114

Application of Multiple Tracers to Characterize Complex Sediment and

Pathogen Transport in Karst
Tiong Ee Ting1, Ralph K. Davis2, J. V. Brahana2, P.D. Hays2,3, and Greg Thoma1
1

Department of Chemical Engineering, 3202 Bell, University of Arkansas, Fayetteville, AR 72701

Department of Geosciences, 113 Ozark, University of Arkansas, Fayetteville, AR 72701
3
US Department of Agriculture, National Water Management Center, 44 Ozark, Fayetteville, AR 72701
2

ABSTRACT
Injections of multiple tracers were conduced to characterize ground-water flow, sediment transport, and
E. coli transport in a mantled-karst aquifer under variable flow conditions at the Savoy Experimental Watershed. Rhodamine WT and fluorescein, used as conservative tracers in this study, were injected two hours
after the injection of lanthanum-labeled clay and europium-labeled E.coli into a losing-stream reach under
a natural hydraulic gradient. The injection occurred on the recessional limb of a major storm pulse, and fate
and transport of the tracers were observed for multiple tests under varying hydrologic conditions using multiple tracers for two springs of an underflow/overflow spring complex. The underflow spring, Langle, is
located approximately 490 meters in a straight-line direction from the injection point, in a different surfacewater catchment than the losing stream. The major overflow spring, Copperhead, is 453 meters in a straightline from the injection point, and it lies in the same surface-water catchment as the losing stream. The altitude of the resurgence of Langle Spring is about 3 centimeters less than the resurgence of Copperhead
Spring, based on multiple surveys using a total station.
Results from the tracer breakthrough for near steady-state conditions showed the arrival of suspended
sediment and E. coli at 10.7- and 5.9 hours respectively before the conservative dye tracers at Langle Spring.
The early arrival of sediment and E. coli is hypothesized to result from gravitational settling velocity coupled with the effect of pore-size exclusion. The conservative dye tracers arrived first at Copperhead Spring,
followed by E. coli and sediment, essentially a reversal of the sequence at Langle Spring. During later
storm-induced tracer tests, all tracers were observed to arrive simultaneously at each spring, with Copperhead Spring, along the shorter flow path, receiving the tracer pulses about an hour before Langle Spring.
This and other tracer tests in this overflow/underflow system suggest that sediment and E. coli are stored in
pools in the subsurface. These pools provide continuous full-conduit flow to Langle, the underflow spring,
and only partially-full conduit flow to Copperhead, the overflow spring. However, during high flows associated with transient storm events, the tracers are flushed from ephemeral storage in the pools and move as
a pulse associated with the rising limb of the hydrograph. The application of multiple tracers proved to be
an invaluable tool in providing mechanisms to fully characterize the subsurface flow.

115

Estimating Ground-Water Age Distribution from CFC and Tritium Data

in the Madison Aquifer, Black Hills, South Dakota
By Andrew J. Long and Larry D. Putnam
U.S. Geological Survey, 1608 Mountain View Rd, Rapid City, South Dakota 57702

Abstract
Ground-water age distribution was estimated for water collected from a well in the karstic Madison
aquifer in the Black Hills of South Dakota using a ground-water mixing model for chlorofluorocarbon
(CFC) and tritium data. Input functions for the model included precipitation concentrations for four tracersCFC12, CFC11, CFC113 (6-month data), and tritium (yearly data). Madison aquifer water often is a
complex mixture of waters of various ages; however, existing ground-water age-dating methods generally
are not well suited for estimating the unique age distributions of ground water that can occur in karst aquifers. CFC data alone often can provide estimates of piston-flow ages or binary mixtures of young and old
water, but generally are inadequate for estimating age distributions at a finer time discretization. However,
if a time series of tritium data is incorporated into an age-dating model along with CFC data, an age distribution discretized to a 6-month time step can be estimated with statistical significance by assuming that
ground-water age fits a probability density function (PDF). This method estimates one age distribution that
satisfies all of the combined tracer data and thus has two advantages. The first of which is that the number
of measured values applied to a single problem is maximized, which helps to constrain the solution, and
second is that confidence in the solution is increased if a single solution satisfies more than one type of data.
The PDF indicates the estimated fraction of water at a site for each 6-month age category. Because results
from multiple age-dating tracers should agree, and because together they may provide complimentary information, combining all of the data into one model can be a powerful method for describing the history of
recharge to a well or spring.
The best fit of CFC and tritium data for samples from a municipal water supply well open to the Madison aquifer was a bimodal age distribution, which was a composite of a uniform and a lognormal PDF. Data
used in the model included the concentrations for each of the 3 CFCs (1 sample) and a time series of tritium
concentrations (4 samples over 10 years). These samples provided a total of 7 tracer concentrations, which
were compared to the corresponding modeled values. Parameter optimization methods, which minimize the
residuals of measured and modeled values, were used to estimate the 4 parameters that describe the bimodal
age distribution. Because there were 7 measured tracer concentrations and only 4 parameters to be estimated, the solution was adequately constrained, and the parameters could be estimated with reasonable confidence. Results indicated that about 33 percent of the mixture was less than 2 years old (uniform PDF
component), 5 percent was 10 to 30 years old (lognormal PDF component), and the remaining 62 percent
was more than 50 years old. Because CFC and tritium concentrations in precipitation were very low before
1950, the age distribution of water more than about 50 years old could not be estimated. The bimodal age
distribution was the only distribution tested that could explain the combined CFC and the tritium data with
acceptable 95-percent confidence limits on the estimated parameter values.

