River-sourced PM
(Fig. 4b). Sample 2–05 also had a peak in Ni to Al co-
occurring with a peak in Cd to Al, Cu to Al, and As to Al
indicating a mixture of sources (Fig. 4b). Peaks in Mn to
Al occurred in samples 2–03, 2–06, and 2–12, suggest-
ing Chaco Creek sourced PM (Supporting Information,
Fig. SI8).
Core 2 sediment color and grain size corresponded to
ranges observed in PM in upstream tributaries. Core 2
sediment was predominantly brown (Munsell color 4/3),
red (Munsell color 4/6), and dark brown (Munsell color
3/4) (Munsell Color x-rite 2009) (Fig. 4a). Similar
colors were observed in PM from the tributaries, where
suspended sediment in the Animas River, Upper San
Juan River, Mancos River, La Plata River, and McElmo
Creek ranged from brown (Munsell color 4/3) to dark
brown (Munsell color 3/4) and suspended sediment in
Chinle Creek was red (4/8) (Munsell Color x-rite 2009)
(Supporting Information, Fig. SI9). The exception was
Chaco Creek, which was light olive brown (Munsell
color 5/4), a color not observed in Core 2. Under the
assumption that diagenetic processes did not alter sedi-
ment color following deposition, the red color in
Water Air Soil Pollut (2019) 230:128 Page 9 of 18 128

<0.05
<0.05
<0.05
<0.05
<0.05
<0.05

<0.05
<0.05
<0.05 <0.05
<0.05
<0.05 <0.05
<0.05

<0.05

<0.05 <0.05
<0.05
<0.05 <0.05

<0.05

c
Fig. 2 Average PM concentration in the tributary watersheds
during rainfall (a), snowmelt (b), and baseflow (c) shaded from
upstream (dark blue) to downstream (red). Statistically significant
differences (p < 0.05) between average PM concentrations are
shown. Tributary PM concentrations are only present for runoff
categories occurring within the tributary (e.g., no Chinle baseflow
PM concentrations exist). PM concentrations not shown in Fig. 2
are given in Supporting Information (Fig. SI6)

