EEG Journal Vol. XXVI, No. 3

EDITORIAL OFFICE: Environmental & Engineering Geoscience journal, Department of Geology, Kent State University, Kent, OH 44242, U.S.A. phone: 330-672-2968, fax: 330-672-7949, ashakoor@kent.edu.

CLAIMS: Claims for damaged or not received issues will be honored for 6 months from date of publication. AEG members should contact AEG, 3053 Nationwide Parkway, Brunswick, OH 44212 Phone: 844-331-7867. GSA members who are not members of AEG should contact the GSA Member Service center. All claims must be submitted in writing.

POSTMASTER: Send address changes to AEG, 3053 Nationwide Parkway, Brunswick, OH 44212. Phone: 844-331-7867.

Include both old and new addresses, with ZIP code. Canada agreement number PM40063731. Return undeliverable Canadian addresses to Station A P.O. Box 54, Windsor, ON N9A 6J5 Email: returnsil@imexpb.com.

DISCLAIMER NOTICE: Authors alone are responsible for views expressed in­­articles. Advertisers and their agencies are solely responsible for the content of all advertisements printed and also assume responsibility for any claims arising therefrom against the publisher. AEG and Environmental & Engineering Geoscience reserve the right to reject any advertising copy.

EDITORIAL BOARD Watts, Chester “Skip” F. Radford University, Chair Hasan, Syed University of Missouri, Kansas City Nandi, Arpita East Tennessee State University

Oommen, Thomas Michigan Technological University Sasowsky, Ira D. University of Akron

ASSOCIATE EDITORS Brankman, Charles, Consultant Boston MA Bruckno, Brian Virginia Department of Transportation Clague, John J. Simon Fraser University, Canada De Graff, Jerome V. California State University, Fresno Fryar, Alan University of Kentucky Hauser, Ernest Wright State University Hutchinson, Jean Queens University, Canada Keaton, Jeff AMEC Americas Marinos, Vassillis Aristotle University of Thessaloniki, Greece

SUBSCRIPTIONS:

McBride, John Brigham Young University Mwakanyamale, Kisa Illinois State Geological Survey Santi, Paul Colorado School of Mines Dee, Seth University of Nevada, Reno Shlemon, Roy R.J. Shlemon & Associates, Inc. Stephenson, William U.S. Geological Survey Stock, Greg National Park Service Sukop, Michael Florida International University Ulusay, Resat Hacettepe University, Turkey Watts, Chester F. “Skip,” Radford University West, Terry Purdue University

Environmental & Engineering Geoscience August 2020 VOLUME XXVI, NUMBER 3

SUBMISSION OF MANUSCRIPTS

Member subscriptions: AEG members automatically receive digital access to the journal as part of their AEG membership dues. Members may order print subscriptions for $75 per year.

GSA members who are not members of AEG may order for $60 per year on their annual GSA dues statement or by contacting GSA.

Environmental & Engineering Geoscience (E&EG), is a quarterly journal devoted to the publication of original papers that are of potential interest to hydrogeologists, environmental and engineering geologists, and geological engineers working in site selection, feasibility studies, investigations, design or construction of civil engineering projects or in waste management, groundwater, and related environmental fields. All papers are peer reviewed.

Nonmember subscriptions are $310 and may be ordered from the subscription department of either organization. A postage differential of $10 may apply to nonmember subscribers outside the United States, Canada, and Pan America. Contact AEG at 844-331-7867; contact GSA Subscription Services, Geological Society of America, P.O. Box 9140, Boulder, CO 80301.

The editors invite contributions concerning all aspects of environmental and engineering geology and related disciplines. Recent abstracts can be viewed under “Archive” at the web site, “http://eeg.geoscienceworld.org”.

Articles that report on research, case histories and new methods, and book reviews are welcome. Discussion papers, which are critiques of printed articles and are technical in nature, may be published with replies from the original author(s). Discussion papers and replies should be concise.

Single copies are $75.00 each. Requests for single copies should be sent to AEG, 3053 Nationwide Parkway, Brunswick, OH 44212.

To submit a manuscript go to https://www.editorialmanager.com/EEG/ default.aspx. If you have not used the system before, follow the link at the bottom of the page that says New users should register for an account. Choose your own login and password. Further instructions will be available upon logging into the system. Please carefully read the “Instructions for Authors”.

© 2020 by the Association of Environmental and Engineering Geologists

THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER EDITORS

Brian G. Katz Environmental Consultant Tallahassee, FL 32309 eegeditorbkatz@gmail.com

Authors do not pay any charge for color figures that are essential to the manuscript. Manuscripts of fewer than 10 pages may be published as Technical Notes.

For further information, you may contact Dr. Abdul Shakoor at the editorial office.

Cover photo Cover photo is of Lodi Marsh Spring in the Lodi Marsh State Natural Area in southern Wisconsin, courtesy of Susan Swanson. See article on page XX

Volume XXVI, Number 3, August 2020

All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage and retrieval system, without permission in writing from AEG.

Abdul Shakoor Department of Geology Kent State University Kent, OH 44242 330-672-2968 ashakoor@kent.edu

ENVIRONMENTAL & ENGINEERING GEOSCIENCE

Environmental & Engineering Geoscience (ISSN 1078-7275) is published quarterly by the Association of Environmental & Engineering Geologists (AEG) and the Geological Society of America (GSA). Periodicals postage paid at AEG, 3053 Nationwide Parkway, Brunswick, OH 44212 and additional mailing offices.

THE JOINT PUBLICATION OF THE ASSOCIATION OF ENVIRONMENTAL AND ENGINEERING GEOLOGISTS AND THE GEOLOGICAL SOCIETY OF AMERICA SERVING PROFESSIONALS IN ENGINEERING GEOLOGY, ENVIRONMENTAL GEOLOGY, AND HYDROGEOLOGY

Environmental & Engineering Geoscience Volume 26, Number 3, August 2020 Table of Contents 271

Foreword to the Environmental & Engineering Geoscience Journal Special Edition on Springs Abe Springer

273

Variability in Groundwater Flow and Chemistry in the Houzhai Karst Basin, Guizhou Province, China Joshua M. Barna, Alan E. Fryar, Le Cao, Benjamin J. Currens, Tao Peng and Chen Zhu

291

Discrimination of Nitrogen Sources in Karst Spring Contributing Areas using a Bayesian Isotope Mixing Model and Wastewater Tracers (Florida, USA) Andy Canion, Katherine M. Ransom, and Brian G. Katz

313

Influence of Desert Springs on Habitat of Endangered Zuni Bluehead Sucker (Catostomus discobolus yarrowi) Rebecca J. Frus, Laura J. Crossey, Clifford N. Dahm, Karl E. Karlstrom, and Livia Crowley

331

Using Reference Springs to Describe Expected Flow, Temperature, and Chemistry Conditions for Geologically Related Groups of Springs Susan K. Swanson, Grace E. Graham, and David J. Hart

345

Common Spring Types in the Valley and Ridge Province: There Is More than Karst Dorothy J. Vesper and Ellen K. Herman

359

Hydrology of a Southern Appalachian Hypocrene Spring-Fed Fen Jeffrey Wilcox, Emily Bradshaw Marino, Adam Warwick, and Megan Sutton

367

Geochemical Variability in Karst-Siliciclastic Aquifer Spring Discharge, Kaibab Plateau, Grand Canyon Alexander J. Wood, Abraham E. Springer, and Benjamin W. Tobin

EEGS-26-03-03_1XO

July 8, 2020

8:23

Foreword to the Environmental & Engineering Geoscience Journal Special Edition on Springs ABE SPRINGER* School of Earth and Sustainability, Northern Arizona University, P.O. Box 4099, Ashurst Building, Room A108, Flagstaff, AZ 86011

This special issue of Environmental & Engineering Geoscience (E&EG) was catalyzed by a special session titled “Springs: Groundwater-Influenced Ecosystems, Gaining Streams, and Wetlands” held at the 2018 Geological Society of America (GSA) annual meeting in Indianapolis, Indiana. Five of the papers in this special edition were presented at the GSA meeting, while two papers resulted from broader interest in the special issue. The papers provide unique and diverse perspectives on the hydrogeology and ecohydrology of springs from six different U.S. states and one basin in China. The studies cover a gradient from arid to humid regions and areas of low elevation to high elevation. Unique interactions of the physical geochemical systems with both plants and animals are documented carefully.

Although springs are ubiquitous on Earth and are essential for supporting people and ecosystems, to our knowledge, this is the first issue devoted to springs not only by a journal of the GSA and the Association of Environmental & Engineering Geologists but also by any geological journal. There have been recent ecological books (Botosaneanu, 1998; Stevens and Meretsky, 2006) and issues of ecological journals devoted to springs or books focused on the sub-category of karst springs (Stevanovic and Kresic, 2009) but not geology journals. This issue is timely, too, as we approach the 100th anniversary of Meinzer’s (1923) publication of the first widely used classification system of springs. Unfortunately, as documented by Springer and Stevens (2009), publications on springs by geologists since Meinzer’s (1923) bulletin focused mainly on large-magnitude or water supply springs, not on smaller and more widely distributed springs, which are important for ecosystems. Nor do most geologyrelated publications on springs include an analysis or interpretation of the interactions between the physical and the biological dimensions. Consistently, there have been sessions at the GSA annual meeting on springs, and this issue of E&EG is timely in bringing modern studies of the ecohydrology of springs to the geology literature.

* Corresponding author email: abe.springer@nau.edu

Swanson, Graham, and Hart provide an important summary of reference springs from their broader studies of springs across Wisconsin. They selected seven reference springs from six previously defined groups that represent more than 400 springs in Wisconsin.

The process they describe for developing reference springs can be broadly applicable to those studying springs in other regions.

Wilcox, Marino, Warwick, and Sutton provide a unique ecohydrologic study of a springs ecosystem managed through fire. This may be the first detailed publication for a hypocrene spring sphere of discharge and the first published study on the hydrogeology of a pitcher plant ecosystem. The description of the hydraulic response of the aquifer to prescribed burning is a unique and important contribution to the literature and broadly applicable to other spring ecosystems where fire is an important management tool.

Frus, Crossey, Dahm, Karlstrom, and Crowley document the hydrogeological and geochemical conditions necessary to support an endangered species of fish in New Mexico. The research is innovative and broadly applicable to understanding groundwater conditions necessary to support other desert fishes.

Vesper and Herman describe the importance of springs other than karst springs to the Valley and Ridge Province of the eastern United States. They provide new studies of ephemeral headwater springs and warm springs and their responses to structural geology and storm events. They make the case for why similar information should be collected at other springs to assist with management and decision making.

Wood, Tobin, and Springer incorporate unique geological and geochemical methods from springs to characterize the groundwater flow in remote, datapoor aquifers of the southwestern United States. The geological and geochemical methods they demonstrate can be used to interpret flow directions and magnitudes in other data-poor aquifers.

Barna, Fryar, Le Cao, Currens, Peng, and Zhu provide the international contribution to the special issue. Their study applies and refines widely applicable methods for studying karst systems to the Houzhai catchment in Guizhou province, China, and provides a foundation for future studies in similar settings.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 271–272

271

EEGS-26-03-03_1XO

July 8, 2020

8:23

Springer

Canion, Ransom, and Katz describe the influences of nitrate on algae and alteration of trophic structure in response to multiple anthropogenic stressors for springs in Florida. Sources of nitrate were determined through nitrogen isotope analyses. The study is broadly applicable through the demonstration of using a multi-tracer approach to develop locally relevant remediation strategies for nitrogen sources that impact springs ecosystems.

The authors of this special edition are indebted to E&EG co-editor Brian Katz. He was instrumental in providing the suggestion for this special edition and leading the submission and review process of the manuscripts to publication. His support is greatly appreciated. The authors are also grateful for the reviewers of the papers for this special edition who improved the quality of the publications.

We hope this is not the last special edition of E&EG related to springs. We challenge other GSA journals to include more research on these special locations at

272

the interface between geologic systems and ecological systems. Global increasing pressure on withdrawals of water from aquifers to support increasing human populations puts additional challenges on effective stewardship of springs.

REFERENCES Botosaneanu, L., 1998, Studies in Crenobiology: The Biology of Springs and Springbrooks: Backhuys, Leiden, The Netherlands.

Meinzer, O. E., 1923, Outline of Ground-Water Hydrology, with Definitions: U.S. Geological Survey Water-Supply Paper 494.

Springer, A. E. and Stevens, L. E., 2009, Spheres of discharge of springs: Hydrogeology Journal, Vol. 17, No. 1, pp. 83–93, doi:10.1007/s10040-008-0341-y.

Stevanovic, Z. and Kresic, N., 2009, Groundwater Hydrology of Springs: Elsevier, Amsterdam, The Netherlands.

Stevens, L. E. and Meretsky, V. J. Editors, 2008, Aridland Springs in North America: Ecology and Conservation: University of Arizona Press, Tucson, AZ.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 271–272

EEGS-26-02-08_1XO

July 8, 2020

9:3

Variability in Groundwater Flow and Chemistry in the Houzhai Karst Basin, Guizhou Province, China JOSHUA M. BARNA1, * ALAN E. FRYAR Department of Earth and Environmental Sciences, University of Kentucky, 101 Slone Building, Lexington, KY 40506-0053

LE CAO State Key Laboratory of Environmental Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences, Guiyang, 550002, China

BENJAMIN J. CURRENS2 Department of Earth and Environmental Sciences, University of Kentucky, 101 Slone Building, Lexington, KY 40506-0053

TAO PENG State Key Laboratory of Environmental Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences, Guiyang, 550002, China

CHEN ZHU Department of Earth and Atmospheric Sciences, Indiana University, 1001 E. Tenth Street, GY129, Bloomington, IN 47405-1405

Key Terms: Karst, Hydrogeology, Stable Isotope, Geochemistry, Dye Trace, China ABSTRACT Understanding how karst aquifers store and transmit water and contaminants is an ongoing problem in hydrogeology. Multiple flow paths and recharge heterogeneity contribute to the complexity of these systems. This study explored karst-conduit connectivity and waterchemistry variability within the Houzhai catchment in Guizhou Province, China. Artificial tracer tests were conducted during both the monsoon and dry seasons to understand temporal variability in connectivity and water velocity between karst features. Multiple flow paths through the catchment were activated during the monsoon season and partially abandoned during the dry season. Additionally, gradient reversals during monsoonal high-flow events and as a result of pumping were observed. Synoptic water samples from several karst features taken during both monsoon and dry seasons 1 Present address: ARM Group LLC, 9175 Guilford Road, Suite 310, Columbia, MD 21046 2 Present address: Kentucky Division of Water, 300 Sower Boulevard, 3rd Floor, Frankfort, KY 40601 *Corresponding author email: jmba286@g.uky.edu

elucidated spatial and temporal variability within the catchment. Water residence time was generally longer during the dry season, and flow within the Houzhai catchment was determined to be temporally dependent.

Time-series sampling at the outlet spring following a monsoonal storm event captured chemical variability and identified multiple flow paths. Overall, this study refines widely applicable methods for studying karst systems to this catchment and provides a foundation for future studies in similar settings.

INTRODUCTION Water movement through karst systems is complex and often not well understood, predominantly due to flow-path and recharge heterogeneity. This study explored groundwater flow by monitoring water parameters at multiple locations with varying temporal and spatial frequencies within the karstic Houzhai catchment. This 73.5 km2 basin in Guizhou Province, China, is located in the center of the largest karst area in the world, the Southeast Asian karst region (Wang and Zhang, 2001; Li et al., 2010) (Figure 1). Exposed carbonate rocks account for just over 20 percent of the land surface in China, making this location a key area of karst research (Cao et al., 2015).

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

273

EEGS-26-02-08_1XO

July 8, 2020

9:3

Barna, Fryar, Cao, Currens, Peng, and Zhu

Figure 1. Context map of China with Guizhou Province in dark gray and the Houzhai catchment indicated by the circle.

Approximately 32,400 people lived within the Houzhai catchment as of 2010, and agriculture was the main livelihood for upwards of 95 percent of them (Li et al., 2010). Runoff from organic fertilizers within karst terrain has the potential to travel rapidly through the subsurface, carrying contaminants such as excess nutrients and fecal bacteria to springs, and thereby posing significant risk to drinking-water quality (Li et al., 2010). Another major threat to agriculture in this area is a thinning of the soil layer, known as rocky desertification, which is often exacerbated by a combination of climate change and deforestation (Liu et al., 2010). Soil loss not only reduces vegetation coverage and makes farming less productive, but it also decreases epikarst groundwater storage, thus making springs less reliable sources of water supply (Liu et al., 2010).

This study extended our understanding of the Houzhai catchment by using artificial tracers to identify primary flow paths and calculate groundwater velocities during both monsoon and dry seasons. It also examined spatial and temporal variations in water chemistry to draw inferences about system function. The study area was chosen because the karstconduit network has been reasonably well delineated for at least two decades (Zhang et al., 1998). However, information regarding responses of springs in the catchment to storms is limited, and there are no known results of dye tracing within the catchment.

This study provides insights into how this system may react to changes in environmental conditions such as land use/land cover and rainfall. Ideally, the results

274

will contribute to knowledge needed for local policy makers to implement effective source-water protection practices.

BACKGROUND Site Geology and Hydrology The Houzhai catchment is classified as a typical mountain karst river basin (Wang and Zhang, 2001) (Figure 2). Elevation ranges from 1552 m above sea level (masl) in the headwaters to 1212 masl at the outlet, with slope decreasing from southeast to northwest (Zhang et al., 2017). Numerous karst peaks add to the relief, especially in the eastern part of the catchment.

Geomorphically, the Houzhai catchment is characterized by cockpit karst, which is composed of conical hills and star-shaped valleys that formed during the Neogene under tropical climatic conditions (Yu and Zhang, 1988; Chen et al., 2018). Catchment topography transitions from peak-cluster depressions, characterized by closely spaced peaks and sinkholes that transfer runoff quickly underground, within the eastern, higher-elevation part of the catchment towards peak-cluster basin and hill combination topography (Figure 3), characterized by widely spaced peaks and springs located near the outlet of the catchment.

Karst features within the Houzhai catchment are developed in the Triassic Guanling Formation. Lithology consists mainly of limestone and dolomite, with beds dipping between 5° and 25° northwest (Liu et al., 2010). Carbonate rocks are usually exposed on the

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

EEGS-26-02-08_1XO

July 8, 2020

9:3

Groundwater Variability in Houzhai Basin

Figure 2. Houzhai catchment map with monitored karst features (letters refer to locations in Table 1), inferred karst conduit network, surface stream network, and bodies of water.

surface (Li et al., 2010). Soils are generally thin (average <50 cm) and discontinuous in the eastern half of the catchment and thicker in the western half (Li et al., 2010). These clayey soils have relatively low waterretention capacity overall (Yue et al., 2018). High clay content helps prevent water leakage from rice paddies.

Figure 3. Peak-cluster valley in Houzhai catchment between Sites C and D.

Due to the nature of karstic terrain, basin boundaries often cannot be well delineated and frequently overlap (White, 1988). Within the Houzhai catchment, shale and marlite units act as topographic boundaries to the north and southwest and provide a base for the karst aquifer (Chen et al., 2013). The Yuguan Fault acts as a no-flow boundary to the east and southeast (Chen et al., 2013). The Mugong River lies to the southwest, and Chen et al. (2013) stated that there is little exchange between the river and the karst conduit system. Water leaves the Houzhai catchment at Site K, in the northwest corner (Figure 2). Mean annual water temperature is 16.7°C (Yan et al., 2012). Surface water pH averages 7.5, fluctuating between 7.2 in the summer and 7.9 in the winter (Yan et al., 2012).

The Houzhai catchment has a subtropical humid climate, which is typical of southern China (Zhang et al., 2016). Around 80 percent of precipitation occurs during the monsoon season, which lasts from May to October (Li et al., 2010). Total annual precipitation in 2018 amounted to 1290 mm (basin annual average is ∼1300 mm; Li et al., 2010). Climate change has impacted southwest China by increasing flood and drought frequency (Lian et al., 2015), which is likely to have adverse effects on agricultural productivity.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

275

EEGS-26-02-08_1XO

July 8, 2020

9:3

Barna, Fryar, Cao, Currens, Peng, and Zhu

Water resources are used mainly for drinking and irrigation (Li et al., 2010). Corn and rice are the main crops grown in the wet season, while canola and various vegetables are grown in the dry season (Yue et al., 2018). “Cultivated area” accounts for approximately 20 percent of the total catchment area, two thirds of which consist of rice paddies (Hu et al., 2001).

Both corn and rice are water-intensive to produce, and groundwater pumping from spring orifices for irrigation and domestic use was commonly observed in this study.

Previous Research Significant efforts have been made to understand the karstic groundwater system of the Houzhai catchment. With regard to physical hydrology, Chen et al.

(2018) conducted electrical resistivity tomography surveys to quantify the spatial heterogeneity of epikarst and aquifer permeability within the 1.25 km2 Chenqi sub-catchment near the headwaters of the Houzhai basin. Hourly hydrometric and water isotope (δ18 O and δ2 H) data were used to estimate storage and water age from fast-flow, hillslope, and slow-flow reservoirs in the Chenqi sub-catchment (Zhang et al., 2019). In the Houzhai catchment as a whole, Wang and Zhang (2001) documented the results of four “pulse tests,” which are analogous to large-scale slug tests, between 1988 and 1991. Those authors found that water velocity ranged from 200 to 800 m/hr, with higher velocities during the wet season. These pulse tests were also used to identify three different “types of aquifer media” based on their response and flow recession curve (Yang, 2001). Zhang et al. (2016) found that karstaquifer storage capacity increases along catchment flow paths, with more conduits and well-connected fissures upstream and more matrix and poorly connected fissures downstream. Li et al. (2010) reported that average annual discharges from underground and surface streams in the catchment are roughly equivalent, with 23.3 × 106 m3 /yr discharging from underground conduits and 24.9 × 106 m3 /yr discharging from surface streams. By comparison, using numerical simulations, Chen et al. (2013) found that surface streams only drain around 8 percent of total water discharge from the catchment during flooding periods, meaning that the bulk of flow occurs as groundwater discharge.

Yue et al. (2018) found that water isotope values deviate from the global meteoric water line (GMWL) due to dry-season evaporative losses.

Previous hydrochemical studies focusing on the Houzhai catchment have included monitoring of dissolution rates and exchange between fast- and slowflow systems during wet and dry periods (Zhang et al., 2017). In the Chenqi sub-catchment, Yang et al. (2012)

276

Table 1. Monitoring locations shown in Figure 2 (PKERS = Puding Karst Ecosystem Research Station [Barna, 2019]).

Letter A B C D E F G H I J K L M

Name

PKERS ID#

Xiao Shanba Laoheitan Daxing Sanjianfang Tian Guan Aliangzhai Liugu Trash Spring A Jiu Zhai Maokeng Maoshuikeng Hillslope Spring Chenqi Outlet

98 109 121 123 132 117 129 127 NA 302 294 167 171

found that soil CO2 and rainfall play major roles in epikarst spring electrical conductivity (EC), partial pressure of carbon dioxide (PCO2 ), and pH variability.

Yan et al. (2012) conducted a long-term (21 year) study of surface-water chemistry at the Houzhai outlet and found that fluxes of dissolved Ca2+ , Mg2+ , HCO3 − , and SO4 2− slowly increased while Na+ , K+ , and Cl− fluxes slowly decreased as a result of enhanced karst weathering. Those authors also found that rainfall had the most important influence on dissolved inorganic carbon (DIC) flux, an important parameter in constructing carbon budgets, which averaged 24.2 g C m−2 yr−1 . Using carbon stable-isotope (δ13 CDIC ) compositions, Li et al. (2010) showed that the production of CO2 from organic matter oxidation significantly impacts carbonate dissolution within the catchment. Diammonium phosphate, urea, and organic fertilizers are commonly applied to fields from April through July (Yue et al., 2018). Recent surveys by Buckerfield et al.

(2019) showed that rural residents in the catchment were relatively unaware of the potential water-quality impacts associated with the usage of organic fertilizers.

METHODS Synoptic Sampling Summer synoptic samples were collected on June 16, 2018 at 11 locations throughout the southern portion of the Houzhai catchment (Sites A–K; Figure 2 and Table 1). Winter synoptic samples were collected on December 3, 2018 at the same locations, except Sites H and I, because they were dry. In addition to collecting samples for anion, metal(loid), and stable-isotope analyses, water parameters including temperature (T), specific conductance (SC; temperature-compensated EC), pH, and carbonate alkalinity were measured in the field. Water samples were filtered, acidified (for

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

EEGS-26-02-08_1XO

July 8, 2020

9:3

Groundwater Variability in Houzhai Basin Table 2. Summary of summer and winter dye-tracing methods. Volumes of water for dissolving dye and flushing are approximate.

Name Number of dye injections Date and time Injection location Dye type Dye amount (kg, solid) Volume of water for dissolving dye (L) Volume of flushing water (L) Most recent precipitation event (days ago) Most recent precipitation event amount (mm)

Summer Base Flow

Summer Storm Flow

1 6/17/18 13:30 hr Site A Fluorescein 5.5 300 80 5 19.2

1 6/23/18 13:00 hr Site C Eosine 3.2 175 80 <1 60

preservation of metal[loid] samples), and refrigerated until analysis.

Dye Tracing and Storm-Flow Sampling Passive samplers (“bugs”) containing granular activated charcoal were deployed and then tested in order to identify the presence of fluorescent dyes in water samples. Before the summer and winter dye injections, background bugs were deployed to identify any pre-existing fluorescent dye within the system. During summer, one set of bugs was used to measure dye presence following the base-flow dye trace, and another set was used to identify storm-flow dye- trace presence. During winter, an effort was made to capture the first arrival of dye as it moved through the conduit network. To achieve this, bugs at each downgradient karst feature were swapped out daily or every other day. Charcoal from the bugs was analyzed using a scanning spectrofluorometer at the Kentucky Geological Survey (KGS) laboratory on the University of Kentucky campus, following Currens (2013).

At 1:30 pm on June 17, fluorescein dye powder dissolved in water was injected into the sinkhole at Site A and flushed with additional water (Table 2). Simultaneously, an ∼900 kg aliquot of NaCl dissolved in ∼4,000 L of water pumped from this sinkhole the previous day was also injected. The co-injection of saltwater with the dye solution likely induced mixing with groundwater in the underlying conduit because of density contrasts (Schincariol and Schwartz, 1990). The most recent precipitation event occurred on June 12 (total 19.2 mm), and there was no rain on the 17th, so this was considered a base-flow dye injection. In addition to charcoal bugs, water-level and EC loggers were deployed at Site B to measure any pressure or conductivity pulse from the salt injection (Barna, 2019).

At 1:00 pm on June 23, eosine dye powder dissolved in water was injected at Site C and flushed with additional water (Table 2). Approximately 60 mm of precipitation was recorded at Site B within 48 hours prior

Winter Base Flow 2 12/5/18 10:35 hr Site D Eosine 1.8 100 50 19 1.2

12/5/18 11:25 hr Site C Sulforhodamine B 1.8 100 50 19 1.2

to injection. Unlike the base-flow dye trace at Site A, dye was no longer visible 50 minutes after injection.

Sampling at Site K, the outlet spring of the Houzhai basin, began at 3:00 pm on the 23rd, continued every 2 hours for 24 hours, and then switched to hourly intervals for another 24 hours. In addition to collecting and filtering water samples for dye, alkalinity, anion, metal(loid), and stable-isotope analyses, water parameters including T, pH, and SC were also recorded from grab samples. EC and water-level loggers were deployed along the spring run ∼370 m downstream of the orifice at Site K. Two base-flow dye injections were conducted on December 5, 2018 (Table 2). At 10:35 am, eosine dye powder dissolved in water was injected at Site D and flushed with additional water, followed at 11:25 am by injection of sulforhodamine B powder dissolved in water and flushing with additional water at Site C. These dye injections began at base flow, with the last recorded precipitation event >0.2 mm/hr occurring 19 days prior (November 16). Because these injections were conducted during winter, water levels at both locations were significantly lower than those observed during summer.

Solute and Isotope Analyses Anion, metal(loid), and δ13 CDIC analyses were performed at the Institute of Geochemistry, Chinese Academy of Sciences, in Guiyang, China. Anions were measured using ion chromatography, while cations and SiO2 0 were measured using inductively coupled plasma–optical emission spectrometry. Stable-isotope analyses for DIC were performed using a Thermo Fisher Scientific MAT 252 instrument. The δ13 CDIC data are reported in per mil (‰) notation relative to Vienna Pee Dee Belemnite (VPDB). Water isotopes (δ18 O and δ2 H) were measured at the Kentucky Stable Isotope Geochemistry Laboratory at the University of Kentucky by isotope-ratio infrared spectroscopy with a Los Gatos instrument. The data are reported in per

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

277

EEGS-26-02-08_1XO

July 8, 2020

9:3

Barna, Fryar, Cao, Currens, Peng, and Zhu Table 3. Ranges of summer and winter synoptic sample field parameters, analytes, and saturation indices with median values in parentheses.

Field Parameter or Analyte T (°C) SC (μS/cm) pH Carbonate alkalinity (mg/L CaCO3 ) Ca2+ (mg/L) K+ (mg/L) Mg2+ (mg/L) Na+ (mg/L) SiO2 0 (mg/L) Sr2+ (mg/L) Mg/Ca δ13 CDIC (‰ PDB) δ2 H (‰ VSMOW) δ18 O (‰ VSMOW) F− (mg/L) Cl− (mg/L) NO3 − (mg/L) SO4 2− (mg/L) NO2 − (mg/L) Br− (mg/L) SIcalcite SIdolomite SIstrontianite SIgypsum SIcelestite

Summer

Winter

17.64–22.94 (18.45) 429–592 (492) 6.77–8.10 (7.02) 141–192 (183) 61.09–93.93 (81.59) 2.32–8.70 (3.51) 10.67–19.20 (17.94) 1.87–6.08 (3.71) 4.32–9.34 (5.27) 0.98–2.30 (1.65) 0.21–0.45 (0.35) −13.21 to −8.71 (−12.24) −67.1 to −47.5 (−50.2) −9.76 to −6.89 (−7.57) 0.18–0.34 (0.24) 6.30–12.82 (8.52) 13.74–27.56 (24.68) 46.63–110.63 (64.26) 0.03–0.53 (0.28) <0.2 −0.53 to 0.78 (−0.24) −1.62 to 1.24 (−0.92) −1.85 to −0.46 (−1.45) −1.88 to −1.46 (−1.74) −2.03 to −1.37 (−1.73)

14.81–17.76 (17.26) 366–701 (578) 6.72–7.36 (7.16) 190–257 (209) 59.92–117.89 (104.23) 1.11–11.20 (4.79) 22.91–28.17 (24.95) 1.97–35.01 (7.53) 3.01–5.45 (5.05) 0.27–3.82 (3.09) 0.35–0.69 (0.41) −12.27 to −8.82 (−11.77) −57.0 to −52.0 (−55.59) −8.36 to −7.41 (−8.12) 0.19–0.78 (0.34) 4.79–57.34 (11.91) 14.52–20.72 (16.67) 28.20–191.28 (136.88) <0.05 <0.2 −0.377 to 0.11 (−0.11) −1.15 to −0.14 (−0.56) −1.95 to −0.85 (−1.30) −2.18 to −1.19 (−1.36) −2.84 to −0.99 (−1.18)

mil notation relative to Vienna Standard Mean Ocean Water (VSMOW).

Solute data and additional parameters, including alkalinity, T, and pH, were entered into the geochemical modeling program PHREEQC (version 3.4.0 with phreeqc.dat database file; Parkhurst and Appelo, 2013) to calculate charge balances and saturationindex (SI) values. Eighty-five percent of the water samples charge-balanced to within 2 percent error, and the largest error was <5 percent. Phases of interest for SI calculations included calcite (CaCO3 ), dolomite (CaMg(CO3 )2 ), strontianite (SrCO3 ), gypsum (CaSO4 •2H2 O), and celestite (SrSO4 ). Results of analyses and modeling calculations are tabulated in Barna (2019).

RESULTS Synoptic Sampling Physical and chemical water parameters fluctuated spatially and temporally throughout the Houzhai catchment. Ranges of summer and winter synoptic values for field parameters and analytes are listed in Table 3. Water T was, on average, 1.2°C cooler during winter than during summer. Specific conductance and alkalinity values were higher overall in the upper (southeast) part of the catchment and during the win-

278

ter, whereas pH was also higher during the winter, but without a definitive spatial trend. The Na+ and Cl− concentrations at Site A (35.01 mg/L and 57.34 mg/L, respectively) were anomalously high during the winter.

Mg/Ca ratios were generally higher during the winter (Table 3), especially at Sites J and K. Values of δ13 CDIC tended to be higher during the winter, especially in the upper part of the catchment, whereas the opposite trend holds for δ2 H and δ18 O values. Maps detailing spatial variability in individual groundwater parameters measured during summer and winter synoptic sampling can be found in Barna (2019). Ca-HCO3 was the dominant hydrochemical facies, which is typical of karst groundwater systems (Shuster and White, 1971; Liu et al., 2004).

Synoptic sample δ2 H and δ18 O values fell along the GMWL (Figure 4), with winter samples generally more depleted than summer samples. Nine of the summer samples clustered around average values of −50.8‰ for δ2 H and −7.63‰ for δ18 O, but Site A was considerably more depleted (δ2 H = −67.1‰, δ18 O = −9.76‰), and Site H was somewhat more enriched (δ2 H = −47.5‰, δ18 O = −6.89‰). Winter synoptic samples clustered around average values of −56.0‰ for δ2 H and −8.17‰ for δ18 O, with only Site D markedly enriched (δ2 H = −52.0‰, δ18 O = −7.41‰).

Figure 4 includes a summary of isotopic data collected within the Chenqi sub-catchment by Chen et al. (2018)

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

EEGS-26-02-08_1XO

July 8, 2020

9:3

Groundwater Variability in Houzhai Basin

Figure 4. Plot of synoptic sample, time-series, and Chen et al. (2018) δ18 O versus δ2 H values relative to the global meteoric water line (dashed).

Error bars represent one standard deviation; colored bars represent recorded data values. The isotopic range of the Chen et al. (2018) rainfall data (δ2 H = −120.2‰ to −17.9‰, δ18 O = −16.4‰ to 0‰) was too large to be plotted.

for comparison with this study. These data show isotopic variability of wells, a hillslope spring (Site L), rainfall, and outlet discharge (Site M). Average isotopic values are depleted relative to the data in this study, except for Site A, with average rainfall values being most depleted.

Time-Series Analysis The storm-flow trace at Site C occurred on June 23 at 1:00 pm, 3 hours after water level peaked at Site K (Figure 5). Therefore, the samples capture the falling limb of the storm event. Precipitation during this event, measured at the Site B weather station, occurred in two main pulses: 35.5 mm of rain peaking at 3:30 pm June 21, followed by 24.1 mm of rain peaking at 3:30 pm June 22 (Figure 5). A total of 6.5 mm of rainfall occurred after those two pulses, with only 2.0 mm during the sampling period at Site K. Minimum EC during the recession was logged at 9:35 am on June 24, nearly 24 hours after the stage peak (Figure 5), while pH reached a maximum at 11:00 am. Concentrations of several solutes (K+ , Na+ , SiO2 0 , Cl− , Sr2+ , SO4 2− ) fell to minimum values 19–27 hours after the stage peak (Figure 6). Magnesium increased notably during the afternoon and evening of the 24th, and Mg/Ca values reached their maximum at 8:00 pm. Alkalinity generally trended upward (Barna, 2019), while NO3 − generally decreased over the sampling interval (Figure 6C). There were no discernible trends for F−

or Ca2+ (Barna, 2019), nor for δ13 CDIC , which fluctuated within 0.8‰ (−11.55 to −12.34‰).

Values of δ2 H and δ18 O co-varied, increasing at the beginning of the sampling period and then sharply decreasing during the first half of June 24, followed by a saw-shaped pattern of fluctuations (Figure 6D). Site K δ2 H and δ18 O time-series sample values fell between winter and summer synoptic samples, with averages of −52.4‰ for δ2 H and −7.82‰ for δ18 O (Figure 4). Two of the three Site C time-series samples plotted within the summer synoptic sample cluster, but the first sample, which was taken immediately following the stormflow dye trace, was significantly more enriched (δ2 H = −49.9‰, δ18 O = −7.09‰).

Dye Tracing On June 18, 1 day after the base-flow dye injection at Site A, dye was visible in water discharging from all three of the observed outlets at Site B. Dye was still visible on June 19 at Site B and was visible at Site C on June 20, but dye was not visible at Site D or Site G. Except at Site K, none of the background bugs produced any positive dye spectra. The peak for fluorescein recorded by the Site K background bug was significantly lower than the peak observed from the bug that was deployed to capture the base-flow dye trace.

Analyses of the charcoal bugs showed dye presence at Sites B, C, D, G, I, and K (Figure 7). The bug deployed at Site J was vandalized at an unknown time between deployment and retrieval, so even though enough

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

279

EEGS-26-02-08_1XO

July 8, 2020

9:3

Barna, Fryar, Cao, Currens, Peng, and Zhu

Figure 5. Plot showing continuously logged EC and water level at Site K leading up to and during the time of sampling in relation to precipitation data from the Site B weather station.

charcoal was recovered to test for dye, its absence does not conclusively indicate that no dye passed through Site J. The bugs at Sites E and F were lost prior to recovery and could not be tested.