116

A Multi-Tracer Approach for Evaluating the Transport of Whirling

Disease to Mammoth Creek Fish Hatchery Springs, Southwestern Utah
By Lawrence E. Spangler1, Meiping Tong2, and William Johnson2
1U.S. Geological Survey, 2329 Orton Circle, Salt Lake City, Utah 84119
2Department of Geology and Geophysics, University of Utah, Salt Lake City, Utah 84112

ABSTRACT
The Utah Division of Wildlife Resources has been concerned about the vulnerability of selected springfed fish hatcheries to whirling disease caused by the microscopic parasite Myxobolus cerebralis. Whirling
disease is typically transmitted from one water body to another by birds or fishermen but can potentially
migrate along underground flow paths in areas where aquifer permeability is high and ground-water movement is rapid enough to allow passage and survival of the parasite. Mammoth Creek Fish Hatchery in southwestern Utah tested positive for whirling disease in 2002. Because adjacent Mammoth Creek also tested
positive, a study was begun to evaluate potential hydrologic connections between the creek, an irrigation
canal off the creek, and the hatchery springs.
Dye-tracer studies indicate that water lost through the channel of Mammoth Creek discharges from the
west and east hatchery springs. Ground-water time of travel to the springs was about 7.5 hours, well within
the 2-week timeframe of viability of the parasite. Results of studies using soil bacteria and club moss spores
as surrogate particle tracers indicate that the potential for transport of the parasite through the fractured
basalt may be low. Bacteria concentrations in spring water generally were below reporting limits, and club
moss spores were recovered from only a few samples. However, peak concentrations for the bacteria and
club moss spores in water from the east hatchery spring coincided with peak dye recovery. No particle tracers were recovered from the west hatchery spring.

INTRODUCTION
The Utah Division of Wildlife Resources operates 10 fish hatcheries in Utah that use water from
large springs and has been concerned about the vulnerability of these hatcheries to whirling disease
caused by the microscopic parasite Myxobolus cerebralis. Whirling disease is typically transmitted
from one water body to another by birds or fishermen. However, the triactinomyxon spores (TAMs)
produced by the parasite can potentially migrate
along underground flow paths in areas where aquifer
permeability is high, such as in karst and volcanic
terrains, and the movement of ground water is sufficiently rapid to allow viable passage of the spores.
In 2000, whirling disease was detected in the
Midway Fish Hatchery, about 30 miles (mi) southeast of Salt Lake City. Results of investigations by
Carreon-Diazconti and others (2003) showed that
the likely source of the parasite in the spring water

supplying the hatchery was the Provo River. Water

diverted from the river, which also tested positive
for the disease, was used to irrigate farmland upgradient from the hatchery and subsequently moved
downward into the karst (travertine) aquifer supplying the springs. Use of cultured soil bacteria as a surrogate tracer for the parasite showed that transport of
the spore to the springs through open conduits and
fractures in the limestone was possible (Stephen
Nelson and Alan Mayo, Brigham Young University,
written commun., 2000).
In 2002, Mammoth Creek Fish Hatchery in
southwestern Utah became the second State- operated facility to become infected by whirling disease.
Because adjacent Mammoth Creek also tested positive, the U.S. Geological Survey, in cooperation
with the Utah Division of Wildlife Resources, began
a study to evaluate potential hydrologic connections
and determine ground-water travel times between
the creek, an irrigation canal off the creek, and the

117

hatchery springs, and to assess the potential for

transport of the parasite along underground flow
paths to the springs. This paper summarizes the
results of tracer studies.
DESCRIPTION OF STUDY AREA
Mammoth Creek State Fish Hatchery is located
about 2 mi southwest of Hatch, Utah, at the mouth of
Mammoth Creek Valley, at an altitude of 7,000 feet
(ft) (fig. 1). The hatchery is situated at the base of a
40-ft-high basalt cliff, from which two major (west
and east) springs discharge. Total discharge of the
springs averages about 3 cubic feet per second
(ft3/sec), with a variability of less than 1 ft3/sec.
Flow from the springs is diverted through the hatchery for fish-rearing operations and is then discharged
into Mammoth Creek, which flows past the hatchery. McCormick spring also discharges from near

the base of the basalt cliff about 750 ft northeast of

the hatchery springs, on private land (fig. 1). Discharge of this spring was about 50 gallons per
minute (gpm) during the study and appeared to be
fairly constant. Bonanza spring emerges from talus
alongside the channel of Mammoth Creek about
1,200 ft upstream from the hatchery (fig. 1) and discharged about 40 gpm. Discharge of this spring was
observed to vary with changes in streamflow in
Mammoth Creek.
During the summer, water is diverted from
Mammoth Creek into a canal about 2 mi west of the
hatchery (fig. 1) for irrigation in the lower part of the
valley. During the study, all water from the creek
was diverted into the canal and only a small amount
of inflow from springs was observed downstream in
the channel, which subsequently was lost through
the streambed (fig. 2).

Figure 1. Location of injection sites and springs and general directions of ground-water movement
in the Mammoth Creek study area, southwestern Utah.

118

Quaternary-age basaltic lava partly fills Mammoth Creek Valley and caps adjacent ridges. In the
vicinity of the hatchery, the basalt has been
entrenched by Mammoth Creek to a depth of as
much as 40 ft (fig. 2). Vertical and horizontal fracturing is pervasive throughout the basalt. Limestones, marls, and calcareous shales of the Tertiaryage Claron Formation underlie the basalt and adjacent hillsides and are locally cavernous.
METHODOLOGY
Major-ion chemistry, tritium age-dating,
streamflow measurements, spring discharge variability, and tracer studies were used to determine
hydrologic relations in the Mammoth Creek hatchery area. Fluorescent dyes (sodium fluorescein and
rhodamine WT) and sodium bromide were used to
establish ground-water connections between Mammoth Creek, an irrigation canal off the creek, and the
springs at, and in the vicinity of, the fish hatchery.
Automatic samplers collected water directly from
the springs for analysis. Dye samples were analyzed
by filter fluorometry (Wilson and others, 1986).
Sodium bromide samples were analyzed by ion
chromatography (Fishman and Friedman, 1989).
Non-pathogenic cultured soil bacteria (Acidovorax)
and club moss (Lycopodium) spores were used as
surrogate particle tracers to simulate the size (10 to
100 microns) and transport characteristics of the
whirling disease parasite through the fractured
basalt aquifer. Bacteria samples were collected manually in centrifuge vials, magnetically tagged, and
analyzed by ferrographic techniques (Johnson and
McIntosh, 2003). Club moss spores were collected
in plankton nets (fig. 3), isolated by filtration, and
analyzed by standard microscopic techniques (Gardner and Gray, 1976).

Figure 2. Mammoth Creek channel at the Sartini tracerinjection site, looking downstream. Surface water seeps
into the streambed and appears to move along fractures
within the basalt to the springs.