samples 2–01, 2–12, 2–13, 2–15, and 2–18 (Fig. 4a)
may reflect sediment sourced from Chinle Creek.
Median (P50) grain size in Core 2 ranged from 4 to
150 μm with a majority of the sediment < 10 μm in size
(Fig. 4c). Median grain size was used because particles
were observed over the entire size range (0.25 μm to
87.5 μm or 0.5 μm to 350 μm, dependent on lens used)
and were not normally distributed (Supporting
Information, Fig. SI 10). The median suspended sedi-
ment size in Chinle and Chaco Creeks was smaller than
those observed in sediment Core 2 (P50 2.9 ± 0.12and
3.4 ± 0.05 μm, respectively), while Upper San Juan
River, Animas River, Mancos River, La Plata River,
and McElmo Creek mobilized particles with median
 128 Page 10 of 18
<0.05
Fig. 3 Average PM concentrations (mgmetal/mgsed) in the Animas River during rainfall (red), snowmelt (blue), and baseflow (gray).
Statistically significant differences (p < 0.05) between average PM concentrations are shown
size ranging from 8.5 to 39.0 μm (Supporting
Information, Fig. SI11). Core 2 sphericity measure-
ments were not useful in differentiating tributary source
(mean sphericity 0.71 ± 0.01) and are provided in the
Supporting Information (Fig. SI12).
Sediment elemental concentrations in the extractable
phase were 1 to 1000 times lower than those determined
by dissolution of the matrix (Hornewer 2014) (Fig. 4d).
In general, over 50% of elemental concentration was in
Sample Color/ID Metal:Al rao
a b
Fig. 4 a Dominant Munsell color of sediment (Munsell Color x-
rite 2009) and sample ID as a function of depth from the top of the
core (m); b ratio of extractable As (gold), Cd (orange), Cu (green),
Pb (light blue), Mo (dark blue), Ni (purple), and Se (pink) to Al as
a function of depth from the top of the core (m); c Median (P50)
grain size (μm) for samples (black line) as a function of depth from
Water Air Soil Pollut (2019) 230:128
<0.05
the sediment matrix and not on the coating, with less
than 10% of Al, Cr, Sb, and V on the coating (low ratio
of extractable to digested, Fig. 4d).
Similar to elemental concentrations, color, and grain
size, sediment Pb isotope ratios linked deposited sedi-
ment to PM in upstream tributaries. Suspended PM Pb
isotope ratios from mined San Juan River tributaries
generally reflected Pb isotopes in the mined mineralized
vein (Eureka Graben) (Church et al. 1997), whereas Pb
P50 Grain Size (μm) Extractable:Digested (%)
c d
the top of the core (m) with the lens size range in gray; and d
Percent of metal in extractable phases relative to full digestion as a
function of depth from the top of the core (m). The lesser number
of samples in D reflects lesser samples fully digested (Hornewer
2014)
Water Air Soil Pollut (2019) 230:128
isotope ratios in unmined tributaries reflected the pre-
dominant underlying geologic unit (Fig. 5). Suspended
sediment in the Animas, Upper San Juan, La Plata, and
McElmo Rivers had Pb isotope ratios that reflected the
mineralized vein (Church et al. 1997), which is present
in the headwaters of each tributary except McElmo
Creek (Fig. 5). The Dolores River, headwaters underlain
by the mineralized vein, is diverted to McElmo Creek to
maintain year-round flow (Larrick 2010) and appears to
dominate the McElmo Pb signature. The Mancos River
Pb isotope ratios reflected a combination of the miner-
alized vein and the Mancos Shale, which has relatively
enriched 208Pb/204Pb and 206Pb/204Pb (Church et al.
1997) (Fig. 5). Chinle and Chaco Creeks Pb isotope
ratios were enriched, potentially reflecting the Chinle
Sandstone (Fig. SI5).
Pb isotope ratios in core 2 spanned Pb isotope ratios
measured in tributary suspended sediment (Fig. 6),
where Pb isotope ratios for samples 2–02 and 2–03
reflected the mineralized vein, and samples 2–01, 2–
12, 2–13, 2–15, and 2–18 reflected the Chinle Sand-
stone (Fig. 6). Pb isotopes measured in 2–04 and 2–05
represented a transition, similar to the Pb isotope ratios
measured in the Mancos River (Fig. 6).
4.2 Sediment Matrix Signatures in PCA Space
Runoff category signatures were established through
PCA of sediment trap PM concentrations (via total
digestion) (mgmetal/mgsed), which were deployed during
baseflow with intermittent rainfall and peak snowmelt
(UDWQ et al. 2018). Sediment trap periods occupied
distinct PC spaces (Fig. 7a); where sediment deposited
during snowmelt had higher Pb, Zn, V, Mo, and As
Fig. 5 Lead isotope ratios
(208Pb/204Pb) for sediment
(Church et al. 1997) (black
squares), PM in the San Juan
River (open triangles), and PM in
tributaries (open circles) as a
function of decreasing elevation.
Church et al. (1997) measured
lead isotopes in the Eureka
Graben, Mancos Shale, as well as
in the bed sediment along the
length of the Animas River
Page 11 of 18 128
concentrations (lower PC1) than sediment deposited
during baseflow with intermittent rainfall, which were
higher in Fe, Al, and Ni concentrations (higher PC1)
(Fig. 7a). Rainfall/baseflow sediment deposited in 2015
had higher Fe, Al, and Ni concentrations (higher PC1)
than sediment deposited during rainfall/baseflow in
2016 (Fig. 7a), possibly reflecting the larger 2015 rain-
fall events resulting in significant flow in Chinle Creek
(Supporting Information, Fig. SI2) (UDWQ et al. 2018).
Additionally, the significantly higher sedimentation rate
captured in 2015 rainfall/baseflow trap than in 2016
rainfall/baseflow trap (1.97 m/year versus 0.21 m/year,
respectively) indicates the activation of a large sediment
source (e.g., Chinle Creek) during the 2015 deployment.
Sediment trap periods were not further differentiated
in PC2, as sediment deposited during both snowmelt
and baseflow with intermittent rainfall spanned PC2
(Fig. 7a). Unlike sediment trap period, sediment trap
location (upstream, midstream, and downstream) did
not occupy distinct PCA spaces (Fig. 7a), where colors
corresponding to sediment trap location were mixed
throughout the PCA space.
Core 2 sediment PM concentrations (total digestion),
occupied similar PCA space to sediment traps, except-
ing 2–07 which was lower in all PM concentrations
(Fig. 7b). Samples 2–01, 2–13, 2–15, 2–17, and 2–18
occupied PC space within the probability ellipse
assigned to 2015 rainfall/baseflow sediment traps, indi-
cating that these sample concentrations were similar
(95% confidence) (Fig. 7b). Whereas samples 2–03
and 2–16 occur outside the probability ellipse for peak
snowmelt sediment, these samples occupied a similar
PC2 space (higher Pb, Zn, Mo, and As) as that occupied
by early and peak snowmelt samples (Fig. 7b). Samples
 128 Page 12 of 18 Water Air Soil Pollut (2019) 230:128