Continuous loggers at Site B measured a waterlevel rise beginning ∼40 minutes prior to the saltwater and dye injection (Figure 8). The water level initially crested 25 minutes before the injection and began to fall and then rebounded for ∼2.5 hours before decreasing. Beginning ∼17 hours after the ultimate water-level peak, a broad EC peak was superimposed on an increasing trend (Figure 8).

For the summer storm-flow trace, no dye was detected in the Site K water samples. However, all the bugs that were redeployed before the storm-flow trace indicated the presence of dye (Figure 9). The bug that was redeployed at Site F was positive for eosine. The bugs at Sites G and K were positive for both fluorescein and eosine, as was the bug that was deployed at Site I before the base-flow trace.

Before any of the winter dye injections, fluorescein still appeared to be visible at Site A when winter synoptic samples were taken (>5 months after injection).

This result, which is consistent with elevated Na+ and Cl− values measured at the same time, could reflect slow leaching of co-injected dye and salt from soil around the sink. No dye was visible at Site B, and all the background bugs deployed tested negative for dye.

One day after the winter base-flow injections, which occurred on December 5, dye was still visible but much less concentrated at the injection locations (Sites C and D). Three days after the injections, residual dye ad-

280

sorbed to the soil and sediment substrate was observed at each injection site, but no dye was visible in the water. Pumping was observed during each visit to Site E but not at other sites. Site G was the only downgradient location where dye was visibly present in the water after injection. The first observation was 3 days after dye injection; the color became darker on day 4 and fainter on day 5. Based on this visible change, the main dye pulse arrived at Site G 4 days after injection.

One day after the injections, the bug deployed at Site F recorded the arrival of sulforhodamine B, the dye used at Site C (Figure 10). The following day, sulforhodamine B was detected in bugs from Sites E and F, and bugs confirmed the presence of dye at Site G 4 days after injection (Figure 10). In order to extend the monitoring time as long as possible, bugs were deployed at Sites J and K on December 10 and retrieved on December 13th. Sulforhodamine B was positively identified from the Site K bug, but no dye was observed at Site J at any point during the winter (Figure 10).

Based on the bug results for sulforhodamine B, and assuming straight-line flow paths, winter base-flow velocities from Site C to several other karst features were estimated (Table 4).

DISCUSSION Synoptic Sampling Specific conductance and carbonate alkalinity can be seen as analogues for water residence time (Hess and White, 1988; Chen et al., 2018). These parameters

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

EEGS-26-02-08_1XO

July 8, 2020

9:3

Groundwater Variability in Houzhai Basin

Figure 6. Plots showing pH and water level (A), sulfate and sodium (B), nitrate and chloride (C), δ2 H and δ18 O (D), magnesium and strontium (E), and potassium and silica (F) values during time-series sampling. Error for δ2 H = 0.3‰. Error for δ18 O = 0.05‰.

are generally higher during winter. This is likely because less water is flowing within the conduit system during the dry season, meaning that hydraulic heads and groundwater velocities are lower.

Residence time and lithology seem to be the two main controls on Mg/Ca ratios (Langmuir, 1971; Wigley, 1973; Plummer, 1977; Lohmann, 1988; and Fairchild et al., 1996). Longer residence times, either as a result of slower water velocity or longer flow paths, correspond with higher Mg/Ca ratios. This partly explains the increased Mg/Ca ratios in the winter and towards the outlet. Additionally, Sites J and K are lo-

cated within the uppermost unit of the Guanling Formation, which is composed chiefly of dolomite. This lithologic variability likely accounts for some of the change in water chemistry. Similarly, higher SO4 2− concentrations during winter may reflect a greater contribution of deep groundwater that has dissolved more gypsum along its flow path (Yan et al., 2012; Yang et al., 2012).

Site H seems to be an outlier compared to the other karst features. It was the only karst feature at which no dye was observed during the summer baseflow trace (Figure 7). Additionally, water parameter

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

281

EEGS-26-02-08_1XO

July 8, 2020

9:3

Barna, Fryar, Cao, Currens, Peng, and Zhu

Figure 7. Map showing summer base-flow dye-trace bug results.

measurements such as SC, alkalinity, and Mg/Ca ratios were all notably lower than other springs. Water temperature was close to the diurnal range of temperatures logged at Site C (Barna, 2019). Overall, this suggests that flow-path lengths at Site H are relatively short and that it is perched above the main conduit network.

A comparison of δ13 CDIC data from the synoptic samples shows a statistically significant (p < 0.01) increase in δ13 CDIC from summer to winter (from −12.24 to −11.77‰ for median values; Table 3).

The δ13 CDIC values during the summer were roughly midway between C3 plant δ13 C values (−28‰) and rock δ13 C values (0‰), and close to C4 plant δ13 C

Figure 8. Plot showing water level (gray) and EC (black) at Site B, with the Site A injection time marked (vertical dotted line).

282

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

EEGS-26-02-08_1XO

July 8, 2020

9:3

Groundwater Variability in Houzhai Basin

Figure 9. Map showing summer storm-flow dye-trace bug results.

values (−14‰) (O’Leary, 1988). Winter synoptic samples shifted closer towards the rock δ13 C end of the spectrum, as observed by Li et al. (2010). This is consistent with longer residence times, increased dissolution, and less oxidation of organic material during winter. While the main crop within the catchment is rice (a C3 plant), corn (a C4 plant) is also grown. Oxidation of other plant matter also likely influences δ13 C values.

One notable outlier in the data is Site J, which showed a more enriched δ13 CDIC value than the other loca-

tions in both summer and winter. The dissolution of more dolomite around Site J could explain this anomalous δ13 C value. Sheppard and Schwarcz (1970) found that dolomite δ13 C values are ∼1‰ higher than cooccurring low-Mg calcite. Site J showed the highest winter Mg/Ca ratio and third highest Mg/Ca ratio in the summer. Given that alkalinity values at Site J were not particularly high, isotopic variation as a result of lithologic variability, rather than the total amount of dissolution, is likely the cause of this enrichment.

Figure 10. Winter base-flow dye-trace map with numbers representing the number of days after the trace that dye arrived at that karst feature.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

283

EEGS-26-02-08_1XO

July 8, 2020

9:3

Barna, Fryar, Cao, Currens, Peng, and Zhu Table 4. Summary of inferred groundwater flow velocities under different hydrologic conditions.

Velocity Summer base flow Site A to Site B Site A to Site K Summer storm flow Site C to Site K Winter base flow Site C to Site F Site C to Site E Site C to Site G Site C to Site K Site D to Site J

Value (km/d) 2 ࣙ2.6 >3.7 >0.75 1–2 0.4–0.5 0.9–1.5 <0.6

The observation that water samples generally fell along the GMWL on a plot of δ2 H versus δ18 O suggests that evaporation was minimal at most locations.

This is in contrast with the observations of Yue et al.

(2018) and may reflect the fact that those authors sampled surface water, which is more prone to evaporation, as well as groundwater. The most depleted sample, from Site A during the summer, appears to have been a mixture of rainwater and groundwater, even though it had not rained for 5 days before sampling.

The relatively low alkalinity and Cl− concentration of the sample support the inference of groundwater dilution by rainwater, consistent with the inference that Site A is a sinkhole that subaerially exposes the water table. The two most enriched samples, from Sites H and C (at the time of injection), both fell below the GMWL, which suggests evaporation (Fritz et al., 1976; Lakey and Krothe, 1996). The anomalously enriched sample at Site C may have resulted from overflow of partly evaporated paddy water into the pool during the previous storm, which is consistent with observations during dye injection. The summer synoptic sample from Site K fell within the field of timeseries storm samples from that spring and was more depleted than other summer synoptic samples (apart from Site A). This is consistent with Site K discharging a substantial component of deeper groundwater recharged at higher elevation within the basin. The relatively enriched values for the winter sample at Site D may indicate that the site is not well connected to the main conduit network during the dry season. The northern branch of the network may transmit water more slowly, which could explain why the eosine injected at Site D was not observed at any other karst feature. Except for Site A in summer, δ2 H and δ18 O values in this study were more enriched than average values of rainfall and groundwater samples collected from June to August 2017 in the higher-elevation Chenqi sub-catchment (Chen et al., 2018; Zhang et al., 2019) (Figure 4).

284

Based on geochemical modeling calculations, details of which are provided in Barna (2019), synoptic water samples were closer to saturation with respect to calcite during winter, especially in the southwesternmost part of the catchment. This was likely a result of less dilution and slower flow, allowing the water to approach equilibrium with the matrix, as well as retrograde solubility (Ford and Williams, 2007). Water samples were generally more undersaturated with respect to dolomite. Like calcite, SIstrontianite winter values were somewhat closer to saturation than summer values. All water samples were undersaturated with respect to gypsum and celestite, and both indices were generally more negative during the summer, except at Sites J and K. Higher sulfate concentrations as a result of slower flow, especially in the upper catchment, can account for higher gypsum and celestite saturation indices during the winter. Overall, the geochemical analysis supports the inference of seasonal variability in groundwater velocity within the catchment.

Time-Series Analysis The arrival of storm flow at Site K was indicated by minimum values of SC, pH, and concentrations of various solutes (Cl− , SO4 2− , Na+ , K+ , Mg2+ , Sr2+ , SiO2 0 ) ∼1 day after the stage peak (Figure 6). The fact that secondary minima and maxima followed the EC minimum and pH maximum suggests the arrival of a second storm pulse (Figures 5 and 6). The distinctive Mg2+ response could have resulted from spatial variability in carbonate lithology within the basin. The broad decline in NO3 − concentration during the recession (Figure 6C) suggests a gradual flushing of fertilizer, which is liberally applied to rice paddies throughout the catchment.

More distinctly than solutes and field parameters, the δ2 H and δ18 O time-series plot for Site K appears to show three pulses of recharge, as indicated by successive drops and rebounds (Figure 6D). This may represent recharge from progressively farther up-gradient within the watershed, particularly since δ2 H is most depleted for the third pulse, consistent with higherelevation recharge and with rainout (Darling et al., 2005). This is supported in comparison with the relatively depleted values measured within the Chenqi sub-catchment (Chen et al., 2018; Zhang et al., 2019) (Figure 4). Alternatively, the three pulses observed at Site K could represent contributions from different branches of the conduit network. The best way to test this hypothesis would be to conduct a tracer test, coupled with isotopic sampling, along each branch during a summer storm event.

The time-series SI plots for carbonate minerals (calcite, dolomite, and strontianite) have similar shapes

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

EEGS-26-02-08_1XO

July 8, 2020

9:3

Groundwater Variability in Houzhai Basin

Figure 11. Saturation indices for carbonate minerals (A) and sulfate minerals (B).

(Figure 11A), which mirror pH time-series measurements. The SI plots of the sulfate minerals, gypsum and celestite (Figure 11B), both mirror the U-shaped SO4 2− pattern. For carbonate minerals, maximum SI values occurred 4 hours after breakthrough of stormwater pulses, as indicated by local EC and T minima (Barna, 2019), and these maximum SI values were greater than values observed during synoptic (baseflow) sampling. For sulfate minerals, all SI values were less than those observed during synoptic sampling, and minimum values occurred 6 hours after breakthrough of storm-water pulses.

Dye Traces Multiple conduit flow paths remained active at base flow during the monsoon season, as indicated by the presence of dye at both Sites C and D (Figure 12A).

Both the Site B and Site C bugs produced eluent that had enough dye to be visible to the naked eye, consistent with visual observations in the field. The bug

recovered from Site D during summer base flow also produced eluent with a visible amount of dye, although not as bright as Site B or Site C, which suggests that Site D is located on a secondary pathway. The negative dye result at Site H is consistent with other factors suggesting that this spring is a high-level overflow and is not continuously connected to the conduit system.

The negative result at Site J may be due to vandalism of the bug before the arrival of the dye because, based on the inferred conduit flow paths, dye likely traveled via both conduit pathways (Figure 2).

The low-level detection of fluorescein in the background bug at Site K could be a false positive as a result of cross-contamination. However, neither the field blank nor any of the other background samples tested positive. It is more likely that the background detection is an artifact of fluorescein from a source such as radiator coolant, which is plausible given that the orifice is located beneath a highway bridge. The bug deployed to capture the base-flow dye trace at Site K showed a fluorescein peak approximately twice the intensity of

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

285

EEGS-26-02-08_1XO

July 8, 2020

9:3

Barna, Fryar, Cao, Currens, Peng, and Zhu

Figure 12. Conceptual maps showing summer base flow (A), summer storm flow (B), and winter base flow (C) through the Houzhai catchment.

Thicker arrows correspond to faster groundwater velocities.

286

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

EEGS-26-02-08_1XO

July 8, 2020

9:3

Groundwater Variability in Houzhai Basin

the background peak, which indicates the tracer velocity from Site A to Site K was ࣙ2.6 km/d (10.4 km in 4 days; Table 4).

The immediate water-level peak from the salt injection suggests a pressure pulse that moved through the system rapidly after the injection, displacing water already in the conduit system to discharge from the springs at Site B (Figure 8). The initial water-level rise beginning ∼40 minutes before the injection may indicate a coincidental input of water, such as from one of the paddies around the sensor at Site B. The occurrence of a pressure pulse also suggests that the conduit system between Sites A and B experienced pipe-full conditions during the injection (Ryan and Meiman, 1996). Because water arriving at Site B during the pressure pulse was already moving through the system, there was no decrease in EC through dilution. The arrival of injected water was signaled by the EC peak ∼21 hours later. Based on this arrival and an inferred conduit length of 1.75 km, the monsoon-season baseflow velocity between Site A and Site B was ∼2 km/d (Table 4).

The bugs deployed at Sites G and K to monitor the storm-flow trace captured both fluorescein and eosine, suggesting that fluorescein had not been completely flushed from the system. The detection of both fluorescein and eosine for the bug at Site I, which was deployed from the start of the fluorescein trace until 48 hours after the eosine trace (a total of 8 days), suggests that this dry valley is connected to the trunk conduit network during high flow. Detection of eosine at Site F, which is inferred to be located on a tributary conduit flowing northeast to Site C, suggests a gradient reversal in this area during storm events (Figure 12B).

The lower-bound distance on this gradient reversal is ∼750 m.

An unexpected result of the winter dye traces was the relatively rapid flow from Site C towards Sites F and E (Figure 12C). Based on pre-existing maps of the conduit network, both karst features were thought to be up-gradient of Site C (Figure 2), especially during base-flow conditions. The reason for the inferred gradient reversal was likely the significant pumping observed at Site E. Because the rice paddies were dormant during this time, the water was likely used for other purposes, such as domestic use or construction.

In general, inferred groundwater velocities were faster in summer than in winter, which is consistent with greater recharge and steeper hydraulic gradients during the wet season (Table 4).

There are several possible interpretations for the lack of dye observed at Site J. Groundwater velocity could have been <0.6 km/d (4.4 km over 8 days). Alternatively, Sites D and J may not be hydraulically connected during the dry season, but water from Site D

would still be expected to discharge at Site K. Because Sites C and D are comparable in distance from Site K, the lack of detection of eosine (injected at Site D) and the detection of sulforhodamine B (injected at Site C) suggest slower velocities along the northern branch of the conduit network than the southern branch. The observation of sulforhodamine B at Site K but not at Site J suggests that Site J may not be located along the trunk conduit linking Sites G and K (Figure 12C).

Sulforhodamine B may have flowed through a conduit located further west, possibly in connection with Site I. In general, karst conduit connectivity and groundwater velocity increase in response to precipitation (Shuster and White, 1971; Ryan and Meiman, 1996).

Therefore, future changes in rainfall could increase flow variability between wet and dry seasons.

CONCLUSIONS The Houzhai catchment in Guizhou Province, China, contains a complex karstic drainage network that shows variable behavior at both event (hours to days) and seasonal timescales. During summer, monsoon rains cause the conduit network to fill and overflow pathways to become active. During winter, overflow pathways remain dry, and some flow paths, such as the one that connects Site D to the rest of the conduit network, seem to decrease in velocity significantly or become hydrologically disconnected.

Carbonate dissolution is likely greater during the summer monsoon season due to the greater circulation of groundwater within the system. Saturation-index values, especially with respect to calcite, are generally more negative throughout the catchment during baseflow conditions in summer than in winter. This implies that flushing promotes mineral dissolution, as observed for DIC in surface water in the basin (Yan et al., 2012). Summer δ13 CDIC samples are more depleted than winter samples, as observed by Li et al. (2010), indicating shorter residence times and increased oxidation of organic matter, especially from agricultural byproducts such as rice leaves and corn stalks, during the summer. Summer storm events flush nitrate from the catchment; nitrate concentrations did not exceed either the Chinese standard of 20 mg/L NO3 -N (National Standards of the People’s Republic of China, 1997) or the U.S. standard of 10 mg/L NO3 -N (U.S. Environmental Protection Agency, 2019). Organic fertilizer, possibly including human waste, is likely making its way into the karst conduit system. This and the relatively rapid groundwater velocities observed throughout the catchment pose significant contamination risks for local populations relying on groundwater for drinking water, given the findings of Buckerfield et al. (2019) regarding fecal pathogens.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

287

EEGS-26-02-08_1XO

July 8, 2020

9:3

Barna, Fryar, Cao, Currens, Peng, and Zhu

This research builds on previous studies within the Houzhai catchment by constraining values of groundwater velocity during summer and winter baseflow conditions as well as during summer storm-flow events. Additionally, understanding of hydrologic connectivity within the southern portion of the Houzhai catchment has become more established. These collected data align well with preceding work completed at this site. Further application of dye-tracing techniques could significantly expand upon insights derived from this study. One potentially fruitful area of exploration could probe conduit connectivity and flow velocity during infrequent dry-season storm events.

While likely more challenging to capture due to their relative rarity, comparing the results of that type of dye trace to those conducted in this study could significantly increase overall comprehension of groundwater flow variability. The heterogeneous nature of the Houzhai catchment conduit network likely generates responses to storm events that vary nonlinearly with fluctuations in the amount, intensity, and location of precipitation within the catchment, even within a single monsoon season.

The findings of this study have several practical implications for water-resource management in the catchment. Tracer tests provide estimates of travel times along the conduit network under various conditions, which are useful for understanding the susceptibility of water supplies to contamination and for parameterizing groundwater flow models (Chen et al., 2013; Zhang et al., 2017, 2019). In addition, the observed gradient reversal at Sites E and F in December indicates the potential for changes in flow rates and directions as a result of pumping. Consequently, there may be localized dewatering of conduits during the dry season. Overall, the combination of tracer injection with water-quality monitoring during base flow (synoptically and seasonally) and storm flow provides a foundation for further research on conduit connectivity and groundwater velocity.

ACKNOWLEDGMENTS Thanks go to Andrea Erhardt and Junfeng Zhu for helping to shape the direction of this project. Laboratory assistance was provided by Jordan Munizzi, Andrea Conner, and Zhikang Wang. The field support of Driver Song, Guanru Zhang, Qianyun Cheng, Sarah Buckerfield, Qiangshan Gao, Xuemei Liu, and Yang Changan was invaluable. Funding for this project was provided through the University of Kentucky Ferm and Brown McFarlan funds, the University of Kentucky Confucius Institute, and the Chinese Academy of Sciences through the Opening Fund of the State Key Laboratory of Environmental Geochemistry (SK-

288

LEG2016903). Three anonymous reviewers provided thoughtful comments that improved the manuscript.

REFERENCES Barna, J. M., 2019, Variability in Groundwater Flow and Chemistry in the Houzhai Karst Basin, Guizhou Province, China: Unpublished M.S. Thesis, Department of Earth & Environmental Sciences, University of Kentucky, Lexington, KY. Electronic document, available at https://doi.org/10.13023/etd.2019.109.

Buckerfield, S. J.; Waldron, S.; Quilliam, R. S.; Naylor, L. A.; Li, S.; and Oliver, D. M., 2019, How can we improve understanding of faecal indicator dynamics in karst systems under changing climatic, population, and land use stressors?— Research opportunities in SW China: Science Total Environment, Vol. 646, pp. 438–447.

Cao, J.; Yuan, D.; Tong, L.; Mallik, A.; Yang, H.; and Huang, F., 2015, An overview of karst ecosystem in southwest China: Current state and future management: Journal Resources Ecology, Vol. 6, No. 4, pp. 247–256.

Chen, X.; Zhang, Y. F.; Zhou, Y. Y.; and Zhang, Z. C., 2013, Analysis of hydrogeological parameters and numerical modeling groundwater in a karst watershed, southwest China: Carbonates Evaporites, Vol. 28, No. 1–2, pp. 89–94.

Chen, X.; Zhang, Z.; Soulsby, C.; Cheng, Q.; Binley, A.; Jiang, R.; and Tao, M., 2018, Characterizing the heterogeneity of karst critical zone and its hydrological function: An integrated approach: Hydrological Processes, Vol. 32, pp. 2932–2946.

Currens, J. C., 2013, Kentucky Geological Survey Procedures for Groundwater Tracing Using Fluorescent Dyes: Kentucky Geological Survey, Series XII, Information Circular 26. Electronic document, available at https://kgs.uky.edu/kgsweb/olops/pub/kgs/IC26_12.pdf Darling, W. G.; Bath, A. H.; Gibson, J. J.; and Rozanski, K., 2005, Isotopes in water. In Leng, M. J. (Editor), Isotopes in Palaeoenvironmental Research: Springer, Dordricht, The Netherlands, pp. 1–66.

Fairchild, I. J.; Tooth, A. F.; Huang, Y.; Borsato, A.; Frisia, S.; and McDermott, F., 1996, Spatial and temporal variations in water and stalactite chemistry in currently active caves: A precursor to interpretations of past climate. In Botrell, S. (Editor), Proceedings of the Fourth International Symposium on the Geochemistry of the Earth’s Surface: University of Leeds, Ilkley, U.K., pp. 229–233.

Ford, D. C. and Williams, P. W., 2007, Karst Hydrogeology and Geomorphology: John Wiley & Sons, Chichester, U.K. Fritz, P.; Cherry, J. A.; Sklash, M.; and Weyer, K. U., 1976, Storm runoff analysis using environmental isotopes and major ions. In Interpretation of Environmental Isotope and Hydrochemical Data in Groundwater Hydrology: International Atomic Energy Agency, Vienna, Austria, pp. 111–130.

Hess, J. W. and White, W. B., 1988, Storm response of the karstic carbonate aquifer of southcentral Kentucky: Journal Hydrology, Vol. 99, pp. 235–252.

Hu, X. J.; Chen, B.; Hu, X. H.; and He, G. H., 2001, Study on the model of rational land use in the karst areas of the Houzhai River Basin: Carsologica Sinica, Vol. 20, pp. 305–309.

Lakey, B. and Krothe, N. C., 1996, Stable isotopic variation of storm discharge from a perennial karst spring, Indiana: Water Resources Research, Vol. 32, No. 3, pp. 721–731.

Langmuir, D., 1971, The geochemistry of some carbonate ground waters in central Pennsylvania: Geochimica Cosmochimica Acta, Vol. 35, pp. 1023–1045.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

EEGS-26-02-08_1XO

July 8, 2020

9:3

Groundwater Variability in Houzhai Basin Li, S.-L.; Liu, C.-Q.; Li, J.; Lang, Y.-C.; Ding, H.; and Li, L. B., 2010, Geochemistry of dissolved inorganic carbon and carbonate weathering in a small typical karstic catchment of southwest China: Isotopic and chemical constraints: Chemical Geology, Vol. 277, No. 3–4, pp. 301–309.

Lian, Y.; You, G.; Lin, K.; Jiang, Z.; Zhang, C.; and Qin, X., 2015, Characteristics of climate change in southwest China karst region and their potential environmental impacts: Environmental Earth Sciences, Vol. 74, pp. 937–944.

Liu, L. H.; Shu, L. C.; Chen, X. H.; and Oromo, T., 2010, The hydrologic function and behavior of the Houzhai underground river basin, Guizhou Province, southwestern China: Hydrogeology Journal, Vol. 18, No. 2, pp. 509–518.

Liu, Z. H.; Groves, C.; Yuan, D. X.; Meiman, J.; Jiang, G. H.; He, S. Y.; and Li, Q. A., 2004, Hydrochemical variations during flood pulses in the south-west China peak cluster karst: Impacts of CaCO3 -H2 O-CO2 interactions: Hydrological Processes, Vol. 18, No. 13, pp. 2423–2437.

Lohmann, K. C., 1988, Geochemical patterns of meteoric diagenetic systems and their application to studies of paleokarst. In James, N.P., Choquette, P.W. (Editors), Paleokarst: SpringerVerlag, New York, pp. 58–80.

National Standards of the People’s Republic of China, 1997, Environmental Quality Standards for Surface Water: Electronic document, available at http://english.mee.gov.cn/SOE/ soechina1997/water/standard.htm O’Leary, M. H., 1988, Carbon isotopes in photosynthesis: Bioscience, Vol. 38, No. 5, pp. 328–336.

Parkhurst, D. L. and Appelo, C. A. J., 2013, Description of Input and Examples for PHREEQC Version 3: A Computer Program for Speciation, Batch-Reaction, One-Dimensional Transport, and Inverse Geochemical Calculations: U.S. Geological Survey Techniques and Methods 6-A43.

Plummer, L. N., 1977, Defining reactions and mass transfer in part of the Floridan Aquifer: Water Resources Research, Vol. 13, pp. 801–812.

Ryan, M. and Meiman, J., 1996, An examination of short-term variations in water quality at a karst spring in Kentucky: Groundwater, Vol. 34, No. 1, pp. 23–30.

Schincariol, R. A. and Schwartz, F. W., 1990, An experimental investigation of variable density flow and mixing in porous media: Water Resources Research, Vol. 26, No. 10, pp. 2317– 2329.

Sheppard, S. M. and Schwarcz, H. P., 1970, Fractionation of carbon and oxygen isotopes and magnesium between coexisting calcite and dolomite: Contributions Mineralogy Petrology, Vol. 26, pp. 161–198.

Shuster, E. and White, W. B., 1971, Seasonal fluctuations in the chemistry of limestone springs: A possible means for

characterizing carbonate aquifers: Journal Hydrology, Vol. 14, pp. 93–128.

U.S. Environmental Protection Agency, 2019, National Primary Drinking Water Regulations: Electronic document, available at https://www.epa.gov/ground-water-anddrinking-water/national-primary-drinking-water-regulations Wang, L. C. and Zhang, Y. Z., 2001, Karst conduit flow and its hydrodynamic characteristics—Houzhai River drainage basin in Puding, Guizhou, China, as an example: Chinese Science Bulletin, Vol. 46, pp. 45–51.

White, W. B., 1988, Geomorphology and Hydrology of Karst Terrains: Oxford University Press, New York.

Wigley, T. M. L., 1973, The incongruent solution of dolomite: Geochimica Cosmochimica Acta, Vol. 37, pp. 1397–1402.

Yan, J. H.; Li, J. M.; Ye, Q.; and Li, K., 2012, Concentrations and exports of solutes from surface runoff in Houzhai Karst Basin, southwest China: Chemical Geology, Vol. 304, pp. 1–9.

Yang, R.; Liu, Z. H.; Zeng, C.; and Zhao, M., 2012, Response of epikarst hydrochemical changes to soil CO2 and weather conditions at Chenqi, Puding, SW China: Journal Hydrology, Vol. 468, pp. 151–158.

Yang, Y., 2001, A study of the structure of karst aquifer medium and the groundwater flow in Houzhai underground river basin: Carsologica Sinica, Vol. 20, No. 1, pp. 17–20.

Yu, J., and Zhang, H., 1988, Karst geomorphology in Puding, Guizhou Province: Carsologica Sinica, Vol. 7, No. 2, pp. 163– 172.

Yue, F.-J.; Li, S.-L.; Zhong, J.; and Liu, J., 2018, Evaluation of factors driving seasonal nitrate variations in surface and underground systems of a karst catchment: Vadose Zone Journal, Vol. 17, 170071.

Zhang, R. R.; Shu, L. C.; Zhu, J. T.; Yu, Z. B.; and Jiang, P., 2016, Storage and drainage characteristics of a highly heterogeneous karst aquifer in Houzhai basin: Groundwater, Vol. 54, No. 6, pp. 878–887.

Zhang, Z.; Chen, X.; Cheng, Q.; and Soulsby, C., 2019, Storage dynamics, hydrological connectivity and flux ages in a karst catchment: Conceptual modelling using stable isotopes: Hydrology Earth System Sciences, Vol. 23, pp. 51–71.

Zhang, Z.; Chen, X.; and Soulsby, C., 2017, Catchment-scale conceptual modelling of water and solute transport in the dual flow system of the karst critical zone: Hydrological Processes, Vol. 31, No. 19, pp. 3421–3436.

Zhang, Z.; Zhang, J.; Yang, J.; and Shen, P., 1998, Research on the spatial structure of karst massif—Taking the basin of the Houzhai subterranean stream in Puding County for example: Chinese Geographical Science, Vol. 8, No. 3, pp. 256– 263.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 273–289

289

EEGS-26-02-08_1XO

July 8, 2020

9:3

0

EEGS-26-01-05_1XO

July 8, 2020

9:29

Discrimination of Nitrogen Sources in Karst Spring Contributing Areas Using a Bayesian Isotope Mixing Model and Wastewater Tracers (Florida, USA) ANDY CANION* St. Johns River Water Management District, 4049 Reid Street, Palatka, FL 32177

KATHERINE M. RANSOM Department of Land, Air, and Water Resources, University of California, One Shields Avenue, Davis, CA 95616

BRIAN G. KATZ Florida Department of Environmental Protection, 2600 Blair Stone Road, Tallahassee, FL 32399

Key Terms: Springs, Karst, Nitrogen, Wastewater, Mixing Model ABSTRACT Many springs in Florida have experienced a proliferation of nuisance algae and alteration of trophic structure in response to increases in nitrate concentration concurrent with rapid population growth and land use intensification beginning in the mid-20th century. While loading targets and remediation plans have been developed by state agencies to address excess nitrogen inputs, further confirmation of the relative contribution of nitrogen sources to groundwater is necessary to optimize the use of resources when implementing projects to reduce nitrogen loads. In the present study, stable isotopes of nitrate and wastewater indicators were used to discriminate sources of nitrogen in wells and springs in central Florida. Sampling was performed in 50 wells at 38 sites and at 10 springs with varying levels of agriculture and urban development. Nitrate isotope values were used to develop Bayesian mixing models to estimate the probability distribution of the contributions of nitrate sources in wells. Prior probabilities for the fractional contribution of each source were adjusted based on land use and density of septic tanks. Sucralose and the Cl:Br mass ratio were used as confirmatory indicators of wastewater sources. In residential areas, mixing model results indicated that fertilizer or mixed fertilizer and wastewater (septic tank effluent and reuse water) were the primary sources, with sucralose detections corresponding to wells with elevated contributions from wastewater. Sources of nitrogen in pasture and field crop areas were pri*Corresponding author email: acanion@sjrwmd.com

marily fertilizer and manure; however, model posterior distributions of δ15 N indicated that manure sources may have been overpredicted. The present study demonstrates the utility of a multi-tracer approach to build multiple lines of evidence to develop locally relevant remediation strategies for nitrogen sources in groundwater.

INTRODUCTION Florida has over 700 documented springs, the majority of which are karst springs fed by the Floridan Aquifer System (Scott et al., 2004). Florida’s springs are concentrated in the northwest and north-central areas of the state and contribute significant flows to river systems in these areas. Many of these springs have experienced varying levels of ecosystem degradation in response to anthropogenic stressors. The most widely observed impact in spring systems is the proliferation of epiphytic and benthic algae, which has been attributed primarily to increases in nitrate concentrations at spring vents (Stevenson et al., 2007; Quinlan et al., 2008). Across Florida springs, nitrate-N concentrations have risen from pre-development values of less than 0.1 mg L−1 to present day values of between 0.5 and 5.0 mg L−1 , with nitrate-N concentrations in some small springs exceeding 30 mg L−1 (Katz, 1992; Katz et al., 1999). In response to the impact of elevated nitrate at springs, the state of Florida established a numeric nutrient criterion that nitrate plus nitrite not exceed 0.35 mg L−1 at spring vents (Fla. Admin. Code R. 62-302.531), based on limiting concentrations for algal growth.

Determining the relative contribution and locations of nitrogen sources to groundwater within springs contributing areas (springsheds) is critical for planning of

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

291

EEGS-26-01-05_1XO

July 8, 2020

9:29

Canion, Ransom, and Katz

cost-effective reduction of nitrogen loading to springs and downstream water bodies. Previous work has established springshed-scale nitrogen budgets using a combination of land use, specific source data (e.g., septic tanks, wastewater application), hydrogeologic data, and nitrogen attenuation factors based on literature reports (Katz et al., 2009; Tucker and Stroehlen, 2010; and Eller and Katz, 2017). Although these approaches provide spatially explicit estimates of nitrogen sources and loads at the land surface and top of aquifer, they have limited ability to account for spatial heterogeneity in soil biogeochemical transformations, aquifer transport, and attenuation by aquifer denitrification. Given that large springsheds may range in size from hundreds to thousands of square kilometers, further spatial refinement of groundwater impacts may be accomplished using geospatial and statistical models to predict groundwater nitrate concentrations, which implicitly account for spatial heterogeneity (Nolan et al., 2002; Almasri and Kaluarachchi, 2007; and Canion et al., 2019). Further discrimination of differential transport and attenuation of nitrogen sources may be accomplished using geochemical tracers. This approach is often desirable where heterogeneous land use and multiple nitrogen sources impact groundwater.

The most widely used method for determining nitrate sources in groundwater is the dual isotope approach, where the isotopic ratios of nitrogen (δ15 N) and oxygen (δ18 O) are measured simultaneously in nitrate (Xue et al., 2009). Because of differences in source isotopic ratios, this approach may be used to quantify, based on mass balance, nitrate contributions from atmospheric deposition, nitrate fertilizer, ammonia/organic nitrogen fertilizer, natural soil organic matter, wastewater, and manure (Kendall et al., 2008).

However, the dual isotope approach still has limitations for identification and quantification of sources.

Assumptions must be made about the extent of isotopic overprinting by biogeochemical cycling in the soil and aquifer (Kendall et al., 2008), and quantification by linear mixing models is complicated by significant overlap in the distributions of source isotopic values and undetermined systems (>n + 1 sources, where n is the number of isotope tracers). Statistical summaries that provide a range of solutions were proposed by Phillips and Gregg (2003) in order to overcome scenarios where no unique mixing model solution existed.

Recently, the incorporation of a Bayesian framework into isotope mixing models has furthered statistical estimation of nitrate sources by incorporating variation in source isotopic values and prior information on source contribution. The SIAR (Stable Isotope Analysis in R) Bayesian mixing model (Par-

292

nell et al., 2010) has been applied in agriculturallydominated watersheds in Belgium and France to determine contributions from fertilizer, manure, and wastewater to springs and surface water (Xue et al., 2012; El Gaouzi et al., 2013). The SIAR model was also recently applied to stormwater runoff data from residential catchments in Florida, where it was determined that atmospheric deposition and synthetic fertilizers were the primary sources of nitrate (Yang and Toor, 2016). Ransom et al. (2016) used a Bayesian framework with two additional tracers (δ11 B and iodine concentration) to discriminate septic tank, manure, and fertilizer nitrogen sources in shallow domestic wells in the Central Valley in California, and source contributions were validated by an analysis of surrounding land use. Unlike SIAR, the Central Valley model was able to assign non-normal prior distributions for source isotopic ratios.

Additional geochemical markers may also be used to supplement nitrate stable isotopes in determining groundwater nitrogen sources. Boron stable isotope measurements (δ11 B) have been used to improve discrimination between wastewater and animal manure sources (Vengosh, 1998; Widory et al., 2004, 2005).

The ratio of chloride to bromide, along with chloride concentration, was shown to be a marker for septic tank nitrogen inputs in multiple aquifers in the United States (Katz et al., 2011). A number of wastewater organic microconstituents have also been evaluated as conservative, co-migrating tracers of nitrogen (Badruzzaman et al., 2013). Of these organic tracers, sucralose has been identified as one of the most conservative because it is highly soluble, has low rates of microbial degradation (half-life ∼1 year), and has low potential sorption to soils (Oppenheimer et al., 2011; Soh et al., 2011). Statewide surveys in Florida have seen high detection rates for sucralose, indicating its usefulness as a widespread wastewater tracer in surface and groundwater (Silvanima et al., 2018). Sucralose does not, however, discriminate between septic tank nitrogen sources and land-applied treated wastewater (reuse); therefore, tracers unique to wastewater treatment facilities, including gadolinium anomaly and iohexol, have been proposed to further discriminate between wastewater sources (Schmidt et al., 2013).