RESULTS AND DISCUSSION

On the basis of dye-tracer tests completed in
October 2002 and October 2003 (table 1), water lost
through the channel of Mammoth Creek about 3,000
ft southwest of the hatchery (at Sartini) discharges
from the west and east hatchery springs and from
McCormick spring (fig. 1). However, water lost
through the channel farther downstream appears

Figure 3. Plankton nets were used to collect club moss

spores from the west and east hatchery springs. Spores
were used as surrogate tracers (33 microns) for the whirling disease parasite.

Table 1. Summary of tracer injections in the Mammoth Creek study area, southwestern Utah.
[g, grams; kg, kilograms; , no data]

Tracer-injection
site

Date-time of
tracer injection

Type of tracer

Mammoth Creek

10/02/02

1300

Rhodamine WT
dye

Mammoth Creek at
Sartini

10/12/02

1300

Fluorescein dye

Mammoth Creek at
Sartini

Mammoth Creek
canal
Mammoth Creek
canal

10/09/03

1630

Rhodamine WT
dye

Amount
of
tracer
1 liter

Tracer-recovery
site

Date-time of
tracer recovery
(first arrival)

Travel
time to
first
arrival
(hours)
1
7.25

Linear
distance
(feet)

Bonanza spring

10/02/02

2015

Hatchery springs
(combined)

10/12/02

2300

McCormick spring

10/13/02

1100

West hatchery spring

10/10/03

0000

7.5

2,800

East hatchery spring

10/10/03

0100

8.5

3,000

McCormick spring

10/10/03

1045

18.25

3,300

1014 cells

East hatchery spring

10/10/03

0700

14.75

3,000

East hatchery spring

10/10/03

1205

19.25

3,000

454 g

1 liter

750

10

2,800

22

3,300

10/09/03

1615

Bacteria
(OY-107 strain)

10/09/03

1645

Club moss spores

1 kg

10/10/02

1340

Sodium bromide

25 kg

No recovery

10/11/02

1720

Sodium bromide

25 kg

No recovery

07/31/03

2200

Fluorescein dye

1.36 kg

McCormick spring

08/19/03

1700

451

(7)

Samples collected downstream of spring; maximum travel time.

Samples collected daily; maximum travel time.
3
Samples collected twice daily; maximum travel time.
4
Recovered near peak dye concentration; maximum travel time.
5
Represents composite sample over previous 13.5 hours.
6
Dye recovered on activated charcoal; maximum travel time.
7
Exact location of loss zone along canal unknown.
2

119

119
119

119
119

120

to discharge only from Bonanza spring. Groundwater travel time (first arrival) from Mammoth
Creek (at Sartini) to the west hatchery spring was
about 7.5 hours with a lag of about 1 hour between
the west and east springs (fig. 4). Time to peak dye
concentration (about 7 parts per billion) occurred
about 8 hours after first arrival. Total dye-mass
recovery for both springs was about 22 percent of
that injected.
Ground-water movement from Mammoth
Creek to the hatchery springs appears to be along
flow path(s) that are separate from those to Bonanza
spring and are probably related to fracturing within
the basalt. However, because water from the
hatchery springs and McCormick spring discharges
from multiple outlets along the same horizon, flow
appears to be, at least in part, along lateral zones of
high permeability within the basalt. These zones
could include horizontal fractures, interflow
horizons between successive lava flows, or possibly
the contact between the base of the basalt and the
original valley floor.
Although pathways of rapid ground-water flow
exist between the losing reach along Mammoth
Creek and the hatchery springs, low variability in

spring flow indicates that this is probably a small

component of total discharge and that average
ground-water travel time within the aquifer is likely
to be considerably longer. The concentration of
tritium (15.4 picocuries per liter) in water from the
west hatchery spring indicates, however, a
substantial component of modern (post-1960s)
water.
Results of dye-tracer studies indicate that
ground-water time of travel between Mammoth
Creek and the west and east hatchery springs is well
within the 2-week timeframe of viability of the
whirling disease parasite. However, results of
studies using bacteria and club moss spores as
surrogate tracers to simulate the size and movement
of the parasite underground indicate that the
potential for transport of the parasite through the
fractured basalt aquifer from the creek may be low.
Bacteria concentrations in water samples from the
springs generally were below reporting limits (less
than 10 cells per milliliter), and club moss spores
were recovered from only a few samples.
Substantial losses of the particle tracers probably
occurred during infiltration through the streambed
sediments and during transport within the aquifer.

Dye concentration (parts per billion)

8
7
6
5

West sprin
g
4

Peak dye recovery (9:00 am)

Club moss spore recovery (12:05 pm)

3
2
1

Bacteria recovery
(7:00 am)

Tracer injectio
n
(4:30 pm)

East spring

0
10/8/03 10/9/03 10/10/0310/11/03 10/12/0310/13/0310/14/0310/15/0310/16/0310/17/0310/18/03

Figure 4. Rhodamine WT dye-recovery curves, and bacteria and club moss spore peak recoveries for
the east hatchery spring. No particle tracers were recovered from the west hatchery spring.

121

Although the vast majority of particle tracers were

not recovered, peak concentrations for the bacteria
(about 10 cells per milliliter) and club moss spores
(about 60 spores per milliliter) in water from the east
hatchery spring coincided with peak dye recovery
(fig. 4). No particle tracers were recovered from the
west hatchery spring.
Streamflow measurements along the irrigation
canal off Mammoth Creek showed substantial losses
along selected reaches, particularly in the upper part
of the canal (fig. 1). Measured streamflow losses
along a 2-mi reach below the diversion were as
much as 2 ft3/sec, or about 22 percent of the flow.
Bromide and dye tracers injected in the canal just
below the diversion in October 2002 and July 2003,
respectively, were not detected at the hatchery
springs, but dye was detected at McCormick spring
(table 1). Non-detection of the tracers at the hatchery
springs probably resulted from dispersion and dilution within the matrix of the basalt aquifer, resulting
in ground-water travel times greater than the 6-week
monitoring period and (or) tracer concentrations
below the detection limits. Although water lost
along the upper reaches of the canal probably discharges at the hatchery springs, ground-water travel
times likely exceed the timeframe of viability for
transport of the parasite through the basalt.
SUMMARY
Dye-tracer studies at the Mammoth Creek Fish
Hatchery indicate that water lost through the channel
of Mammoth Creek discharges from the west and
east hatchery springs. Ground-water time of travel to
the springs was about 7.5 hours, well within the 2week timeframe of viability of the whirling disease
parasite. However, results of studies using soil bacteria (Acidovorax) and club moss (Lycopodium)
spores as surrogate particle tracers for the parasite
indicate that the potential for transport through the
fractured basalt from the creek may be low. Substantial losses of the particle tracers occurred during
streambed infiltration and aquifer transport. Bacteria
concentrations generally were below reporting limits and club moss spores were recovered from only a
few samples. However, peak concentrations for the
bacteria and club moss spores in water from the east
hatchery spring coincided with peak dye recovery.