Fig. 6 a Dominant Munsell color

of sediment (Munsell Color x-rite
2009) and sample ID as a function
of depth from the top of the core
(m); b Pb isotope ratios for the
tributaries (closed circles), San Chinle
Chaco
Juan River (open circles), and 2-01, 2-12, 2-13,
Lake Powell Sediment (squares). Mancos
2-15, 2-18
Sediments are colored based on
dominant Munsell color of 2-02
sediment McElmo
2-03
Lower
La Plata Mid
Upper
Animas

a b

2–09 and 2–19 occurred in the 2016 rainfall/baseflow
probability ellipse, indicating similar concentrations
(Fig. 7b).
4.3 Average Tributary Suspended PM Loads
Whereas PM concentrations link geochemical signa-
tures in suspended sediment to geochemical signatures
in deposited sediment, suspended PM loads, in contrast,
link suspended material mass to deposited layer
thickness. Tributary suspended PM loads were com-
pared to Lower San Juan River suspended PM loads to
estimate the relative contribution of PM from different
tributaries (both mining and non-mining sourced) to PM
loads directly upstream of Lake Powell.
The contribution of upstream tributaries to the Lower
San Juan River suspended PM load changed with runoff
category (Fig. 8), even though overall Lower San Juan
River PM loads were similar regardless of runoff category
(p > 0.05, Supporting Information Fig SI13). During

Trap Period
Rainfall/baseflow
Early snowmelt Cores
Peak snowmelt Core 2
Trap Locaon Trap Period
Upstream Rainfall/baseflow
Mid-stream Early snowmelt
Downstream Peak snowmelt

a b
Fig. 7 a PCA for USGS sediment traps deployed over three 29.2% of the variance, respectively. PC3 explained 14.5% of the
periods: 2015 rainfall/baseflow (circles), 2016 snowmelt variance; b Sediment core samples were assigned PC1 and PC2
(squares), and 2016 rainfall/baseflow (triangles), collected at three weights generated by sediment traps (blue diamonds) and graphed
locations: upstream (red), mid-stream (green), and downstream with sediment deposited during 2015 rainfall/baseflow (black
(blue) (Fig. 1). Graphical representation of elemental PC weights circles), 2016 snowmelt (squares), and 2016 rainfall/baseflow
(solid lines) are given, where vectors denote weight and direction (triangles). Probability ellipses, representing the 95% confidence
of PC1 and PC2 weights. PC1 and PC2 explained 48.3% and interval, are given for the distinct sediment trap categories
Water Air Soil Pollut (2019) 230:128 Page 13 of 18 128

Fig. 8 Average percent Snowmelt

particulate metal loads in the
Upper San Juan (checkered bar),
Animas (black bar), La Plata
(speckled bar), Mancos (gray
bar), McElmo (light gray bar),
Chaco (diagonal striped bar), and
Chinle (vertical striped bar)
tributaries during, snowmelt,
rainfall, and baseflow events.
Open bars represent the percent of
particulate load measured in the
Lower San Juan River not
accounted for by the upstream a
tributary watersheds. PM load Rainfall
(mgmetal/s) contributions for each
tributary during snowmelt,
rainfall, and baseflow are
provided in the Supporting
Information (Fig. SI16)