Although land surface nitrogen load targets have been developed to address excess nitrogen loading to many springs in Florida, further confirmation of the relative contribution of local nitrogen sources to groundwater is necessary to optimize the use of resources when implementing projects to reduce nitrogen loads. In the present study, stable isotopes of nitrate and wastewater indicators were used to discriminate local sources of nitrogen in wells and springs in central Florida. A Bayesian mixing model was fit to the dual

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

EEGS-26-01-05_1XO

July 8, 2020

9:29

Nitrogen Sources in Springsheds

Figure 1. (A) Location of springs sampled: (1) Silver Springs, (2) Silver Glen Springs, (3) Alexander Springs, (4) Rock Springs, (5) Wekiva Springs, (6) Sanlando Spring, (7) Starbuck Spring, (8) Gemini Springs, (9) Blue Spring, (10) Ponce DeLeon Springs. Dashed lines indicate contributing areas based on the Upper Floridan Aquifer potentiometric surface. (B) Well sites (corresponding to Table 3) and land use (2014) in the study area. The St. Johns River Water Management District (SJRWMD) boundary is shown in red.

nitrate isotope data from wells using the JAGS software (Plummer, 2017), based on the methods described

by Ransom et al. (2016). Land use data were applied to set informative prior distributions on the nitrate source contributions to wells in the mixing models. Analysis of wastewater indicators provided additional evidence for contributions of wastewater nitrogen. In wells with only reduced nitrogen or where denitrification had occurred, the mixing model could not be applied, and wastewater indicators provided an alternative indicator of nitrogen source.

STUDY AREA AND METHODS Study Area The area of study was in central Florida along the middle St. Johns and Ocklawaha rivers (Figure 1), an area characterized by mantled karst topography. These two river valleys lie between ridges with sandy soils and high recharge potential, and significant groundwater inputs to the rivers from artesian Upper Floridan Aquifer springs as well as diffuse groundwater flow occur in this region (Belaineh et al., 2012). The Upper

Floridan Aquifer is highly permeable in most places and includes the Ocala Limestone and Avon Park formations, where transmissivity values are between 900 and 9,000 m2 d−1 (Kuniansky et al., 2012). The Upper Floridan Aquifer is mostly unconfined in the northwestern parts of the study area (near Silver Springs) and is semi-confined to confined by the intermediate confining unit, comprised of Miocene Hawthorn Group clays, in the southern and eastern parts of the study area. In confined areas, a surficial aquifer is present that is capable of recharging to the Upper Floridan Aquifer due to leakance and solution features (Boniol et al., 1993).

Four first-magnitude springs (>2.8 m3 s−1 discharge), six second-magnitude springs (>0.28 m3 s−1 discharge), and 50 wells within contributing areas (springsheds) were selected for study (Table 1). The wells sampled withdraw water from the Upper Floridan Aquifer (33 wells), producing zones within the intermediate confining unit (intermediate aquifer, five wells), and the surficial aquifer (12 wells). Dominant land use between the springsheds is varied, including forested, urban/residential, and mixed urban/residential and agriculture. Additionally, some

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

293

EEGS-26-01-05_1XO

July 8, 2020

9:29

Canion, Ransom, and Katz Table 1. Hydrologic characteristics of springs and land cover of contributing areas.1

Spring Alexander Springs Blue Spring Gemini Springs Ponce De Leon Springs Rock Springs Sanlando Spring Silver Glen Springs Silver Springs Starbuck Spring Wekiwa Springs

Estimated Springshed Area (km2 )

Average Discharge (m3 s−1 )

Apparent Age2 (yr)

Chemistry Type

Agriculture (%)

Urban and Residential (%)

Forest, Water, Wetlands (%)

Total Septic Tanks

620 636 110 263 7413 804 491 2,324 804 7413

3.0 4.2 0.3 0.8 1.7 0.5 3.0 21.7 0.4 1.9

24.7 22.2 <5 21.7 21.3 22.8 26.1 34.5 19.1 13.5

Na-Cl Na-Cl Na-Cl Na-Cl/Ca-HCO3 Ca-HCO3 Ca-HCO3 Na-Cl Ca-HCO3 Ca-HCO3 Ca-HCO3

8 7 3 17 9 0 1 23 0 9

6 28 52 9 39 77 5 21 73 39

86 65 45 74 52 23 94 56 27 52

7,200 51,400 4,600 4,400 39,300 7,600 4,400 63,000 8,000 39,300

1 Land cover percentages are from the 2014 Land Use/Land Cover map (SJRWMD, 2018). Total septic tanks are from the Florida Water Management Inventory Project (FDOH, 2016).

2 Based on 3 H/3 He (Toth and Katz, 2006; Walsh, 2009; and Knowles et al., 2010).

3 Wekiwa and Rock Springs delineated as a combined springshed.

4 Approximated based on a 5-km-radius circular buffer; size chosen based on ratio of springshed area to discharge at Wekiva Springs.

springsheds have a high percentage of developed areas with septic tanks (Florida Department of Health [FDOH], 2016). The springs vary in major ion composition, with springs near the St. Johns River having a Na-Cl water type due to relict (or residual connate water) seawater inputs and springs farther from the St. Johns River having a Ca-HCO3 water type consistent with dissolution of limestone that forms the Upper Floridan Aquifer. The apparent water age of all study springs is young, ranging from less than 5 years old to 36 years old (Table 1). Previous work has demonstrated that apparent ages in these springs is a result of mixing of 30%–70% recently recharged water (<10 years) with older (>60 years) water (Toth and Katz, 2006).

Sampling and Analytical Methods Sampling at the springs and wells occurred between February and July 2018 and was performed in accordance with Florida Department of Environmental Protection (FDEP) standard protocols. Field measurements of conductivity and dissolved oxygen were recorded at the time of sampling. Samples collected for nitrate dual isotope analysis were filtered through a 0.45-µm filter into high-density polyethylene bottles at the time of sampling and frozen until analysis. Samples for sucralose and iohexol were collected unfiltered in glass containers (amber glass for iohexol) and refrigerated until analysis.

Nitrogen species, chloride, and bromide were analyzed at the St. Johns River Water Management District (SJRWMD) using analytical methods approved by the U.S. Environmental Protection Agency. Method

294

detection limits for NO3 − , NH4 + , and TKN, were 0.01, 0.005, and 0.05 mg L−1 , respectively. Chloride and bromide had detection limits of 3 mg L−1 and 0.1 mg L−1 , respectively. Analysis of δ15 N and δ18 O of nitrate was performed at the University of California at Davis Stable Isotope Facility using the denitrifier method (Sigman et al., 2001; Casciotti et al., 2002). Reference standards for δ15 N and δ18 O were air and Vienna Standard Mean Ocean Water, respectively, and accepted analytical precision was 0.4‰ for δ15 N and 0.5‰ for δ18 O. Sucralose was analyzed via high-performance liquid chromatography/tandem mass spectrometry (LC-MS/MS) at the FDEP Central Laboratory (Tallahassee, FL) using the methods described in Silvanima et al. (2018). The method detection limit for sucralose was 10 ng L−1 . Iohexol was analyzed by Eurofins Eaton Analytical (Monrovia, CA) with solid phase extraction followed by LC-MS/MS. The minimum reporting level for iohexol was 100 ng L−1 .

Land Use and Wastewater Nitrogen Sources Land use surrounding each monitoring well was calculated from the St. Johns River Water Management District 2014 Land Cover/Land Use spatial layer (SJRWMD, 2018). Land use polygons were clipped in a 1-km-radius circular buffer using ArcGIS 10.6.1.

Polygon areas were aggregated based on the highest level (level 1) of the Florida Land Use, Cover and Forms Classification System to calculate the percentage of each land use category. To confirm whether changes in the dominant land use surrounding a well had occurred, a 1989 land use layer from SJRWMD

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

EEGS-26-01-05_1XO

July 8, 2020

9:29

Nitrogen Sources in Springsheds

was analyzed using the same buffers as the 2014 land use. The land use with the largest area within the buffer was compared between 1989 and 2014.

Wastewater sources to groundwater were not captured by land use coverages, and thus the proximity of wastewater-derived sources to groundwater was determined using other spatial data sets. Septic tank locations were derived from a parcel-based model developed by the FDOH (2016). In each buffer area used for land use analysis, the centroids were calculated for parcels classified as having a septic tank. The total number of septic tanks was then calculated for each buffer area. Application of treated wastewater (reuse water) for irrigation and recharge is a highly utilized strategy for water conservation in central Florida. Due to difficulties in accurately determining the quantities of water reuse application at spatial scales relevant to this study, a qualitative analysis was used to determine reuse application within well buffer areas. A spatial data set with reuse water service lines and destination areas (SJRWMD, unpublished) was overlaid with the 1-km-radius well buffers to determine the presence and type of reuse. Three types of reuse water locations were identified: (1) residential irrigation, (2) golf course irrigation, and (3) direct application of reuse (e.g., wastewater sprayfields, rapid infiltration basins, percolation ponds). The source reuse water was expected to have similar chemical composition, but the reuse locations varied in timing and intensity of water application and have variable capacity for nitrogen attenuation in the soil zone.

Bayesian Mixing Model A Bayesian mixing model was developed with the nitrate isotope values (δ15 N, δ18 O) as tracers and four sources of nitrate, including wastewater (septic tank effluent and reuse water), manure, NO3 − fertilizer, and NH4 + fertilizer (includes synthetic organic nitrogen fertilizer). Atmospheric deposition of nitrogen was not included based on the observation that δ18 O values indicated minimal atmospheric deposition contribution.

Nitrogen from naturally occurring organic matter in soils was likewise excluded from the analysis based on the rationale that all the wells chosen for study had elevated nitrogen from anthropogenic sources. Also, most soils in the study area contain low amounts of organic matter. Background nitrate concentrations in Floridan Aquifer water are generally less than 0.1 mg L−1 (Katz, 1992; Cohen et al., 2007), and thus natural soil organic nitrogen is expected to contribute minimally to wells sampled in the present study. Wells with evidence of denitrification were excluded from the analysis. A threshold concentration of 0.5 mg L−1 dissolved oxygen was used to exclude wells where significant

denitrification had occurred. After removal of wells with presumed denitrification, a total of 31 wells were selected for analysis in the mixing model.

The mixing model was based on two mass balance equations from Ransom et al. (2016) and initially from Accoe et al. (2008) where each measured isotope value was modeled as a linear combination of the fractional contribution of each source of nitrate to the well multiplied by the isotopic signature of each corresponding source: 4 δ15 Nmixture = fsource ∗ δ15 Nsource + εN (1) source=1

δ18 Omixture =

4 source=1

fsource ∗ δ18 Osource + εO (2)

where ε is a Gaussian error term reflecting stochastic variation in the isotope value and is unique to each well. For use in the model, the mass balance equations were combined into matrix form. The matrix form of the mixing model initially comes from Massoudieh and Kayhanian (2013) and is written as C = YF + E

(3)

where C is a 2-by-31 matrix of two measured isotope values in 31 wells, Y is a 2-by-4 matrix of two isotopic signatures in four sources, F is a 4-by-31 matrix of four fractional source contributions to each of the 31 wells, and E is a 2-by-31 matrix of errors for each tracer in each well. A Dirichlet distribution was used as the prior distribution for the fractional contributions of each source of nitrate to each well. The Dirichlet distribution is a multivariate generalization of the beta distribution for K classes (K = 4 sources) and ensures that the fractional source contributions from each well will sum to one.

Informative prior means for the fractional contributions of wastewater and manure were set for each well based on land use and the number of septic tanks in a 1-km radius (Appendix, Table A1). The Dirichlet parameter, α, was adjusted where there was less prior evidence for either wastewater or manure sources. The resulting prior mean fractional contribution for each source is calculated as αsource (4) μsource = 4 source=1 αsource The wastewater source α prior was adjusted according to the following rules for the percentage of urban land use with a given number of septic tanks: a. 1/5 for ࣘ 20% urban land use with ࣘ 100 septic tanks b. 1/2 for ࣙ 20% urban land use with ࣘ 100 septic tanks c. 1 for ࣙ 20% urban land use with ࣙ 100 septic tanks

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

295

EEGS-26-01-05_1XO

July 8, 2020

9:29

Canion, Ransom, and Katz

Similarly, the manure source α prior was adjusted based on the percentage of agricultural land use according to the following rules: a. b. c. d.

1/10 for < 1% agricultural land use 1/5 for 1%–10% agricultural land use 1/2 for 10%–20% agricultural land use 1 for > 20% agricultural land use

Dirichlet priors for fertilizer sources were not adjusted based on land use because fertilizer application occurs at both agricultural and residential sites and because limited information on the relative use of nitrate versus ammonium/organic fertilizers was available. The alpha parameters were assigned to the Dirichlet distribution for each fractional source contribution of nitrate to each well in order to achieve the desired informative prior mean and relatively low prior variance. Low prior variance assisted model convergence and required strong evidence of the isotopic measurements to move the posterior fractional source contribution distributions away from the priors.

Literature values were compiled to establish empirical prior distributions for each combination of tracer and source (Appendix, Table A2). Prior distributions were fit using the fitdistrplus and MASS packages within the R statistical computing environment (Delignette-Muller et al., 2019; R Core Team, 2019; and Ripley et al., 2019). A gamma or a Student t distribution was used as the prior distribution for the δ15 N and δ18 O source signatures. The gamma distribution was chosen for δ15 N in manure and wastewater, as the isotopic δ15 N values in these two sources are unlikely to be negative and the distributions tend to be right-skewed. We chose not to include the dairy manure source δ15 N values referenced in Ransom et al.

(2016) because the livestock in the present study area were almost entirely beef cattle and horses. This led to a lower mean δ15 N value for the manure source, which more accurately reflects lower ammonia volatilization and lower isotopic enrichment associated with pasture applied and stockpiled manure. Prior distributions for δ15 N in NO3 − and NH4 + fertilizer, as wells as prior distributions for all δ18 O source values, were fit using a Student t distribution with 10 degrees of freedom. A Gaussian distribution was chosen as the likelihood distribution to represent Eq. 3. The choice of the Gaussian distribution ensured the real space, linear combination of isotopic values from each of the four sources as described in Eqs. 1 and 2.

Markov chain Monte Carlo methods were used to estimate the posterior probability distribution of each of the four sources’ fractional contribution of nitrate to each well. Analysis was performed with the Gibbs sampler JAGS (Plummer, 2017) within the R statistical computing environment (R Core Team, 2019)

296

using the package rjags (Plummer, 2018). We used an adaptation and burn-in phase of 2 million iterations for each of two chains. The sampling phase consisted of 50,000 samples with a thinning rate of 100 for a total of 500 posterior realizations retained from each chain. Finally, the posterior realizations from each chain were combined for each monitored parameter.

The final model run took approximately 2 hours on an Intel Xeon E3-1505M chipset running at 3.00 GHz with 32 GB of ECC DDR4 RAM at 2,400 MHz. Trace, autocorrelation, running mean, and density plots were visually inspected for model convergence and proper mixing of chains. Normalized central tendencies were calculated for each set of four fractional source contributions for each well as a summary measure of the likely proportional contribution of each source to each well. Normalized central tendencies were calculated as the geometric mean of each posterior distribution of fractional source contributions, normalized by well.

RESULTS AND DISCUSSION Nitrogen Forms and Concentrations Nitrogen in Spring Vents Nitrate-N concentrations in springs ranged from 0.04 to 0.05 mg L−1 in springs with unimpacted, forested watersheds (Alexander Springs, Silver Glen Springs) up to 0.39–1.4 mg L−1 for the remainder of springs (Table 2). The range of nitrate-N for the current study springs, which have springsheds influenced primarily by urban and residential land use, is lower than a previously reported range of 1.0–4.2 mg L−1 for Florida springs with predominantly agricultural land use (Katz, 2004). It is important to note that some of the difference in nitrate concentrations between the current study springs with residential land use and previous results from springs in agricultural areas may be due to regional differences in aquifer geochemistry. Previous work by Heffernan et al. (2012) in Florida springs found significant denitrification (35%–90% N removal) in all of the current study springs based on dissolved gas and other chemical data (Table 2), whereas the agriculturally influenced springs studied by Katz (2004) were found to have less nitrogen removed by denitrification (0%–40%). This difference in denitrification is most likely due to a lack of confinement of the Upper Floridan Aquifer and oxic conditions in the agriculturally influenced springs, whereas in all the current study springs except Silver Springs, some degree of confinement was present within their springsheds, allowing for the development of low dissolved oxygen and subsequent denitrification.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

EEGS-26-01-05_1XO

July 8, 2020

9:29

Nitrogen Sources in Springsheds Table 2. Water quality parameters and source indicators at springs.

Site Alexander Springs Blue Spring Gemini Springs Ponce De Leon Springs Rock Springs Sanlando Spring Silver Glen Springs Silver Springs Starbuck Spring Wekiwa Springs

NO3 -N (mg L−1 )

NH4 -N (mg L−1 )

TKN (mg L−1 )

δ15 NNO3 Air (‰)

0.04 0.79 1.40 0.81

0.01 ND 0.03 0.04

0.05 0.07 0.10 0.10

8.01 11.82 7.09 9.80

1.28 0.62 0.05 1.24 0.39 1.12

ND 0.11 0.01 0.01 0.04 ND

0.04 0.20 ND 0.08 0.06 0.08

6.60 14.88 5.43 7.30 19.99 11.49

δ18 ONO3 VSMOW (‰)1

Dissolved Oxygen (mg L−1 )

Percent of Nitrogen Remaining2

Sucralose (ng L−1 )

Cl:Br

Iohexol (ng L−1 )

2.34 6.95 4.28 9.54

1.9 1.4 1.0 1.1

12% 13% 50%–65% 21%–31%

ND3 320 330 130

331 304 345 290

ND ND ND ND

8.08 10.39 − 0.39 5.74 8.91 9.67

1.0 0.6 3.6 2.3 0.8 0.5

43% 10% 11% 50%–93% 10% 30%

160 960 ND 36 460 870

270 371 340 327 341 360

ND ND ND ND ND ND

VSMOW = Vienna Standard Mean Ocean Water.

Data from Heffernan et al. (2012).

3 ND = not detected.

1 2

Reduced forms of nitrogen were also present at spring vents but were significantly lower than nitrate.

Total Kjeldahl nitrogen (TKN) concentrations were between 0.04 and 0.2 mg L−1 , and ammonium-N concentrations were between 0 and 0.11 mg L−1 . The presence of reduced forms of nitrogen may be an indication of (1) assimilatory nitrate reduction and dissimilatory nitrogen reduction to ammonium (DNRA) in reducing areas of the surficial and Upper Floridan aquifers or (2) mixing of deep Upper Floridan and Lower Floridan water with shallow upper Floridan water (Toth and Katz, 2006; Knowles et al., 2010).

Recently, in situ push-pull tracer experiments provided rate estimates and microbial functional gene evidence for DNRA as a significant nitrate reduction pathway in the Upper Floridan Aquifer that co-occurs with denitrification (Henson et al., 2017). An analysis of deeper, Lower Floridan Aquifer wells in the study area revealed consistently higher ammonia concentrations (0.30 ± 0.28 mg L−1 NH4 -N) than nitrate concentrations (0.01 ± 0.003 mg L−1 ), indicating the DNRA may be a significant nitrate reduction pathway in the deeper groundwater that mixes with more recently recharged water (n = 10; SJRWMD unpublished data).

Nitrogen in Wells Surficial and intermediate groundwater well NO3 N concentrations were generally low and ranged from 0.01 to 1.32 mg L−1 , with a median value of 0.06 mg L−1 . In Upper Floridan wells, NO3 -N exhibited two patterns dependent on the springshed where they were located (Table 3). Median concentrations of NO3 -N were 1.87 mg L−1 in wells from the

Silver Springs springshed and <0.01 mg L−1 all other springsheds. For the Silver Springs springshed, NO3 N in wells ranged from 0.35 to 12.79 mg L−1 , while in the remainder of the wells, only two wells had elevated NO3 -N above 0.02 mg L−1 . The observed difference in groundwater nitrate concentrations between the Silver Springs springshed and the other springsheds is most likely a result of differences in nitrogen removal by denitrification due to differences in the confinement and depth to the Upper Floridan Aquifer. The Floridan Aquifer is largely unconfined, shallow, and oxic in the western half of the Silver Springs springshed, and the western springshed receives most of the nitrogen loading. The Wekiwa, Blue, Gemini, and DeLeon springsheds have a thicker confining layer, the Floridan Aquifer is generally deeper, and low dissolved oxygen is more prevalent (Boniol et al., 2014).

Reduced nitrogen forms were higher than NO3 -N in most of the surficial and intermediate aquifer wells as well as in Upper Floridan wells in confined (Wekiwa, Blue, Gemini, and De Leon) springsheds. In surficial and intermediate aquifer wells, concentrations of TKN were between 0 and 2.17 mg L−1 , with a median value of 0.22 mg L−1 (Table 3). Concentrations of NH4 -N were between 0 and 2.06 mg L−1 , with a median value of 0.14 mg L−1 . Upper Floridan wells in the Wekiwa, Blue, Gemini, and De Leon springsheds both had similar ranges (0–1.92 mg L−1 ) and median values (0.19 mg L−1 ) of TKN and NH4 -N, and ammonium made up 90% or more of the TKN in Upper Floridan wells with elevated reduced nitrogen concentrations (>0.2 mg L−1 TKN). The predominance of high ammonium concentrations in this region of the aquifer provides further evidence of the potential for DNRA to be an important mechanism for

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

297

EEGS-26-01-05_1XO

July 8, 2020

9:29

Canion, Ransom, and Katz

Figure 2. Biplot of nitrogen and oxygen stable isotopes for springs and wells. Source isotopic ranges are represented using boxes bounded by the emprical mean ± 2 standard deviations of the literature values used in the present study to model prior distributions. The approximate trajectory of denitrification (2:1, δ15 N:δ18 O) is overlaid for reference.

nitrate reduction that may compete with denitrification. Although DNRA is expected to be less favored than denitrification under the low organic carbon conditions found in the Floridan Aquifer, the presence of sulfide under reducing conditions in the aquifer may simultaneously inhibit denitrification and provide an electron donor for chemolithoautotrophic DNRA (Rye et al., 1981; Burgin and Hamilton, 2007).

Nitrate Isotope Measurements Nitrate isotope values in springs samples were consistent with a denitrification influence for most springs (Table 2 and Figure 2). Four springs had isotope values that were not significantly elevated and had signatures that appeared consistent with either soil organic nitrogen (Alexander and Silver Glen Springs) or a mixture of fertilizer and wastewater/manure sources (Silver and Rock Springs). However, spring vent samples were not included in the Bayesian mixing model because of previously determined denitrification (Heffernan et al., 2012) and evidence that the spring vents are comprised of a mixture of old and new water (Toth and Katz, 2006). Dual isotope analysis of nitrate was performed in 38 of the 50 study wells, as 12 wells had nitrate concentrations that were below the quantitation limit for dual isotope analysis (Table 3). A majority of wells exhibited nitrate isotope signatures consistent with a mixture of fertilizer and wastewater/manure (Figure 2), and soil nitrogen was assumed to be a minor contributor to nitrate in wells because wells were

298

selected for the study based on elevated nitrate concentrations from anthropogenic sources.

Heavy δ15 N and δ18 O values consistent with denitrification were observed in wells with low dissolved oxygen (<0.5 mg L−1 ). Four surficial wells had low dissolved oxygen concentrations and isotope values that indicated that denitrification had occurred (Table 3). This was an unexpected result for a shallow, unconfined aquifer; generally, such aquifers are oxic, have high nitrate concentrations, and experience limited denitrification (Burow et al., 2010). Previous sampling of shallower (approximately 3 m) wells in the Wekiwa Springs area yielded higher nitrate concentrations than the present study (2.4 ± 0.3 mg L−1 ), with less influence of denitrification (Tucker et al., 2014).

The isotopic evidence for denitrification and the presence of reduced forms of nitrogen in surficial wells in the present study suggest that reducing conditions are more prevalent than previously known in the surficial aquifer and may allow for attenuation of nitrogen by denitrification in water prior to its recharging of the Floridan Aquifer.

Upper Floridan wells exhibited a spatial pattern in isotope values consistent with the forms of nitrogen.

Wells in the mostly unconfined Silver Springs springshed showed no evidence for denitrification, consistent with the high dissolved oxygen concentrations and predominance of nitrate. Most Upper Floridan wells in the remaining springsheds had nitrate concentrations that were too low for dual isotope analysis. Based on the isotopic evidence for denitrification, wells with dissolved oxygen concentrations below 0.5 mg L−1 were

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

EEGS-26-01-05_1XO

July 8, 2020

9:29

Nitrogen Sources in Springsheds

excluded from mixing model analysis. These results highlight the need to verify the assumptions of limited influence of denitrification in isotopic nitrogen source studies as well as the value in adding additional tracers for nitrogen sources when reduced nitrogen species dominate in groundwater and preclude dual isotope analysis.

Wastewater Markers Wastewater Markers in Springs The wastewater markers sucralose, iohexol, and the chloride-to-bromide mass ratio (Cl:Br) were analyzed in springs and wells for confirmation of wastewater sources. Sucralose was detected at all springs except those with forested watersheds (Table 2), and concentrations were between 36 and 960 ng L−1 . The three springs with the highest sucralose concentrations (Sanlando, Starbuck, and Wekiwa) have high densities of septic tanks close to the springs but also have extensive application of reuse water in the area. Iohexol was analyzed in order to discriminate septic tank effluent from reuse water; however, it was not detected at any spring sites. Although iohexol has been used to detect runoff of reuse water irrigation in surface waters, there are still uncertainties in the photodegradation and soil biodegradation rates (Oppenheimer et al., 2018). Further confirmation that iohexol is an appropriate conservative tracer in groundwater may be necessary before its routine implementation as a reuse water tracer. The Cl:Br ratios were analyzed as an additional marker for wastewater contributions. Springs with Na-Cl water type had ratios reflective of a seawater Cl:Br ratio (Figure 3). This is consistent with previous work that identified relict seawater as the source of major ions in the middle reach of the St. Johns River (Belaineh et al., 2012). The remaining springs fell close to the mixing line between septic leachate and dilute groundwater.

An approximation of the contribution of wastewater (both septic effluent and reuse water) by volume to springs can be made based on dilution of sucralose assuming that (1) sucralose has approximately equal endmember concentrations in all wastewater sources and (2) soil adsorption and attenuation is negligible. Sucralose concentrations have been shown to be similar between septic tanks and reuse system effluent, with average concentrations between 40,000 and 50,000 ng L−1 (Oppenheimer et al., 2011; Schmidt et al., 2013; and SJRWMD, unpublished data). Using the range of observed sucralose concentrations at the spring sites, an approximate volumetric contribution of 0.1%–2.4% for wastewater sources is estimated for the study springs. Similar contributions by septic tank effluent (0.1%–1.0%) have been estimated using the

artificial sweetener acesulfame in groundwater seeps in Ontario, Canada (Spoelstra et al., 2017). Under a hypothetical situation where only septic tank effluent or reuse water was the nitrogen source to a spring, an upper bound on the spring nitrate concentrations can be estimated. Based on literature values of total nitrogen concentrations in septic tank effluent (58 mg L−1 ; Lusk et al., 2017) and reuse water (3–12 mg L−1 ; Badruzzaman et al., 2012), nitrate concentrations of 1.4 and 0.29 mg L−1 are estimated as the upper bound for springs by septic effluent and reuse water, respectively.

Actual contributions from each source are expected to be lower due to soil zone and aquifer attenuation processes. Attenuation of nitrogen from septic tank effluent and reuse water ranges from 40% to 75% and 50% to 85% in Florida soils, respectively (Eller and Katz, 2017, and references therein).

Wastewater Markers in Wells In well samples, sucralose was widely distributed in unconfined aquifers, with detections in 50% and 65% of wells in the surficial aquifer and Upper Floridan wells within the Silver Springs springshed, respectively (Table 3). A statewide survey of sucralose in Florida water bodies (Silvanima et al., 2018) had a similar detection percentage for unconfined aquifers (30%, n = 118). Sucralose was detected in only three of 13 (23%) Upper Floridan wells in the springsheds with widespread confinement (Wekiwa, Blue, Gemini, and DeLeon). Iohexol was analyzed in five wells where reuse water was a potential source of nitrogen but was not detected (Table 3).

Concentrations of sucralose in unconfined wells were between 15 and 730 ng L−1 , except for one result of 16,000 ng L−1 (well L-1026). This high value of sucralose was measured in the surficial aquifer directly under a small wastewater treatment plant with direct land application of treated effluent and indicated that the well was directly in the plume of effluent. Two other wells (OR0894 and M-0771) also had notably high sucralose concentrations (580 and 730 ng L−1 ) but likely differed in the source of nitrogen. At OR0894 (a surficial well), no reuse application was identified, but the surrounding number of septic tanks was 416.

A paired Upper Floridan well at the same site also had detectable sucralose (220 ng L−1 ), indicating a local region of poor confinement. At M-0771, there were 428 septic tanks in the surrounding area, but a golf course utilizing reuse water for irrigation was immediately adjacent and upgradient of the well. Additionally, most of the septic tanks at this site were downgradient on the potentiometric surface. Overall, sucralose concentrations in wells were lower than the respective springs, which suggests there may be a disproportionate

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

299

EEGS-26-01-05_1XO

July 8, 2020

9:29

Canion, Ransom, and Katz

Figure 3. Mass ratio of chloride to bromide versus chloride concentration in springs and study wells. Dilute groundwater endmember is from Davis et al. (2004), and septic leachate endmember is from Panno et al. (2006). The target area for septic leachate influence is represented by the black box (Katz et al., 2011). Binary mixing lines between septic leachate and dilute groundwater and between seawater and dilute groundwater are shown in purple and red, respectively.

contribution from wastewater nitrogen sources near the springs that was not adequately sampled rather than elevated nitrogen from wastewater sources throughout the springsheds.

Surficial and intermediate aquifer well Cl:Br mass ratios fell between the seawater-dilute groundwater and septic leachate-dilute groundwater mixing lines.

Based on a previous comparison of nitrogen isotope data and Cl:Br mass ratios in groundwater and springs (FDEP, unpublished data), sites with an inorganic nitrogen source generally had Cl:Br ratios below 400.

Nitrogen isotope data that were consistent with an organic nitrogen source typically corresponded to more variable Cl:Br ratios but were generally greater than 400. Four Upper Floridan Aquifer wells had high Cl:Br ratios (>400). All four wells were in areas with high densities of septic tanks and, in the case of two wells (M-0771 and M-0786), extensive application of reuse water. Two surficial wells (S-0716 and V-0814) had Cl:Br ratios >400; however, sucralose was not detected in either well. Other wells where sucralose was detected (up to 16,000 ng L−1 ) all had Cl:Br ratios less than 300, which may have resulted due to dilution of conservative ions by rainwater.

to estimate the fractional contribution of four nitrogen sources: wastewater, ammonium fertilizer and synthetic organic N fertilizers, nitrate fertilizer, and manure. The use of informative Dirichlet priors derived from land use is in contrast to previous mixing model studies of nitrate sources that implemented either vague (equal contributions of sources) Dirichlet priors (Xue et al., 2012; Yang et al., 2013; and Yang and Toor, 2016) or an informative prior for a single nitrogen source (soil organic nitrogen) across all sites (Ransom et al., 2016). Unequal prior means here were justified by relatively homogeneous distributions of land uses around well sites as well as similar agricultural use (pasture and field crops) at most agricultural sites. The present use of informative priors is analogous to isotope-based Bayesian mixing models of animal diets, where informative priors were assigned based on gut content analysis (Moore and Semmens, 2008). An advantage of using informative priors is that nitrogen sources with overlapping distributions of isotope values may be better discriminated based on prior knowledge of their contributions from an independent source of information.

Wastewater Sources

Bayesian Mass Balance Mixing Model Predictions of Nitrate-Nitrogen Sources A Bayesian mass-balance–based mixing model was developed using dual nitrate isotope data from wells

300

Posterior geometric means for the fractional contribution of wastewater sources (defined as both septic effluent and land-applied reuse water) were between <0.01 and 0.27 for agricultural areas and between

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

EEGS-26-01-05_1XO

July 8, 2020

9:29

Nitrogen Sources in Springsheds

Figure 4. Mixing model predicted fractional source contributions of groundwater nitrate in study wells. The geometric mean (black dot), 68% credible interval (thick line), and 95% credible interval (thin line) are shown for each well. Detection of sucralose (†) and Cl:Br ratios >400 (‡) are indicated on the wastewater panel. Wastewater refers to both septic tank effluent and land-applied treated wastewater.

<0.01 and 0.60 for residential areas (Figure 4). Evaluation of the central tendencies of the model results across wells allows for generalization, but the utility of Bayesian mixing models over traditional mixing models is the ability to characterize uncertainty around estimates of source contribution. For wastewater, the 95% credible intervals were large for most wells and were likely a result of the overlap in source isotopic distributions.

Contributions of wastewater in agricultural areas were further confirmed by wastewater markers at all sites that were classified primarily as agricultural.

These sites were primarily in the Silver Springs springshed and had septic tanks (44–152) in the 1-km buffer area, and sucralose concentrations were relatively low (15–41 ng L−1 ). For wells in residential areas, the mean fractional contribution of wastewater showed a weakly increasing trend with higher densities of septic tanks, and there was good agreement between

wastewater indicators and elevated contributions of wastewater sources (sucralose concentrations between 15 and 730 ng L−1 ). Wells directly below areas with land-applied treated wastewater had contrasting results: well S-1310 had a mean wastewater contribution close to the prior mean (0.30), and no wastewater markers were detected, whereas well M-0786 had higher predicted mean contribution from wastewater (0.48), and both sucralose and Cl:Br ratios indicated wastewater influence. These differences may be due to differences in nitrogen attenuation between sites but also may be due to well locations, as the monitoring wells were not sited with the specific intent of monitoring treated wastewater application.

Fertilizer Sources In agricultural areas, central tendencies were between 0.07 and 0.36 for ammonium fertilizer and

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

301

302

OR0107 S-1015 L-1026 OR0894 OR0661 OR1108 V-0197 V-1151 V-0814 S-0716 S-1310

V-1028 OR0651 OR0546 OR1109 V-0837 S-0717 M-0776

M-0782

M-0778

M-0041

M-0031

M-0527

M-0063

M-0217 M-0039 M-0213 M-0785 M-0419 M-00402 M-0465 M-07712 M-0774

12 13 14 6 15 10 16

17

18

19

20

21

22

23 24 25 26 27 28 29 30 31

Well ID

1 2 3 4 5 6 7 8 9 10 11

Well Site

UFA UFA UFA UFA UFA UFA UFA UFA UFA

UFA

UFA

UFA

UFA

UFA

UFA

Silver Silver Silver Silver Silver Silver Silver Silver Silver

Silver

Silver

Silver

Silver

Silver

Silver

DeLeon Wekiwa Wekiwa Wekiwa Blue Gemini Silver

Wekiwa Wekiwa Wekiwa Wekiwa Wekiwa Wekiwa Blue Blue Blue Gemini Gemini

60 40 30 90 64 90 120 72 90

120

120

66

80

55

195

50 72 60 90 60 64 50

40 50 47 20 44 39 30 37 42 20 35

Residential Residential Residential Residential Residential Residential Residential Residential Residential Residential Land-applied wastewater Dairy and field crops Forested Residential Residential Residential Residential Pasture and field crops Pasture and field crops Pasture and field crops Pasture and field crops Pasture and field crops Pasture and field crops Pasture and field crops Residential Residential Residential Residential Residential Residential Residential Residential Residential

Primary N-Contributing Land Use

65 81 108 135 138 347 413 428 658

144

111

96

81

76

52

27 0 82 618 334 196 44

29 297 338 416 596 618 18 56 302 196 0

Septic Tanks (1-km radius)

SF — — — — GCI — GCI —

—

—

—

—

—

—

— — — RI GCI SF —

RI — RIB, Perc — — RI RI — Perc, RI SF RIB

Water Reuse (1-km radius)

1.88 2.64 0.39 0.65 0.95 0.35 0.73 7.99 3.35

1.62

2.49

1.86

1.60

0.86

10.10

0.06 * 0.72 0.04 0.13 0.04 12.79

0.16 0.03 * 0.04 0.11 1.32 0.71 0.59 0.25 0.02 0.05

0.01 * * * * * * * *

0.02

*

*

0.02

0.01

*

0.14 0.2 0.04 0.12 * 1.14 *

0.01 0.36 2.06 0.35 0.66 0.01 * 0.03 0.19 0.39 0.06

0.08 0.09 0.09 0.08 * * * 0.06 0.05

*

*

*

0.06

0.06

*

0.34 0.22 0.26 0.15 * 1.57 0.05

0.18 1.52 2.17 1.39 0.86 0.04 0.08 0.15 0.22 1.25 0.15

4.26 1.50 3.77 7.71 3.15 2.68 3.85 7.24 5.21

4.35

7.79

4.79

5.90

16.03

4.75

16.04 ** 7.75 3.49 6.44 ** 6.56

3.79 5.37 ** 4.69 42.42 8.57 22.82 3.61 19.07 ** 5.99

3.78 0.70 1.91 3.30 1.05 5.04 3.52 5.53 1.79

3.27

6.64

4.27

5.69

13.89

11.51

14.98 ** 3.97 4.52 4.82 ** 4.06

− 0.44 18.15 ** 11.92 22.98 9.22 18.71 4.20 11.06 ** − 2.02

δ15 Nδ18 ONO3 Air NO3 VSMOW NO3 -N NH4 -N TKN (mg L−1 ) (mg L−1 ) (mg L−1 ) (‰) (‰)1

3.5 5.1 3.4 4.7 6.3 2.9 5.5 1.7 5.5

5.6

4.3

5.7

4.3

0.6

7.2

0.2 0.2 5.7 0.9 6.3 0.2 5.6

7.5 0.2 0.2 0.9 0.2 2.8 0.3 4.9 0.4 0.2 1.7

DO (mg L−1 )

26 * * 41 15 * * 730 *

19

41

34

15

16

26

* * 44 * * * 15

* 48 16,000 580 46 38 240 * * * *

147 * * 243 129 * * 596 335

*

367

*

*

247

303

214 290 291 154 229 133 *

207 152 184 176 129 236 262 65 394 519 200

Sucralose (ng L−1 ) Cl:Br

July 8, 2020

SA IA IA IA IA IA UFA

SA SA SA SA SA SA SA SA SA SA SA

Well Depth Aquifer Springshed (ft)

Table 3. Water quality parameters and source indicators in study wells (* = not detected, ** = not analyzed). Water Reuse abbreviations are: RIB: Rapid Infiltration Basin (land application at wastewater treatment facilities); Perc: Percolation Pond (Land application of treated wastewater from small treatment facilities); RI: Residential Irrigation; GCI: Golf Course Irrigation; SF: Sports Facilities.