No particle tracers were recovered from the west

hatchery spring. In addition, bromide and dye tracers
injected in an irrigation canal were not detected at
the hatchery springs.
REFERENCES
Carreon-Diazconti, C., Nelson, S.T., Mayo, A.L., Tingey,
D.G., and Smith, M., 2003, A mixed groundwater
system at Midway, Utah: Discriminating superimposed
local and regional discharge: Journal of Hydrology, v.
273, p. 119-138.
Fishman, M.J., and Friedman, L.C., 1989, Methods for
determination of inorganic substances in water and fluvial sediments: U.S. Geological Survey Techniques of
Water-Resources Investigations, Book 5, Chap. A1,
545 p.
Gardner, G.D., and Gray, R.E., 1976, Tracing subsurface
flow in karst regions using artificially colored spores:
Bulletin of the Association of Engineering Geologists,
v. 13, no. 3, p. 177-197.
Johnson, W.P., and McIntosh, W.O., 2003, Tracking of
injected and resident (previously injected) bacterial
cells in groundwater using ferrographic capture: Journal of Microbiological Methods, v. 54, p. 153-164.
Wilson, J.F., Jr., Cobb, E.D., and Kilpatrick, F.A., 1986,
Fluorometric procedures for dye tracing: U.S. Geological Survey Techniques of Water-Resources Investigations, Book 3, Chap. A12, 34 p.

122

National Evaporite Karst--Some Western Examples

By Jack B. Epstein
U.S. Geological Survey, National Center, MS 926A, Reston, VA 20192

ABSTRACT
Evaporite deposits, such as gypsum, anhydrite, and rock salt, underlie about one-third of the United
States, but are not necessarily exposed at the surface. In the humid eastern United States, evaporites exposed
at the surface are rapidly removed by solution. However, in the semi-arid and arid western part of the United
States, karstic features, including sinkholes, springs, joint enlargement, intrastratal collapse breccia, breccia
pipes, and caves, locally are abundant in evaporites. Gypsum and anhydrite are much more soluble than
carbonate rocks, especially where they are associated with dolomite undergoing dedolomitization, a process
which results in ground water that is continuously undersaturated with respect to gypsum. Dissolution of
the host evaporites cause collapse in overlying non-soluble rocks, including intrastratal collapse breccia,
breccia pipes, and sinkholes. The differences between karst in carbonate and evaporite rocks in the humid
eastern United States and the semi-arid to arid western United States are delimited approximately by a zone
of mean annual precipitation of 32 inches. Each of these two rock groups behaves differently in the humid
eastern United States and the semi-arid to arid west. Low ground-water tables and decreased ground water
circulation in the west retards carbonate dissolution and development of karst. In contrast, dissolution of
sulphate rocks is more active under semi-arid to arid conditions. The generally thicker soils in humid climates provide the carbonic acid necessary for carbonate dissolution. Gypsum and anhydrite, in contrast, are
soluble in pure water lacking organic acids. Examples of western karst include the Black Hills of South
Dakota and the Holbrook Basin in Arizona. A draft national map of evaporite karst is presented here.

INTRODUCTION
The present, the map indicating engineering
aspects of karst (Davies and others, 1984, scale
1:7,500,000) adequately shows the distribution of
carbonate karst in the United States, but the widespread distribution of evaporite karst is inadequately
portrayed. The map depicts areas of karstic rocks
(limestone, dolomite, and evaporites), and
pseudokarst, classified as to their engineering and
geologic characteristics (size and depth of voids,
depth of overburden, rock/soil interface conditions,
and geologic structure).
In the eastern United States, where average
annual precipitation commonly is greater than 30
inches, gypsum deposits generally are eroded or dissolved to depths of at least several meters or tens of
meters below the land surface. So, although gypsum
in the east may locally be karstic, the lack of exposures makes it difficult to prove this without

subsurface study of the gypsum and its dissolution

features. In the semi-arid western part of the United
States, however, in areas where the average annual
precipitation commonly is less than about 32 inches,
gypsum tends to resist erosion and typically caps
ridges, mesas, and buttes. In spite of its resistance to
erosion in the west, gypsum commonly contains visible karst features, such as cavities, caves, and sinkholes, attesting to the importance of ground-water
movement, even in low-rainfall areas. Salt karst is
less common at the earths surface than gypsum
karst because it is so soluble that it survives at the
surface only in very arid areas.
While the distribution of carbonate karst on the
Davies map is generally adequate, the map only
depicts gypsum karst in a few areas (fig. 1). In an
extensive text on the back of the map, caves and fissures in gypsum in western Oklahoma and the eastern part of the Texas Panhandle is mentioned. The

123

Figure 1. Map showing areas of evaporite karst in the United States, as depicted
by Davies and others (1984).

map does not show the distribution of salt or salt

karst even though the text mentions natural subsidence and man-induced subsidence due to solution
mining in salt beds in south-central and southwestern Kansas.
Several national maps of evaporite deposits
were summarized by Epstein and Johnson (2003)
who prepared a map showing the present perception
of evaporite distribution and evaporite karst in the
United States (fig. 2). The map includes gypsum/
anhydrite and halite basins, and incorporates the
limited areas of evaporite karst depicted by Davies
and others (1984) compared to the larger areas of the
same shown by Johnson (1997). Collapse due to
human activities, such as solution mining, are also
shown, as well as a line of mean annual precipitation
(32.5 in.) that approximates the boundary between
distinctively different karst characteristics, between
the humid eastern United States and the semi-arid
west. Also shown are the Holbrook Basin in Arizona and the Black Hills of South Dakota and

Wyoming, whose variety of surface and subsurface

evaporite-karst features are described here.
HOLBROOK BASIN, ARIZONA
Many workers have reported a variety of
evaporite- and carbonate-karst features in Arizona
(fig. 3A) that are not found on Davies (1984) map
(fig. 3B). Subsurface halite deposits were mapped
by Eaton and others (1972), Johnson and Gonzales
(1978), Ege (1985), and Neal and others (1998);
more detailed mapping of salt deposits in the Holbrook Basin was done by Peirce and Gerrard (1966)
and Rauzi (2000). An area of breccia pipes was
delimited in northwest Arizona by Harris (2002);
they were probably the result of collapse over carbonate rocks, but evaporite collapse could not be
ruled out. Scattered gypsum and anhydrite localities
were shown by Withington (1962). Comparing this
composite map with the Davies map (fig. 3B) shows
that many types of karstic features could be shown
in both evaporite and carbonate rocks in Arizona.