b
Baseflow

snowmelt, the Animas River was the dominant source of
Lower San Juan River PM loads (> 50% contribution of
As, Cd, Cr, Cu, Pb, Se, Tl, and Zn) (Fig. 8a). The Upper
San Juan River contributed the second largest loads of Cr
and Cu in the Lower San Juan River during snowmelt
(30% and 36% contribution, respectively), whereas
Mancos River snowmelt PM loads were even less signif-
icant (< 20% contribution for each element). Unaccounted
PM loads (Fig. 8 open bars) are examined in Supporting
Information (Appendix I).
During rainfall events, Chinle Creek was the dominant
source (> 50% contribution) of Lower San Juan River PM
loads for Al, Cr, Cu, Pb, Ni, V, and Zn (Fig. 8b). During
rainfall events, the Animas River contributed the largest
PM loads of As, Be and Se (32%, 89%, and 78%, respec-
tively), the Upper San Juan River contributed the second
largest PM loads of Se and Zn (20% and 22%, respective-
ly), McElmo Creek contributed the largest PM loads of Fe
(48%) and Tl (93%), and the Chaco Creek contributed the
largest PM load of Cd (30%). The La Plata River and
Mancos River did not contribute any significant rainfall
PM loads. Whereas the Animas River contributed the
largest loads of Be and Tl during baseflow conditions
(99% and 68%, respectively), a majority of the Lower
San Juan River baseflow PM loads were not attributable
to an upstream tributary (Fig. 8c).
 128 Page 14 of 18
The overall percent PM load contributions from each
tributary to the Lower San Juan River during a given
runoff category did not vary significantly when one
standard deviation of the average PM load was consid-
ered (Supporting Information Fig. SI14). The propor-
tionality of average PM loads to standard deviation PM
loads suggests that PM loads were consistent within
runoff events and that PM source was not limited. When
considering discrete samples, PM load and discharge
were directly related (power law, R2 > 0.68, e.g., Al
PM load = 221.32*discharge0.75) in each tributary re-
gardless of runoff category, indicating consistency of
PM concentrations among runoff categories. An excep-
tion was rainfall in the Animas River, where both high
and low PM loads occurred at low discharge
(Supporting Information, Fig. SI15).
5 Discussion
5.1 Mined versus Unmined Tributary Signatures
During snowmelt, discharge (Supporting Information
Fig. SI12), PM load (Fig. 8), and Pb isotope ratio
(Fig. 5) of the Lower San Juan River each reflected the
Animas River. The association of Animas snowmelt PM
loads (Fig. 8) and Pb-isotope with mining activity
(Fig. 5) (Church et al. 1997), as well as the fact that
snowmelt runoff predominantly originated in high ele-
vation, mining-intensive terrain, links Animas, and
Lower San Juan River snowmelt PM to mining legacy.
Animas River snowmelt PM load contributions domi-
nated those of the other mined tributaries (Upper San
Juan, La Plata, and Mancos Rivers) (Fig. 8) due to
relatively insignificant flows in the La Plata and Mancos
Rivers, as well as trapping by the Navajo Dam in the
Upper San Juan River (Supporting Information Fig.
SI2). That the Upper San Juan River Pb isotope ratio
was representative of the mineralized vein (Fig. 5) sug-
gests that not all mining-sourced PM is trapped by
Navajo Dam and the ~ 30% contributions of Upper
San Juan River snowmelt PM loads of Cu, Cr, and Cd
are potentially mining-sourced (Fig. 8).
That the Mancos River Pb isotope ratio reflected a
combination of Mancos Shale and mineralized vein
(Fig. 5) (Church et al. 1997) suggests that both mining
and non-mined terranes contributed to Mancos River
PM loads. Reported elevated dissolved concentrations
of Cu and Ag attributed to mining in the east fork of the
Water Air Soil Pollut (2019) 230:128
Mancos Watershed (Clow and Stephens 2005) were not
apparent in downstream Mancos River PM loads
(Fig. 8), possibly due to dilution by loads from the
unmined West Fork tributaries as well as differences in
partitioning among dissolved versus particulate phases.
Reported elevated As and Se concentrations (mgmetal/L)
in Mancos Shale (Dyer et al. 2016; U.S. Department of
Energy 2011; Larrick and Ashmore 2012) were
reflected in Mancos As and Se PM concentrations,
which rivaled those of the Animas River (Supporting
Information Fig. SI7). In contrast, As and Se PM loads
from the Mancos River were minor fractions of the As
and Se PM load of the Lower San Juan River, reflecting
the relatively low flow in the Mancos River compared to
the Animas River (Fig. 8).
In contrast to snowmelt-dominated watersheds,
Chinle Creek and other unmined tributaries dominated
PM contributions to the San Juan River during rainfall
events (e.g., Al, As, Cr, Co, Cu, Fe, Pb, Ni, V, and Zn)
(Fig. 8). The association of Chinle Creek rainfall PM
loads (Fig. 8) to elements associated with the Chinle
Sandstone (Newman 1962) and Chinle and Chaco
Creek Pb isotope with the Chinle Sandstone (Fig. 5)
links Chinle and Chaco Creek rainfall PM loads to
unmined sources. McElmo Creek Fe and Tl PM rainfall
loads (Fig. 8) may reflect both mined and unmined