EEGS-26-01-05_1XO 9:29

Canion, Ransom, and Katz

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

2

1

OR0106 OR0548 S-1014 OR0893 OR1110 V-0196 V-1152 V-0810 V-1169 S-1230

S-1408

V-1030 V-0156

37

12 38

DeLeon DeLeon

Gemini

Wekiwa Wekiwa Wekiwa Wekiwa Wekiwa Blue Blue Blue Blue Gemini

200 195

200

395 155 300 140 180 234 140 311 137 404

Residential Residential Residential Land-applied wastewater Residential Residential Residential Residential Residential Residential Residential Residential Residential Land-applied wastewater Land-applied wastewater Dairy and field crops Plant nursery

VSMOW = Vienna Standard Mean Ocean Water.

Iohexol analyzed but not detected.

UFA UFA

UFA

UFA UFA UFA UFA UFA UFA UFA UFA UFA UFA

49 47 135 52

1 14 2 4 6 7 8 9 36 11

Silver Silver Silver Silver

32 33 34 35

UFA UFA UFA UFA

M-0772 M-07732 M-05282 M-07862

Well Site

Primary N-Contributing Land Use

27 152

94

29 82 297 416 618 18 56 302 365 0

865 1085 183 106

Septic Tanks (1-km radius)

— Perc

RIB, RI

RI — — — RI RI — Perc, RI RIB, RI RIB

GCI — SF Sprayfield

Water Reuse (1-km radius)

* 9.87

*

0.02 0.01 0.01 0.01 0.02 * * 0.02 1.30 *

0.40 1.88 2.18 2.58

0.42 *

0.83

0.05 0.02 1.31 1.72 0.18 0.04 0.01 1.19 * 0.47

* * * *

0.60 *

0.90

0.12 0.09 1.44 1.92 0.19 0.07 * 1.27 * 0.53

* * * *

** 1.40

**

2.81 ** ** 46.37 8.85 ** ** ** 13.15 **

4.29 7.78 7.22 8.02

** 6.99

**

1.70 ** ** 26.83 15.18 ** ** ** 5.25 **

0.83 3.99 3.73 3.55

δ15 Nδ18 ONO3 Air NO3 VSMOW NO3 -N NH4 -N TKN (mg L−1 ) (mg L−1 ) (mg L−1 ) (‰) (‰)1

0.1 4.4

0.1

0.8 0.1 0.1 0.1 0.1 0.2 0.2 0.6 6.6 0.2

5.8 3.4 3.2 3.9

DO (mg L−1 )

* *

120

* 190 * 220 * * * * * *

* * 98 180

365 424

347

* 300 207 215 * 217 279 309 470 293

* 359 345 512

Sucralose (ng L−1 ) Cl:Br

July 8, 2020

Well Depth Well ID Aquifer Springshed (ft)

Table 3. Continued.

EEGS-26-01-05_1XO 9:29

Nitrogen Sources in Springsheds

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

303

EEGS-26-01-05_1XO

July 8, 2020

9:29

Canion, Ransom, and Katz

Figure 5. Source nitrate isotope empirical distributions (blue bars), prior probability density (black lines), and posterior probability densities predicted by the mixing model (blue lines).

between 0.15 and 0.80 for nitrate fertilizer (Figure 4).

Taken together, the two types of fertilizer had modeled mean contributions of 50%–80% in agricultural areas.

High contributions from synthetic fertilizer are consistent with previous work in Florida springs within agricultural springsheds that demonstrated that synthetic fertilizer was a dominant nitrogen source over organic nitrogen sources (i.e., wastewater and manure) (Katz, 2004; Heffernan et al., 2012). In some cases, high contributions of nitrate fertilizer may have been a result of legacy agricultural impacts. For example, regionally high nitrate from defunct citrus operations are known to be present in the groundwater near well M-0782 as well as in converted citrus groves in the greater Orlando area, where Wekiwa, Rock, Sanlando, and Starbuck springs are found (Canion, 2017).

In residential wells, central tendencies were between 0.14 and 0.83 for ammonium fertilizer and between 0.08 and 0.71 for nitrate fertilizer (Figure 4). The combined mean contributions of fertilizer sources were modeled between 40% and 100%, with landscape fertilization as the most likely source. The 95% credible intervals were large for ammonium fertilizer but

304

were smaller for nitrate fertilizer, which could be attributed to the unique range of nitrate fertilizer δ18 O values. Contributions from ammonium fertilizer were markedly lower at sites where elevated contributions from wastewater were predicted. Contributions from nitrate fertilizer were generally close to prior distributions, indicating that there was little evidence in the data to shift the prior distribution. Wells OR894 and OR1108 were notable exceptions and had posterior mean contributions of 0.71 and 0.56 for nitrate fertilizer, respectively. Well OR894 illustrates a potential pitfall in relying on nitrate isotope data for source attribution when reduced forms of nitrogen have higher concentrations. Nitrogen in this well was mostly in the form of TKN (1.39 mg L−1 ), and the sucralose concentration was relatively high (580 ng L−1 ), which indicated that the wastewater contribution was higher than predicted by the mixing model.

Manure Sources Modeled mean manure source contributions were between <0.01 and 0.37 at agricultural sites and

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

EEGS-26-01-05_1XO

July 8, 2020

9:29

Nitrogen Sources in Springsheds

between <0.01 and 0.10 at residential sites (Figure 4). Manure posterior means at agricultural sites were similar to prior means, indicating that there was insufficient evidence in the data to shift the informative prior distributions of the manure sources. Low contributions from manure may be attributed to the predominance of pasture for beef cattle and horses. Pasture operations in Florida have been shown to have limited nitrogen loading to groundwater when managed properly (Sigua, 2010). Concentrated manure sources, including dairies and horse manure stockpiles, may still contribute nitrogen locally but were not targeted for sampling in the present study. One well cluster was adjacent to a dairy (V-1028 and V-1030); however, low dissolved oxygen prevented the use of nitrate isotope data from these wells. For areas with agricultural influence, better constraints on the relative contribution of manure and wastewater may be improved by the addition of boron isotope measurements, which have greater distinction between the two sources (Komor, 1997; Vengosh, 1998).

Posterior Distributions of Source Isotope Values In addition to probabilistic estimates of fractional source contributions, the Bayesian mixing model produces posterior distributions for the source isotope values that can be used to evaluate model results (Figure 5). For wastewater sources, both δ15 N and δ18 O had posterior distributions that closely followed prior distributions. Manure sources showed a clear shift to a lower mean (2.9‰) and variance for δ15 N in the posterior distribution. These lower values are not consistent with literature data and indicate that the model may be overpredicting manure contributions.

Both NO3 − and NH4 + fertilizers exhibited some degree of alteration in posterior distributions of δ15 N and δ18 O. For NO3 − fertilizer, the prior mean for δ15 N was shifted from 0.49‰ to 6.3‰ in the posterior distribution, and the mean for δ18 O was shifted from 22.7‰ to 16.8‰. This shift was likely caused by the influence of denitrification, as groundwater that has experienced partial denitrification may have δ15 N values between the fertilizer and wastewater ranges and δ18 O values slightly below the NO3 − fertilizer range. Smaller shifts in posterior distribution were observed for NH4 + fertilizer. Mean δ15 N and δ18 O were shifted from 0.49‰ to 2.3‰ and 3.2‰ to −0.86‰, respectively.

SUMMARY AND CONCLUSIONS In the present study, we used multiple lines of evidence, including dual nitrate isotope analysis and wastewater markers, to attribute sources of elevated groundwater nitrate in wells and springs. The

influence of denitrification precluded the modeling of nitrate sources at spring vents; however, sucralose concentrations indicated that wastewater contributed to all springs with developed springsheds. For wells, Bayesian mixing models indicated that, in general, fertilizer sources were the largest contributor to groundwater nitrate or had approximately equal contributions with wastewater and manure sources.

Legacy fertilizer from past citrus production contributes an as-yet-undetermined amount of this fertilizer nitrogen in some regions of the study area.

Detections of sucralose at agricultural sites and a shifted posterior δ15 N distribution for manure sources suggests that manure sources may contribute less than predicted by the model in agricultural areas. The addition of manure specific tracers in future studies (e.g., boron isotopes) may provide better predictions of manure contributions. At residential sites, agreement between predicted wastewater contributions, high total nitrogen concentrations, and sucralose detections provided evidence for localized impacts from wastewater sources. In the areas where many of the residential wells were sited, low concentrations of nitrate and the presence of reduced nitrogen (NH4 -N and TKN) precluded dual nitrate isotope analysis, which illustrates the utility of additional wastewater tracers. Septic effluent and reuse water were lumped into a single wastewater category for the mixing model, but based on higher total nitrogen concentrations in septic tank effluent than reuse water, it is expected that septic tanks contribute more to nitrogen in groundwater. Exceptions may occur where overapplication of reuse water occurs in geologically sensitive areas.

Further development of reuse-specific tracers would help determine whether reuse water contributes any appreciable nitrogen to groundwater.

ACKNOWLEDGMENTS The authors wish to acknowledge Christy Akers, Rick Breed, Dean Dobberfuhl, and Erich Marzolf for sample collection and for providing valuable feedback on the manuscript.

REFERENCES Accoe, F.; Berglund, M.; Duta, S.; Hennessy, C.; Taylor, P.; Van Hoof, K.; and De Smedt, S., 2008, Source Apportionment of Nitrate Pollution in Surface Water Using Stable Isotopes of N and O in Nitrate and B: A Case Study in Flanders (Belgium): JRC Scientific and Technical Report EUR 23425 EN. Almasri, M. N. and Kaluarachchi, J. J., 2007, Modeling nitrate contamination of groundwater in agricultural watersheds: Journal Hydrology, Vol. 343, pp. 211–229. https://doi.org/ http://dx.doi.org/10.1016/j.jhydrol.2007.06.016.

Badruzzaman, M.; Oppenheimer, J. A.; and Jacangelo, J. G., 2013, Impact of environmental conditions on the

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

305

EEGS-26-01-05_1XO

July 8, 2020

9:29

Canion, Ransom, and Katz suitability of microconstituents as markers for determining nutrient loading from reclaimed water: Water Research, Vol. 47, pp. 6198–6210. https://doi.org/10.1016/j.watres.

2013.07.029.

Badruzzaman, M.; Pinzon, J.; Oppenheimer, J.; and Jacangelo, J. G., 2012, Sources of nutrients impacting surface waters in Florida: A review: Journal Environmental Management, Vol. 109, pp. 80–92. https://doi.org/ 10.1016/j.jenvman.2012.04.040.

Belaineh, G.; Stewart, J.; Sucsy, P.; Motz, L. H.; Park, K.; Rouhani, S.; and Cullum, M., 2012, Groundwater hydrology. In Lowe, E.; Battoe, L.; Wilkening, H.; Cullum, M.; and Bartol, T. (Editors), St. Johns River Water Supply Impact Study: St. Johns River Water Management District, Palatka, FL, pp. 4.12–4.100.

Boniol, D.; Davis, J.; Jeannee, N.; and Stokes, J., 2014, Top of the Floridan Aquifer System in Peninsular Florida: Technical Fact Sheet SJ2014-FS1. St. Johns River Water Management District, Palatka, FL. Boniol, D.; Williams, M.; and Munch, D., 1993, Mapping Recharge to the Floridan Aquifer Using a Geographic Information System: St. Johns River Water Management District, Palatka, FL. Burgin, A. J. and Hamilton, S. K., 2007, Have we overemphasized the role of denitrification in aquatic ecosystems? A review of nitrate removal pathways: Frontiers Ecology Environment, Vol. 5, pp. 89–96. https://doi.org/10.1890/15409295(2007)5[89:hwotro]2.0.co;2.

Burow, K. R.; Nolan, B. T.; Rupert, M. G.; and Dubrovsky, N. M., 2010, Nitrate in groundwater of the United States, 1991–2003: Environmental Science Technology, Vol. 44, pp. 4988–4997. https://doi.org/10.1021/es100546y.

Canion, A., 2017, Predictive Modeling of Elevated Groundwater Nintrate in Major Springsheds of the St. Johns River Water Management District: Technical Memorandum 58. St. Johns River Water Management District, Palatka, FL. Canion, A.; McCloud, L.; and Dobberfuhl, D., 2019, Predictive modeling of elevated groundwater nitrate in a karstic spring-contributing area using random forests and regressionkriging: Environmental Earth Sciences, Vol. 78, p. 271. https://doi.org/10.1007/s12665-019-8277-1.

Casciotti, K. L.; Sigman, D. M.; Hastings, M. G.; Böhlke, J. K.; and Hilkert, A., 2002, Measurement of the oxygen isotopic composition of nitrate in seawater and freshwater using the denitrifier method: Analytical Chemistry, Vol. 74, pp. 4905–4912. https://doi.org/10.1021/ac020113w.

Cohen, M. J.; Lamsal, S.; and Korhnak, L. V., 2007, Sources, Transport and Transformation of Nitrate-N in the Florida Environment: Special Publication SJ2007-SP10, St. Johns River Water Management Disctrict, Palatka, FL. Davis, S. N.; Fabryka-Martin, J. T.; and Wolfsberg, L. E., 2004, Variations of bromide in potable ground water in the United States: Groundwater, Vol. 42, pp. 902–909. https://doi.org/10.1111/j.1745-6584.2004.t01-8-.x.

Delignette-Muller, M.-L.; Dutang, C.; Pouillot, R.; Denis, J.-B.; and Siberchicot, A., 2019, fitdistrplus: Help to Fit of a Parametric Distribution to Non-Censored or Censored Data: R package version 1.0-14. https://cran.rproject.org/package=fitdistrplus.

El Gaouzi, F. Z. J.; Sebilo, M.; Ribstein, P.; Plagnes, V.; Boeckx, P.; Xue, D.; Derenne, S.; and Zakeossian, M., 2013, Using δ15N and δ18O values to identify sources of nitrate in karstic springs in the Paris basin (France): Applied Geochemistry, Vol. 35, pp. 230–243. https://doi.org/10.1016/j.apgeochem.2013.04.015.

306

Eller, K. T. and Katz, B. G., 2017, Nitrogen source inventory and loading tool: An integrated approach toward restoration of water-quality impaired karst springs: Journal Environmental Management, Vol. 196, pp. 702–709. https://doi.org/10.1016/j.jenvman.2017.03.059.

Fla. Admin. Code R. 62-302.531, 2016, Numeric Interpretations of Narrative Nutrient Criteria. Florida Administrative Code Chapter 62: Surface Water Quality Standards.

Florida Department of Health, 2016, Florida Water Management Inventory Project: Florida Department of Health, Tallahassee, FL. Heffernan, J. B.; Albertin, A. R.; Fork, M. L.; Katz, B. G.; and Cohen, M. J., 2012, Denitrification and inference of nitrogen sources in the karstic Floridan Aquifer: Biogeosciences, Vol. 9, pp. 1671–1690.

Henson, W. R.; Huang, L.; Graham, W. D.; and Ogram, A., 2017, Nitrate reduction mechanisms and rates in an unconfined eogenetic karst aquifer in two sites with different redox potential. Journal Geophysical Research: Biogeosciences, Vol 122, pp. 1062–1077. https://doi.org/10.1002/2016JG003463.

Katz, B. G., 1992, Hydrochemistry of the Upper Floridan Aquifer, Florida: USGS Water-Resources Investigations Report 914196, U.S. Geological Survey, Tallahassee, FL. Katz, B. G., 2004, Source of nitrate contamination and age of water in large karstic springs of Florida: Environmental Geology, Vol. 46, pp. 689–706.

Katz, B. G.; Eberts, S. M.; and Kauffman, L. J., 2011, Using Cl/Br ratios and other indicators to assess potential impacts on groundwater quality from septic systems: A review and examples from principal aquifers in the United States: Journal Hydrology, Vol. 397, pp. 151–166. https://doi.org/10.1016/j.jhydrol.2010.11.017.

Katz, B. G.; Hornsby, H. D.; Bohlke, J. F.; and Mokray, M. F., 1999, Sources and Chronology of Nitrate Contamination in Spring Waters, Suwannee River Basin, Florida: WaterResources Investigations Report 99-4252, U.S. Geological Survey, Tallahassee, FL. Katz, B. G.; Sepulveda, A. A.; and Verdi, R. J., 2009, Estimating nitrogen loading to ground water and assessing vulnerability to nitrate contamination in a large karstic springs basin, Florida: Journal American Water Resources Association, Vol. 45, pp. 607–627.

Kendall, C.; Elliott, E. M.; and Wankel, S. D., 2008, Tracing anthropogenic inputs of nitrogen to ecosystems. In Michener, M. and Lathja, K. (Editors), Stable Isotopes in Ecology and Environmental Science, 2nd ed.: Blackwell, New York, pp. 375–449. https://doi.org/10.1002/9780470691854.ch12.

Knowles, L.; Katz, B. G.; and Toth, D. J., 2010, Using multiple chemical indicators to characterize and determine the age of groundwater from selected vents of the Silver Springs Group, central Florida, USA: Hydrogeology Journal, Vol. 18, pp. 1825–1838. https://doi.org/10.1007/s10040-010-0669-y.

Komor, S. C., 1997, Boron contents and isotopic compositions of hog manure, selected fertilizers, and water in Minnesota: Journal Environmental Quality, Vol. 26, pp. 1212– 1222.

Kuniansky, E. L.; Bellino, J. C.; and Dixon, J. F., 2012, Transmissivity of the Upper Floridan Aquifer in Florida and Parts of Georgia, South Carolina, and Alabama: U.S. Geological Survey Scientific Investigations Map 3204. https://pubs.usgs.gov/sim/3204.

Lusk, M. G.; Toor, G. S.; Yang, Y.-Y.; Mechtensimer, S.; De, M.; and Obreza, T. A., 2017, A review of the fate and transport of nitrogen, phosphorus, pathogens, and trace organic chemicals in septic systems: Critical Reviews

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

EEGS-26-01-05_1XO

July 8, 2020

9:29

Nitrogen Sources in Springsheds Environmental Science Technology, Vol. 3389, pp. 1–87. https://doi.org/10.1080/10643389.2017.1327787.

Massoudieh, A. and Kayhanian, M., 2013, Bayesian chemical mass balance method for surface water contaminant source apportionment: Journal Environmental Engineering, Vol. 139, pp. 250–260. https://doi.org/10.1061/(ASCE)EE.19437870.0000645.

Moore, J. W. and Semmens, B. X., 2008, Incorporating uncertainty and prior information into stable isotope mixing models: Ecology Letters, Vol. 11, pp. 470–480. https://doi.org/10.1111/j.1461-0248.2008.01163.x.

Nolan, B. T.; Hitt, K. J.; and Ruddy, B. C., 2002, Probability of nitrate contamination of recently recharged groundwaters in the conterminous United States: Environmental Science Technology, Vol. 36, pp. 2138–2145. https://doi.org/10.1021/es0113854.

Oppenheimer, J.; Eaton, A.; Badruzzaman, M.; Haghani, A. W.; and Jacangelo, J. G., 2011, Occurrence and suitability of sucralose as an indicator compound of wastewater loading to surface waters in urbanized regions: Water Research, Vol. 45, pp. 4019–4027. https://doi.org/10.1016/j.watres.2011.05.014.

Oppenheimer, J. A.; Jacangelo, J. G.; Dong, B.; Lee, D.; Schwab, K.; Janicki, A.; and Schmidt, H., 2018, Developing Tools for Surface Water Nutrient Loading Attributable to Reclaimed Water: Water Research Foundation, Denver, CO, 142 p.

Panno, S. V.; Hackley, K. C.; Hwang, H. H.; Greenberg, S. E.; Krapac, I. G.; Landsberger, S.; and O’Kelly, D. J., 2006, Characterization and identification of Na-Cl sources in ground water: Groundwater, Vol. 44, pp. 176–187. https://doi.org/10.1111/j.1745-6584.2005.00127.x, Parnell, A. C.; Inger, R.; Bearhop, S.; and Jackson, A. L., 2010, Source partitioning using stable isotopes: Coping with too much variation: PLoS One, Vol. 5, No. 3, p. e9672. https://doi.org/10.1371/journal.pone.0009672.

Phillips, D. L. and Gregg, J. W., 2003, Source partitioning using stable isotopes: Coping with too many sources. Oecologia, Vol. 136, pp. 261–269. https://doi.org/10.1007/s00442-0031218-3.

Plummer, M., 2017, JAGS: Just Another Gibbs Sampler: Version 4.3.0. http://mcmc-jags.sourceforge.net.

Plummer, M., 2018, rjags: Bayesian Graphical Models Using MCMC: R package version 4-8. https://CRAN.Rproject.org/package=rjags.

Quinlan, E. L.; Phlips, E. J.; Donnelly, K. A.; Jett, C. H.; Sleszynski, P.; and Keller, S., 2008, Primary producers and nutrient loading in Silver Springs, FL, USA: Aquatic Botany, Vol. 88, pp. 247–255.

R Core Team, 2019, R: A Language and Environment for Statistical Computing: R Core Team, Vienna, Austria.

Ransom, K. M.; Grote, M. N.; Deinhart, A.; Eppich, G.; Kendall, C.; Sanborn, M. E.; Souders, A. K.; Wimpenny, J.; Yin, Q. Z.; Young, M.; and Harter, T., 2016, Bayesian nitrate source apportionment to individual groundwater wells in the Central Valley by use of elemental and isotopic tracers: Water Resources Research, Vol. 52, pp. 5577–5597. https://doi.org/10.1002/2015WR018523.

Ripley, B.; Venables, B.; Bates, D. M.; Hornik, K.; Gebhardt, A.; and Firth, D., 2019, MASS: Support Functions and Datasets for Venables and Ripley’s MASS: R package version 7.3-51.4. https://cran.r-project.org/package= MASS. Rye, R. O.; Back, W.; Hanshaw, B. B.; Rightmire, C. T.; and Pearson, F. J., 1981, The origin and isotopic composition of dissolved sulfide in groundwater from carbon-

ate aquifers in Florida and Texas: Geochimica Cosmochimica Acta, Vol. 45, pp. 1941–1950. https://doi.org/10.1016/00167037(81)90024-7.

Schmidt, H.; Waller, P.; Oppenheimer, J.; Badruzzaman, M.; Pinzon, J.; and Jacangelo, J., 2013, No sweetener in your stormwater, but what about your reclaimed water?: Florida Water Resources Journal, February, pp. 35–47.

Scott, T. M.; Means, G. H.; Meegan, R. P.; Means, R. C.; Upchurch, S. B.; Copeland, R. E.; Jones, J.; Roberts, T.; and Willet, A., 2004, Springs of Florida: Florida Geological Survey Bulletin No. 66, Florida Geological Survey, Tallahassee, FL. Sigman, D. M.; Casciotti, K. L.; Andreani, M.; Barford, C.; Galanter, M.; and Böhlke, J. K., 2001, A bacterial method for the nitrogen isotopic analysis of nitrate in seawater and freshwater: Analytical Chemistry, Vol. 73, pp. 4145–4153. https://doi.org/10.1021/ac010088e.

Sigua, G. C., 2010, Sustainable cow-calf operations and water quality: A review: Agronomy Sustainable Development, Vol. 30, pp. 631–648. https://doi.org/10.1051/agro/2010001.

Silvanima, J.; Woeber, A.; Sunderman-Barnes, S.; Copeland, R.; Sedlacek, C.; and Seal, T., 2018, A synoptic survey of select wastewater-tracer compounds and the pesticide imidacloprid in Florida’s ambient freshwaters: Environmental Monitoring Assessment, Vol. 190, p. 435. https://doi.org/10.1007/s10661-018-6782-4.

Soh, L.; Connors, K. A.; Brooks, B. W.; and Zimmerman, J., 2011, Fate of sucralose through environmental and water treatment processes and impact on plant indicator species: Environmental Science Technology, Vol. 45, pp. 1363–1369. https://doi.org/10.1021/es102719d.

Spoelstra, J.; Senger, N. D.; and Schiff, S. L., 2017, Artificial sweeteners reveal septic system effluent in rural groundwater: Journal Environmental Quality, Vol. 46, No. 6, pp. 1434–1443. https://doi.org/10.2134/jeq2017.06.0233.

Stevenson, R. J.; Pinowska, A.; Albertin, A.; and Sickman, J. O., 2007, Ecological Condition of Algae and Nutrients in Florida Springs: The Synthesis Report: Florida Department of Environmental Protection, Tallahassee, FL. St. Johns River Water Management District, 2018, 2014 Land Cover Update: Photointerpretation Key: St. Johns River Water Management District, Palatka, FL. Toth, D. J. and Katz, B. G., 2006, Mixing of shallow and deep groundwater as indicated by the chemistry and age of karstic springs: Hydrogeology Journal, Vol. 14, pp. 827–847. https://doi.org/10.1007/s10040-005-0478-x.

Tucker, W. A.; Diblin, M. C.; Mattson, R. A.; Hicks, R. W.; and Wang, Y., 2014, Nitrate in shallow groundwater associated with residential land use in central Florida: Journal Environmental Quality, Vol. 43, pp. 639–646. https://doi.org/10.2134/jeq2012.0370.

Tucker, W. A. and Stroehlen, C. A., 2010, Wekiva River Basin Nitrate Sourcing Study: MACTEC, Newberry, FL. Vengosh, A., 1998, The isotopic composition of anthropogenic boron and its potential impact on the environment: Biological Trace Element Research, Vol. 66, pp. 145–151. https://doi.org/10.1007/BF02783134.

Walsh, S. J.; Knowles, L.; Katz, B. G.; and Strom, D. G., 2009, Hydrology, Water Quality, and Aquatic Communities of Selected Springs in the St. Johns River Water Management District, Florida: USGS Scientific Investigations Report 2009– 5046, U.S. Geological Survey, Reston, VA. Widory, D.; Kloppmann, W.; Chery, L.; Bonnin, J.; Rochdi, H.; and Guinamant, J. L., 2004, Nitrate in groundwater: An

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

307

EEGS-26-01-05_1XO

July 8, 2020

9:29

Canion, Ransom, and Katz isotopic multi-tracer approach: Journal Contaminant Hydrology, Vol. 72, pp. 165–188. https://doi.org/10.1016/j.jconhyd.

2003.10.010.

Widory, D.; Petelet-Giraud, E.; Négrel, P.; and Ladouche, B., 2005, Tracking the sources of nitrate in groundwater using coupled nitrogen and boron isotopes: A synthesis: Environmental Science Technology, Vol. 39, pp. 539–548. https://doi.org/10.1021/es0493897.

Xue, D.; Botte, J.; De Baets, B.; Accoe, F.; Nestler, A.; Taylor, P.; Van Cleemput, O.; Berglund, M.; and Boeckx, P., 2009, Present limitations and future prospects of stable isotope methods for nitrate source identification in surface- and groundwater: Water Research, Vol. 43, pp. 1159–1170. https://doi.org/10.1016/j.watres.2008.

12.048.

308

Xue, D.; De Baets, B.; Van Cleemput, O.; Hennessy, C.; Berglund, M.; and Boeckx, P., 2012, Use of a Bayesian isotope mixing model to estimate proportional contributions of multiple nitrate sources in surface water: Environmental Pollution, Vol. 161, pp. 43–49. https://doi.org/10.1016/ j.envpol.2011.09.033.

Yang, L.; Han, J.; Xue, J.; Zeng, L.; Shi, J.; Wu, L.; and Jiang, Y., 2013, Nitrate source apportionment in a subtropical watershed using Bayesian model: Science Total Environment, Vol. 463–464, pp. 340–347. https://doi.org/10.1016/j.scitotenv.2013.06.021.

Yang, Y. Y. and Toor, G. S., 2016, δ15N and δ18O reveal the sources of nitrate-nitrogen in urban residential stormwater runoff: Environmental Science Technology, Vol. 50, pp. 2881– 2889. https://doi.org/10.1021/acs.est.5b05353.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

0.42

0.23 0.39 0.81 0.07 0.35 0.85 0.87

0.34 0.79 0.21 0.42 0.57 0.72 0.83 0.79 0.20 0.22 0.10 0.23 0.21

0.25

0.25

0.48 0.48 0.01 0.88 0.55

M-0031 M-0039 M-0040 M-0041 M-0063 M-0213 M-0217

M-0419 M-0465 M-0527 M-0528 M-0771 M-0772 M-0773 M-0774 M-0776 M-0778 M-0782 M-0785 M-0786

OR0106

OR0107

OR0546 OR0548 OR0651 OR0661 OR0893

0.00 0.00 0.00 0.01 0.00

0.11

0.11

0.19 0.01 0.50 0.09 0.19 0.00 0.02 0.00 0.45 0.74 0.56 0.38 0.71

0.68 0.10 0.01 0.56 0.53 0.00 0.00

0.03

0.01 0.01 0.14 0.03 0.00

0.07

0.07

0.39 0.02 0.07 0.02 0.22 0.02 0.04 0.00 0.04 0.01 0.00 0.02 0.00

0.03 0.11 0.02 0.03 0.01 0.04 0.02

0.03

Agriculture Upland

0.11 0.11 0.50 0.05 0.27

0.29

0.29

0.05 0.11 0.20 0.38 0.02 0.11 0.08 0.19 0.24 0.02 0.32 0.31 0.03

0.05 0.31 0.07 0.33 0.12 0.07 0.07

0.05

Forest

0.01 0.01 0.00 0.00 0.00

0.00

0.00

0.00 0.00 0.00 0.00 0.00 0.06 0.00 0.00 0.05 0.00 0.00 0.01 0.00

0.00 0.00 0.01 0.00 0.00 0.00 0.01

0.31

Water

0.39 0.39 0.35 0.00 0.17

0.09

0.09

0.00 0.01 0.00 0.01 0.00 0.08 0.01 0.01 0.00 0.00 0.02 0.01 0.00

0.00 0.01 0.03 0.00 0.00 0.01 0.00

0.12

Wetland

0.00 0.00 0.00 0.03 0.00

0.20

0.20

0.03 0.05 0.03 0.08 0.00 0.02 0.03 0.01 0.02 0.01 0.00 0.05 0.06

0.00 0.07 0.06 0.00 0.00 0.02 0.02

0.03

82 82 0 596 416

29

29

138 413 111 183 428 865 1085 658 44 76 52 135 106

96 81 347 81 144 108 65

338

Septic Tanks Transportation/ within 1-km Barren Radius

Large wastewater sprayfield Residential irrigation Residential irrigation — — — — —

— — — — —

Percolation pond; RIB — — Golf course — — — Recreational facilities — — — — Golf course Golf course

Water Reuse within 1-km Radius

0.5 — — — —

0.5

0.5

1 1 1 1 1 1 1 1 0.5 0.5 0.2 1 1

0.5 0.5 1 0.2 1 0.5 1

—

1 — — — —

1

1

1 1 1 1 1 1 1 1 1 1 1 1 1

1 1 1 1 1 1 1

—

Nitrate Wastewater Fertilizer

1 — — — —

1

1

1 1 1 1 1 1 1 1 1 1 1 1 1

1 1 1 1 1 1 1

—

Ammonia Fertilizer

Dirichlet Parameter (alpha)

0.2 — — — —

0.1

0.1

0.5 0.1 1 0.2 0.2 0.1 0.2 0.1 1 1 1 0.2 0.2

1 0.5 0.2 1 1 0.1 0.1

—

Manure

July 8, 2020

L-1026

Site

Urban/ Built-Up

2014 Land Use Fraction

Table A1. Land use percentages and Dirichlet prior parameters for wells.

EEGS-26-01-05_1XO 9:29

Nitrogen Sources in Springsheds

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

309

310

0.55 0.78

0.78

0.78

0.39

0.39

0.76 0.76 0.00 0.00 0.74

0.23

0.66

0.66

0.43

0.43

0.42 0.05 0.05 0.06 0.06 0.41

OR0894 OR1108

OR1109

OR1110

S-0716

S-0717

S-1014 S-1015 S-1230 S-1310 S-1408

V-0156

V-0196

V-0197

V-0810

V-0814

V-0837 V-1028 V-1030 V-1151 V-1152 V-1169

Site

Urban/ Built-Up

0.00 0.27 0.27 0.00 0.00 0.07

0.03

0.03

0.01

0.01

0.27

0.00 0.00 0.09 0.09 0.00

0.07

0.07

0.02

0.41 0.06 0.06 0.14 0.14 0.16

0.00

0.00

0.02

0.02

0.06

0.02 0.02 0.21 0.21 0.02

0.00

0.00

0.03

0.03

0.00 0.03

0.11 0.35 0.35 0.64 0.64 0.33

0.29

0.29

0.14

0.14

0.40

0.03 0.03 0.46 0.46 0.01

0.12

0.12

0.04

0.04

0.27 0.04

Forest

0.00 0.06 0.06 0.00 0.00 0.00

0.13

0.13

0.06

0.06

0.00

0.02 0.02 0.02 0.02 0.02

0.22

0.22

0.02

0.02

0.00 0.02

Water

0.02 0.21 0.21 0.16 0.16 0.00

0.13

0.13

0.05

0.05

0.05

0.13 0.13 0.12 0.12 0.07

0.20

0.20

0.08

0.08

0.17 0.08

Wetland

0.04 0.00 0.00 0.00 0.00 0.03

0.00

0.00

0.07

0.07

0.00

0.04 0.04 0.10 0.10 0.13

0.00

0.00

0.03

0.03

0.00 0.03

334 27 27 56 56 365

302

302

18

18

152

297 297 0 0 94

196

196

618

618

416 618

Septic Tanks Transportation/ within 1-km Barren Radius — Residential irrigation Residential irrigation Residential irrigation Recreational facilities Recreational facilities — — RIB RIB Multiple ribs and irrigation Percolation pond Residential irrigation Residential irrigation Percolation pond; residential Irrigation Percolation pond; residential irrigation Golf course — — — — RIB; residential irrigation

Water Reuse within 1-km Radius

1 — — 0.2 — 1

—

—

—

—

1

— — — 1 —

—

—

—

1

1 1

1 — — 1 — 1

—

—

—

—

1

— — — 1 —

—

—

—

1

1 1

Nitrate Wastewater Fertilizer

1 — — 1 — 1

—

—

—

—

1

— — — 1 —

—

—

—

1

1 1

Ammonia Fertilizer

Dirichlet Parameter (alpha)

0.1 — — 0.1 — 0.1

—

—

—

—

0.2

— — — 0.1 —

—

—

—

0.2

0.1 0.1

Manure

July 8, 2020

0.02

0.00 0.02

Agriculture Upland

2014 Land Use Fraction

Table A1. Continued.

EEGS-26-01-05_1XO 9:29

Canion, Ransom, and Katz

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

EEGS-26-01-05_1XO

July 8, 2020

9:29

Nitrogen Sources in Springsheds Table A2. List of references used to model source prior distributions.

Tracer

Nitrogen Source

n

δ15N

Wastewater

38

NH4 fertilizer

14

NO3 fertilizer

282

Manure δ18O

Wastewater NH4 fertilizer NO3 fertilizer Manure

18 23 23 292 23

References Widory et al., 2004; Accoe et al., 2008; Hinkle et al., 2008; Panno et al., 2008; Katz et al., 2010 Briand et al., 2017 Vitòria et al., 2004; Widory et al., 2004; Bateman & Kelly, 2007; Briand et al., 2017 Vitòria et al., 2004; Accoe et al., 2008; Davis et al., 2015; Michalski et al., 2015 Widory et al., 2004; Bateman and Kelly, 2007; Accoe et al., 2008; Panno et al., 2008; Briand, 2017 Accoe et al., 2008; Panno et al., 2008; Katz et al., 2010; Briand et al., 2017 Accoe et al., 2008; Panno et al., 2008; Katz et al., 2010; Briand et al., 2017 Vitòria et al., 2004; Accoe et al., 2008; Davis et al., 2015; Michalski et al., 2015 Accoe et al., 2008; Panno et al., 2008; Briand et al., 2017; Katz et al., 2010

REFERENCES FOR APPENDIX Accoe, F.; Berglund, M.; Duta, S.; Hennessy, C.; Taylor, P.; Van Hoof, K.; and De Smedt, S., 2008, Source Apportionment of Nitrate Pollution in Surface Water Using Stable Isotopes of N and O in Nitrate and B: A Case Study in Flanders (Belgium): JRC Scientific and Technical Report EUR 23425 EN. Luxembourg.

Bateman, A. S. and Kelly, S. D., 2007, Fertilizer nitrogen isotope signatures: Isotopes Environmental Health Studies, Vol. 43, pp.