124

Figure 2. Distribution of outcropping and subsurface evaporite rocks in the United States
and areas of reported evaporite karst (from Epstein and Johnson, 2003). The 32.5" meanannual-precipitation line approximates a diffuse boundary between eastern and western
karst.

The Holbrook Basin in east-central Arizona

demonstrates that dissolution of deeply buried
evaporites can cause subsidence of overlying nonsoluble rocks. The basin is more than 100 miles
wide and contains an aggregate of about 1,000 feet
of salt, anhydrite, and sylvite interbedded with clastic red beds in the Permian Sedona Group (formerly
the Supai Group) (Peirce and Gerrard, 1966; Neal
and others, 1998; Rauzi, 2000). The top of the salt is
between 600 and 2,500 ft below the surface (Mytton,
1973). These workers describe the removal of
evaporites at depth along a northwest-migrating dissolution front, causing the development of presently
active collapse structures in the overlying Coconino
Sandstone and Moenkopi Formation.
For example, in the area about 10 mi northwest
of Snowflake, AZ, the Coconino and other rocks dip
monoclinally southward along the Holbrook anticline towards a large depression enclosing a dry

lake. The depression is the result of subsidence due

to evaporite removal. Collapse extends upwards
from the salt, forming a network of spectacular sinkholes in the overlying Coconino Sandstone (fig. 4,
5A) (Neal and others, 1998; Harris, 2002). Draping
of the Coconino has caused opening of extensive
tension fissures, some of which are many tens of feet
deep (Neal and others, 1998; Harris, 2002) (Fig. 5B).
If the definition of karst is allowed to include subsidence structures due to the dissolution and
removal of soluble rocks below and extending
upwards into non-soluble rocks, then a separate map
category may be needed to delineate these rocks.
Thus, any karst map must show non-soluble rocks
whose collapse structures are the result of dissolution of evaporite rocks below. A somewhat similar
situation prevails in the Black Hills of South Dakota
and Wyoming and was alluded to at Stops 3 and 4 of
the Southern Field Trip (Epstein, Agenbroad, and
others, this volume, 2005).

125

Figure 3. Maps comparing types and distribution of karst features in Arizona: A) distribution
of evaporite and carbonate karst as presented by various authors; B) distribution of carbonate karst (no evaporite karst was shown) as presented by Davies and others (1984).

BLACK HILLS, WYOMING AND SOUTH

DAKOTA
In the semi-arid Black Hills of Wyoming and
South Dakota, significant deposits of gypsum and
lesser anhydrite are exposed at the surface in several
stratigraphic units (Table 1, Epstein and Putnam,
this volume), including the Minnelusa, Spearfish,
and Gypsum Spring Formations. The field guides
that accompany this volume document many of the
karst features found in these rocks. The outcrop pattern of sedimentary units in the Black Hills is controlled by erosion on an irregular domal uplift about
130 mi long and 60 mi wide. The central core of
Precambrian rock is surrounded by four zones of
sedimentary rock with contrasting lithologies and
differing karst features. These are, from the center
(oldest) outwards: (1) The limestone plateau, made
up of Cambrian to Pennsylvanian limestone, dolo-

mite, and silici-clastic rocks, and containing worldclass caves such as Wind and Jewel Caves in the
Pahasapa Limestone. Overlying these limestones is
the Minnelusa Formation, which contains as much
as 235 ft of anhydrite in its upper half in the subsurface. This anhydrite has been dissolved at depth,
producing a variety of dissolution structures (Stop 1,
Epstein, Agenbroad, and others, this volume; and
Stop 4, Epstein, this volume). (2) The Red Valley,
predominantly underlain by red beds of the
Spearfish Formation of Triassic and Permian age
and containing several gypsum beds totaling more
than 75 ft thick in places. Dissolution of these
evaporites, and those in underlying rocks, has produced shallow depressions and sinkholes, some of
which are more than 50 ft deep (Stops 8 and 9, of
Epstein, Davis, and others, 2005, this volume). The
Gypsum Spring Formation, which overlies the
Spearfish, contains a gypsum unit, as much as

126

15 feet thick, that has developed abundant sinkholes

in places. (3) The "Dakota" hogback, held up by
resistant Sandstone of the Inyan Kara Group of Cretaceous age, and underlain by shales and sandstones
of the Sundance and Morrison Formations. Collapse
structures, such as breccia pipes and sinkholes,
extend up through some of these rocks from underlying soluble rocks, probably in the Minnelusa Formation. (4) Impure limestone and shale extending
outward beyond the hogback, some of which are
shown as karstic units on the map of Davies and others (1984), but such features are unknown in those
rocks.
Figure 6 compares the map of limestone outcrops shown by Davies and others (1984) for the
Black Hills of South Dakota with the more detailed
categories that are proposed here, including carbonate and evaporite karst, intrastratal karst, and nonsoluble rocks with collapse features due to dissolution of other rock units at depth. This characterization may also be suitable for other areas of the
western United States.
Anhydrite in the Minnelusa is generally not
seen in surface outcrops. It has been, and continues
to be, dissolved at depth, forming collapse breccias,
breccia pipes, and sinkholes that extend upwards
more than 1,000 ft into overlying units. Mapping
this intrastratal karst is fairly easy, because the outcrop distribution of the Minnelusa is well known. At
depth, brecciation of the upper part of the Minnelusa
has developed significant porosity, resulting in an
important aquifer. Ground water migrates along
breccia pipes that extend upwards through overlying
formations. A migrating dissolution front (Epstein,
2001; 2003) similar to the situation reported for the
Holbrook Basin in Arizona, is summarized below.
Dissolution Front in the Minnelusa Formation
The upper half of the Minnelusa Formation contains abundant anhydrite in the subsurface, and
except for a few areas near Beulah and Sundance,
Wyoming (Brady, 1931), and in Hell Canyon in the
southwestern Black Hills (Braddock, 1963), no