terranes. Whereas McElmo Watershed is unmined, it is
supplemented by water from the Dolores River, which is
mined (Larrick 2010), and this was reflected in the Pb
isotope ratio (Fig. 5).
5.2 Linking Sediment Core Layers to PM of Upstream
Tributaries
Elemental and isotopic signatures in Lake Powell sedi-
ment were linked to those of PM in upstream tributaries.
The 0.66-m layer near the top of core 2, represented by
samples 2–02 and 2–03, likely reflected Animas River
snowmelt PM, where the Pb isotope ratios represented
the mineralized vein present in the headwaters of the
Animas River (Fig. 5). As well, ratios of Pb to Al, Cd to
Al, Cu to Al, and Zn to Al were characteristic of the
Animas River (Fig. 4b), and elemental concentrations
occupied similar PC space as sediment deposited during
snowmelt (Fig. 7).
The dominance of the Animas River PM signature in
the Lower San Juan River during snowmelt (Supporting
Information, SI17) in addition to the low deposition
rates modeled in the San Juan River during high flow
Water Air Soil Pollut (2019) 230:128
typical of snowmelt (Sullivan et al. 2017) would allow
Animas River snowmelt PM signatures to be deposited
as a layer in the sediment of the San Juan River Delta.
The 0.24-m layer deposited before the snowmelt layer,
represented by samples 2–04 and 2–05, possibly
reflected Animas River early snowmelt PM as the Pb
isotope ratio was mixed (Fig. 6), but no excursions in
elemental concentration were noted.
Sediment core layers with enriched Pb isotope ratios
(Fig. 6), red color (Fig. 4), and enriched Mn to Al
(Supporting Information Fig. SI8) likely reflected
Chinle and Chaco Creek PM sources (i.e., layers repre-
sented by samples 2–01, 2–12, 2–13, 2–15, 2–17, and
2–18). These samples also had elemental concentrations
that occupied similar PC space as sediment deposited
during 2015 rainfall/baseflow (Fig. 6), suggesting the
dominance of Chinle Creek PM during large rainfall
events. That Chinle Creek (farthest downstream tribu-
tary at ~ 20-km upstream of Lower San Juan) dominated
rainfall layers deposited in the San Juan River Delta
sediment, indicates that short duration high intensity
flow events transmit PM across (~20 km) the system.
Whereas the red color of core 2 layers could potentially
reflect post-depositional oxidation of iron, it more likely
reflects the red color of Chinle Creek PM, since PCA
and Pb isotope ratios for red sediment layers also sug-
gested Chinle Creek as a source.
The 0.57-m-thick sandy layer, represented by sam-
ples 2–06 and 2–07, cannot be attributed to a runoff
category because PM concentrations were lower than
those measured in any sediment trap (Fig. 7); however,
the Pb isotope ratios were representative of the Chinle
Sandstone (Fig. 5). The median grain size of the sandy
layer (154 μm and 90 μm, for sample 2–06 and 2–07,
respectively) (Fig. 4C) was larger than any median
tributary particle size, suggesting that events mobilizing
coarse sand particles were rare, or that coarse particles
were not successfully collected using the grab sampling
technique in upstream tributaries. Either possibility is
consistent with the good agreement in PM concentra-
tions measured between sampling composited and grab
samples (UDEQ et al. 2018), since coarse particles have
low extractable elemental concentrations (Fig. 4d). No-
tably, no sand layers were observed in sediment traps,
suggesting that coarse particles were not suspended in
the water column.
Other layers likely reflect a mixture of sources de-
posited during baseflow, where sediment PM concen-
trations were lower than both baseflow with intermittent
Page 15 of 18 128
rainfall and snowmelt (Fig. 7) and Pb isotopes were
enriched compared to the Chinle Sandstone (Fig. 6)
(i.e., samples 2–08, 2–11, 2–14, and 2–19). Sediment
core sample 2–16 occupied similar PC space as sedi-
ment deposited during snowmelt but had a Pb isotope
ratio reflective of the Chinle Sandstone, potentially in-
dicating a mixture of sources.
Because Animas River snowmelt PM loads in the
San Juan River can be attributed to mining in the head-
waters of the Animas River, sediment layers reflecting
the Animas River snowmelt signature may also be at-
tributed to mining sources. Conversely, because rainfall
PM loads sourced from the Chinle and Chaco tributaries
can be attributed to non-mining sources, sediment
reflecting Chinle and Chaco Creek rainfall signatures
may be attributed to non-mining sources.
Therefore, the percent of PM sourced from mining
and non-mining sources was estimated from the percent
contribution of mining and non-mining sourced PM in
the Lower San Juan River (Fig. 8), the length of each
sediment layer (Hornewer 2014), and the bulk density
(Supporting Information, Fig. SI19) of each sediment
layer. Assuming each subsample PM concentration rep-
resented the average PM concentration for a given layer,
the percent of mining and non-mining sourced PM over
Core 2 was calculated. For Core 2, 0.2% Al, 4% As, 5%
Cu, 1% Fe, 10% Pb, and 8% Zn was attributed to mining