237–247.

Briand, C.; Sebilo, M.; Louvat, P.; Chesnot, T.; Vaury, V.; Schneider, M.; and Plagnes, V., 2017, Legacy of contaminant N sources to the NO3 − signature in rivers: A combined isotopic (δ15 N-NO3 − , δ18 O-NO3 − , δ11 B) and microbiological investigation: Scientific Reports, Vol. 7, pp. 1–11.

Davis, P.; Syme, J.; Heikoop, J.; Fessenden-Rahn, J.; Perkins, G.; Newman, B.; Chrystal, A. E.; and Hagerty, S. B., 2015, Quantifying uncertainty in stable isotope mixing models: Journal Geophysical Research: Biogeosciences, Vol. 120, pp. 903–923.

Hinkle, S. R.; Böhlke, J. K.; and Fisher, L. H., 2008, Mass balance and isotope effects during nitrogen transport

through septic tank systems with packed-bed (sand) filters: Science Total Environment, Vol. 407, pp. 324–332.

Katz, B. G.; Griffin, D. W.; McMahon, P. B.; Harden, H. S.; Wade, E.; Hicks, R. W., and Chanton, J. P., 2010, Fate of effluent-borne contaminants beneath septic tank drainfields overlying a karst aquifer: Journal Environmental Quality, Vol. 39, pp. 1181–1195.

Michalski, G.; Kolanowski, M.; and Riha, K. M., 2015, Oxygen and nitrogen isotopic composition of nitrate in commercial fertilizers, nitric acid, and reagent salts: Isotopes Environmental Health Studies, Vol. 51, pp. 382–391.

Panno, S. V.; Kelly, W. R.; Hackley, K. C.; Hwang, H. H.; and Martinsek, A. T., 2008, Sources and fate of nitrate in the Illinois River Basin: Illinois Journal Hydrology, Vol. 359, pp. 174– 188.

Vitòria, L.; Otero, N.; Soler, A.; and Canals, A., 2004, Fertilizer characterization: Isotopic data (N, S, O, C, and Sr): Environmental Science Technology, Vol. 38, 3254–3262.

Widory, D.; Kloppmann, W.; Chery, L.; Bonnin, J.; Rochdi, H.; and Guinamant, J. L., 2004, Nitrate in groundwater: An isotopic multi-tracer approach: Journal Contaminant Hydrology, Vol. 72, pp. 165–188.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 291–311

311

EEGS-26-01-05_1XO

July 8, 2020

9:29

0

EEGS-26-02-05_1XO

July 8, 2020

10:2

Influence of Desert Springs on Habitat of Endangered Zuni Bluehead Sucker (Catostomus discobolus yarrowi) REBECCA J. FRUS* Nevada Water Science Center, U.S. Geological Survey, 160 N. Stephanie Street, Henderson, NV 89002, USA

LAURA J. CROSSEY Department of Earth and Planetary Sciences, University of New Mexico, Albuquerque, NM 87131, USA

CLIFFORD N. DAHM Biology Department, University of New Mexico, Albuquerque, NM 87131, USA

KARL E. KARLSTROM Department of Earth and Planetary Sciences, University of New Mexico, Albuquerque, NM 87131, USA

LIVIA CROWLEY Cibola National Forest and National Grasslands, U.S. Forest Service, Albuquerque, NM 87113, USA

Key Terms: Hydrochemistry, Groundwater, Spring, Aquatic Habitats, Spatial and Temporal Geochemistry, Endangered Fish ABSTRACT Located on the southeastern part of the Colorado Plateau, the Zuni Mountains are home to the endangered Zuni Bluehead Sucker (ZBS) (Catostomus discobolus yarrowi). A 4-year study was conducted on a low-flow (<80 cm3 /s) hillslope spring and intermittent stream, that are home to one of the three remaining ZBS populations. Seasonal measurements of physical and hydrochemical parameters were used to estimate the contribution of groundwater to the stream and to identify geologic and hydrologic controls for the spring discharge. Seasonal concentrations and standard deviations (s) of Mg2+ were used to determine that the spring water (5.6 mg/L; s = 0.4) and surface water upgradient from the spring input (10.7 mg/L; s = 11.2) is from different sources. Surface water down-gradient from the spring input maintain ZBS populations and is a mixture of spring water and up-gradient surface water. Mass solution mixing was used to determine spring water contributes up to 99 percent of the down-gradient water during drier seasons. Isotopes (δD, δ18 O, 3 H) in*Corresponding author email: rfrus@usgs.gov

dicate that the spring water has been recharged primarily from snowmelt within the last 70 years, while upgradient surface water is seasonal runoff from rain and snowmelt. Continuous monitoring of dissolved oxygen (DO) mean concentrations (up-gradient = 1.6 mg/L and down-gradient = 5.7 mg/L) indicated that surface water up-gradient from the spring input are anoxic and unable to support ZBS. Surface water down-gradient from the spring input maintain appropriate DO concentrations due to perennially discharging spring waters reaerating downstream habitats.

INTRODUCTION Persistent aridity in the twenty-first century (MacDonald, 2010) has caused available water in the semiarid southwestern United States to decline and not recover (Overpeck and Udall, 2010). As the human population continues to grow and expand, urban, agricultural, and rangeland areas continue to deplete water supplies that are already stressed (Zektser et al., 2005). The practice of over-pumping groundwater supplies has led to significant drops in local and regional water tables throughout the U.S. Desert Southwest (Jacobs and Holway, 2004).

Spring waters sustain ecosystem habitats that concentrate and isolate unique species relative to the surrounding desert environment. Ephemeral to perennial

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

313

EEGS-26-02-05_1XO

July 8, 2020

10:2

Frus, Crossey, Dahm, Karlstrom, and Crowley

Figure 1. Image of an adult male Zuni Bluehead Sucker (Catostomus discobolus yarrowi).

desert springs and resulting wetlands have been identified as biodiversity hotspots (Stevens and Meretsky, 2008). In the Grand Canyon landscape, spring wetlands maintain 11 percent of the plant species but account for less than 0.01 percent of the area (Perla and Stevens, 2008). In both the Sonoran Desert and Mojave Desert, state and federal officials recognize that during dry months, low-flow springs and seeps can support both warm- and cold-water fish species (Minckley and Deacon, 1991; Minckley and Douglas, 1991; and Pool and Olden, 2014). Desert fish and aquatic species are at risk of losing critical habitat and possible full extinction with the loss of surface water discharge from springs due to groundwater withdrawal and drought conditions (Ruhí et al., 2014).

The combination of current land management practices and severe drought and increasing air temperatures (Overpeck and Udall, 2010) has caused many springs of the U.S. Southwest to dry (Unmack and Minckley, 2008). Recently, springs have been recognized by management agencies as important indicators of groundwater sustainability. Baseline inventories of these resources are now being completed on some public lands to measure future impacts to groundwater availability and the habitats that depend on them (Springer and Stevens, 2009; Frus, 2016).

To evaluate the effects of groundwater discharge at springs on desert aquatic habitats, a first-order desert stream, Agua Remora, was investigated. Agua Remora (AR) provides refugia to the endangered Zuni Bluehead Sucker (ZBS) (Catostomus discobolus yarrowi), a small (average length = 200 mm) algae-eating fish (Figure 1) (USFWS, 2014). ZBS is the only remaining species of Bluehead Sucker in the Zuni River. This desert fish has only four populations remaining in the wild, all within the Zuni Mountains, NM. One ZBS population is located on lands managed by the Cibola

314

National Forest. The population is isolated to a 200 m stretch of perennial spring-fed water in AR. The goal of this study was to determine if the groundwater discharge is critical to the survival of the ZBS. We report on the spatial and temporal hydrogeochemical processes of the groundwater and surface water within the 200 m ZBS habitat. A 4 year study used a multi-disciplinary approach to understand groundwater flow paths and residence time, groundwater and surface water recharge conditions, and seasonal variations of field parameters to determine the impact that groundwater has on the habitat. Results indicate that during dry seasons, groundwater discharge provides up to 99 percent of water to the ZBS habitat and that discharge and subsequent re-aeration of groundwater from the hillslope spring maintain dissolved oxygen levels high enough to sustain ZBS. STUDY AREA AR is located on the Colorado Plateau in westcentral New Mexico (Figure 2A). AR is a first-order stream to the Rio Nutria, which flows into the Zuni River. The Zuni River is a tributary to the Little Colorado River (HUC6 150200) (Propst, 1999; Propst et al., 2001, 2008), with its headwaters in the Zuni Mountains, west of the Continental Divide (Figure 2A).

We report here on a low-flow (<80 cm3 /s) spring and associated habitat located in the first-order stream, AR. AR is an intermittent stream with four perennially wet pools and a perennial hillslope spring (Figure 2B). For the study area, the stream channel length is about 120 m long with a change in elevation of about 13 m. Flow in the stream channel is towards the westsouthwest. The pools can range in size from 3 to 20 m long and 1 to 3 m wide. During the wet season, the deepest pool is less than 2 m deep.

The ZBS individuals are found in the perennial pools where the spring water enters the stream channel and down-gradient from that location (Carman, 2004). The unnamed hillslope spring at AR is about 4 m higher and 6 m away (north) from the stream channel. The spring area is grassy year-round with 2 to 4 small (<1 m) channels where groundwater discharges and flows south into the stream channel (Figure 2C).

The largest of the spring channels is where springwater samples were collected and spring discharge was measured. Spring water flows into a perennial pool in the stream channel. Upstream from where spring water enters the stream channel, there is another perennial pool. Downstream from where spring water enters the stream channel, there are two additional perennial pools. ZBS populations are unable to survive in perennial water upstream from where spring

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

EEGS-26-02-05_1XO

July 8, 2020

10:2

Groundwater Influence on Endangered Fish Habitat

Figure 2. Study site on regional and local scales. Inset: Four Corners region, southwestern USA. Red line indicates boundaries for the Colorado Plateau; blue line is Little Colorado River. (A) Digital elevation model of Zuni Mountains and Rio Nutria, headwater of the Zuni River. (B) Agua Remora site in Zuni Mountains, NM, with perennial hillslope spring (star) and perennial pools (blue) in stream channel.

Icons identify water sample and continuous monitoring locations (circle) and Zuni Bluehead Sucker populations (fish). (C) Photo of the largest spring channel that discharges into perennial pools.

water enters the stream channel (Figure 2B), even if ZBS are introduced (Carman, 2004; Gilbert and Carman, 2013).

Depending on the season, connectedness and greenness of the stream channel vary. Each of the pools has a similar geometry because they were all dug by a backhoe to capture spring flow. Upstream from where spring water enters the stream channel, there is a wetted perimeter along the streambed no larger than 2 to 3 m wide. Downstream from where spring water enters the stream channel, there is a 15-m-wide meadow that is grassy year-round. Downstream from the lowest pool, the stream channel dries during the warm season.

The Zuni River has seen a 90 percent reduction in ZBS habitat (Gido and Propst, 1999; USFWS, 2014), and populations are limited to four spring-fed sections of the Rio Nutria (Carman, 2004; Gilbert and Carman, 2013; Turner and Wilson, 2009; and Unmack and Minckley, 2008). In 2014, the ZBS habitat was designated critical by the U.S. Fish and Wildlife Service due to habitat destruction, stream drying, predation by non-native species, and small population sizes with restricted ranges (USFWS, 2013, 2014). Additionally, ZBS was listed as endangered by the U.S. Fish and Wildlife Service due to the limited population size (USFWS, 2013, 2014).

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

315

EEGS-26-02-05_1XO

July 8, 2020

10:2

Frus, Crossey, Dahm, Karlstrom, and Crowley

Climate Weather stations for the western Zuni Mountains have been recording daily precipitation measurements since 1949 (Figure 2A, MCGAFFEY 5 SE and KNMNEWME3). The weather stations are located in the higher elevations of the Rio Nutria watershed, within 5 km of the AR site. Annual precipitation (350 mm) is bimodal, with peaks occurring as snowfall (112 mm, 25 percent of annual) from January through March and as monsoonal rainfall (131 mm, 29 percent of annual) from August through September. Large rain and snow events infiltrate shallow aquifers, but snow precipitation is the primary source for infiltration and recharge of confined aquifers of the Desert Southwest (Anderson et al., 2003).

A U.S. Geological Survey (USGS) streamflow gage station at Ramah (09386900) measures daily discharge on the Rio Nutria upstream of the Nutria Monocline. The streamflow gauge is located within the Rio Nutria watershed, about 20 km downstream from the AR site. This study was conducted from May 2012 through May 2015, and the highest streamflow discharge during this time was recorded on May 1, 2016, with a peak of 2.8 m3 /s. Overall, the average daily discharge during the study was 0.008 m3 /s, with 48 percent of the flow below 0.003 m3 /s.

Hydrogeologic Setting Water chemistry can be impacted by the geologic history of the surrounding rocks (Springer and Stevens, 2009; Wilson, 2004). Geologic structures, such as faults, can act as barriers or conduits to vertical groundwater flow (Gudmundsson, 2000; Crossey et al., 2006; Stevens and Meretsky, 2008; Apaydin, 2010; Banerjee et al., 2011; and Connell, 2011) and can allow for mixing of water from different sources (Glynn and Plummer, 2005; Plummer et al., 2004). In areas with upwelling water along fault zones, spring densities can be high in relation to adjacent non-faulted areas (Hubbs, 1995). Spring-water chemistry can identify discrete hydrochemical zones (Crossey et al., 2006, 2009; Banerjee et al., 2011; and Williams et al., 2013), making springs ideal indicators for mixing of groundwater from underlying geological structures (Manga, 1996, 1999).

Geologic evidence indicates that the Zuni Mountains underwent repeated deformation and magmatism from the early Proterozoic through Cenozoic times.

The Zuni Mountains have a northwest-southeast– trending core of uplifted basement rocks including 1,655 Ma quartz monzonite (Xm) (Strickland et al., 2003). The Precambrian basement is overlain unconformably by Permian strata (Krainer et al., 2003)

316

(Figure 3), indicating the Zuni Mountains were an uplifted region during the Pennsylvanian Ancestral Rocky Mountains orogeny. Additional deformation and fault movement took place during the Laramide orogeny about 70 m.y. ago. The overall geometry of the Zuni Uplift is a broad arch or domal anticline with steeply dipping bedding of monoclines along the northeast and southwest sides (Kelley, 1967; Karlstrom et al., 1997; and Strickland et al., 2003). The steeply dipping limb of the southwestern monocline is Oso Ridge (Figures 3 and 4), which is primarily formed of thinned Permian rocks that dip as much as 75° to the west (Anderson et al., 1998, 2003).

The deformation of the bedding has created pathways for vertical movement of fluids along faults and contacts. The hydrogeology of the western Zuni Mountains reflects the complex geologic history of the area (Figures 3 and 4). As groundwater flows through aquifers, the water chemistry can be altered by rockwater interactions (Drever, 1982; Wilson, 2004; Glynn and Plummer, 2005; and Crossey et al., 2009), where older water has more time to interact with lithology, resulting in higher salinity and increased total dissolved solids (Plummer et al., 2004; Phillips et al., 2013) relative to rain water.

The hydrologic properties of different geologic units within the Zuni Mountains vary both locally and regionally. The Precambrian units are not considered to be a major groundwater resource, except where fractured along fault zones or weathered along erosional unconformities. The Precambrian section is overlain

by Permian strata of the Abo Formation (Pa) and Yeso Formation (Py) (Figure 5). Both Pa and Py produce poor-quality groundwater due to low permeability and porosity, as well as high salinity due to the concentration of evaporites within each of the units (Hood and Kister, 1962; Cooley et al., 1969; Baldwin and Rankin, 1995; and Connell, 2011).

In conformable contact with Py (Figure 5), the white, fine- to medium-grained, cross-bedded siliceous Glorieta Sandstone (Pg) is overlain by a gray fossiliferous interbedded limestone of the San Andres Formation (Psa) (Colpitts, 1969; Baars, 1974; Anderson et al., 2003; and Connell, 2011). The Psa and Pg are hydrologically connected (Psg). The Psg is the most productive aquifer that provides the best water quality in the region (Colpitts, 1969; Baars, 1974; Baldwin and Anderholm, 1992; Baldwin and Rankin, 1995; Anderson et al., 2003; Connell, 2011; and Robertson et al., 2013).

The Triassic Moenkopi (TRm) and Chinle Group (TRc) locally overlie the Permian strata. The TRm is composed of relatively thin (<25 m thick), darkred micaceous sandstones interbedded with mudstones and siltstones (Repenning et al., 1969; Connell, 2011),

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

EEGS-26-02-05_1XO

July 8, 2020

10:2

Groundwater Influence on Endangered Fish Habitat

Figure 3. Surficial hydrogeologic map of northern Zuni Mountains. Yellow star indicates spring at Agua Remora where Zuni Bluehead Sucker population is found. Cross-section lines A-A and B-B are found in Figure 4.

and it is considered to be a poor source for groundwater (Cooley et al., 1969). The TRc can be as much as 600 m thick regionally and is dominated by mudstones and siltstones with some sandstones and carbonates (Heckert and Lucas, 2003). The different sections of the TRc are exposed at distinct locations throughout the western Zuni Mountains, including the McGaffey Member, which is a white, well-sorted sandstone that forms cliffs found near the AR site (Heckert and Lucas, 2003; Robertson et al., 2013). The different TRc sections are known to locally store water that ranges in water quality and hydrologic capabilities (Robertson et al., 2013).

Quaternary alluvial aquifers also store and provide water locally in the Zuni Mountains. The alluvium is found in stream channels and in valley bottoms not occupied by streams. Alluvial aquifers transport and store high-intensity rain-event water and spring snowmelt (Harrington et al., 2002). Groundwater supplies in alluvial aquifers are not only recharged

through different precipitation events, but they are also affected by interactions with the atmosphere during dry times when water is lost due to evapotranspiration (Barnes, 1988).

To further understand the hydrologic spatial and temporal processes within the semiarid landscape that impact the endangered ZBS and its critical habitat, we examined AR, one of four remaining habitat sites. Groundwater discharges at a hillslope spring at AR from a contact between the Proterozoic basement rocks and the Permian Abo, just east of the Oso Ridge Monocline (Figures 3 and 4). Using the best available geologic evidence, results in the cross section were inferred to indicate a fault at depth resulting from a reverse fault structure. The faulting dies out before reaching the surface, as is common in monoclines (Erslev, 1991; Paul and Mitra, 2012). Groundwater vertically flows up this fault structure and is expressed at the surface at the AR spring site (Figure 4).

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

317

EEGS-26-02-05_1XO

July 8, 2020

10:2

Frus, Crossey, Dahm, Karlstrom, and Crowley

Figure 4. Geologic cross sections of Zuni Mountains across Oso Ridge and Agua Remora spring site. Locations of A-A and B-B are shown on Figure 3.

Upstream from the spring, the stream channel is composed of shallow grus alluvium with granite bedrock (Xp) exposed. Downstream from the spring, the channel incises the Pa and Py units and becomes muddy and silty during wet times. The stream channel then cuts thru Oso Ridge and remains in the bedded Pg and Psa until its confluence with the Rio Nutria 10 km to the southwest.

METHODS To describe the temporal and spatial variations of groundwater and surface water at AR, seasonal visits [spring (N = 4), summer (N = 3), fall (N = 3), and winter (N = 3)] began in May 2012 and lasted through May 2015 (Table 1). Initially, collection sites were established within the stream channel where endangered ZBS individuals were found. In addition, a shallow

318

(1 m) temporary well was installed at the spring site for collection of spring water. In May 2013, the site protocol was changed to determine the influence spring water has on the ZBS habitat. Sample sites were established within the stream channel, with one up-gradient from the spring-water discharge into the stream channel, and one below where the spring discharge enters the stream channel (Figure 1). Sample water was collected seasonally from the spring (N = 15), the upgradient pool (N = 12), and the down-gradient pool (N = 14).

For each site visit, USGS procedures were followed for sample collection (Myers, 2006). Untreated sample water was collected in polypropylene bottles for alkalinity, anions, and isotope analysis. To analyze for metals, sample water was collected and treated in the field by pushing the water through a 0.45 μm hydrophilic polyethersulfone filter into a polypropylene

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

EEGS-26-02-05_1XO

July 8, 2020

10:2

Groundwater Influence on Endangered Fish Habitat Table 1. Locations and descriptions for hill-slope spring and first-order stream, Agua Remora, Zuni Mountains, New Mexico.

Latitude DMS

Sample Location Up-gradient from spring Spring at AR Down-gradient from spring

Longitude DMS

35°19 34.99 N 35°19 36.71 N 35°19 36.04 N

108°29 56.105 W 108°29 57.35 W 108°30 00.22 W

Elevation (m)

Type

2,344 2,348 2,342

Stream Spring Stream

Abbreviations: DMS, degrees minutes seconds; Elevation (m), elevation in meters; AR, Agua Remora.

bottle and treated with concentrated nitric acid (69.5 percent HNO3 ) to a pH <4. In October 2014, a third 500 mL polypropylene bottle of untreated sample water was collected at the spring site for tritium analysis.

Samples were stored and transported in a designated cooler with ice. Upon returning from the field, samples were stored in a refrigerator in the water chemistry laboratory on the University of New Mexico campus.

Alkalinity was determined in the laboratory using the end-point titration method (Rounds, 2012) with the necessary volume of weak acid (0.02 per-

Figure 5. Hydrostratigraphic column of the Zuni Mountains geologic units with hydrologically permeable layers in blue.

cent H2 SO4 ). Cation concentrations were analyzed using inductively coupled plasma/optical emission spectroscopy (ICP/OES) (USEPA, 1994), and anion concentrations were analyzed using ion chromatography (IC) (USEPA, 1997) at the Department of Earth and Planetary Sciences at the University of New Mexico, Albuquerque, NM. The charge balance of major ions was then computed for all surface and groundwater samples (N = 41) and determined to be less than ±5 percent for all samples. Duplicate analyses were routinely performed on 10 percent of samples, and external reference standards were used to ensure accuracy.

Isotopic ratios of liquid water (δD and δ18 O) were analyzed using untreated spring and surface water. Reported values for δD and δ18 O in per mil (‰) were determined by laser ring-down cavity spectrometry (Romanini et al., 1997; Gupta et al., 2009), using Vienna Standard Mean Ocean Water (VSMOW) as the standard. Analyses were performed at the Center for Isotopic Studies at the University of New Mexico.

Mean tritium content, weighted for volume, was determined at the Environmental Isotope Laboratory at the University of Arizona using a LKB Wallac Quantulus 1220 spectrophotometer (Eastoe et al., 2012) with a detection limit of 0.6 tritium units (TU) ±2σ for lowcounting samples, which applies for 9-fold enrichment and 1500 minutes of counting.

Field parameters were gathered at the spring and surface water locations using either an Oakton pH/CON 300 meter and/or a Yellow Springs Instruments (YSI) Professional Plus meter. The Oakton pH/CON 300 meter is a waterproof microprocessorbased meter that used a single-junction 12 mm pH electrode (−2.00 to 16.00 pH, ±0.01 pH) with built-in temperature sensor (0.0 to 100.0°C, ±0.5°C) and a specific conductance (SC) (normalized to 25°C) electrode (0 to 199.9 mS/cm, ±1 percent) (Oakton, 2000). The YSI Professional Plus handheld multiparameter meter was used with a Quatro cable that included a specific conductance sensor (0 to 200 μS/cm, ±0.5 percent), a pH glass combination electrode (0 to 14 pH units, ±0.2 percent), a galvanic dissolved oxygen sensor (0 to 20 mg/L, ±2 percent), and a field-grade water temperature sensor (−5 to 70°C, ±0.2°C) (YSI, 2009).

Prior to field work, all sensors were calibrated using

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

319

EEGS-26-02-05_1XO

July 8, 2020

10:2

Frus, Crossey, Dahm, Karlstrom, and Crowley

manufacturers’ guidelines using certified pH and specific conductance (SC) standards. In the field, the dissolved oxygen optical sensor was again calibrated using the local barometric pressure in 100 percent watersaturated air.

In situ measurements of dissolved oxygen concentrations were made using YSI model 6920 V2 sondes (YSI Incorporated, 2008). Measurements were made at 30 minute intervals from May 27, 2013, to December 19, 2013, and re-deployed from May 23, 2014, to February 25, 2015. Equipment checks and calibrations of the YSI sondes were performed about every three months with 100 percent water-saturated air for the optical dissolved oxygen (DO) sensor (USEPA, 2002; Wagner et al., 2006; Solinst, 2012).

Continuous monitoring measurements were uploaded to an external YSI 650 multi-parameter display system. AQUARIUS software (Wagner et al., 2006) was used to perform quality assurance (corrections were documented), address equipment failure, delete erroneous values during calibration visits, and run statistical analyses. Measurements were verified based on calibration data and field parameter measurements made with the YSI Professional Plus. In general, discrete measurements of DO were consistent with the in situ 30 minute interval DO measurements.

While post-deployment calibrations were not taken on the YSI sonde, the YSI DO Handbook (YSI Incorporated, 2008) indicates that the optical DO sensors maintain and hold their calibration for several months (YSI Incorporated, 2008). Therefore, it was assumed there was little drift in the data over the course of the deployment and that the accuracy of the sensor over time was within the reported 1% (YSI Incorporated, 2008). Equipment failure resulting in loss of DO data occurred at the site up-gradient from the spring on September 13, 2013, at 19:00 hours, through December 19, 2013, on August 28, 2014, 02:00 hours, through October 12, 2014, 16:00 hours, and on January 14, 2015, 13:00 hours. Equipment failure occurred at the down-gradient site on May 23, 2014, through July 9, 2014, on September 4, 2014, 12:00 hours, through October 12, 2014, 16:00 hours, and on January 8, 2015 at 16:30 hours. Data are available in Frus (2016).

Geochemical modeling of results was conducted using mass solution mixing scenarios in Geochemists Workbench (Bethke, 2008). Major ion chemistry values from two or more water samples were combined in various proportions to converge on a final mixing scenario that was representative of the chemistry of an end-member water sample. The process of mass solution mixing was iterative. Saturation indices were also calculated for each of the iterations to aid in determining the most representative mixing scenario. Mass solution mixing was used to determine the seasonal in-

320

fluence that spring water had on the fish habitat. Additionally, mass solution mixing was used to determine the groundwater flow path and source for the spring water.

RESULTS Surface water and groundwater were collected seasonally (May 2012 to October 2015) at the AR site.

A whisker-plot diagram (Figure 6A) represents the variability of the major ion concentrations between water collected at each site. Major ion concentrations analyzed from water samples collected at the spring (black) showed little variation between seasons and years. Surface water samples collected upgradient from the spring input (white) showed major ion concentrations with the largest variability (Figure 6A). The variability in the major ion concentrations from water samples collected up-gradient from the spring input encompass the major ion concentrations from water samples collected at the spring and down-gradient for most of the major ions, except magnesium (Mg2+ ). The magnesium concentration for water samples collected from the spring had a mean and standard deviation of 5.6 ± 0.33 mg/L, water samples collected up-gradient from the spring had a mean and standard deviation of 9.5 ± 1.71 mg/L, and water samples collected down-gradient from the spring had a mean and standard deviation of 7.5 ± 0.60 mg/L. Water samples collected from the spring were the most depleted in δD and δ18 O of all site water (Figure 7). Water samples collected from the spring site also showed the smallest variability in δD and δ18 O ratios, with a maximum range of −3.16‰ and −1.09‰, respectively. Up-gradient from the spring input, surface water samples showed the largest variations, with ranges of −16.06‰ for δD and −3.01‰ for δ18 O. The seasonal differences of the surface water samples upgradient from the spring showed that only winter values were as negative as the spring water samples. Additionally, the seasonal variability of the up-gradient surface water samples moved away from the Global Meteoric Water Line (GMWL) at a slope of 4 in spring and summer seasons. Water samples down-gradient from the spring input were more enriched in δD and δ18 O than the spring water but were depleted relative to the surface water up-gradient from the spring input. The water down-gradient from the spring input showed a range of −9.52‰ and −1.68‰, respectively.

Discrete measurements completed at all locations indicated that throughout the study, each site maintained acid neutrality, with pH averages and standard deviations for all sites of 7.24 ± 0.59 pH units (Table 2). Discrete measurements of specific conductance of the spring water showed the least

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

EEGS-26-02-05_1XO

July 8, 2020

10:2

Groundwater Influence on Endangered Fish Habitat

Figure 6. Hydrochemical results from water samples collected at Agua Remora, Zuni Mountains, NM. (A) Whisker-plot diagram of major ions where whiskers represent the minimum and maximum values, the box represents the first and third quartile, and the center line is the median of the data. (B) Piper diagram showing relative proportions of major ion concentrations for seasonal samples collected at spring and surface water locations.

Figure 7. Stable isotopes δD and δ18 O of sample waters collected seasonally at Agua Remora site waters (N = 41), Zuni Mountains, NM, plotted relative to the global meteoric water line (GMWL; Craig 1961). Results are relative to the Vienna Standard Mean Ocean Water (VSMOW).

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

321

322 pH

n.a.

2.4 3.9 n.a.

3.1 n.a.

0.1 0.9 4.9 1.6 4.8 1.6 0.2 0.1 0.1 n.a. n.a.

6.3 n.a.

4.9 n.a.

6.3 9 7.9 5.5 6.7 9.7 6.3 8.9

11.3 11.4 15.5 8.1 11.9 17.4 14.1 5.5 11.1 11.5 10.8 10.9 6.3 8.9 6.8 17 17.7 16.9 3.2 12.7 18 14.9 3.9 13.5 15.4 9 4.2 6.6 13.5

406 577 563 615 432 539 550 427 456 543 536 377 292 376

501 521 534 510 479 500 569 565 504 439 238 482 543 501 522

322 686 341 485 474 307 399 510 500 268 281 323

Specific Conductivity (μS/cm)

288.6 369.6 393.4 363.5 336.9 365.4 376.7 298.9 319.8 402 380.2 255.9 227.2 285.8

347 351.1 340.9 352.3 384.3 348.8 353.7 358.2 349.1 410.2 372.2 360.7 356.5 388.3 342.4

250.7 444 267.2 335.6 347.5 213.5 285.2 404.5 372.9 194.1 190.9 231.1

Dissolved Solids (mg/L)

26.9 29.6 29.9 32.2 32.4 32 32.7 30.3 31.5 34.6 33.6 29 25.5 30.4

29.7 28.9 31.3 32.4 31.7 32.1 31.9 31.2 31.7 33 32.9 32.8 34.5 30.9 31.9

32.6 32.9 33.7 31.5 36.9 28.1 35.8 37.2 28.8 26.6 25.5 29.9

Ca++ (mg/L)

6.8 7 6.7 7.8 7.4 6.6 7.1 7.9 7.3 8 7.6 8.7 7.3 8.3

5.3 5.2 5.5 5.8 5.4 5.5 5.4 5.3 5.4 6.1 6.1 6.2 6.1 5.4 5.6

10.8 10.6 11.6 9.9 11.5 9.7 10.9 13.4 9.6 9.5 8.2 10

Mg++ (mg/L)

43.6 64.8 79.4 70.4 50.7 71 68.8 37.7 53.9 72.4 69.7 30.2 19.9 34.7

65.1 63.3 60.4 64.3 73 65.7 66.1 67.6 64 68.7 60.7 59.6 63.7 63.4 58.6

15.3 88.8 21 50.2 36.6 10.6 26 64.1 70.2 11.2 9.2 13.8

Na+ (mg/L)

1.7 2 1.8 2.2 1.6 2.3 1.5 1.6 1.5 1.8 2.4 1.7 1.5 1.8

2.1 2.2 2 2.3 1.9 2.1 1.7 1.9 1.9 2.2 1.8 2.2 2.1 2.1 2

2.5 2.8 2.3 2.7 2.2 2.1 2.7 3 2.7 2 1.9 2.2

K+ (mg/L)

131.8 144.6 167.3 116.2 141.6 142 148.9 136.7 147.8 158 114.3 120.8 119.5 144.6

128.7 128.5 130.7 129 133 136.7 132.4 131.2 127.5 151 123.9 127.8 129.4 133.6 133.6

185.5 114.1 171.5 146.4 207.5 147.7 185.6 148.3 98.7 114.7 119 155.1

HCO3 − (mg/L)

58.4 98.5 84.1 130.8 85 103.5 90.1 64 70.2 108.4 123.7 46.4 40.8 47.9

92.7 96.6 86.2 94.7 115.6 94.1 95.1 94.7 91 123.4 116.8 97.9 95.4 127.1 96.7

16.7 158.1 26.9 79.9 40 10.8 32.3 117 128.4 19.6 18.2 16.7

SO4 2− (mg/L)

10.8 16.5 18.4 19.8 11.4 17.8 15.4 9.2 11.8 18.6 20.6 9 6.4 8.4

16.3 15.7 15.9 14.6 15.4 15.5 16.1 15.5 14.6 16.8 16.6 20.6 16.4 16.3 11.5

6.4 26.8 5.4 16.1 10 4 7.1 17.4 23 3.8 3.3 3.9

Cl− (mg/L)

−79.74 −92.13 −85.15 −79.63 −76.07 −82.01 −80.58 −85.1 −87.53 −90.71 −85.93 −84.23 −93.28 −92.7 −91.66 −92.98 −93.25 −93.44 −93.48 −94.83 −92.57 −93.86 −93.13 −92.9 −92.03 −94.32 −94.33 −84.53 −86.56 −87.86 −94.05 −88.64 −88.9 −86.67 −87.02 −85.41 −89.83 −89.34 −92.4 −86.57 −87

−12.65 −12.62 −12.14 −12.69 −12.75 −12.74 −13.23 −12.84 −12.62 −12.97 −13 −13.21 −12.55 −13.48 −13.13 −11.21 −11.38 −11.81 −12.82 −11.63 −11.87 −11.77 −12.05 −11.38 −12.32 −12.68 −12.89 −12.39 −12.32

δD (‰ VSMOW)

−10.44 −12.42 −11.16 −10.16 −9.82 −11.58 −10.04 −10.76 −11.88 −12.83 −11.77 −12.28

δ18 O (‰ VSMOW)

July 8, 2020

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

Abbreviations: °C, degrees centigrade; mg/l, milligram per liter; ‰ VSMOW, per mil relative to Vienna Standard Mean Ocean Water.

n.a. n.a.

1.7 n.a.

0.7 4.7 3 2.7 4.5 10 2.5 3.1

Dissolved Oxygen (mg/L)

12.9 1.2 11.9 19 15.5 0.8 12.7 17.5 8.5 3.9 6.5 11.2

Water Temperature (°C)

Up-gradient from spring 5/17/2012 6.28 12/2/2012 8.77 5/27/2013 6.83 7/22/2013 6.87 9/7/2013 6.88 12/19/2013 7.07 5/23/2014 6.42 7/9/2014 6.93 10/12/2014 7.45 2/21/2015 7.68 4/3/2015 7.47 5/31/2015 6.88 Spring at AR 5/17/2012 7.05 7/4/2012 7.52 9/1/2012 7.24 12/2/2012 8.94 5/27/2013 7.63 7/22/2013 6.63 9/7/2013 6.89 12/19/2013 7.22 5/23/2014 n.a.

7/9/2014 7.25 10/12/2014 7.81 10/18/2014 7.83 2/21/2015 7.07 4/3/2015 7.36 5/31/2015 6.83 Down-gradient from spring 5/17/2012 7.31 7/4/2012 7.13 9/1/2012 7.05 12/2/2012 5.93 5/27/2013 7.58 7/22/2013 6.38 9/7/2013 7.7 12/19/2013 7.79 5/23/2014 6.6 7/9/2014 6.75 10/12/2014 7.49 2/21/2015 7.8 4/3/2015 7.59 5/31/2015 7.51

Sample Location and Date

Table 2. Hydrogeochemical results for surface and spring waters at Agua Remora (AR), first-order stream to Zuni River, Zuni Mountains, NM.

EEGS-26-02-05_1XO 10:2

Frus, Crossey, Dahm, Karlstrom, and Crowley

EEGS-26-02-05_1XO

July 8, 2020

10:2

Groundwater Influence on Endangered Fish Habitat

Figure 8. Field parameters (air temperature [°C], daily precipitation [mm], daily discharge [m3 /s], and water temperature [°C]) plotted with time for sampling locations at the Agua Remora site.

amount of variability, with an average and standard deviation of 493 ± 75.54 µS/cm. The water up-gradient of the spring input showed the largest variability, with average and standard deviations of 408 ± 120 µS/cm, and the water down-gradient from the spring input had an average and standard deviation of 478 ± 92 µS/cm (Table 2).

Continuous monitoring of air temperatures for the AR site showed a season signal with daily mean temperatures having a minimum of −5°C for winter and a maximum of 24°C for the summer (Figure 8). Water temperatures for spring water showed slight seasonal variability, with an average temperature of 11.2°C and a standard deviation of 3.3°C. Water temperature for surface water up-gradient from the spring input reflected more of the seasonal air temperature changes, with a maximum temperature of 19.5°C in the summer and a minimum temperature of 0.8°C in the winter (Figure 8). This yields an average temperature of 10.4°C, which is close to that of the spring water, but with double the standard deviation at ±6.6°C. Discrete and continuous measurements of DO concentrations for the surface water up-gradient and down-gradient from the spring were measured from August 2012 to May 2015. The daily mean DO concentrations from the continuous measurements for the surface water up-gradient from the spring input had an average of 1.5 mg/L, with a maximum DO concen-

tration of 4.6 mg/L. The surface water down-gradient from the spring input maintained a higher DO concentration overall with the average of the daily means at 5.7 mg/L and a maximum DO concentration of 7.7 mg/L (Figure 9).