anhydrite or gypsum crops out. A log of the upper

part of the Minnelusa from Hell Canyon contains
235 ft (72 m) of anhydrite and gypsum (Braddock,
1963; Brobst and Epstein, 1963). Where anhydrite
is present in the Minnelusa, its rocks are not brecciated or only slightly so. Where the rocks are brecciated in outcrop, anhydrite is absent. Clearly, the
brecciation is the result of collapse following subsurface dissolution of anhydrite.
The Madison and Minnelusa are the major aquifers in the Black Hills. They are recharged by rainfall on and by streams flowing across their up-dip
outcrop area. In the Minnelusa, removal of anhydrite progresses downdip with continued dissolution
of the anhydrite (fig. 7), collapse breccia is formed,
breccia pipes extend upwards, and resurgent springs
develop at the sites of sinkholes. Cox Lake, Mud
Lake, Mirror Lake, and McNenny Springs (Stop 8,
Epstein, Davis, and others, this volume), are near the
position of the dissolution front. As the Black Hills
is slowly lowered by erosion, the anhydrite dissolution front in the subsurface Minnelusa moves downdip and radially away from the center of the uplift.
The resurgent springs will dry up and new ones will
form down dip as the geomorphology of the Black
Hills evolves. Abandoned sinkholes on canyon
walls (see figure 8, Epstein, Davis, and others, this
volume) attest to the former position of the dissolution front.
Because ground water has dissolved the anhydrite in the Minnelusa in most areas of exposure, and
because anhydrite is present in the subsurface, a
transition zone should be present where dissolution
of anhydrite is currently taking place. A model of
this zone has been presented by Brobst and Epstein
(1963, p. 335) and Gott and others (1974, p. 45) and
is shown here in figure 7. Consequences of this
model include (1) the up-dip part of the Minnelusa is
thinner than the downdip part because of removal of
significant thickness of anhydrite, (2) the upper part
of the Minnelusa should be continually collapsing,
even today, and (3) the properties of the water in this
transition zone may be different than elsewhere.

127

Figure 4. Topographic map of "The Sinks", a series of depressions in the Coconino

Sandstone, about 10 miles northwest of Snowflake, Arizona, in the Holbrook Basin.
Wide depressions are highlighted with a dashed line; the one in the southwest corner
of the map is about one mile long. Other depressions are as much as 100 feet deep.

128

Figure 5. Collapse structures in clastic rocks overlying the salt-bearing Sedona Group in the Holbrook Basin, 810 mi northwest of Snowflake, AZ. A) Steep-sided sinkhole in a hole-pocked area called "The Sinks," located in
the Coconino Sandstone. Note the variable amount of subsidence along major joints. B) Open tension fractures
in the Moekopi Formation caused by flexure of the Holbrook "anticline" (actually a monocline) due to dissolution
of salt at depth. Also see figures in Harris (2002).

If this process is correct, then present resurgent

springs should be eventually abandoned and new
springs should develop down the regional hydraulic
gradient of the Black Hills. One example might be
along Crow Creek where a cloud of sediment from
an upwelling spring lies 1,000 ft (300 m) north of
McNenny Springs (See figure 15, Stop 8, Northern
Field trip Guide). This circular area, about 200 ft
across, might eventually replace McNenny Springs.
Age of Brecciation
Solution of anhydrite in the Minnelusa probably
began soon after the Black Hills was uplifted in the
early Tertiary and continues today. Recent subsidence is evidenced by sinkholes opening up within
the last 20 years (See figure 26, Epstein, Davis and
others, this volume), collapse in water wells and natural springs resulting in sediment disruption and
contamination, fresh circular scarps surrounding
shallow depressions, and calcium sulphate and
sodium chloride issuing from spring water throughout the Black Hills.
The brecciation of the upper part of the Minnelusa formation occurred after the up-dip portion of
the Minnelusa was breached following uplift.
Ground water was then able to penetrate the impermeable layers overlying the Minnelusa and the
anhydrite was dissolved. Darton and Paige (1925)
found "older terrace deposits" with Oligocene fossils on the Minnelusa, indicating that the breaching

of the Minnelusa occurred before the Oligocene and

after the Late Cretaceous uplift.
An earlier alternative explanation for the cause
and timing of brecciation was given by Bates (1955)
who believed the brecciation occurred almost concurrently with the deposition of the Minnelusa when
ground water converted the anhydrite to gypsum and
the resulting expansion heaved and shattered the surrounding strata leaving jumbled blocks that were
reworked by the sea. However, field evidence supports the conclusion that brecciation occurred after,
not during, the deposition of the Minnelusa. Disruption of bedding in the Minnelusa and in higher
stratigraphic units becomes less intense upwards
from the zone of anhydrite removal in the Minnelusa
(Stop 1, Epstein, Agenbroad, and others, 2005, this
volume). Subsidence effects in the overlying formations also become less dramatic. The resistant, thin,
Minnekahta Limestone, lying between the red beds
of the Opeche Shale and Spearfish Formation, contains few collapse features; some sinkholes penetrate the entire thickness of the Minnekahta and are
therefore result of collapse from below (Stop 3,
Epstein, Davis, and others, this volume). The most
significant effect on the Minnekahta is the undulations seen in outcrop everywhere in the Black Hills.
Breccia pipes and sinkholes are known as high as the
Lakota Formation, about 1,000 feet above the
Minnelusa.

129

Figure 6. Maps comparing depiction of karst areas in the Black Hills of South Dakota: A) carbonate-karst units, as
presented by Davies and others (1984); B) carbonate- and evaporite-karst units, as herein proposed for a new national karst map.