sources and 5% Al, 2% As, 7% Cu, 1% Fe, 3% Pb, and
4% Zn was attributed to non-mining sources. The re-
maining PM was likely sourced from a mixture of
mining and non-mining sources. As this is a first attempt
at attributing mining and non-mining source PM in the
sediment of Lake Powell, future work is needed to refine
estimates made here.
5.3 Age of the Core
Our results suggest that Core 2 was deposited in only ~
1.3 years, which reflects the high sedimentation rates
reported to occur during some years in the San Juan
River delta. Core 2 was collected in mid-August 2010,
indicating that the 2010 snowmelt signature was repre-
sented in samples 2–03 to 2–04 (near the top of the
record). No other snowmelt layer was detected in core 2,
suggesting that the 2009 snowmelt was not represented.
2009 was a low snowmelt year in the Lower San Juan
River, with peak flow ending in early May rather than
the typical June or July (Supporting Information Fig.
SI20), (USGS 2018). Core 2 was collected directly after
 128 Page 16 of 18
a large precipitation event (08/03/2010) where flow in
the Lower San Juan River exceeded 5000 cfs (USGS
2018) and flow in Chinle Creek reached ~ 100 cfs, likely
yielding the layer represented by sample 2–01. A similar
age for this core, 1.4 years, was determined using bathy-
metric survey elevation and average sedimentation rates
(UDWQ et al. 2018).
Whereas Pb isotopes, particle sizes, and enrichment
factors were determined for core 2, these attributes were
not determined for cores 1 and 3. PCA was performed
for cores 1 and 3 (Supporting Information, Fig. SI21).
Both core 1 and core 3 had a single sample fall within
the probability ellipse established by 2016 snowmelt,
and multiple samples fall within the probability ellipses
established by 2015 and 2016 rainfall/baseflow sedi-
ment traps. Whereas future work will examine the re-
cord provided in these sediment cores, the detection of
an Animas River snowmelt layer in core 2 suggests that
high flow, typical of spring runoff, is needed for up-
stream signatures (above Chinle Creek) to be preserved
in the downstream sediment record.
6 Conclusions
San Juan River Delta sediment PM reflects mining and
non-mining sources from the tributaries of the San Juan
River. Tributary Pb isotopes and PM concentrations,
which differ with runoff category, were used to link
sediment core layers to upstream tributaries, where the
3.37 m core examined here represented approximately
1.3 years of sediment deposition. Approximately 10%
of the overall PM deposited in the sediment core was
attributed to mining sources, with ~ 80% of the overall
PM reflecting a mixture of mining and non-mining
sources.
Acknowledgements This research was supported by the Utah
Department of Environmental Quality (Contract #171990). The
authors wish to thank the scientists at the NNEPA, USGS, and
DEQ for their support.
References
Abell, R. (1994). San Juan River Basin water quality and contam-
inants review (p. 1). Museum of Southwestern Biology.
Arditsoglou, A., & Samara, C. (2005). Levels of total suspended
particulate matter and major trace elements in Kosovo: a
Water Air Soil Pollut (2019) 230:128
source identification and apportionment study.
Chemosphere, 59, 669–678.
Axtmann, E. V., & Luoma, S. N. (1991). Large-scale distribution
of metal contamination in the fine-grained sediments of the
Clark Fork River, Montana, U.S.A. Applied Geochemistry, 6,
75–88.
Besser, J. M., & Leib, K. J. (2007). Toxicity of metals in water and
sediment to aquatic biota. In Integrated investigations of
environmental effects of historical mining in the Animas
River Wa ters ehd , S an Jua n Cou nt y, C olo ra do .
(Professional Paper 1651) (pp. 837–849). Reston: U.S.
Department of the Interior, U.S. Geological Survey.
Bliesner, R., & Lamarra, V. (2000). San Juan River basin recovery
implementation program: hydrology, geomorphology and
habitat studies. U.S: Fish and Wildlife Services.
Bloesch, J., & Evans, R. D. (1982). Lead-210 dating of sediments
compared with accumulation rates estimated by natural
markers and measured with sediment traps. Hydrobiologia,
92, 579–586.
Borrok, D. M., Wanty, R. B., Ridely, W. I., Lamothe, P. J.,
Kimball, B. A., Verplanck, P. L., & Runkel, R. L. (2009).
Application of iron and zinc isotopes to track the sources and
mechanisms of metal loading in a mountain watershed.
Applied Geochemistry, 24, 1270–1277.
Christensen, V. G., & Juracek, K. E. (2001). Variability of metals
in reservoir sediment from two adjacent basins in the central
Great Plains. Environmental Geology, 40(4–5), 470–481.
Church, S. E., Kimball, B. A., Fey, D. L., Ferderer, D. A., Yager, T.
J., & Vaughn, R. B. (1997). Source, transport, and
partitioning of metals between water, colloids, and bed sed-
iment of the Animas River, Colorado. United States
Geological Survey, 97–151.
Clow, S., & Stephens, D. B. (2005). Nonpoint source assessment
of the Ute Mountain Ute Reservation of Colorado, New
Mexico and Utah. Ute Mountain Ute Tribe and Stephens
and Associates, Inc.