DISCUSSION To determine the influence that spring water has on the habitat for the endangered ZBS, a small hillslope spring and the perennial surface water up-gradient and down-gradient from the spring were studied. To characterize the seasonal variations in spring and surface water at the AR site, major ions for the different collection sites were plotted on a Piper diagram (Piper, 1944) to allow for comparison (Figure 6B). Water samples collected at the spring (black) varied little between seasons, and water up-gradient from the spring input (white) varied over the seasons, while the water samples down-gradient from the spring input (gray) were intermediate. Magnesium (Mg2+ ) concentrations in the spring and surface water samples were used to infer that spring water comes from a different source than water up-gradient from the spring input. Water down-gradient from the spring input represents a mixture of the two up-gradient waters.

The percent of water that the spring provides to the ZBS habitat for each season was determined us-

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

323

EEGS-26-02-05_1XO

July 8, 2020

10:2

Frus, Crossey, Dahm, Karlstrom, and Crowley

Figure 9. Plot of continuous monitoring and discrete field measurements of dissolved oxygen concentrations (DO, mg/L) with time for sampling locations at the Agua Remora site. Data are plotted against U.S. Environmental Protection Agency water quality standards of DO concentrations for non-salmonid cool-water fish.

ing the mass chemistry of major ions from water collected at the spring and up-gradient surface water (Figure 9). Results indicated that during wet seasons (spring and winter), the spring and up-gradient surface water are contributing equally to the down-gradient ZBS habitat. During dry seasons (summer and fall), the spring is providing as much as 99 percent of the water that is found down-gradient in the ZBS habitat (Figure 10).

Stable isotopes (δD and δ18 O) can be useful to understand recharge and atmospheric mechanisms for a hydrologic system (Glynn and Plummer, 2005; Sharp, 2006). The GMWL represents a linear relationship (slope of 8) between the stable isotopes of water (Craig, 1961). Movement along the GMWL represents fractionation of isotopes of water due to seasonal temperature differences, movement of atmospheric water onto the continent, and movement of atmospheric water to higher latitudes. Precipitation that falls during cold seasons tends to be depleted (more negative) relative to precipitation that falls during warm seasons (Craig, 1961). Movement away from the GMWL represents alteration of surface water due to evapora-

tion, geothermal water-rock interaction, or gaseous exchange. Evaporated water moves away from the GMWL at a slope between 4 and 6 (Craig and Gordon, 1965; Gibson et al., 2008).

An investigation of the seasonal differences of δD and δ18 O in spring water and surface water up-gradient from the spring input offered insight into the timing of recharge events. Water samples collected from the spring varied little across the seasons and were the most depleted of all site waters. Surface water upgradient from the spring input varied, where winter samples (Figure 7, diamonds) were most depleted relative to samples collected in other seasons. The seasonal differences in δD and δ18 O for surface water collected up-gradient of the spring input showed movement along or near the GMWL for fall (Figure 7, upward triangle) and spring (Figure 7, square) seasons and then movement away from the GMWL during the summer (Figure 7, circle). A linear regression (r2 = 0.762, slope = 4) for the water up-gradient from the spring input showed that, in most seasons, there are evaporation effects in the surface water up-gradient from spring input.

Figure 10. Chart of seasonal mixing of spring water (black) and up-gradient surface water (white) for seasonal down-gradient ZBS pools (gray line), where mass solution mixing model was used to determine seasonal mixing proportions (Bethke, 2008).

324

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

EEGS-26-02-05_1XO

July 8, 2020

10:2

Groundwater Influence on Endangered Fish Habitat

When accounting for these variations, a trend for winter and summer (monsoon) recharge begins to take shape (Figure 7), indicating that spring water is recharged from winter precipitation (snowmelt). The isotopic variability of the surface water up-gradient from the spring input suggests that evaporation and recharge occur in both warm and cold seasons (Figure 7). The down-gradient surface water, which is the habitat for the ZBS, is then a mixture of these two different water sources.

Spring-Water Influence on ZBS Habitat Continuous monitoring data of DO concentrations (DO, mg/L) were acquired to help understand the impact that the spring water has on the fish habitat (Figure 9). The U.S. Environmental Protection Agency (USEPA) has established federal water quality standards for the protection of aquatic life. DO concentrations above 6 mg/L are ideal to maintain populations of cold-water fish. DO below 3 mg/L causes acute mortality in different life stages of cold-water fish (USEPA, 2014). At DO between 6 and 3 mg/L, there is no production impairment. The daily mean DO concentrations at the down-gradient ZBS habitat (gray) showed an average DO concentration of 5.7 mg/L. In contrast, within 7 days of the first deployment (July 2013), there was a large rain event that corresponded with a drop in the DO concentration levels of the surface water up-gradient from the spring (white) to below 3 mg/L, and the pool did not recover again through September 2013. While this rain event was also associated with lowered DO levels in the down-gradient ZBS habitat, recovery to pre-event levels took less than 15 days, and DO concentrations never went below the acute mortality line established by the USEPA (Figure 9).

The low values of the DO concentrations at both sites is interpreted to be the result of the delivery of labile organic carbon (both dissolved and particulate) by the rain event. The organic carbon stimulated metabolism that lowered DO concentrations (Dahm, 1981; Grimm et al., 1997; and Dahm et al., 2015). With the delivery of oxygenated spring water to the downgradient surface water, the ZBS habitat recovered, and appropriate DO concentration levels were maintained.

In contrast, the surface water up-gradient from the spring input was essentially stagnant, with limited reaeration, making recovery slow. In the up-gradient surface water, diffusion processes were overwhelmed by organic decomposition (Dahm, 1981; Dahm et al., 2015). This makes the surface water up-gradient from the spring anoxic during low-flow times, and suitable habitat for the fish species is not available due to the lack of spring input delivering re-aerated water.

Additional declines of DO concentrations for both sites were found at the beginning of winter 2014. The surface water up-gradient from the spring began to decline in early November 2014 and bottomed out at 0.5 DO mg/L, while the down-gradient ZBS habitat began declining in early December 2014 and fell below 2 DO mg/L. Through the remaining deployment (January 2015), neither of the pools recovered from this winter sag. This sag is interpreted as a direct effect of the ice covering the pools. Ice greatly reduces reaeration, and DO sags are commonplace under the ice (Baehr and Degrandpre, 2002). Metabolism does not stop under the ice, but re-oxygenation does (Bertilsson et al., 2013).

Regional Hydrogeochemistry for Spring Water To determine groundwater flow paths and mixing scenarios, the spring water chemistry collected in October 2014 was compared to the hydrochemistry of local and regional aquifers, including the Psg, which is the region’s most productive aquifer. A regional data set gathered from the literature was used to identify the variability for each of the possible hydrogeologic sources for the spring water. Throughout the seasons, the spring water had total dissolved solids (TDS) <400 ppm, and therefore regional water values from the Triassic Chinle (TRc) unit, with TDS below 600 ppm (N = 22), were averaged and mixed with regional water values from the Permian Abo (Pa) unit, which had TDS <1,200 ppm. Using Geochemist WorkBench (Bethke, 2008), the mass solution mixing model of major ion chemistry as well as saturation indices (comparable to the spring water) identified two end members for the AR spring sample. Mixing results indicated that the spring at the AR site is 80 percent TRc and 20 percent Pa. To determine the amount of young water recharging the spring, tritium (3 H) sampling was performed on October 18, 2014. 3 H has a half-life of 12.32 years and is found naturally in Earth’s atmosphere due to cosmic rays interacting with nitrogen-14 nuclei.

Tritium can be used as an anthropogenic marker due to the influx of 3 H during the testing of nuclear weapons in the 1950s and 1960s, with peak amount-weighted mean atmospheric concentration (>1,000 TU) in 1963 and 1964 (Plummer et al., 2004; Glynn and Plummer, 2005). Nuclear weapons−generated tritium concentrations rained out of the atmosphere in the 1970s and 1980s with precipitation collected from Albuquerque having amount-weighted mean concentrations at 200 TU and 23 TU, respectively (Plummer et al., 2004).

Since 1992, seasonal mean tritium concentrations in precipitation is reported to be 9.1 ± 3.7 TU for Albuquerque, NM (Plummer et al., 2004; Eastoe et al.,

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

325

EEGS-26-02-05_1XO

July 8, 2020

10:2

Frus, Crossey, Dahm, Karlstrom, and Crowley

2012). Although measurements of tritium concentration before the 1950s are lacking, natural atmospheric 3 H concentrations are estimated to be close to that of post-1992 values. Because of the short half-life, 3 H can be used to determine the amount of young (post1950) water that is stored in aquifers and can help with understanding mixing scenarios (Glynn and Plummer, 2005). Accordingly, water with 3 H < 0.5–1.1 TU is interpreted to have principally/entirely been recharged before 1952; water with 3 H < 1.5 TU was predominately recharged before 1952; water with 3 H = 1.5 TU to >4 TU is a mix of pre-1952 and modern recharge (5–10-year-old recharge); water with 4 to 10 TU has primarily modern recharge; water with 3 H > 10 TU indicates some nuclear weapons testing precipitation is present (1950s–1960s recharge) (Eastoe et al., 2012; Drakos et al., 2013). The amount-weighted mean concentration of tritium for the spring water at AR was found to be 1.5 ± 0.30 TU and is interpreted as being recharged by predominantly older (>70 years) water sources.

Hydrogeochemical data (stable isotopes δD and δ18 O, mass mixing, and tritium analysis) were combined with geologic data (geologic mapping, cross section, and stratigraphy) to understand the flow path and residence time of the spring water at AR. More than 70 years ago, snowmelt infiltrated into the TRc in the northeast portions of the Zuni Mountains, outside of the Rio Nutria watershed (Figure 3). This water then flowed horizontally toward the west, remaining in the TRc. Near the northern boundary of Oso Ridge, the water intersects a fault and flows vertically down into the Precambrian basement rock. The groundwater then flows south and vertically up to mix with the Pa water and emerges to the surface at the AR site (Figure 4).

Additional evidence for this flow path along the inferred basement faults is also supported by the specific conductance of the two water sources (Figure 8).

While both the spring water and up-gradient water have relatively low ion concentrations, the flow path for the spring water is much longer. The explanation for the longer flow path associated with low conductivity is linked to minimal interaction with the Precambrian quartz monzonite (Xm). While geothermal waters are able to interact with basement rock and undergo significant chemical modification, cooler waters have very little ion exchange with these rocks over shorter timescales (Drever, 1982). We interpret that the spring water at AR carried the initial ion concentration from the TRc interaction upon infiltration and then flowed through the Xm fault, with little chemical modification, to emerge at the Pa/Xm contact, where waters mixed and discharged at AR spring.

326

CONCLUSIONS Using spatial and temporal analyses of major ions, stable isotopes (δD, δ18 O), and DO concentration of spring water and surface water both up-gradient and down-gradient from spring discharge, two sources of water were identified at the AR site. The downgradient surface water is home to the endangered ZBS and has been listed as critical habitat for the species.

The spring water is found to have little variability in the proportions of major ion concentrations and to be primarily recharged from snowmelt that has had limited interaction with the atmosphere over the last 70 years. The second water source, the surface water up-gradient from the spring, has major ion concentrations that vary throughout the seasons, is recharged with both rain and snow events, and undergoes evaporation. The surface water up-gradient from the spring is interpreted to be that of the shallow alluvium that flows in the stream channel from the surrounding area.

The down-gradient surface water is then a mixture of these two water sources. Depending on the season, the spring water provides between 35 and 99 percent of the down-gradient water; in summer and fall, the spring is contributing the majority of the water to the ZBS habitat. Continuous monitoring of DO revealed that the surface water up-gradient from the spring is a stagnant pool that becomes hypoxic and anoxic due to the lack of movement of water through the channel. The water down-gradient from the spring input is recharged by water from the hillslope spring, which provides reaerated water that keeps the DO levels in appropriate ranges to maintain the ZBS population. Hydrochemical mixing models indicate that the spring is a mixture of Triassic Chinle Formation (TRc) and Permian (Pa) Abo Formation water with an 80/20 ratio, respectively. Tritium data analyzed from spring water indicate that the groundwater was recharged more than 70 years ago. Therefore, before 1945, melting snow infiltrated the TRc (located on the eastern and northern flanks of the mountain range) and flowed southwest along an unmapped fault through the Xm. The discharge of the spring is at the contact of the Xm and the Pa, where small amounts of Pa water mixed with TRc water before emerging to the surface. This work provides evidence that the groundwater spring discharge is imperative for the survival of the ZBS. The DO data are a clear indication of the impacts that the spring water has on the ZBS and how essential the perennial spring flow is for maintaining the ZBS habitat at the AR site. The results from this study have informed resource management agencies to move the focus from using springs for consumptive uses (such as development for livestock) to instead drilling wells in other hydrostratigraphic units for livestock. This allows the

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

EEGS-26-02-05_1XO

July 8, 2020

10:2

Groundwater Influence on Endangered Fish Habitat

water in the springs to support the ecosystem and wildlife instead of being piped away.

ACKNOWLEDGMENTS We thank Adam Frus, Elizabeth Frus, Rebecca Wacker, Chad Bryant, Lauren Main, Mariah Kelley, Rachel Swatenson-Franz, Pavel Vakhlamov, and Chris McGibbon for assistance in the field and laboratory.

Analyses were made with the help of Dr. Abdul-Mehdi Ali (Analytical Laboratory in the Department of Earth and Planetary Sciences), Dr. Viorel Atudordi and Dr. Laura Burkemper (Center for Stable Isotopes), and Dr. David Dettman (Environmental Isotope Laboratory). Funding for this project was provided by Cibola National Forest as a shared-cost agreement.

REFERENCES Anderson, O. J.; Maxwell, C. H.; and Lucas, S. G., 1998, Geologic map of Fort Wingate quadrangle, McKinley County, New Mexico, Scale 1:24,000. In Anderson, O. J.; Maxwell, C. H.; and Lucas, S. G. (Editors.), Geology of Fort Wingate Quadrangle, McKinley County, New Mexico: New Mexico Bureau of Geology and Mineral Resources Open-File Report 473, 16 p.

Anderson, O. J.; Maxwell, C. H.; and Lucas, S. G., 2003, Geology of Fort Wingate Quadrangle, McKinley County, New Mexico: New Mexico Bureau of Geology and Mineral Resources Open-File Report 473, 16 p.

Apaydin, A., 2010, Relation of tectonic structure to groundwater flow in the Beypazari region, NW Anatolia, Turkey: Hydrogeology Journal, Vol. 18, No. 6, pp. 1343–1356. https://doi.org/10.1007/s10040-010-0605-1.

Baars, D. L., 1974, Permian rocks of north-central New Mexico. In Woodward, L. A. and Callender, J. F. (Editors) New Mexico Geological Society Guidebook 25th Field Conference: New Mexico Bureau of Geology and Mineral Resources, Socorro, NM, pp. 167–171. Retrieved from https://nmgs.nmt.edu/publications/guidebooks/downloads/ 25/25_p0167_p0169.pdf.

Baehr, M. M. and Degrandpre, M. D., 2002, Under-ice CO2 and O2 variability in a freshwater lake: Biogeochemistry, Vol. 61, pp. 95–114. https://doi.org/10.1023/a:1020265315833.

Baldwin, J. A. and Anderholm, S. K., 1992, Hydrogeology and Ground-Water Chemistry of the San Andres–Glorieta Aquifer in the Acoma Embayment and Eastern Zuni Uplift, West-Central New Mexico: U.S. Geological Survey Water-Resources Investigations Report 91-4033, 313 p.

Baldwin, J. A. and Rankin, D. R., 1995, Hydrogeology of Cibola County, New Mexico.: U.S. Geological Survey WaterResources Investigations 94-4178, 102 p.

Banerjee, A.; Person, M.; Hofstra, A.; Sweetkind, D. S.; Cohen, D.; Sabin, A.; Unruh, J.; Zyvoloski, G.; Gable, C. W.; Crossey, L.; and Karlstrom, K. E., 2011, Deep permeable fault-controlled helium transport and limited mantle flux in two extensional geothermal systems in the Great Basin, United States: Geology, Vol. 39 No. 3, pp. 195–198. https://doi.org/10.1130/G31557.1.

Barnes, C., 1988, Tracing of water movement in the unsaturated zone using stable isotopes of hydrogen and oxygen: Journal Hydrology, Vol. 100, No. 1, pp. 143–176. https://doi.org/10.1016/0022-1694(88)90184-9.

Bertilsson, S.; Burgin, A.; Carey, C. C.; Fey, S. B.; Grossart, H.-P.; Grubisic, L. M.; Jones, I. D.; Kirillin, G.; Lennon, J. T.; Shade, A.; and Smyth, R. L., 2013, The under-ice microbiome of seasonally frozen lakes: Limnology and Oceanography, Vol. 58, No. 6, pp. 1998–2012. https://doi.org/10.4319/lo.2013.58.6.1998.

Bethke, C. M., 2008, Geochemical and Biogeochemical Reaction Modeling, 2nd ed.: Cambridge University Press, Cambridge, U.K., 564 p.

Carman, S. M., 2004, Zuni Bluehead sucker (Catostomus discobolus yarrowi) Recovery Plan: New Mexico Department of Game and Fish, Santa Fe, NM, 41 p.

Colpitts, R. M. J., 1989, Permian reference section for southeastern Zuni Mountains, Cibola County, New Mexico. In Anderson, O.J.; Lucas, S. G., Lover, D. W., and Cather, S. M. (Editors), New Mexico Geological Society 10th Field Conference Guidebook: New Mexico Bureau of Geology and Mineral Resources, Socorro, NM, pp. 177–181.

Connell, S. D., 2011, Preliminary Study of the Geologic Framework of the Colorado Plateau—Middle Rio Grande Basin Transition, New Mexico: New Mexico Bureau of Geology and Mineral Resources Open-File Report 539, 58 p.

Cooley, M. E.; Harshbarger, J. W.; Akers, J. P.; and Hardt, W. F., 1969, Regional Hydrogeology of the Navajo and Hopi Indian Reservations, Arizona, New Mexico, and Utah: U.S. Geological Survey Professional Paper 521-A, 61 p.

Craig, H., 1961, Isotopic variations in meteoric waters: Science, Vol. 133, No. 3465, pp. 1702–1703, https://doi.org/10.1126/science.133.3465.1702.

Craig, H. and Gordon, L. I., 1965, Deuterium and oxygen 18 variations in the ocean and the marine atmosphere. In Tongiorgi, E. (Editor) Paleotemperatures, Conference on Stable Isotopes in Oceanographic Studies: Consiglio Nazionale delle Richerche, Laboratorio de Geologia Nucleare, Pisa, Italy, pp. 8–130.

Crossey, L. J.; Fischer, T. P.; Patchett, P. J.; Karlstrom, K. E.; Hilton, D. R.; Newell, D. L.; Huntoon, P.; Reynolds, A. C.; and de Leeuw, G. A. M., 2006, Dissected hydrologic system at the Grand Canyon: Interaction between deeply derived fluids and plateau aquifer waters in modern springs and travertine: Geology, Vol. 34, No. 1, pp. 25–28. https://doi.org/10.1130/G22057.1.

Crossey, L. J.; Karlstrom, K. E.; Springer, A. E.; Newell, D. L.; Hilton, D. R.; and Fischer, T. P., 2009, Degassing of mantle-derived CO2 and He from springs in the southern Colorado Plateau region—Neotectonic connections and implications for groundwater systems:. Geological Society America Bulletin, Vol. 121, No. 7, pp. 1034–1053. https://doi.org/10.1130/B26394.1.

Dahm, C. N., 1981, Pathways and mechanisms for removal of dissolved organic carbon from leaf leachate in streams: Canadian Journal Fisheries and Aquatic Sciences, Vol. 38, pp. 68–76. https://doi.org/10.1139/f81-009.

Dahm, C. N.; Candelaria-Ley, R. I.; Reale, C. S.; Reale, J. K.; and Van Horn, D. J., 2015, Extreme water quality degradation following a catastrophic forest fire: Freshwater Biology, Vol. 60, No. 12, pp. 2584–2599. https://doi.org/10.1111/fwb.12548.

Drakos, P. G.; Riesterer, J. W.; and Bemis, K., 2013, Recharge sources and characteristics of springs on the Zuni Reservation, New Mexico. In Zeigler, K.; Timmons, M.; Timmons, S.; and Semken, S. (Editors), New Mexico Geological Society 64th Field Conference Guidebook, Geology of Route 66 Region: Flagstaff to Grants, 2013: New Mexico Bureau of Geology and Mineral Resources, Socorro, NM, pp. 205–213.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

327

EEGS-26-02-05_1XO

July 8, 2020

10:2

Frus, Crossey, Dahm, Karlstrom, and Crowley Drever, J. I., 1982, The Geochemistry of Natural Waters, 1st ed.: Prenctice-Hall, Englewood Cliffs, NJ. Eastoe, C. J.; Watts, C. J.; Ploughe, M.; and Wright, W. E., 2012, Future use of tritium in mapping pre-bomb groundwater volumes: GroundWater, Vol. 50, No. 1, pp. 87–93. https://doi.org/10.1111/j.1745-6584.2011.00806.x.

Erslev, E. A., 1991, Trishear fault-propagation folding: Geology, Vol. 19, No. 6, pp. 617–620.

Frus, R.J., 2016, Multidisciplinary Work to Determine Hydrology of Arid Land Springs and How Spring Waters Influence Water Quality and Ecosystem Health for Desert Environments: Ph.D. Dissertation, Department of Earth and Planetary Sciences, University of New Mexico, Albuquerque, NM, 216 p. https://digitalrepository.unm.edu/eps_etds/29/ (accessed on 01/19/2020).

Gibson, J. J.; Birks, S. J.; and Edwards, T. W. D., 2008, Global prediction of δA and δ2 H-δ18 O evaporation slopes for lakes and soil water accounting for seasonality: Global Biogeochemical Cycles, Vol. 22, No. 2, pp. GB2031. https://doi.org/10.1029/2007GB002997.

Gido, K. B. and Propst, D. L., 1999, Habitat use and association of native and nonnative fishes in the San Juan River, New Mexico and Utah: Copeia, Vol. 1999, No. 2, pp. 321–332. https://doi.org/10.2307/1447478.

Gilbert, E. I. and Carman, S. M., 2013, Zuni Bluehead Sucker Monitoring and Conservation Efforts 2011 and 2012: New Mexico Department of Game and Fish, Sante Fe, NM, 46 p.

Glynn, P. D. and Plummer, L. N., 2005, Geochemistry and the understanding of ground-water systems: Hydrogeology Journal, Vol. 13, No. 1, pp. 263–287. https://doi.org/10.1007/s10040004-0429-y.

Grimm, N. B.; Chacon, A.; Dahm, C. N.; Hostetler, S. W.; Lind, O. T.; Starkweather, P. L.; and Wurtsbaugh, W. W., 1997, Sensitivity of aquatic ecosystems to climatic and anthropogenic changes: The Basin and Range, American Southwest and Mexico: Hydrological Processes, Vol. 11, pp. 1023–1041. https://doi.org/10.1002/(SICI)10991085(19970630)11:8<1023::AID-HYP516>3.0.CO;2-A. Gudmundsson, A., 2000, Active fault zones and groundwater flow: Geophysical Research Letters, Vol. 27, No. 18, pp. 2993–2996. https://doi.org/10.1029/1999GL011266.

Gupta, P.; Noone, D.; Galewsky, J.; Sweeney, C.; and Vaughn, B. H., 2009, Demonstration of high-precision continuous measurements of water vapor isotopologues in laboratory and remote field deployments using wavelength-scanned cavity ringdown spectroscopy (WS-CRDS) technology: Rapid Communications Mass Spectrometry, Vol. 23, No. 16, pp. 2534–2542.

Harrington, G. A.; Cook, P. G.; and Herczeg, A. L., 2002, Spatial and temporal variability of groundwater recharge in Central Australia: A tracer approach: GroundWater, Vol. 40, No. 5, pp. 518–528.

Heckert, A. B. and Lucas, S. G., 2003, Triassic stratigraphy in the Zuni Mountains, west-central New Mexico. In Lucas, S. G., Semken, S. C., Berglof, W., Ulmer-Scholle, D. (Editors) New Mexico Geological Society 54th Field Conference Guidebook: New Mexico Bureau of Geology and Mineral Resources, Socorro, NM, pp. 245–263.

Hood, J. and Kister, L., 1962, Saline-Water Resources of New Mexico: U.S. Geological Survey Water-Supply Paper 1601, 70 p.

Hubbs, C., 1995, Springs and spring runs as unique aquatic systems: Copeia, Vol. 4, pp. 989–991.

Jacobs, K. L. and Holway, J. M., 2004, Managing for sustainability in an arid climate: Lessons learned from 20 years of groundwater management in Arizona,

328

USA: Hydrogeology Journal, Vol. 12, No. 1, pp. 52–65. https://doi.org/10.1007/s10040-003-0308-y.

Karlstrom, K. E.; Dallmeyer, R. D.; and Grambling, J. A., 1997, Ar evidence for 1.4 Ga regional metamorphism in New Mexico: Implications for thermal evolution of lithosphere in the southwestern USA: Journal of Geology, Vol. 105, No. 2, pp. 205–224. https://doi.org/10.1086/515912.

Kelley, V. C., 1967, Tectonics of the Zuni-Defiance region, New Mexico and Arizona. In Trauger, F. (Editor), Guidebook of the Defiance–Zuni–Mt. Taylor Region Arizona and New Mexico; 18th Field Conference: New Mexico Bureau of Geology and Mineral Resources, Socorro, NM, pp. 27–33.

Krainer, K.; Lucas, S. G.; and Kues, B. S., 2003, Upper Pennsylvanian strata in the Zuni Mountains, west-central New Mexico. In New Mexico Geological Society 54th Field Conference Guidebook: New Mexico Bureau of Geology and Mineral Resources, Socorro, NM, pp. 219–231.

MacDonald, G. M., 2010, Climate change and water in southwestern North America special feature: Water, climate change, and sustainability in the southwest: Proceedings of the National Academy of Sciences of the United States of America, Vol. 107, No. 50, pp. 21256–21262. https://doi.org/10.1073/pnas.0909651107.

Manga, M., 1996, Hydrology of spring-dominated streams in the Oregon Cascades: Water Resources Research, Vol. 32, No. 8, pp. 2435–2439. https://doi.org/10.1029/96WR01238.

Manga, M., 1999, On the timescales characterizing groundwater discharge at springs: Journal Hydrology, Vol. 219, pp. 56–69. https://doi.org/10.1016/S0022-1694(99)00044-X. Minckley, W. L. and Deacon, J. E., 1991, Battle Against Extinction. Native Fish Management in the American southwest: University of Arizona Press, Tuscon, AZ, 371 p.

Minckley, W. L. and Douglas, M. E., 1991, Discovery and extinction of western fishes: A blink of the eye in geologic time. Chapter 1. In Minckley, W. L. and Deacon, J. E. (Editors), Battle Against Extinction: Native Fish Management in the American Southwest: University of Arizona Press, Tuscon, AZ, pp. 1–7.

Myers, M. D., 2006, National field manual for the collection of water-quality data. Chapter A4, U.S. Geological Survey Techniques of Water-Resources Investigations. 06-A4, Reston, VA, 166 p.

Oakton, 2000, Portable Waterproof pH/Con 300 Meter, R1 ed.: Manual for Use of Meter: Oakton, Vernon Hills, IL, 28 p.

Overpeck, J. and Udall, B., 2010, Dry times ahead: Science, Vol. 328, No. 06, pp. 1642–1644.

Paul, D. and Mitra, S., 2012, Control of basement faults on the geometry and evolution of compressional basement-involved structures: AAPG Bulletin, Vol. 96, No. 10, pp. 1899–1930. https://doi.org/10.1306/02271211140.

Perla, B. S. and Stevens, L. E., 2008, Biodiversity and productivity at an undisturbed spring in comparison with adjacent grazed riparian and upland habitats. In Stevens, L. E. and Meretsky, V. J. (Editors), Aridland Springs in North America Ecology and Conservation: University of Arizona Press and The Arizona-Sonora Desert Museum, Tuscon, AZ, pp. 230– 243.

Phillips, F. M.; Walvoord, M. A.; and Small, E. E., 2013, Effects of environmental change on groundwater recharge in the desert southwest. In Hogan, J. F.; Phillips, F. M.; and Scanlon, B. R. (Editors), Groundwater Recharge in a Desert Environment: The Southwestern United States: American Geophysical Union, Washington, D.C., pp. 273–294.

Piper, A. M., 1944, A graphic procedure in the geochemical interpretation of water analyses: Eos, American

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

EEGS-26-02-05_1XO

July 8, 2020

10:2

Groundwater Influence on Endangered Fish Habitat Geophysical Union Transactions, Vol. 25, pp. 914– 928.

Plummer, L. N.; Bexfield, L. M.; Anderholm, S. K.; Sanford, W. E.; and Busenberg, E., 2004, Geochemical characterization of Ground-Water Flow in the Santa Fe Group Aquifer System, Middle Rio Grande Basin, New Mexico: U.S. Geological Survey Water-Resources Investigation Report 03-4131, 395 p.

Pool, T. K. and Olden, J. D., 2014, Assessing long-term fish responses and short-term solutions to flow regulation in a dryland river basin: Ecology of Freshwater Fish, Vol. 24, pp. 56–66. https://doi.org/10.1111/eff.12125.

Propst, D. L., 1999, Threatened and Endangered Fishes of New Mexico, Technical Report No. 1: New Mexico Department of Game and Fish, Santa Fe, NM, 84 p.

Propst, D. L.; Gido, K. B.; and Stefferud, J. A., 2008, Natural flow regimes, nonnative fishes, and native fish persistence in arid-land river systems: Ecological Applications, Vol. 18, pp.

1236–1252. https://doi.org/10.1890/07-1489.1.

Propst, D. L.; Hobbes, A. L.; and Stroh, T. L., 2001, Distribution and notes on the biology of Zuni Bluehead Sucker, Catostomus discobolus yarrowi, in New Mexico: Southwestern Naturalist, Vol. 46, No. 2, pp. 158–170.

Repenning, C. A.; Cooley, M. E.; and Akers, J. P., 1969, Stratigraphy of the Chinle and Moenkopi Formations, Navajo and Hopi Indian Reservations, Arizona, New Mexico, and Utah: U.S. Geological Survey Professional Paper 521-B, 33 p.

Robertson, A. J.; Henry, D. W.; and Langman, J. B., 2013, Geochemical Evidence of Groundwater Flow Paths and the Fate and Transport of Constituents of Concern in the Alluvial Aquifer at Fort Wingate Depot Activity, New Mexico, 2009: U.S. Geological Survey Scientific Investigations Report 2013-5098, 89 p.

Romanini, D.; Kachanov, A. A.; and Stoeckel, F., 1997, Diode laser cavity ring down spectroscopy: Chemical Physics Letters, Vol. 270, No. 5, pp. 8.

Rounds, S.A., 2012, Alkalinity and acid neutralizing capacity, ver. 4.0 (September 2012). U.S. Geological Survey Techniques of Water-Resources Investigations, book 9, chap. 6.6. https://doi.org/10.3133/twri09A6.6.

Ruhí, A.; Holmes, E. E.; Rinne, J. N.; and Sabo, J. L., 2014, Anomalous droughts, not invasion, decrease persistence of native fishes in a desert river: Global Change Biology, Vol. 21, pp.

1482–1496. https://doi.org/10.1111/gcb.12780.

Sharp, Z., 2006, Priniciples of Stable Isotope Geochemistry, 1st ed.: Pearson Prentice Hall, Upper Saddle River, NJ, 385 p.

Solinst, 2012, User Guide Levelogger Series Software Version 4: Solinst Canada Ltd., Georgetown, ON, Canada, 82 p.

Springer, A. E. and Stevens, L. E., 2009, Spheres of discharge of springs: Hydrogeology Journal, Vol. 17, No. 928, pp. 83–93. https://doi.org/10.1007/s10040-008-0341-y.

Stevens, L. E. and Meretsky, V. J., 2008, Aridland Springs in North America Ecology and Conservation: University of Arizona Press and The Arizona-Sonora Desert Museum, Tuscon, AZ, 432 p.

Strickland, D.; Heizler, M. T.; Selverstone, J.; and Karlstrom, K. E., 2003, Proterozoic evolution of the Zuni Mountains, western New Mexico: Relationship to the Jemez Lineament and implications for a complex cooling history: In Lucas, S. G., Semken, S. C., Berglof, W., Ulmer-Scholle, D. (Editors) New Mexico Geological Society 54th Field Conference Guidebook: New Mexico Geological Society, Socorro, NM, pp. 109– 119.

Turner, T. F. and Wilson, W. D., 2009, Conservation Genetics of Zuni Bluehead Sucker (Catostomus discobolus yarrowi) in New Mexico. Final Report Contract # 09-516-0000-00017:

New Mexico Department of Game and Fish, Santa Fe, NM, 18 p.

Unmack, P. J. and Minckley, W. L., 2008, The demise of desert springs. In Stevens, L. E. and Meretsky, V. J. (Editors), Aridland Springs in North America Ecology and Conservation: University of Arizona Press and The Arizona-Sonora Desert Museum, Tuscon, AZ, pp. 11–34.

U.S. Environmental Protection Agency (USEPA), 1994, Method 200.7: Determination of Metals and Trace Elements in Water and Wastes by Inductively Coupled Plasma–Atomic Emission Spectrometry, Revision 4.4: U.S. Environmental Protection Agency, Cincinnati, OH. U.S. Environmental Protection Agency (USEPA), 1997, Method 300.1: Determination of Inorganic Anions in Drinking Water by Ion Chromatography, Revision 1.0: U.S. Environmental Protection Agency, Cincinnati, OH. U.S. Environmental Protection Agency (USEPA), 1986, Ambient Water Quality Criteria for Dissolved Oxygen (EPA Gold Book 440/5-86-003): U.S. Environmental Protection Agency, Washington, D.C. U.S. Environmental Protection Agency (USEPA), 2002, Standard Operating Procedure for Calibration and Field Measurement Procedures for the YSI Model 6-Series Sondes (including: Temperature, pH, Specific Conductance, Turbidity, and Dissolved Oxygen, Vol. 1: U.S. Environmental Protection Agency, Washington, D.C. U.S. Environmental Protection Agency (USEPA), 2014, Water Quality Standards Handbook (40 CFR 131, Subpart C): U.S. Environmental Protection Agency, Washington, D.C. http://water.epa.gov/scitech/swguidance/standards/ handbook/chapter06.cfm.

U.S. Fish and Wildlife Service (USFWS), 2013, Endangered and Threatened Wildlife and Plants; Proposed Endangered Status for the Zuni Bluehead Sucker. Federal Register, Vol. 78: U.S. Fish and Wildlife Service, Washington, D.C. U.S. Fish and Wildlife Service (USFWS), 2014, Endangered and Threatened Wildlife and Plants; Endangered Species Status for the Zuni Bluehead Sucker; Final Rule, 50 CFR Part 17, Federal Register, Vol. 79: U.S. Fish and Wildlife Service, Washington, D.C. http://www.gpo.gov/fdsys/pkg/FR-201407-24/pdf/2014-17205.pdf.

Wagner, R. J.; Boulger, R. W., Jr.; Oblinger, C. J.; and Smith, B. A., 2006, Guidelines and Standard Procedures for Continuous Water-Quality Monitors: Station Operation, Record Computation, and Data Reporting: U.S. Geological Survey Techniques and Methods 1-D3, 51 p.

Williams, A. J.; Crossey, L. J.; Karlstrom, K. E.; Newell, D. L.; Person, M.; and Woolsey, E., 2013, Hydrogeochemistry of the Middle Rio Grande aquifer system— Fluid mixing and salinization of the Rio Grande due to fault inputs: Chemical Geology, Vol. 351, pp. 281–298. https://doi.org/http://dx.doi.org/10.1016/j.chemgeo.2013.

05.029.

Wilson, M. J., 2004, Weathering of the primary rockforming minerals: Processes, products and rates: Clay Minerals, Vol. 39, No. 3, pp. 233–266. https://doi.org/10.1180/0009855043930133.