There are several places in the Black hills where

bedding in the Minnelusa is moderately or steeply
dipping and intruded by breccia pipes. If the pipes
formed soon after deposition of Minnelusa sediments, they should be nearly right angles to bedding.
Such is not the case, as seen at mileage 59.3 in the
Southern Field trip Guide, where the dip of bedding
is moderate and shown in figure 8 where the dip is
steeper.
HUMID VERSUS SEMI-ARID KARST
Comparing the known locations of surface
evaporite karst with a map showing annual average
rainfall shows a striking relationship between precipitation and the occurrence of evaporite karst (fig.
2). Most occurrences of surficial karst in gypsum
shown in figure 2 lies west of a zone with annual
precipitation of about 32 inches (represented by the
32.5-inch isobar). Many of the karst areas shown in
figure 2 are due to dissolution at depth. In Michigan,
earlier studies suggest that the karstic collapse features there were formed soon after deposition of the
Devonian evaporites (Landes, 1945), but Black

(1997) showed that sinkhole development occurred

after the most recent glaciation.
The degree to which soluble rocks are dissolved
depends, in part, on the amount of rainfall and the
solubility of the rock. Sulphate-bearing rocks--gypsum and anhydrite--are perhaps 10-30 times more
soluble in water than carbonate rocks (Klimchouk,
1996). Both carbonates and sulphates behave differently in the humid eastern United States and the
semi-arid to arid west. Low ground-water tables and
decreased ground-water circulation in the west does
not favor rapid carbonate dissolution and development of karst. In contrast, sulphate rocks are dissolved much more readily and actively than
carbonate rocks, even under semi-arid to arid conditions. The presence of extensive karst in carbonates
in the west probably dates to a more humid history.
Additionally, the generally thicker soils in humid
climates provide the carbonic acid that enhances carbonate dissolution. Gypsum and anhydrite, in contrast, are more readily soluble in water that lacks
organic acids. This relationship suggests an interesting area for future study.

130

Figure 7. Dissolution of anhydrite in the Minnelusa Formation and down-dip migration of the dissolution
front.

In the eastern United States karstification

occurs by acid-charged water percolating downward, altering soluble rocks below. While this is
partly true in the west, much of the karst, such as in
the Black Hills, is produced by artesian water
migrating upward, affecting overlying rocks in a different manner than in the humid east. Thus, as
shown above, non-soluble rocks bear the imprint of
karst.
HUMAN-INDUCED KARST
It is well known that subsidence in karstic rocks
can be exacerbated by human activities. Lowering
of the water table by well-pumping or by draining of
quarries can reduce support of soils overlying sinkholes, thus causing their collapse. Subsurface min-

ing of salt and other evaporites may eventually cause

collapse of overlying rocks, such as at the Retsof
mine in Livingston County, NY (Nieto and Young,
1998; Gowan and Trader, 2003). Localities of subsidence due to solution mining were mapped by
Dunrud and Nevins (1981) and are shown in figure
2. The bibliographic list of ground subsidence due
to evaporite dissolution of Ege (1979) contains
many instances where such subsidence was due to
human activities. Knowing the location of shallow
and deep mines is important to local officials, in
order to understand the potential for such subsidence. For example, abandoned gypsum mines in
western New York are abundant, and recent settlement of many houses near Buffalo, New York,
partly may be the result of subsidence over these
mines.

131

Figure 8. Near-vertical breccia pipe (short dash) in steeply dipping beds (long
dash) of the Minnelusa Formation, Frannie Peak Canyon, Fanny Peak 7.5minute Quadrangle, six miles southeast of Newcastle, Wyoming, NW1/4 SE1/4,
T. 44 N., R. 60 W.

REFERENCES
Bates, R. L., 1955, Permo-Pennsylvanian formations
between Laramie Mountains, Wyoming, and Black
Hills, South Dakota: Am. Assoc. Petroleum Geologists
Bull., v. 39, p. 1979-2002.
Black, T.J., 1997, Evaporite karst of northern lower
Michigan: Carbonates and Evaporites, v. 12, p. 81-83.
Braddock, W.A., 1963, Geology of the Jewel Cave SW
Quadrangle, Custer County, South Dakota: U.S. Geological Survey Bulletin 1063-G, p. 217-268.
Brady, F. H., 1931, Minnelusa formation of Beulah district, northwestern Black Hills, Wyoming: Am. Assoc.
Petroleum Geologists Bull., v. 15, p. 183-188
Brobst, D.A., and Epstein, J.B., 1963, Geology of the
Fanny Peak quadrangle, Wyoming-South Dakota:
U.S. Geological Survey Bulletin 1063-I, p. 323-377.
Daly, Chris, and Taylor, George, 2000, United States
Average Annual Precipitation, 1961-1990: Spatial Climate Analysis Service, Oregon State University;
USDA -NRCS National Water and Climate Center,
Portland, Oregon; USDA -NRCS National Cartography and Geospatial Center, Fort Worth, Texas; Online
linkage: <http:/ /www.ftw.nrcs.usda.gov/prism/prismdata.html>

Darton, N.H., and Paige, Sidney, 1925, Description of the

central Black Hills, South Dakota: U.S. Geological
Survey Geologic Atlas, folio 219, 34 p.
Davies, W.E., Simpson, J.H., Ohlmacher, G.C., Kirk,
W.S., and Newton, E.G., 1984, Map showing engineering aspects of karst in the United States: Reston, Va.,
U.S. Geological Survey National Atlas of the United
States of America, scale 1:7,500,000.
Dunrud, C.R., and Nevins, B.B., 1981, Solution mining
and subsidence in evaporite rocks in the United States:
U.S. Geological Survey Miscellaneous Investigation
Series Map I-1298, 2 sheets.
Eaton, G.P., Peterson, D.L., and Schumann, H.H., 1972,
Geophysical, geohydrological, and geochemical reconnaissance of the Luke salt body, central Arizona: U.S.
Geological Survey Professional Paper 753, 28 p.
Ege, J.R., 1979, Selected bibliography on ground subsidence caused by dissolution and removal of salt and
other soluble evaporites: U.S. Geological Survey
Open-file Report 79-1133, 28 p.
Ege, J.R., 1985, Maps showing distribution, thickness,
and depth to salt deposits of the United States: U.S.
Geological Survey Open-File Report 85-28, 11 p.