Coulthard, T. J., & Macklin, M. G. (2003). Modeling long-term
contamination in river systems from historical metal mining.
Geological Society of America, 31(5), 451–454.
Covelli, S., & Fontolan, G. (1997). Application of a normalization
procedure in determining regional geochemical baselines,
Gulf of Trieste, Italy. Environmental Geology, 30(1–2), 34–
45.
Czuba, J. A., Magirl, C. S., Czuba, C. R., Grossman, E. E., Curran,
C. A., Gendaszek, A. S., & Dinicola, R. S. (2011). Sediment
load from major rivers into Puget Sound and its adjacent
waters. US Department of the Interior, US Geological
Survey.
Dologopolova, A., Weiss, D. J., Seltmann, R., Kober, B., Mason,
T. F. D., Coles, B., & Stanley, C. J. (2006). Use of isotope
ratios to assess sources of Pb and Zn dispersed in the envi-
ronment during mining and ore processing with the Orlovka-
Spokoinoe mining site (Russia). Applied Geochemistry, 21,
563–579.
Dyer, M., Moore, S. A., & Stumpf, S. E. (2016). Water quality
monitoring for the Mancos River in Mesa Verde National
Park: 2012-2013 summary report. In National Park Service:
Natural resource data series NSP/SCPN/NRDS-2016/1221.
Evans, T. D., & Paulson, L. J. (1983). The influence of Lake
Powell on the suspended sediment-phosphorus dynamics of
Water Air Soil Pollut (2019) 230:128
the Colorado River inflow to Lake Mead (p. 57). University
of Nevada Las Vegas.
Graney, J. R., Halliday, A. N., Keeler, G. J., Nriagu, J. O., Robbins,
J. A., & Norton, S. A. (1995). Isotopic records of lead
pollution in lake sediments from northeastern United States.
Geochmica et Cosmochimica Acta, 50(9), 1715–1728.
Hart, R. J, Taylor, H. E., Antweiler, R. C., Graham, D. D., Fisk, G.
G., Riggins, S. G., and Flynn, M. E. (2005). Sediment chem-
istry of the Colorado River delta of Lake Powell, Utah, 2001:
U.S. Geological Survey Open File Report 2005-1178, p. 33.
Hornewer, N. J. (2014). Sediment and water chemistry of the San
Juan River and Escalante River Deltas of Lake Powell, Utah,
2010–2011. United States Geologic Survey. Open File
Report 2014-1096.
Irons, W. V., Hembree, C. H., & Oakland, G. L. (1965). Water
resources of the Upper Colorado River Basin. Technical
report: Washington, D.C., U.S. Geological Survey
Professional Paper, 441, 370.
Kimball, B. A., Runkel, R. L., Walton-Day, K., & Bencala, K. E.
(2002). Assessment of metal loads in watersheds affected by
acid mine drainage by using tracer injection and synoptic
sampling: Cement creek, Colorado, USA. Applied
Geochemistry, 17, 1183–1207.
Kimball, B. A., Walton-Day, K., & Runkel, R. L. (2007).
Quantification of metal loading by tracer injection and syn-
optic sampling, 1996-2000, chap. E9 of Church, S.E., von
Guerard, P., and Finger, S.E., eds., Integrated investigations
of environmental effects of historical mining in the Animas
River Watershed, San Juan County, Colorado. U.S.
Geological Survey Professional Paper, 1651, 417–495.
Kimball, B. E., Mathur, R., Dohnalkova, A. C., Wall, A. J.,
Runkel, R. L., & Brantely, S. L. (2009). Copper isotope
fractionation in acid mine drainage. Geochemica et
Cosmochimica Acta, 73, 1247–1263.
Larrick, C. (2010). McElmo Creek, lower San Juan River,
Colorado and cottonwood wash. Utah Water Quality and
Trends Assessment, 2009–2010.
Larrick, C., & Ashmore, J. (2012). Mancos River water quality
and trends assessment: 2011 – 2012. Ute Mountain Ute tribe.
Leib, K. J., Mast, M. A., & Wright, W. G. (2003). Using water-
quality profiles to characterize seasonal water quality and
loading the upper Animas River Basin, Southwestern
Colorado. U.S. Geological Survey, Water-Resources
Investigations Report 02–4230.
Li, S., & Zhang, Q. (2010). Risk assessment and seasonal varia-
tions of dissolved trace elements and heavy metals in the
upper Han River, China. Journal of Hazardous Materials,
181, 1051–1058.
Malher, B., Van Metre, P. C., & Callender, E. (2006). Trends in
metals in urban and reference lake sediments across the
United States, 1970 to 2001. Environmental Toxicology and
Chemistry, 25(7), 1698–1709.
Mau, D. P., & Christensen, V. G. (2000). Comparison of sediment
deposition in reservoirs of four Kansas watersheds. United
States Geological Survey, Fact Sheet 102-00.
Mineral Resource Data System (MRDS). (2017). United States
Geological Survey. Accessed at: https://mrdata.usgs.
gov/mrds/.
Munsell Color x-rite. Munsell Soil Color Charts with Genuine
Munsell Color Chips. (2009). Year Revised, 2009, Munsell
Color 4300 44th Street, Grand Rapids, MI 49512, USA.
Page 17 of 18 128
Newman, W. L. (1962). Distribution of elements in sedimentary
rocks of the Colorado plateau- a preliminary report. United
States Geological Survey Bulletin 1107-F.
Palau Estevan, C. V., Arregui De La Cruz, F., & Carlos Alberola,
M. D. M. (2012). Burst detection in water networks using
principal component analysis. Journal of Water Resources
Planning and Management, 138(1), 47–54. https://doi.
org/10.1061/(ASCE)WR.1943-5452.0000147.
Paschke, S., Kimball, B. A., & Runkel, R. L. (2005).
Quantification and Simulation of Metal Loading to the