YSI Incorporated., 2008, The Dissolved Oxygen Handbook: Yellow Springs Instruments, Yellow Springs, OH. YSI Incorporated., 2009, Multiparameter Water Quality Sondes User Manual 6-Series: Yellow Springs Instruments, Yellow Springs, OH. Zektser, S.; Loáiciga, H. A.; and Wolf, J. T., 2005, Environmental impacts of groundwater overdraft: Selected case studies in the southwestern United States: Environmental Geology, Vol. 47, pp. 396–404. https://doi.org/10.1007/s00254-004-1164-3.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 313–329

329

EEGS-26-02-05_1XO

July 8, 2020

10:2

0

EEGS-26-02-06_1XO

July 8, 2020

8:28

Using Reference Springs to Describe Expected Flow, Temperature, and Chemistry Conditions for Geologically Related Groups of Springs SUSAN K. SWANSON* Beloit College, Department of Geology, 700 College Street, Beloit, WI 53511

GRACE E. GRAHAM DAVID J. HART Wisconsin Geological and Natural History Survey, 3817 Mineral Point Road, Madison, WI 53705

Key Terms: Springs, Groundwater, Temperature, Geochemistry, Spring Monitoring ABSTRACT Reference springs can result in improved conceptual models of groundwater flow and an expanded understanding of the temporal variations in flow, temperature, and chemistry that may be expected for related springs. The reference spring concept is patterned off of the common practice of establishing reference sites to establish benchmark ecological conditions. We use the term “reference spring” to indicate a spring that is minimally disturbed and representative of a geologically related group of springs. Seven reference springs were selected from six previously defined groups that represent over 400 springs in Wisconsin, United States. Geologic cross-sections were constructed for each reference spring site, and springs were monitored for flow, spring-water temperature, and chemistry for up to 4 years, revealing new relationships within and between groups. Examples include: a range of temperature conditions that can be related to the depth of groundwater circulation for springs discharging from layered bedrock uplands, and differences in thermal patterns and specific conductance that can be related to the permeability and solubility of surficial glacial deposits for springs emanating from uneven glacial terrain. The results suggest that establishing reference springs may be a useful approach in other regions where geologically related groups of spring have been identified.

INTRODUCTION Springs result from the convergence or concentration of groundwater flow due to topography, fault

*Corresponding author email: swansons@beloit.edu

lines, fracture sets, solution channels, layered geologic media, or other systematic structures and conditions within a geologic terrain (Bryan, 1919; Alfaro and Wallace, 1994; Fetter, 2001; Smart and Worthington, 2004; and Springer and Stevens, 2009). Therefore, where one spring occurs, other geologically related springs are likely. Groups or clusters of springs align with surface exposures or subcroppings of the discontinuities that aid in focusing the upwelling or gravity drainage of water. Depending on the complexity of the environment, one or more groups of springs may be present. Such groupings of geologically related springs have been shown to exhibit similar geochemical (Swanson et al., 2001; Deocampo, 2004), temperature (Bundshuh, 1993; Luhmann et al., 2011), and general flow properties (Manga, 1997; Swanson and Bahr, 2004; and Florea and Vacher, 2006).

Particularly for springs discharging from granular or fractured media, as opposed to karst springs, which can have unique recharge processes and conduit networks producing more variable discharge and quality (Karimi, 2012), springs in a given geologically related group may be similarly vulnerable to changes in groundwater recharge or withdrawals (Hu et al., 2007; Leake et al., 2008; and Parsen et al., 2016).

They may also be equally susceptible to the effects of human activities on spring-water quality, such as elevated concentrations of chloride from road salt or nitrate from agricultural fertilizers (Swanson et al., 2001; Foos, 2003; Werner and di Pretoro, 2006; and Valerity et al., 2015).

Monitoring all springs in a given group may be costly or difficult, especially in remote settings (Tobin and Schwartz, 2016). For these reasons, long-term monitoring efforts for springs may be particularly well suited to the concept of a reference site, once groups of geologically related springs are identified. In stream ecology studies, reference sites are commonly used to establish benchmark conditions for a stream within

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

331

EEGS-26-02-06_1XO

July 8, 2020

8:28

Swanson, Graham, and Hart

a particular region. The reference sites are often relatively undisturbed by humans, so that they may be used to estimate attainable ecological conditions or to evaluate temporal and spatial changes in ecological integrity. Regions are commonly defined as areas with similar geomorphology, soils, natural vegetation, and land use (Hughes et al., 1986; Hughes, 1995; Stoddard et al., 2006). Where the concept of a reference site is applied to benthic invertebrate studies or assessments of water quality, minimally disturbed sites are defined by selected chemical, as well as physical and biological characteristics (Reynoldson et al., 1997). In this study, we use the term “reference spring” to indicate a spring that is minimally disturbed and representative of a geologically related group of springs.

Physico-chemical trends detected at reference springs are likely to be useful in evaluating the direct or indirect effects of human activities on spring hydrology, as well as broader aquifer conditions, because springs can integrate the signals of geological and hydrological processes over large spatial areas and long periods of time (Manga, 2001). Tobin and Schwartz (2016) concluded that selective monitoring of hydrogeochemical groups of karst springs could be used to assess overall aquifer behavior, vulnerability, and aquifer conditions. In this study, we tested the use of reference springs discharging primarily from granular or fractured media to improve the understanding of the dominant controls on spring occurrence in a geologically diverse region. We integrated surface and subsurface stratigraphic information, geochemical data, and thermal patterns in order to characterize spatial and temporal groundwater conditions at the reference springs. Changes in the geochemical or thermal conditions in the future may signal the need for monitoring of other geologically related springs, as well as inform their vulnerability to land-use changes, diversions of flow, or changes in recharge.

STUDY AREA The state of Wisconsin, United States, is home to an abundance of springs, found across the state and in a variety of geologic settings. Proterozoic and Archean sandstones, lava flows, and crystalline rocks underlie the region. Above these rocks, mostly undeformed Paleozoic sedimentary rocks gently dip away from the Precambrian high (the Wisconsin Dome) in northcentral Wisconsin. During the Pleistocene, repeated advances of the Laurentide Ice Sheet and meltwater streams deposited glacial and alluvial deposits, covering the bedrock in the northern and eastern portions of the state and creating a variety of glacial landforms such as moraines, drumlins, kettles, and outwash plains (Dott and Attig, 2004; Mudrey et al.,

332

2007). In southwestern Wisconsin (the Driftless Area), Cambrian- and Ordovician-age strata are deeply dissected and regularly exposed in steep and narrow valleys. Pleistocene deposits are absent except for layers of loess on ridges. Holocene hillslope sediment is present on valley sides, and stream sediment fills valley bottoms (Figure 1) (Clayton and Attig, 1997; Mudrey et al., 2007).

The principal aquifers in Wisconsin are composed of Cambrian and Ordovician sandstones and dolomites, Silurian dolomite, and Quaternary sand and gravel deposits. The Cambrian-age strata primarily consist of coarse- to medium-grained sandstone (Mt. Simon Formation) or medium- to fine-grained sandstone (Wonewoc Formation, Tunnel City Group, Jordan Formation), with some shale (Eau Claire Formation) and dolomitic siltstone (St. Lawrence Formation) (Kammerer et al., 1998) (Figure 2). Intergranular groundwater flow dominates, but bedding-plane fractures and high-permeability zones are also known to contribute to the preferential flow of groundwater through some of these rocks (Swanson, 2007). The Ordovician-age strata consist of dolomite (Prairie du Chien Group), sandstone with some shale (Ancell Group), dolomite with some limestone and shale (Sinnipee Group), and shale (Maquoketa Formation).

Intergranular groundwater flow dominates in the clastic rocks, but the carbonates are commonly fractured, with the fractures providing primary pathways for groundwater flow (Kammerer et al., 1998; Bradbury, 2009).

The Silurian dolomite aquifer occurs in eastern Wisconsin (Figure 1). The aquifer has low primary permeability, but it is heavily fractured. Thin, horizontal fracture zones related to bedding planes are laterally extensive and continuous for several kilometers. They have significant transmissivity and control groundwater flow at local and regional scales (Muldoon et al., 2001; Rayne et al., 2001). The sand and gravel aquifer consists of un-lithified sand and gravel within glacial deposits and valley alluvium. It is not continuous like the Cambrian and Ordovician sandstones and dolomites or the Silurian dolomite. Instead, it exists as surficial outwash deposits and valley trains or isolated lenses of sand and gravel within less permeable till or glacio-lacustrine material. Intergranular groundwater flow dominates (Kammerer et al., 1998).

Well-developed karst conditions exist within some of the carbonate bedrock units in the region, such as the Prairie du Chien Group, the Sinnipee Group, and the Silurian dolomite. These rocks are commonly fractured, with the fractures providing the primary pathways for groundwater flow. Some of the fractures have also been enlarged by rock dissolution, primarily in northeastern and southwestern Wisconsin, where the

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

EEGS-26-02-06_1XO

July 8, 2020

8:28

Using Reference Springs to Describe Flow Conditions

Figure 1. General geology of Wisconsin. (a) Surficial deposits modified from Wisconsin Geological and Natural History Survey (1989). (b) Bedrock geology modified from Mudrey et al. (2007).

glacial cover is thin or absent (Muldoon et al., 2001; Bradbury, 2009).

Springs are associated with all of the major aquifers in Wisconsin. They contribute to the region’s vast wetlands, lakes, and world-class trout streams and create habitat for endangered and threatened species (Bradbury and Cobb, 2008; Swanson, 2013). In 2003, the state took steps to protect springs that result “in a current of flowing water with flows of a minimum of one cubic foot per second at least 80 percent of the time” (2003 Wisconsin Act 310, now Wisconsin Statute §381.34), and the Wisconsin Department of Natural Resources (WDNR) was charged with evaluating whether groundwater pumping by new highcapacity wells (wells pumping at least 4.4 L/s, or

100,000 U.S. gal/d) will impact these springs. Initially, a few studies compiled springs data from historical sources or documented springs at the county scale (Fermanich et al., 2006; Grote, 2007; Macholl, 2007; Swanson et al., 2009), but due to the expansion of groundwater-sourced irrigation in the region, concerns over the effects of withdrawals on surface waters and a need for reliable springs data have persisted (Kraft et al., 2012).

METHODS To establish reference springs, a broad understanding of the range of spring resources in a region is necessary. Spring inventory methods developed in the last

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

333

EEGS-26-02-06_1XO

July 8, 2020

8:28

Swanson, Graham, and Hart

Figure 2. Generalized stratigraphic column for Paleozoic rocks comprising the Cambrian and Ordovician sandstone and dolomite aquifer and the Silurian dolomite aquifer in Wisconsin (modified from Clayton and Attig, 1997).

decade provide extensive databases of base-level information. Lower-level surveys provide broad landscape descriptions and information on location. They typically include a few measurements to aid in determining the need and methods for more comprehensive inventories. Higher-level surveys provide detailed biological and physico-chemical information, enabling resource managers to evaluate ecosystem integrity, detect major changes to the spring site over time (if sites are revisited and re-surveyed), and make comparisons among springs (Sada and Pohlman, 2006; Florida Department of Environmental Protection, 2007; USDA Forest Service, 2012a, 2012b; and Stevens et al., 2016). The inventory methodology used in Wisconsin was based on these well-established approaches and resulted in a comprehensive database of spring characteristics, including spring coordinate data, access, environmental conditions on the day of the field survey, site disturbance, geology, geomorphology, spring type, flow rate, water quality (pH, conductivity, and temperature), and vegetation cover (Swanson et al., 2019).

Between 2014 and 2017, 415 springs were identified, mapped, and surveyed. The lower flow threshold for inclusion in the inventory was approximately 7.1 L/s (0.25 ft3 /s) at the time of measurement. Using the full suite of field measurements collected during the statewide inventory, as well as existing information on geology, Swanson et al. (2019) defined six geologically related groups of springs that are thought to be representative of most of the springs in the region.

334

In this study, seven reference springs were selected considering the six previously defined geologic groups and other factors such as the number of springs in each group, the groundwater province (Kammerer et al., 1998), the geographic province (Martin, 1965), and the ecological landscape (Wisconsin Department of Natural Resources, 2015). This ensured that the reference springs are representative of the diversity of physical and biological settings in Wisconsin. Minimally disturbed springs located on public lands and those flowing close to the mean discharge of all inventoried springs (28 L/s or 1 ft3 /s) were also prioritized.

To improve understanding of the reference springs and their geologic groups, geologic cross-sections were prepared for all reference springs using nearby private well construction reports and maps of the un-lithified surficial deposits or bedrock geology. Vertical exaggeration ranged from 5 to 10 times.

Onset TidbiTs (Bourne, MA) recorded spring-water temperature near each reference spring orifice at 1 hour intervals throughout the duration of the project.

Thermographs were qualitatively interpreted in the context of work by Bundshuh (1993), who modeled and described a range of annual variations in springwater temperatures associated with shallow aquifer systems. Bundshuh (1993) assumed a periodic function of recharged groundwater corresponding to May and November recharge events and an average annual recharge of 200 mm, which is similar to conditions in Wisconsin (Swanson and Bahr, 2004). The models also assumed that the depth to the groundwater table is about constant and that groundwater recharge is uniform. Aquifer thickness was assumed to be much less than the horizontal extent, resulting in predominantly horizontal flow (Bundshuh, 1993). The models that simulate aquifer systems of variable thickness are most applicable to the layered bedrock uplands in the Driftless Area of Wisconsin, while the models with constant aquifer thickness are most applicable to the glaciated landscapes in Wisconsin.

Discharge measurements and geochemical sampling took place on a biannual (from 2015 to 2017) and then quarterly basis (from 2018 to 2019). The discharge measurement technique depended on spring channel conditions and flow rate. A 20 cm (8 in.) cutthroat flume was used in narrow and shallow channels with un-lithified bed materials. The velocity-area method was implemented in wider and deeper channels using a wading rod and an electromagnetic meter (0 to 6 m/s ± 2–4 percent of reading) or an acoustic Doppler velocity meter (0 to 4 m/s ± 1 percent of reading).

Water samples were collected at the spring orifice and filtered using a handheld vacuum pump and 0.45 μm filters. Samples were analyzed for major ions and alkalinity at the University of Wisconsin Stevens

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

EEGS-26-02-06_1XO

July 8, 2020

8:28

Using Reference Springs to Describe Flow Conditions

Figure 3. Groups of geologically related springs (after Swanson et al., 2019) and reference spring positions. Specific conductance values were binned using a natural breaks classification.

Point Water and Environmental Analysis Laboratory, and stable isotopes of oxygen (δ18 O) and hydrogen (δ2 H) were analyzed at the Iowa State University Stable Isotope Laboratory. Field measurements included pH, temperature, and electrical conductivity, measured with a HACH H160 meter (Loveland, CO) and an Oakton CON 400 Series conductivity meter (Vernon Hills, IL), respectively. Total alkalinity was also measured in the field using CHEMets test kits (Midland, VA). Summary statistics, analysis of variance (ANOVA), and Tukey-Kramer honest significant difference (HSD) tests were used to discern geochemical patterns among reference spring waters and further describe the initial geological groups of springs.

RESULTS Reference Spring Selection Swanson et al. (2019) found that local variations in topography, surficial geology, and bedrock geology strongly influence the spatial distribution of springs in Wisconsin. Many springs align with glacial landforms; others are highly dependent on stratigraphic position.

Patterns in spring-water conductivity align with those in topographic position and stratigraphy, supporting six groups of geologically related springs (Table 1 and Figure 3).

The seven reference springs chosen and evaluated in this study represent four of the six geologically related groups, as well as a wide range of hydrologic, ge-

ographic, and ecological zones (Table 2 and Figure 3).

Three reference springs were selected from group 1 due to the large number of springs in the group (195).

These springs are located within the Driftless Area or near its margins. They are rheocrene, fracture, or contact springs that emerge along hillslopes or at the base of valleys, where streams have down-cut into Cambrian sandstones. The springs emerging at the base of the valleys sometimes have seepage-filtration morphologies due to overlying, saturated colluvium or alluvium. Springs in this group also emerge near the contact between the Ordovician Prairie du Chien Group and the overlying Ordovician Ancell Group.

Spring-water specific conductance values are moderate relative to other Wisconsin springs and reflect flow through quartz-rich sandstone (Swanson et al., 2019).

Maiden Rock Spring is typical of the stratigraphically lower springs in the group. Big Spring and Kelly Spring both discharge from stratigraphically higher positions.

However, Big Spring is located within the Driftless Area, whereas glacial materials (sandy till and sand and gravel outwash) overlie the Cambrian and Ordovician rocks near Kelly Spring (Kostka et al., 2004).

Lodi Marsh Spring represents springs in group 4, which are located in southern Wisconsin, where they emerge along the subcrop of the Cambrian Tunnel City Group and its upper or lower contact.

Bedding-parallel fractures that promote preferential groundwater flow are truncated by buried valleys, and springs emerge along the margins of the valleys where overlying glacial deposits are thin. These rheocrene

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

335

EEGS-26-02-06_1XO

July 8, 2020

8:28

Swanson, Graham, and Hart Table 1. Characteristics of major classes of springs in Wisconsin (after Swanson et al., 2019).

Spring Group 1

2

3

4

Landscape or Setting

Stratigraphic Unit Present near Spring Orifice

Geologic Material Present at Orifice

Spring Type

Orifice Morphology

Within or near Cambrian System Sandstone Rheocrene margins of Ordovician Prairie du Dolomite, some Rheocrene, the Wisconsin Chien Group; limestone and hillslope Driftless Area Ordovician Ancell shale; Group sandstone Quaternary alluvium Sand and gravel Rheocrene or colluvium

Fracture Fracture, contact

Within or near Ordovician Sinnipee margins of Group the southern Wisconsin Quaternary alluvium Driftless Area or colluvium

Dolomite, some Rheocrene limestone and shale Sand and gravel Rheocrene

Fracture

Ordovician Prairie du Dolomite, some Rheocrene, Chien Group limestone and hillslope shale Quaternary alluvium Sand and gravel Rheocrene or colluvium

Fracture

Central Wisconsin

Southern Wisconsin

Cambrian Tunnel City Sandstone Group Quaternary alluvium Sand, gravel

Number of Springs

Standard Deviation of Mean Specific Mean Specific Conductance Conductance (μmhos/cm, (µmhos/cm, 25°C) 25°C)

195

564

54

35

742

72

11

778

38

15

968

243

Seepage filtration

Seepage filtration

Seepage filtration

Rheocrene

Fracture

Rheocrene, limnocrene

Seepage filtration

5

Niagara Escarpment

Silurian System Quaternary alluvium or colluvium

Dolomite Sand, gravel

Rheocrene Rheocrene

Fracture Seepage filtration

7

716

222

6

Marginal ridges Quaternary alluvium of northern and eastern ice lobes

Sand, gravel

Rheocrene, limnocrene

Seepage filtration

116

511

193

or limnocrene springs form seepage-filtration morphologies with boiling sands and spring pools. Higher specific conductance values reflect long flow paths through the shallow bedrock aquifer, as well as the surrounding urban environment (Swanson et al., 2006a, 2019) (Table 2, Figure 3).

Three Springs represents springs in group 5, which emerge along the Niagara Escarpment where the Silurian dolomite is exposed or shallowly buried.

These rheocrene springs form either fracture or seepage-filtration morphologies, depending on whether the fractured dolomite is exposed at the

Table 2. Reference spring regional properties.

Name

Groundwater Province1

Geographic Region2

Ecological Region3

Spring Group

Maiden Rock Big Spring Kelly Spring Lodi Marsh Sutherland Road Spring Pine River Three Springs

I II I I IV I II

Western Uplands Western Uplands Western Uplands Eastern Ridges and Lowlands Northern Highland Central Plain Eastern Ridges and Lowlands

Western Coulees and Ridges Western Coulees and Ridges Western Prairie Southeast Glacial Plains North Central Forest Central Sand Hills Northern Lake Michigan Coastal

1 1 1 4 6 6 5

1

Kammerer et al. (1998).

Martin (1965).

3 Wisconsin Department of Natural Resources (2015).

2

336

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

EEGS-26-02-06_1XO

July 8, 2020

8:28

Using Reference Springs to Describe Flow Conditions

Figure 4. Geologic cross-sections of reference spring sites.

land surface or shallowly buried. Specific conductance of many springs in group 5 varies widely depending on the frequency and magnitude of precipitation (Swanson et al., 2019) (Table 2 and Figure 3).

Springs in group 6 occur in glaciated regions of northern, central, and southeastern Wisconsin. They discharge near the break in slope along and between end and interlobate moraines or near the margins of former glacial lakebeds. They are rheocrene or limnocrene, seepage-filtration springs with low specific conductance values, suggesting short groundwater residence times and shallow groundwater flow paths

through the sand and gravel aquifer (Swanson et al., 2019). Due to the large number of springs in the group (116) and to adequately represent differences in geography and ecology across the region, two reference springs were selected from group 6, one in northern Wisconsin (Sutherland Road Spring) and one in central Wisconsin (Pine River Spring) (Table 2 and Figure 3).

Groups 2 and 3 are not represented by a reference spring. These groups have lower numbers of springs, so they are less representative of springs across the region. Additionally, only two springs in either group are

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

337

EEGS-26-02-06_1XO

July 8, 2020

8:28

Swanson, Graham, and Hart

Figure 5. Reference spring thermographs (April 2017–June 2019). Spring-water temperature is in black; surface temperature is in gray.

located on public land and have flows of 28 L/s or more. Both of these springs are significantly affected by human activities (e.g., roads, livestock) and are no longer in a natural state (Swanson et al., 2019).

Reference Spring Conditions Geologic cross-sections constructed as part of this study and field observations of exposures at Maiden Rock, Kelly, Big, Lodi Marsh, and Three Springs support the importance of fractures and/or beddingparallel contrasts in permeability in focusing groundwater flow to these springs. Pine River and Sutherland Road Springs form at the break in slope along end moraines or the irregular glacial topography and are dependent on contrasts in permeability in the surficial glacial deposits (Figure 4a–g).

Reference spring thermographs display a range of thermal patterns. The most recent two years of moni-

338

toring are shown in Figure 5; most springs have 4 years of data with similar patterns. Pine River and Three Springs show seasonal variations in spring-water temperature that are damped and out of phase with surface temperatures. The seasonal variations in springwater temperature at Kelly and Sutherland Road Springs are heavily damped and out of phase with surface temperatures. Lodi Marsh, Big, and Maiden Rock Springs approximate temperature stability throughout the duration of monitoring.

Mean reference spring discharge ranges from 30 L/s to 105 L/s (Figure 6). Only four measurements are available for Three Springs because the measurement position was moved in 2018 to improve accuracy. However, the measurements that are available for Three Springs were collected in different seasons, so they still provide a limited measure of flow consistency. With the exception of Big Spring, flow was fairly consistent at the reference springs in different seasons throughout the monitoring period, which supports a lack of

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

EEGS-26-02-06_1XO

July 8, 2020

8:28

23.14 24.29 15.90 19.46 8.24 24.40 28.75 605.7 629.5 311.9 636.2 211.3 410.7 605.1 9.16 11.45 6.51 6.08 5.19 10.27 13.77 287.3 280.0 125.8 283.3 106.3 179.8 291.6 0.32 1.04 0.60 0.56 0.17 0.86 1.71 9.9 14.7 13.1 17.6 4.1 9.9 11.4 0.72 1.16 0.40 0.61 0.42 0.80 1.25 8.0 13.3 6.2 10.7 0.2 3.6 12.7 0.17 0.54 0.16 0.39 0.10 0.62 0.72 3.9 6.7 2.8 6.5 0.1 4.2 0.7 0.09 0.57 0.08 0.19 0.08 0.22 0.70 2.9 5.2 2.7 3.7 2.2 1.7 5.8 0.03 0.56 0.08 0.06 0.01 0.04 0.06 0.8 1.9 1.3 0.8 0.9 0.8 1.0 0.58 1.03 0.62 1.20 0.20 1.23 1.48 33.3 34.3 14.9 37.0 7.5 22.1 32.4 2.12 1.69 1.27 1.44 0.90 2.61 4.57

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

SD = standard deviation.

6 12 12 12 7 12 12 Maiden Rock Big Spring Kelly Spring Lodi Marsh Sutherland Road Pine River Three Springs

68.1 70.0 33.3 69.3 27.3 43.9 67.6

SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean N

Cl− (mg/L) NO3 − (mg/L) Na+ (mg/L) K+ (mg/L) Mg2+ (mg/L) Ca2+ (mg/L)

extensive karst development in the vicinity of these springs.

The stable isotope results for the seven reference springs follow the general direction of storm systems across the region, from west and south to east and north (Figure 7). Heavier (less negative) values are expected in the south or west (e.g., Big Spring), and values should become lighter (more negative) to the north and east (e.g., Sutherland Road, Three Springs). The quarterly samples collected in 2018 to 2019 were useful in discerning slight seasonal variations of 0.2 to 0.4 per mil in δ18 O and 0.6 to 2.7 per mil in δ2 H. The greatest seasonal variation occurred at Pine River, and the least occurred at Lodi Marsh.

Due to the high carbonate content of both the unlithified glacial deposits and the sedimentary bedrock units in the region, all reference spring waters are a calcium–magnesium bicarbonate type. The ion concentrations and specific conductance values in Table 3 and the cation and anion relative molar percentages in Table 4 show that the greatest differences among the spring waters are in the relative dilution of the waters, rather than in relative percentages of ions in solution.

Using specific conductance as a measure of dilution, Table 5 shows the results of ANOVA and Tukey-Kramer HSD tests (α = 0.05). Lodi Marsh spring-water specific conductance is not significantly different from that of Big Spring or Maiden Rock, but it is different from all other spring waters. Big Spring and Maiden Rock spring-water specific conductance is not significantly different from one another or that of Lodi Marsh or Three Springs. Three Springs spring-water specific conductance is not significantly different from that of Big Spring or Maiden Rock, but it is different from all other spring waters. Pine River,

Table 3. Major ion concentrations for reference springs.

Figure 6. Mean reference spring discharge. Error bars represent the standard deviation of the mean. N = number of flow measurements.

Reference Spring

SO4 3− (mg/L)

Alkalinity (mg CaCO3 /L)

Specific Conductance (µmhos/cm, 25°C)

Using Reference Springs to Describe Flow Conditions

339

EEGS-26-02-06_1XO

July 8, 2020

8:28

Swanson, Graham, and Hart

Figure 7. Stable isotope values for reference springs. Local meteoric water line (LMWL) for Madison, WI, is from Swanson et al. (2006b), and global meteoric water line (GMWL) is from Craig (1961).

Kelly Spring, and Sutherland Road spring-water specific conductance is each significantly different from all other spring waters.

Concentrations of anthropogenically sourced ions such as nitrate and chloride are typical of shallow groundwater in the region (Kammerer, 1995). Although seasonal patterns are unclear, their magnitude and temporal variation differ somewhat among the reference springs (Figure 8) and also prove to be useful in distinguishing groundwater flow paths and residence times, as discussed in the following section.

DISCUSSION The regional inventory approach was successful in yielding six groups of springs with similar geological controls on their spatial distribution. Seven reference springs selected from four of the groups represent the vast majority of springs in the region and further in-

form conditions within these groups. Temporal conditions within each group are discussed below.

Group 1 Maiden Rock, Big, and Kelly Springs represent group 1. The geologic cross-sections show that all three springs are associated with the Cambrian and Ordovician sandstones and dolomites, but they also illustrate the range of conditions within the group. In all three cases, it is likely that recharging groundwater flows downward through the stacked sequence of Paleozoic rocks before flowing horizontally—aided by beddingparallel fractures or high-permeability zones—to the springs. Maiden Rock Spring emerges near the contact of the Tunnel City Group sandstones and the overlying St. Lawrence Formation siltstone and dolomite.

Big and Kelly Springs emerge from the stratigraphically higher and heavily fractured Prairie du Chien dolomite. Ridge tops near Maiden Rock and Big

Table 4. Relative molar percentages of dissolved cations and anions.

Cations (%)

Anions (%)

Reference Spring

Ca2+

Mg2+

Na+

K+

SO4 3−

Cl−

Alkalinity

NO3 −

Maiden Rock Big Spring Kelly Spring Lodi Marsh Sutherland Road Pine River Three Springs

52.8 50.9 52.0 50.2 61.2 52.1 51.1

42.7 41.2 38.5 44.5 27.9 43.3 40.4

3.9 6.5 7.5 4.6 8.8 3.5 7.7

0.6 1.4 2.1 0.6 2.2 1.0 0.8

3.2 4.1 8.7 5.5 3.8 5.2 3.5

7.0 9.9 11.2 9.1 0.6 5.1 10.6

89.7 73.4 80.1 85.4 95.1 89.7 85.9

8.7 12.6 12.9 13.9 0.4 14.8 1.5

340

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

EEGS-26-02-06_1XO

July 8, 2020

8:28

Using Reference Springs to Describe Flow Conditions Table 5. Results of specific conductance ANOVA and Tukey-Kramer HSD tests.

Reference Spring

Mean Specific Conductance (µmhos/cm, 25°C)

Lodi Marsh Big Spring Maiden Rock Three Springs Pine River Kelly Spring Sutherland Road

636.2 629.5 605.7 605.1 410.7 311.9 211.3

1

Connecting Letters Report1 A A A

B B B C D E

Levels not connected by same letter are significantly different.

Springs have a thin cover of loess, whereas those near Kelly Spring are capped by sandy till or sand and gravel outwash (Figure 4a–c). Groundwater may circulate to depths of 100 m or more before emerging at Maiden Rock Spring. At Big Spring, the depth of circulation is less, but it may be as high as 50 m. In both cases, the depth to the water table along ridge tops where recharge is likely to occur is unknown due to a lack of wells and may be complex due to the existence of perched aquifers in the region (Carter et al., 2010).

At Kelly Spring, the water table is shallow (around 5 m), and the depth of circulation through the glacial deposits and the Prairie du Chien dolomite is less than 20 m.

Thermographs, flow records, and spring-water chemistry conditions for the group 1 reference springs

complement the improved conceptual models of groundwater flow described above (Figure 5). Bundshuh (1993) showed that in an aquifer system of variable thickness, variations in spring-water temperatures are unlikely where the water table is deep or where the water table is shallow but the circulation depth is greater than 20 m. The temperature measurements at Maiden Rock and Big Springs are very stable, supporting deep-water-table conditions or deep circulation of groundwater. At Kelly Spring, spring-water temperature variations are damped and lag about 4 months behind changes in surface temperature. This pattern supports a conceptual model that includes a shallow water table, high-permeability surficial deposits, and a shallow circulation depth (Bundshuh, 1993). The range in temperature conditions found for group 1 springs is important because springs in this region of Wisconsin are recognized as being especially important and effective at providing thermal refuge for trout and other sensitive aquatic organisms (Gaffield et al., 2005).

Of the reference springs in group 1, Big Spring shows the greatest variability in flow, as well as in nitrate and chloride concentrations (Figures 6 and 8).

These conditions suggest that fractures may be providing the primary pathways for groundwater flow to this spring. Spring-water specific conductance values for Maiden Rock and Big Springs cannot be distinguished from one another. However, Kelly Spring specific conductance is significantly lower, suggesting that although water discharges from the fractured Prairie du Chien dolomite, it is unlikely that residence times within the dolomite are long.

Group 4

Figure 8. Mean nitrate (gray) and chloride (black) concentrations at reference springs. Error bars represent the standard deviation of the mean.

Lodi Marsh Spring represents group 4. Recharging groundwater flows downward at least 50 m through the shallow sand and gravel aquifer, the Prairie du Chien Group dolomite, the Jordan Formation sandstone, and the St. Lawrence Formation siltstone and dolomite, before flowing horizontally through the Tunnel City Group sandstones. The spring emerges near the subcrop of the Tunnel City Group where the overlying glacial deposits are thin (Figure 4d). The temperature measurements at Lodi Marsh Spring are very stable (Figure 5), suggesting that circulation of groundwater exceeds the depth to which the annual temperature variations will propagate (Bundshuh, 1993). Spring flow is also relatively stable (Figure 6), suggesting a long, horizontal groundwater flow path (Swanson and Bahr, 2004). Nitrate and chloride concentrations are elevated relative to other reference springs, but consistent over the monitoring period (Figure 8), which is characteristic of other springs in the region that discharge from the Tunnel City Group

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

341

EEGS-26-02-06_1XO

July 8, 2020

8:28

Swanson, Graham, and Hart

(Swanson et al., 2001). Specific conductance values for Lodi Marsh Spring cannot be distinguished from those of Maiden Rock and Big Springs (Table 5), which also emerge from the Cambrian and Ordovician sandstones and dolomites.

Group 5 Three Springs represents group 5 and emerges near the contact between the shallow Silurian dolomite aquifer and the overlying, discontinuous sandy till and sandy beach and nearshore sediment (Carson et al., 2016) (Figure 4e). Spring-water temperature variations are damped and lag about 2 months behind changes in surface temperature (Figure 5). This pattern supports a conceptual model that includes a shallow water table, high-permeability surficial deposits, and a shallow groundwater circulation depth (Bundshuh, 1993).

Spring flow appears to be relatively stable, but because only four measurements are available, they may not be representative of the range of conditions at the spring (Figure 6). Nitrate and chloride concentrations are highly variable, which is expected for a spring emerging from the heavily fractured Silurian dolomite aquifer (Figure 8). Specific conductance values for Three Springs cannot be distinguished from those of Maiden Rock and Big Springs (Table 5), where water also encounters dolomite along flow paths to springs.

Group 6 Pine River and Sutherland Road Springs represent group 6. Both springs emerge at the base of and near the break in slope within uneven glacial terrain (Figure 4f and g). Spring-water temperature variations are damped to a greater extent at Sutherland Road Spring, but at both springs, the lag behind changes in surface temperature is 1 to 2 months (Figure 5).

The difference in patterns may be explained by the existence of lower-permeability, near-surface materials at the Sutherland Road Springs. Bundshuh (1993) showed that a thin, low-permeability layer above an aquifer causes a strong reduction in amplitude of spring-water temperature relative to surface temperature, without a substantial phase difference. Surficial materials in the vicinity of Sutherland Road Spring are composed of clayey and silty till, whereas those near Pine River Spring are composed of sand and gravel outwash (Clayton, 1984). Both springs exhibit steady flow and have dilute waters (Figures 6 and 8), reflecting shallow groundwater circulation through a granular aquifer. The greatest seasonal variation in δ18 O and δ2 H also occurred at Pine River, which complements 342

the conceptual model of shallow groundwater flow to this spring.

CONCLUSION Evaluation and monitoring of reference springs resulted in improved conceptual models of groundwater flow to springs and an expanded understanding of temporal variations in flow, temperature, and chemistry that may be expected for springs in each geologic group. In Wisconsin, where most springs emerge from either un-lithified glacial materials or relatively shallow siliciclastic or carbonate bedrock aquifers, springwater specific conductance proved to be especially useful in refining groundwater flow paths to springs. Additionally, due to wide variations in air temperature and seasonal recharge, spring-water thermographs proved to be informative about the depth of groundwater circulation.

This research suggests that the reference spring approach may be useful in designing long-term monitoring programs for springs and tailoring them to the most pressing management questions in a particular region. For example, where water managers are concerned about the effects of climate change on springs, efforts to continuously monitor spring discharge may be included. Monitoring of springs in Wisconsin transferred to the Wisconsin Department of Natural Resources in fall 2019. While outcomes of future monitoring will reveal whether the reference springs can serve as beacons for change, the approach has already improved our understanding of the dominant controls on spring occurrence in the region, and this informs our understanding of spring vulnerability to significant land-use changes, diversions of flow, or changes in recharge.

Unlike the use of the “reference condition” term in the ecological literature, the reference springs established in this study are not necessarily restricted to those with a complete absence of human disturbance or alteration (Stoddard et al., 2006), because very few large springs in the region have escaped human influence. Even so, potential also exists for ecological monitoring of reference springs in the future.

ACKNOWLEDGMENTS This work was funded by the Wisconsin Department of Natural Resources. We thank the many local land managers, fishery and wildlife biologists, foresters, county extension agents, and others who contributed background information for the project.

Sincere thanks go to Rick Blonn, Tyler Burgett, Emma Hall, Emma Koeppel, Dexter Kopas, Ava Krahn,

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

EEGS-26-02-06_1XO

July 8, 2020

8:28

Using Reference Springs to Describe Flow Conditions

Monica Norton, and Christine Shonnard, who provided valuable assistance in the field.

REFERENCES Alfaro, C. and Wallace, M., 1994, Origin and classification of springs and historical review with current applications: Environmental Geology, Vol. 24, No. 2, pp. 112–124.

Bradbury, K. R., 2009, Karst and Shallow Carbonate Bedrock in Wisconsin: Wisconsin Geological and Natural History Survey Factsheet, Madison WI, 2 p.

Bradbury, K. R. and Cobb, M. K., 2008, Delineation of Areas Contributing Groundwater to Springs and Wetlands Supporting the Hine’s Emerald Dragonfly, Door County, Wisconsin: Wisconsin Geological and Natural History Survey Open-File Report 2008-04, 34 p.

Bryan, K., 1919, Classification of springs: The Journal of Geology, Vol. 27, No. 7, pp. 522–561.

Bundshuh, J., 1993, Modeling annual variations of spring and groundwater temperatures associated with shallow aquifer systems: Journal of Hydrology, Vol. 142, pp. 427–444.

Carson, E. C.; Brown, S. R.; Mickelson, D. M.; and Schneider, A. F., 2016, Pleistocene Geology of Door County, Wisconsin: Wisconsin Geological and Natural History Survey Bulletin 109, 44 p.

Carter, J. T. V.; Gotkowitz, M. B.; and Anderson, M. P., 2010, Field verification of stable perched groundwater in layered bedrock uplands: Ground Water, Vol. 49, No. 3, pp. 383–392.

Clayton, L., 1984, Pleistocene Geology of the Superior Region, Wisconsin: Wisconsin Geological and Natural History Survey Information Circular 46, 40 p.

Clayton, L. and Attig, J. W., 1997, Pleistocene Geology of Dane County: Wisconsin: Wisconsin Geological and Natural History Survey Bulletin 95, 64 p.