132

Epstein, J.B., 2001, Hydrology, hazards, and geomorphic

development of gypsum karst in the northern Black
Hills, South Dakota and Wyoming: U.S. Geological
Survey, report WRI 01-4011, p. 30-37.
Epstein, J.B., 2003, Gypsum karst in the Black Hills,
South Dakota-Wyoming; Geomorphic development,
hazards, and hydrology: Oklahoma Geological Survey, report 109, p. 241-254.

Dakota and Wyoming: U.S. Geological Survey Professional Paper 763, 57 p.

Gowan, S. W., and Trader, S. M., 2003, Mechanism of
Sinkhole Formation in Glacial Sediments above the
Retsof Salt Mine, Western New York in, Johnson,
K.S., and Neal, J.T., eds, Evaporite karst and engineering/environmental problems in the United States:
Oklahoma Geological Survey Circular 109, p. 21-30.

Epstein, J.B., 2005, Field Trip Guide 3, Karst Field Trip

to the Western Black Hills: in, Kuniansky, E.L., editor,
U.S. Geological Survey, Karst Interest Group Proceedings, Rapid City, South Dakota, September 12-15,
2005, this volume: U.S. Geological Survey Scientific
Investigations Report Series 2005-5160, this volume.

Harris, R.C., 2002, A review and bibliography of karst

features of the Colorado Plateau, Arizona: Arizona
Geological Survey, Open-File Report 02-07, 43p.

Epstein, J.B., Agenbroad, Larry, Fahrenbach, Mark, Horrocks, R.D., Long, A.J., Putnam, L.D., Sawyer, J.F.,
and Thompson, K.M., 2005, Field Trip Guide 1, Karst
Field Trip to the Southern Black Hills: in, Kuniansky,
E.L., editor, U.S. Geological Survey, Karst Interest
Group Proceedings, Rapid City, South Dakota, September 12-15, 2005, this volume: U.S. Geological
Survey Scientific Investigations Report Series 20055160, this volume.

Johnson, K.S., and Gonzales, S., 1978, Salt deposits in the

United States and regional geologic characteristics
important for storage of radioactive waste; Prepared for
Union Carbide Corporation, Nuclear Division, Office
of Water Isolation, Y/OWI/SUB-7414/1, 188 p.

Epstein, J.B., Davis, A.D., Long, A.J., Putnam, L.D., and

Sawyer, J.F., 2005, Field Trip Guide 2, Karst Field Trip
to the Northern Black Hills: in, Kuniansky, E.L., editor, U.S. Geological Survey, Karst Interest Group Proceedings, Rapid City, South Dakota, September 12-15,
2005, this volume: : U.S. Geological Survey Scientific
Investigations Report Series 2005-5160, this volume.
Epstein, J.B., and Putnam, L.D., 2005, Introduction to
three field trip guides: karst features in the Black Hills,
Wyoming and South DakotaPrepared for the Karst
Interest Group Workshop, September 2005: in,
Kuniansky, E.L., editor, U.S. Geological Survey, Karst
Interest Group Proceedings, Rapid City, South Dakota,
September 12-15, 2005, this volume: U.S. Geological
Survey Scientific Investigations Report Series 20055160, this volume.
Epstein, J.B., and Johnson, K.S., 2003, The need for a
national evaporite-karst map, in, Johnson, K.S., and
Neal, J.T., eds, Evaporite karst and engineering/environmental problems in the United States: Oklahoma
Geological Survey Circular 109, p. 21-30.
Gott, G.B., Wolcott, D.E., and Bowles, C.G., 1974,
Stratigraphy of the Inyan Kara Group and localization
of uranium deposits, southern Black Hills, South

Johnson, K.S., 1997, Evaporite karst in the United States:

Carbonates and Evaporites, v. 12, p. 2-14.

Johnson, K. S., Gonzales, S., and Dean, W. E, 1989, Distribution and geologic characteristics of anhydrite
deposits in the United States, in Dean, W.E.; and
Johnson, K.S. (eds.), Anhydrite deposits of the United
States and characteristics of anhydrite important for
storage of radioactive wastes: U.S. Geological Survey
Professional Paper 1794, p. 990.
Klimchouk, Alexander, 1996, The dissolution and
conversion of gypsum and anhydrite, in Klimchouk,
Alexander; Lowe, David; Cooper, Anthony; and
Sauro, Ugo (eds.), Gypsum karst of the World: International Journal of Speleology, v. 25, nos. 3-4, p. 21-36.
Landes, K.K., 1945, The Mackinac Breccia: Michigan
Geological Survey Publication 44, p.121-154.
Laury, R.L., 1980, Paleoenvironment of a late Quaternary
mammoth-bearing sinkhole deposit, Hot Springs,
South Dakota: Geological Society of America Bulletin,
v. 91, p. 465-475
Mytton, J. W., 1973, Two salt structures in Arizona--the
Supai salt basin and the Luke salt body: U.S. Geological Survey Open-file report, 40 p.
Nieto, A.S., and Young, R. A., 1998, Retsof salt mine collapse and aquifer dewatering, Genesee Valley, Livingston County, New York, in Borchers, J. W. (ed.), Land
subsidence case studies and current research; Proceedings of the Dr. Joseph F. Poland Symposium on Land

133

Subsidence: Association of Engineering Geologists

Special Publication 8, p. 309-325.
Neal, J.T., Colpitts, R., and Johnson, K.S., 1998, Evaporite karst in the Holbrook Basin, Arizona: in, Borchers,
J.W., ed., Land subsidence case studies and current
research: Proceedings of the Dr. Joseph F. Poland
Symposium on Land Subsidence: Association of Engineering Geologists Special Publication No. 8, p.
373-384.
Peirce, H. W., and Gerrard, T. A., 1966, Evaporite deposits of the Permian Holbrook basin, Arizona, in Rau, J.
L., ed., Second symposium on salt, 1973: Cleveland,
Ohio, Northern Ohio Geological Society, v. 1, p. 1-10.
Rauzi, S. L., 2000, Permian salt in the Holbrook Basin,
Arizona: Arizona Geological Survey, Open-File
Report 00-03, 20 p.
Withington, C. F., 1962, Gypsum and anhydrite in the
United States, exclusive of Alaska and Hawaii: U.S.
Geological Survey Mineral Investigations Resource
Map MR-33.