Upper Animas River, Eureka to Silverton, San Juan
County, Colorado, September 1997 and August 1998.
United States Geological Survey, Scientific Investigations
Report 2005–5054.
Potter, L. D., & Drake, C. L. (1989). Lake Powell. Virgin flow to
dynamo. Albuquerque: University of New Mexico Press.
R Core Team. (2013). R: a language and environment for statis-
tical computing. Vienna: R Foundation for Statistical
Computing.
Riba, I., DelValls, T. A., Forja, J. M., & Gomez-Parra, A. (2002).
Influence of the Aznaclollar mining spill on the vertical distri-
bution of heavy metals in sediments from the Guadalquivir
estuary (SW Spain). Marine Pollution Bulletin, 44, 39–47.
Rodriguez-Ferire, L., Avasarala, S., Abdul-Mehdi, S. A., Agnew,
D., Hoover, J. H., Artyushkova, K., Latta, D. E., Peterson, E.
J., Lewis, J., Crossey, L. J., Brearley, A. J., & Cerrato, J. M.
(2016). Post gold king mine spill investigation of metal stabil-
ity in water and sediments of the Animas River watershed.
Environmental Science and Technology, 50(2), 11539–11548.
Runkel, R. L., Crawford, C. G., & Cohn, T. A. (2004). Load
estimator (LOADEST): a FORTRAN program for estimating
constituent loads in streams and rivers: U.S. Geological
Survey techniques and methods book 4, chapter A5, 69.
Runkel, R. L., Kimball, B. A., Walton-Day, K., Verplanck, P. L., &
Broshears, R. E. (2012). Evaluating remedial alternatives for
an acid mine drainage stream: a model post audit.
Environmental Science and Technology, 46(1), 340–347.
https://doi.org/10.1021/es2038504.
Runkel, R. L., Verplanck, P. L., Kimball, B. A., & Walton-Day, K.
(2018). Cinnamon Gulch revisited: another look at separating
natural and mining-impacted contributions to instream metal
load. Applied Geochemistry, 95, 206–217. https://doi.
org/10.1016/j.apgeochem.2018.04.010.
Stanford, J. A., & Ward, J. V. (1991). Limnology of Lake Powell
and the chemistry of the Colorado River. In Colorado River
ecology and dam management: proceedings of a symposium
May 24–25, 1990 (p. 75). Santa Fe: New Mexico National
Academies Press.
Sullivan, K., Cyterski, M., Knightes, C., Kraemer, S.R.,
Washington, J., Preto, L., & Avant, B. (2017). Analysis of
the transport and fate of metals released from the Gold King
Mine in the Animas and San Juan Rivers. U.S.
Environmental Protection Agency. EPA/600/R-16/296.
Thompson, K., & Mundorff, J. C. (1982). Characteristics of
suspended sediment in the San Juan River near bluff (pp.
82–4104). Utah: United States Geological Survey. Water
Research Investigations.
U.S. Bureau of Reclamation (BOR). (2015). Technical Evaluation of
the Gold King Mine Incident. San Juan County, Colorado.
Accessed 2016 at http://www.usbr.gov/docs/goldkingminereport.
pdf. pp 132.
 128 Page 18 of 18
U.S. Bureau of Reclamation (BOR). (2019). About us. In
Accessed 2019 at https://www.usbr.gov/main/about/fact.
html.
U.S. Department of Energy (DOE). (2011). Natural contamination
from the Mancos Shale. Prepared by the Environmental
Sciences Laboratory Legacy Management, April 2011. In
LMS/S07480 ESL-RPT-2011-01.
U.S. Environmental Protection Agency (EPA). (2016). One year
after the Goldin King Mine Incident: a retrospective of EPA’s
efforts to restore and protect impacted communities.
U.S. Environmental Protection Agency (EPA). (2017). Storage
and retrieval database. Water quality data. National Water
Quality Monitoring Council. Accessed at https://www.
waterqualitydata.us/portal/
U.S. Environmental Protection Agency (EPA). Bonita Peak
Mining District Unincorporated, CO. Superfund Site.
Accessed 2018 at https://cumulis.epa.
gov/supercpad/SiteProfiles/index.cfm?fuseaction=second.
Cleanup&id=0802497#bkground
U.S. Geological Survey (USGS). (2016). Determining water-
quality trends using sediment cores, White Rock Lake,
Dallas. United States Geological Survey. Accessed at
https://pubs.usgs.gov/circ/circ1171/html/cores.htm
U.S. Geological Survey (USGS). Method 19–42 Element induc-
tively coupled plasma – atomic emissions spectrometry –
Water Air Soil Pollut (2019) 230:128
mass spectrometry (ICP-AES-MS), multi-acid (ICP-42).
United States Geological Survey. Accessed 2018 at
https://minerals.usgs.gov/science/analytical-
chemistry/method19.html
Utah Division of Water Quality (UDWQ), Utah Department of
Health, United States Geological Survey. (2018). Integration
of long-term impacts from historical mine drainage and epi-
sodic release of metals to sediments along the San Juan River
and Lake Powell System in Utah. Accessed at: https://deq.
utah.gov/legacy/topics/water-quality/gold-king-mine/index.
htm
Vernieu, W. S. (1997). Effects of reservoir drawdown on resus-
pension of deltaic sediments in Lake Powell. Journal of Lake
and Reservoir Management, 13(1), 67–78.
Wang, Y., Yang, L., Kong, L., Liu, E., Wang, L., & Zhu, J. (2015).
Spatial distribution, ecological risk assessment and source
identification for heavy metals in surface sediments from
Dongping Lake, Shandong, East China. Catena, 125, 200–
205.
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional
affiliations.