Craig, H., 1961, Isotopic variations in meteoric waters: Science, Vol. 133, pp. 1702–1703.

Deocampo, D. M., 2004, Hydrogeochemistry in the Ngorongoro Crater, Tanzania, and implications for land use in a World Heritage Site: Applied Geochemistry, Vol. 19, pp. 755–767.

Dott, R. H. and Attig, J. W., 2004, Roadside Geology of Wisconsin: Mountain Press, Missoula, MT, 344 p.

Fermanich, K.; Zorn, M.; Stieglitz, R.; and Waltman, C., 2006, Mapping and Characterization of Springs in Brown and Calumet Counties: University of Wisconsin, Green Bay, WI, Final Report to the Wisconsin Department of Natural Resources, 56 p. plus appendices.

Fetter, C. W., 2001, Applied Hydrogeology, 4th ed.: Prentice Hall, Upper Saddle River, NJ, 598 p.

Florea, L. J. and Vacher, H. I., 2006, Springflow hydrographs: Eogenetic vs. telogenetic karst: Ground Water, Vol. 44, No. 3, pp. 352–361.

Florida Department of Environmental Protection, 2007. Florida Springs Initiative, Program Summary and Recommendations, 2007: Florida Department of Environmental Protection, Tallahassee, FL, 43 p.

Foos, A., 2003, Spatial distribution of road salt contamination of natural springs and seeps, Cuyahoga Falls, Ohio, USA: Environmental Geology, Vol. 44, pp. 14–19.

Gaffield, S. J.; Potter, K. W.; and Wang, L., 2005, Predicting the summer temperature of small streams in southwestern Wisconsin: Journal of American Water Resources Association, February, Vol. 41, No. 1, pp. 25–36.

Grote, K., 2007, Identification and Characterization of Springs in West-Central Wisconsin: University of Wisconsin, Eau Claire,

WI, Final Report to the Wisconsin Department of Natural Resources, 33 p. plus appendices.

Hu, L.; Chen, C.; Jiao, J.; and Wang, Z., 2007, Simulated groundwater interaction with rivers and springs in the Heihe River basin: Hydrological Processes, Vol. 21, pp. 2794–2806.

Hughes, R. M., 1995, Defining acceptable biological status by comparing with reference conditions. In Davis, W. S. and Simons, T. P. (Editors), Biological Assessment and Criteria: Tools for Water Resource Planning and Decision Making: Lewis Publishers, Boca Raton, FL, pp. 31–47.

Hughes, R. M.; Larsen, D. P.; and Omernik, J. M., 1986, Regional reference sites: A method for assessing stream potentials: Environmental Management, Vol. 10, No. 5, pp. 629–635.

Kammerer, P. A., 1995, Ground-Water Flow and Quality in Wisconsin’s Shallow Aquifer System: U.S. Geological Survey WaterResources Investigations Report 90-4171, 42 p.

Kammerer, P. A.; Trotta, L. C.; Krabbenhoft, D. P.; and Lidwin, R. A., 1998, Geology, Ground-Water Flow, and DissolvedSolids Concentrations in Ground Water along Hydrogeologic Sections through Wisconsin Aquifers: U.S. Geological Survey Hydrologic Atlas 731, 4 plates.

Karimi, H., 2012, Hydrogeology of karstic area. In Kazemi, G. A. (Editor), Hydrogeology—A Global Perspective: InTech, London, U.K., 222 p.

Kostka, S. J.; Hinke, H. J.; Mickelson, D. M.; and Baker, R. W., 2004, Preliminary Quaternary Geologic Map of St. Croix County, Wisconsin: Wisconsin Geological and Natural History Survey Open-File Report 2004–22.

Kraft, G. J.; Clancy, K.; Mechenich, D. J.; and Haucke, J., 2012, Irrigation effects in the northern lake states: Wisconsin Central Sands revisited: Ground Water, Vol. 50, No. 2, pp. 308–318.

Leake, S. A.; Pool, D. R.; and Leenhouts, J. M., 2008, Simulated Effects of Ground-Water Withdrawals and Artificial Recharge on Discharge to Streams, Springs, and Riparian Vegetation in the Sierra Vista Subwatershed of the Upper San Pedro Basin, Southeastern Arizona: U.S. Geological Survey Scientific Investigations Report 2008-5207, 15 p.

Luhmann, A. J.; Covington, M. D.; Peters, A. J.; Alexander, S. C.; Anger, C. T.; Green, J. A.; Runkel, A. C.; and Alexander, E. C., 2011, Classification of thermal patterns at karst springs and cave streams: Ground Water, Vol. 49, No. 3, pp.

324–335.

Macholl, J. A., 2007, Inventory of Wisconsin’s Springs: Wisconsin Geological and Natural History Survey Open-File Report 2007-03, 20 p. plus appendices.

Manga, M., 1997, A model for discharge in spring-dominated streams and implications for the transmissivity and recharge of Quaternary volcanics in the Oregon Cascades: Water Resources Research, Vol. 33, No. 8, pp. 1813–1822.

Manga, M., 2001, Using springs to study groundwater flow and active geologic processes: Annual Review of Earth and Planetary Sciences, Vol. 29, pp. 201–228.

Martin, L., 1965, The Physical Geography of Wisconsin: University of Wisconsin Press, Madison, WI, 608 p.

Mudrey, M. G., Jr.; Brown, B. A.; and Greenberg, J. K., 2007, Bedrock Geologic Map of Wisconsin: Wisconsin Geological and Natural History Survey State Map 18-DI, v. 1.0, 1 CDROM. Muldoon, M. A.; Simo, J. A.; and Bradbury, K. R., 2001, Correlation of hydraulic conductivity with stratigraphy in a fractured-dolomite aquifer, northeastern Wisconsin, USA: Hydrogeology Journal, Vol. 9, No 6, pp. 570–583.

Parsen, M. J.; Bradbury, K. R.; Hunt, R. J.; and Feinstein, D. T., 2016, The 2016 Groundwater Flow Model for Dane County,

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

343

EEGS-26-02-06_1XO

July 8, 2020

8:28

Swanson, Graham, and Hart Wisconsin: Wisconsin Geological and Natural History Survey Bulletin 110, 56 p.

Rayne, T. W.; Bradbury, K. R.; and Muldoon, M. A., 2001, Delineation of capture zones for municipal wells in fractured dolomite, Sturgeon Bay, Wisconsin, USA: Hydrogeology Journal, Vol. 9, No. 5, pp. 432–450.

Reynoldson, T. B.; Norris, R. H.; Resh, V. H.; Day, K. E.; and Rosenberg, D. M., 1997, The reference condition: A comparison of multimetric and multivariate approaches to assess water-quality impairment using benthic macroinvertebrates: Journal of the North American Benthological Society, Vol. 16, No. 4, pp. 833–852.

Sada, D. W. and Pohlman, K. F., 2006, U.S. National Park Service Mojave Inventory and Monitoring Network Spring Survey Protocols: Level I and Level II: Desert Research Institute, Inc., Reno, NV, 95 p.

Smart, C. and Worthington, S. R. H., 2004, Springs. In Gunn, J. (Editor), Encyclopedia of Caves and Karst Science: Fitzroy Dearborn, New York, pp. 1495–1505.

Springer, A. E. and Stevens, L. E., 2009, Spheres of discharge of springs: Hydrogeology Journal, Vol. 17, No. 1, pp. 83–93.

Stevens, L. E.; Springer, A. E.; and Ledbetter, J. D., 2016, Springs Ecosystem Inventory Protocols: Springs Stewardship Institute, Museum of Northern Arizona, Flagstaff, AZ, 60 p.

Stoddard, J. L.; Larsen, D. P.; Hawkins, C. P.; Johnson, R. K.; and Norris, R. H., 2006, Setting expectations for the ecological condition of streams: The concept of reference condition: Ecological Applications, Vol. 16, No. 4, pp. 1267–1276.

Swanson, S. K., 2013, Wisconsin’s spring resources: An overview: Geoscience Wisconsin, Vol. 21, 13 p.

Swanson, S. K. and Bahr, J. M., 2004, Analytical and numerical models to explain steady rates of spring flow: Groundwater, Vol. 42, No. 5, pp. 747–759.

Swanson, S. K.; Bahr, J. M.; Bradbury, K. R.; and Anderson, K. M., 2006a, Evidence for preferential flow through sandstone aquifers in southern Wisconsin: Sedimentary Geology, Vol. 184, pp. 331–342.

Swanson, S. K.; Bahr, J. M.; and Potter, K. W., 2006b, A Local Meteoric Water Line for Madison, Wisconsin: Wisconsin Geological and Natural History Survey Open-File Report 2006-01, 4 p.

Swanson, S. K.; Bahr, J. M.; Schwar, M. T.; and Potter, K. W., 2001, Two-way cluster analysis of geochemical data to

344

constrain spring source waters: Chemical Geology, Vol. 179, No. 1–4, pp. 73–91.

Swanson, S. K., 2007, Lithostratigraphic controls on bedding plane fractures and the potential for discrete groundwater flow through a siliciclastic sandstone aquifer, southern Wisconsin, Sedimentary Geology, Vol. 197, pp. 65–78.

Swanson, S. K.; Bradbury, K. R.; and Hart, D. J., 2009, Assessing the vulnerability of spring systems to groundwater withdrawals in southern Wisconsin: Geoscience Wisconsin, Vol. 20, No. 1, pp. 1–13.

Swanson, S. K.; Graham, G. E.; and Hart, D. J., 2019, An Inventory of Springs in Wisconsin: Wisconsin Geological and Natural History Survey Bulletin 113, 24 p.

Tobin, B. W. and Schwartz, B. F., 2016, Using periodic hydrologic and geochemical sampling with limited continuous monitoring to characterize remote karst aquifers in the Kaweah River Basin, California, USA: Hydrological Processes, Vol. 30, pp. 3361–3372.

U.S. Department of Agriculture (USDA) Forest Service, 2012a, Groundwater-Dependent Ecosystems: Level I Inventory Field Guide: Inventory Methods for Planning and Assessment: U.S. Department of Agriculture Forest Service General Technical Report WO-86a, 191 p.

U.S. Department of Agriculture (USDA) Forest Service, 2012b, Groundwater-Dependent Ecosystems: Level II Inventory Field Guide: Inventory Methods for Project Design and Analysis: U.S. Department of Agriculture Forest Service General Technical Report WO-86b, 124 p.

Valerity, Y.; Yuliya, V.; Dmytro, D.; and Yuriy, V., 2015, Nitrates in springs and rivers of East Ukraine: Distribution, contamination and fluxes: Applied Geochemistry, Vol. 53, pp. 71–78.

Werner, E. and di Pretoro, R. S., 2006, Rise and fall of road salt contamination of water-supply springs: Environmental Geology, Vol. 51, pp. 537–543.

Wisconsin Department of Natural Resources, 2015, Ecological Landscapes of Wisconsin: An Assessment of Ecological Resources and a Guide to Planning Sustainable Management: Wisconsin Department of Natural Resources PUB-SS-1131, 2403 p.

Wisconsin Geological and Natural History Survey, 1989, Groundwater Contamination Susceptibility in Wisconsin: Wisconsin Geological and Natural History Survey Page-Size Map 19, scale 1:2,730,000, 2 p.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 331–344

EEGS-26-03-01_1XO

July 8, 2020

8:14

Common Spring Types in the Valley and Ridge Province: There Is More than Karst DOROTHY J. VESPER* Department of Geology and Geography, West Virginia University, 330 Brooks Hall, Morgantown, WV 26506

ELLEN K. HERMAN Department of Geology & Environmental Geosciences, Bucknell University, 231 O’Leary Science Center, Lewisburg, PA 17837

Key Terms: Environmental Geology, Hydrogeology, Springs, Karst, Appalachians ABSTRACT The Valley and Ridge Province (V&R) of the central Appalachians is rich in springs that support ecosystems, provide local water resources, and export water from the region. Although there has been extensive research on springs in the province, the focus has been on chemically variable karst springs. The purpose of this work is to identify common spring types found in the V&R based on an analysis of three regions. Three types of V&R springs are included in this comparison, and their relationship to more general classification systems is included. Headwater springs, located near ridge tops and along ridge flanks, are typically small, may be ephemeral, have localized flow paths, and are associated with siliciclastic units. Karst springs, generally located in the valleys, include both the more chemically variable limestone springs and the more stable dolomite springs.

Thermal warm springs, with temperatures higher than the mean annual air temperature, are less common than the other spring types; they may be large and are typically associated with major thrust faults. The temperature, chemistry, and locations of the springs are controlled by the structural geology and topography as well as the formations and lithologies through which the recharge water travels. There is overlap in the water chemistry and storm responses of the spring groups, but some general trends can be identified, such as lower pH in the headwater springs. The V&R springs are critical resources, but their sustainability, chemistry, and hydrology need to be considered within the local geologic framework.

*Corresponding author email: djvesper@mail.wvu.edu

INTRODUCTION The Appalachians are rich in springs that support ecosystems, provide local water resources, and export water from the region. Although there are many classification systems for springs, there is no widely accepted one that is used across disciplines (LaMoreaux and Tanner, 2001; Kresic and Stevanovic, 2009; Springer and Stevens, 2009; and Glazier, 2014). In a reference/teaching module about springs, Glazier (2014) cited more than 50 publications related to classification systems for springs. Those publications organized spring types by location and geomorphological setting; rock type and structure; water source; “age” of water; direction, persistence, and variability of flow; water chemistry; temperature; relation to flora and fauna; habitat type; biographical isolation; ecology; or use. Springer and Stevens (2009) proposed a spring classification based on “spheres of discharge” that is based on the combination of geomorphology, water flow, water quality, and habitat. This classification system is useful for comparison but does not always fully reflect the common spring types of the Appalachians.

In the central Appalachian Valley and Ridge Physiographic Province (V&R), springs are critical public and private water supplies as well as important agricultural and industrial resources. The eastern continental surface water divide falls within the province, and the waters sourced from the V&R are exported to both the Atlantic Ocean and the Gulf of Mexico. Receiving rivers include the Hudson, Delaware, Susquehanna, Potomac, James, Tar, Cape Fear, and Ohio rivers; some of the downstream cities that make use of these waters include Philadelphia and Pittsburgh, PA, Baltimore, MD, Washington, DC, and Raleigh, NC (Trapp and Horn, 1997).

Hydrogeology research during the 1960s and 1970s established the framework for using physical and chemical data to understand hydrogeologic settings and conditions in springs in the V&R. These studies traced origins of water sources via ion ratios, CO2

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345–358

345

EEGS-26-03-01_1XO

July 8, 2020

8:14

Vesper and Herman

concentrations, and calcite saturation states (Langmuir, 1971; Johnson, 1997); interpreted the nature of karstic flow from the variability in water temperatures and chemistry (Shuster and White, 1971); and determined the association between aquifer yield and fracture traces (Lattman and Parizek, 1964). Karst springs have received much of the attention in the V&R, particularly karst springs that are responsive to storms.

Shuster and White (1971) published one of the first studies undertaken in the United States to systematically categorize karst springs based on the water chemistry. This work, which distributed springs along a continuum between those dominated by conduit flow and those dominated by diffuse or matrix flow, remains among the most highly cited papers on springs in general, and likely the most highly cited work on springs in the V&R (cited 233 times since 1996, SCOPUS 2019).

This work set the tone for spring investigations in the United States; however, it focused on karst springs and excluded the nearby siliciclastic-sourced headwater and warm springs.

Broader surveys of springs include ecological studies and water-quality reports. A series of studies on invertebrate ecosystems in V&R springs included the full breadth of spring types. The characteristic temperatures and chemistries found in different spring types result in different ecosystems (Glazier and Gooch, 1987; Gooch and Glazier, 1991; and Glazier, 1998). The separation of the ecosystems by flow paths results in an excellent opportunity to observe the importance of water chemistry on community evolution (Glazier, 1998).

A U.S. Geological Survey (USGS) ground-water quality report (Johnson et al., 2011), including springs and wells, distinguished between the karst and siliciclastic aquifers in the region, acknowledging that both types of springs are present in the V&R. To address the general lack of attention to noncarbonate springs, we identified general trends in central V&R springs of several types to identify commonalities in terms of source and flow; this assessment was based on data from central Pennsylvania, eastern West Virginia, and southern West Virginia. The goal of this work was not to provide a comprehensive spring classification system but to provide a framework for a generalized understanding of spring types in the V&R and to acknowledge the importance of recognizing these different types of springs.

OVERVIEW OF THE VALLEY AND RIDGE PHYSIOGRAPHIC PROVINCE AND ITS SPRING HYDROLOGY The V&R covers less than 3 percent of the land area in the contiguous United States, but it extends from New York to Alabama (Figure 1). The province is

346

Figure 1. Location of the Appalachian Valley and Ridge (V&R) Province with the three example regions indicated.

characterized by an alternating series of parallel ridges and valleys that trend northeast/southwest (Figures 1 and 2). The area of focus is bounded by the Blue Ridge and Piedmont to the east and by the Appalachian Plateau to the west. The V&R has a maximum width of approximately 120 km in central Pennsylvania (Trapp and Horn, 1997). The Appalachian Great Valley, near the eastern boundary of the V&R, is one of the most distinct features in the province and is typically 15–30 km wide.

The rocks in the geologically complex V&R vary in lithology, unit thickness, structure, fracturing, weathering, and geomorphology. The regions considered in this work are from central Pennsylvania (Figure 3), eastern West Virginia (Figure 4), and southern West Virginia (Figure 5). The strata in these regions range from Cambrian to Pennsylvanian in age. Regional fold structures are found in the V&R, including the Nittany Anticlinorium underlying central Pennsylvania (Butts and Moore, 1936; Hollyday and Hileman, 1996) and the Massanutten Synclinorium in West Virginia and Virginia (Kulander and Dean, 1986; Shultz et al., 1995; and Hollyday and Hileman, 1996). Major thrust faults, like the Birmingham Thrust in central Pennsylvania and the St. Clair Thrust Fault in southeastern West Virginia, shift older rocks on top of younger ones. The orogenic history has resulted in a complex geometry, where bedding planes can be horizontal, vertical, overturned, or anywhere in between.

Subsequent differential erosion of the Appalachian Mountains left behind ridges capped by younger, resistant sandstones (Figure 2) and valleys of older, more easily weathered carbonate units. Although mostly underlain by carbonate units, parts of the Great

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345–358

EEGS-26-03-01_1XO

July 8, 2020

8:14

Spring Types in Valley and Ridge Province

Figure 2. Views of the (A) Tuscarora Formation at the top of Peters Mountain in southern West Virginia (photo D. Vesper) and (B) the valleys and ridges in eastern West Virginia. The town at the base of the closest ridge is Bath, WV, and the view is looking west (photo courtesy of J. Donovan).

Figure 3. Maps of the southern Pennsylvania region: (A) location of the region with Center, MiďŹ&#x201E;in, Huntingdon, and Blair counties in black and the Valley and Ridge in gray; (B) shaded relief map; and (C) generalized geology map.

Figure 4. Maps of the eastern West Virginia region: (A) location of the region with Berkeley and Morgan counties, WV, in black and the Valley and Ridge in gray; (B) shaded relief map; and (4C) generalized geology map.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345â&#x20AC;&#x201C;358

347

EEGS-26-03-01_1XO

July 8, 2020

8:14

Vesper and Herman

Figure 5. Maps of the southern West Virginia region: (A) location of the region with Monroe County, WV, in black and the Valley and Ridge in gray; (B) shaded relief map; and (C) generalized geology map.

Valley in West Virginia and Maryland are underlain by shales (Trapp and Horn, 1997). Overall, the geologic structures control the spatial distribution of formations, which, in turn, inform the locations and types of ground-water flow (Trapp and Horn, 1997). In general, the valleys underlain by soluble carbonate rocks host karst springs (Table 1); these systems are among the most productive aquifers in the V&R and are often used as public water supplies. Sandstone, shale, and other siliciclastic rock units commonly outcrop on the ridge flanks and in upper valleys between the ridges.

Johnson et al. (2011) reported that shale was the most common rock type in the V&R, followed by dolomite, sandstone, and then limestone; the siliciclastic rock types underlie 65 percent of the V&R footprint as de-

fined in that report. Ground water and spring flow in the siliciclastic units tend to be controlled by fractures and, in some areas, primary porosity. In general, flow in the clastic units is more localized and structurally controlled than flow in the carbonate units (Trapp and Horn, 1997). While residence times for ground water in the siliciclastic and carbonate aquifers have similar medians and ranges, a greater percentage of the siliciclastic aquifers have a 25 year or longer residence time (Johnson et al., 2011).

Although there has been little research focused on the non-karst springs in the V&R, there have been some studies on thermal springs due to their geothermal potential (Hobba et al., 1979; Perry et al., 1979) or their historic value (Cohen, 1981). Thermal springs

Table 1. Comparison of spring types in the Valley and Ridge Physiographic Province (V&R).

Spring Groupa and Terminology (A) Headwater springs

(B) Karst springs (carbonate springs)

(C) Thermal warm springs

a b

Common Locations and Geology Flanks and tops of ridges; sandstone, shale, mixed-sedimentary units but primarily siliciclastic In valleys; limestone and dolomite carbonate units

In valleys; mixed lithologies, likely associated with faulting

General Description of Setting Mostly small springs with localized basins and relatively short flow paths; ground-water flow may be confined to units of higher permeability; these springs form the headwaters for the streams flowing off the ridges Typically larger and less common than headwater springs; recharge is both dispersed (infiltration) and concentrated (via sinkholes and/or sinking streams); streams flowing from the ridge flanks can recharge these springs via sinkholes Thermal warm springs are defined as warmer than the mean annual temperature but lower than body temperature (Pentacost, 2005); although both hot and warm thermal springs exist in the V&R region, our example areas only included warm springs

Refers to the spring types illustrated on conceptual Figure 7.

Spring types according to the classification of Springer and Stevens (2009).

348

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345–358

Sphere(s) of Dischargeb Hillslope

Cave, exposure, limnocrene

Limnocrene

EEGS-26-03-01_1XO

July 8, 2020

8:14

Spring Types in Valley and Ridge Province

are much less common than karst springs, but they can be found throughout the Appalachian Mountains. This group includes all springs with temperatures higher than the mean annual air temperature; warm springs have temperatures lower than body temperature (37°C), and hot springs have temperatures above that threshold (Kresic and Stevanovic, 2009).

According to Trapp and Horn (1997), most of the thermal springs in the eastern United States are in the V&R and are associated with faulting. The thermal mineral waters in the V&R played a historical role in the development of the region, and many were used for spas prior to the Civil War (Cohen, 1981).

DATA AVAILABILITY AND BIAS The available data about springs are biased due to access issues (landowner permission, security of watersupply sources), a focus on storm-driven changes, limited knowledge of spring locations (especially for small springs), and a lack of natural settings due to modifications. Access to springs granted by landowners and water-supply companies has played an important role in controlling the type of spring studied. Of the 14 karst springs studied by Shuster and White (1971) in central Pennsylvania, five springs are no longer accessible for monitoring, including one large watersupply spring (Big Spring, Bellefonte, PA). Similar large water-supply springs in the V&R of Pennsylvania are also inaccessible or sufficiently modified to make monitoring difficult (e.g., Roaring Spring, Roaring Spring, PA).

In addition to the limitation of restricted access, some of these larger springs are infrequently studied because of their chemical and discharge consistency. Nearly all recent works on central Pennsylvania springs have examined storm-driven changes in the chemistry of spring discharge or other storm-modified changes (Herman et al., 2008, 2009; Toran and Reisch, 2013; Reisch and Toran, 2014; Longenecker et al., 2017; Berglund et al., 2019; and Toran et al., 2019).

None of the Pennsylvania springs examined in these recent works is currently used for water supply. One exception is the work of Hurd et al. (2010) on seven springs in Pennsylvania, including one water-supply spring, which focused on dye tracing under non-storm conditions. Large public water-supply springs have been included in studies in both eastern and southern West Virginia (Vesper et al., 2009; Bausher, 2018).

Other research has similar biases. For example, Johnson et al. (2011) included 35 springs throughout the V&R in their report on water quality. Although they had wells in both siliciclastic and carbonate aquifers, all of the springs were from the carbonate aquifers. Saad and Hippe (1990) examined the large

springs of Pennsylvania and identified 137 springs discharging 100 gal/min (6.3 L/s) or more in the V&R. Of those springs, 123 were in carbonate rocks, 9 were in sandstone, and 3 were in shale or other layered sedimentary rocks. These numbers may reflect that fact that karst springs tend to be larger, but they are unlikely to reflect the overall distribution of all (large and small) springs. Furthermore, few of the research studies compared the functioning of the different spring types.

Nearly all V&R public water-supply springs and most private water-supply springs have been modified from their original setting. Historically, spring houses and spring boxes were built around springs to protect the water supply, contain the flow, direct the output, keep animals out, and provide cool storage (Long, 1960). More recently, structures have been built around springs in response to regulatory needs and to direct water flow to downstream users or treatment systems. Although the modifications are important to spring users, they can limit researchers’ ability to collect samples, determine spring discharge, or evaluate the water source.

Public databases of springs exist at the federal (USGS, 2016) and state (McColloch, 1986; PA Geological Survey, 2019) levels; however, these databases almost exclusively contain information on large, perennial springs or have single or few measurements of parameters at smaller springs. For example, the USGS database includes current daily flow data for only one spring in Pennsylvania and no springs in West Virginia. In both federal and state databases, smaller or ephemeral springs are rarely included unless they have been incorporated into a specific research project.

The Pennsylvania (PA Geological Survey, 2019) and West Virginia (McColloch, 1986) spring data sets were compiled for the counties included in this work (Figure 6). Most of the springs included in the data sets were karst springs, with fewer sandstone and shale springs included.

Although there has been a significant amount of research on V&R springs, it has not been equally distributed by spring type, characteristics, or use. Better protection of springs and spring resources requires a more comprehensive assessment of the type, number, and characteristics of the springs. The data bias may be partly due to the number of springs of each type; however, there is not currently sufficient information to evaluate that possibility.

EXAMPLE REGIONS IN THE CENTRAL V&R The data and examples used here come from three regions in the central V&R (Figure 1). This summary is based on published literature as well as graduate

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345–358

349

EEGS-26-03-01_1XO

July 8, 2020

8:14

Vesper and Herman

Central Pennsylvania

Figure 6. Distribution of spring sources in six counties. Data for Pennsylvania are from PA Geological Survey (2019), and data for West Virginia are from McColloch (1986). The Monroe County, WV, data set excludes all springs in the Greenbrier Group that are in the county but not in the V&R. Both limestone and dolomite springs are counted as carbonate springs.

student theses and unpublished undergraduate research. The spring nomenclature used here is related to the specific geology of the V&R region, but it may have wider applicability in similar geologic settings with a mix of carbonate and clastic bedrock.

r The central Pennsylvania region includes the springs studied by Shuster and White (1971) as well as other nearby karst, headwater, and warm springs (Figure 3). r The eastern West Virginia springs are in the Great Valley (Berkeley County) and V&R areas to the west of the Great Valley (Morgan County) (Figure 4). r The southern West Virginia springs are in the eastern half of Monroe County (Figure 5).

In addition to the three regions being in the same physiographic province, each region has multiple types of springs. There are also springs in each of the regions that are used for private and/or public water supplies.

Although this discussion is based on these three regions, it can be used a framework for understanding spring types throughout the Appalachians.

350

The central Pennsylvania region includes areas of Blair, Centre, and Huntingdon counties, which are located within the Nittany Arch Anticlinorium (Figure 3) (Butts and Moore, 1936). Ridges in the region are typically capped by the Tuscarora Formation or another resistant sandstone; ridge flanks contain the Juniata, Oswego, and Reedsville formations; and the valleys are underlain by Cambrian to Ordovician carbonates where the karst springs rise (Butts and Moore, 1936). A major structural feature of the region is the Birmingham Thrust Fault, which has up to 1.5 km of displacement.

Many of the significant karst springs in this region have been studied extensively since the 1960s, and much of our early understanding of karst hydrogeology and geochemistry was based on this region (Lattman and Parizek, 1964; Langmuir, 1969, 1971; Jacobson and Langmuir, 1970, 1974; Shuster and White, 1971; and Siddiqui and Parizek, 1971).

More recently, researchers in this region have examined sediment transport, which only occurs in dissolutionally enlarged conduits commonly associated with karst springs (Herman et al., 2008, 2009; Berglund et al., 2018, 2019). Storm-based work using isotopes, rare earth elements, and calcium/zirconium ratios has highlighted the variability in recharge and changes in flow in karst springs that arise in close proximity to each other (Herman et al., 2008, 2009; Berglund et al., 2018, 2019). Other storm-based work on Pennsylvania V&R karst springs has explored what can be learned about recharge and flow based on hysteresis in storm response (Herman et al., 2008, 2009; Toran and Reisch, 2013; Reisch and Toran, 2014; Longenecker et al., 2017; Berglund et al., 2019; and Toran et al., 2019).

A few headwater springs near Huntingdon, PA, have been studied as part of ecological surveys (Glazier and Gooch, 1987; Glazier, 1991; and Gooch and Glazier, 1991). These springs are located on the western limb of a secondary plunging syncline within the larger Nittany Anticlinorium. Two ambient-temperature (9–12°C) headwater springs and one warm (16–18°C) spring from the ecological studies are included in this data compilation. Early reports of the Huntingdon warm spring noted that a bathing pool and hotel had been built nearby at one time but were already abandoned by 1885 (White, 1885; Myers, 1959). These headwater and warm springs discharge from the calcareous shale of the Onondaga Formation in the valley between Warrior Ridge and Stone Creek Ridge.

Other headwater springs are known to exist near the ridge crest and on the flanks of Tussey Mountain (Downey, 2017). Little attention has been paid to

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345–358

EEGS-26-03-01_1XO

July 8, 2020

8:14

Spring Types in Valley and Ridge Province

these springs, though Jacobson and Langmuir (1970) included the sandstone waters to illustrate why sinkholes develop when the aggressive waters reach the shale-carbonate formational contacts near the valley floor.

Spring chemistry data (Table 1) from central Pennsylvania used for this compilation came from Shuster and White (1971), Glazier and Gooch (1987), Barna (2017), Downey (2017), Fink (2018), and Bliss (2019).

Eastern West Virginia The eastern West Virginia region includes springs (1) in the Great Valley in Berkeley County and (2) along and close to Cacapon Mountain in Morgan County (Figure 4). The Great Valley in Berkeley County is underlain largely by Cambrian to Ordovician carbonate rocks that are folded as part of the Massanutten Synclinorium (Kulander and Dean, 1986; Shultz et al., 1995). At the center of the structure, shale of the Martinsburg/Reedsville Formation underlies Opequon Creek as it flows northward into the Potomac River. The area is bounded to the west by North Mountain, a faulted anticline capped by the Tuscarora Sandstone (Grimsley and White, 1916). Thirteen springs, all flowing from carbonate units, are included in this data set. The data from this region suggest that springs closest to the ridge are more likely to be influenced by rapid surface input and shorter flow paths, creating greater variation in their chemistry. The springs located more centrally in the Great Valley may have flow paths that are fed at least partially from flow along strike (Grand, 2005; Vesper et al., 2009).

The Cacapon Mountain area of Morgantown County, WV, is located in the V&R region approximately 20 km west of the Great Valley. Cacapon Mountain is an anticlinal structure with a ridge of Tuscarora Sandstone; a lower ridge (Warm Springs Ridge) to the east runs parallel to Cacapon Mountain and is capped by Oriskany Sandstone (Grimsley and White, 1916). The valley between the ridges (Cold Run Valley) is underlain by near-vertical beds of sandstone and carbonates. To the east of Warm Springs Ridge, there is the Warm Spring Valley. The springs in Morgan County include headwater springs near the ridge tops, karst springs in an upper valley between the ridges, and a warm spring at the base of the ridge (Corder, 2008). The warm spring (Ladies Spring) is one of a set of springs located in Berkeley Springs State Park in Bath, WV. Berkeley Springs was one of the first mineral resorts established in the Virginias (Cohen, 1981; McColloch, 1986) and continues to operate today.

Spring chemistry data (Table 1) from eastern West Virginia used for this compilation came from Grand (2005), Corder (2008), and Vesper et al. (2009).

Southern West Virginia Monroe County, located in the southeastern corner of West Virginia, includes two physiographic provinces—the Allegheny Plateau in the western part of the county, underlain by near-horizontal Mississippian formations, and the V&R in the eastern part of the county (Figure 5). The St. Clair Thrust Fault defines the Allegheny Structural Front between the two provinces. Peters Mountain, the highest peak in this region, is capped by the Tuscarora Sandstone and flanked on the west by mixed-siliciclastic Juniata and Martinsburg (Reedsville) formation strata. The valley west of Peters Mountain is underlain by Middle to Lower Ordovician carbonates.

Spring water in Monroe County, WV, is used extensively for public and private supply. The county’s three public service districts (PSDs) obtain water from V&R springs along Peters Mountain; in combination, the PSDs provide water for more than 60 percent of county households. Although the valley karst springs are typically larger, most of the water-supply springs are headwater springs located on the flanks of the mountain and discharge from formations that are mixed-clastic units.

More than 200 springs have been mapped in the study area in eastern Monroe County and are found on the ridge and flanks of Peters Mountain, in the carbonate valley, and close to the St. Clair Thrust Fault (Richards, 2006; Bausher, 2018). Spring chemistry data (Table 1) from southern West Virginia used for this compilation came from Richards (2006), Moore (2012), and Bausher (2018).

DISCUSSION AND COMPARISON OF COMMON SPRING TYPES IN THE V&R Many of the springs found in the central V&R have common features, functions, and settings. Based on available data for the three regions, the springs can be organized into three types: (A) siliciclasticsourced headwater springs; (B) carbonate-sourced karst springs; and (C) warm springs (Table 1 and Figures 7 and 8). Although rock types are tied to this terminology, it should be noted that location, lithology, and topography are closely integrated in the V&R due to differential erosion of the Appalachian Mountains.

This framework differs from previous aquifer categorizations (Hollyday and Hileman, 1996; Johnson et al., 2011) because it is focused on springs; combines limestone and dolomite units into one group; considers function (headwater, karst); and includes warm springs. Though these three types generally can be separated with basic parameters like pH, tempera-

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345–358

351

EEGS-26-03-01_1XO

July 8, 2020

8:14

Vesper and Herman

Headwater (Siliciclastic) Springs

Figure 7. Conceptual model for common spring types found in the V&R. Headwater springs (A) can be at contacts (A1), fed by fractures or layers in the siliciclastic units (A2), or be closely tied to colluvial deposits (A3). Karst carbonate springs (B) are more commonly found in valleys and are fed by both allogenic and autogenic recharge and can be closely associated with stream resurgences (B1) or in other valley locations (B2). Thermal springs (C) are typically associated with deeper faulting.

ture, and specific conductance (SC), there are overlaps among types and significant spread within each type (Table 2).

For this discussion, the siliciclastic-sourced springs are referred to as headwater springs (type A springs, Figure 7) because they form the headwaters of the streams flowing from the ridges. This group includes springs that flow directly from sandstone (A1, Figure 7) on the ridges or from the mixed-sedimentary units that form the ridge flanks (A2, Figure 7). Although the ridge tops are most often formed by sandstones (see Figure 2A), the lithologies that form the ridge flanks are often a mixture of sedimentary materials dominated by siliciclastic shales and siltstones.

Colluvial material on the ridge flanks can obscure the spring source (A3, Figure 7). Some of these springs may be sourced only from boulder fields common on ridge flanks rather than underlying bedrock; however, they are included in this group because they flow from the ridge flanks and act as headwater stream sources.

The springs in this grouping tend to be small and can be ephemeral, but they may be present in large numbers. Given their small size, these springs are rarely included in databases, and their abundance is not quantified.

Some larger headwater springs are known and are used as water supplies and as bottled water sources.

The relatively low hardness of these waters makes them appealing for drinking water. In central Pennsylvania, springs on the south side of Nittany Mountain were used as a private and public water supply for Oak Hall, PA. In Monroe County, WV, two headwater springs

Figure 8. (A) Headwater spring in central Pennsylvania. (B) Karst spring in eastern West Virginia. (C) Karst spring in central Pennsylvania.

(D) A springhouse for a karst spring in eastern West Virginia. (E) The spring pool structure at the warm Old Sweet Springs in southern West Virginia. Photos by D. Vesper.

352

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345–358

EEGS-26-03-01_1XO

July 8, 2020

8:14

Spring Types in Valley and Ridge Province Table 2. Summary of spring data from three regions.

Headwater Springs

Temperature (°C) Central Pennsylvania Eastern West Virginia Southern West Virginia pH Central Pennsylvania Eastern West Virginia Southern West Virginia SC (mS/cm) Central Pennsylvania Eastern West Virginia Southern West Virginia Ca+2 (mg/L) Central Pennsylvania Eastern West Virginia Southern West Virginia Mg+2 (mg/L) Central Pennsylvania Eastern West Virginia Southern West Virginia

Karst Springs

Thermal Springs

#

Range

#

Range

#

Range

5 2 7

8.8–13.4 11.9–12.0 8.9–12.7

22 22 9

9.2–11.4 11.3–13.6 10.2–12.5

1 1 2

16.8 22.1 22–23

5 2 7

4.7–5.5 4.6–4.8 5.3–7.7

22 22 9

7.3–8.2 6.4–7.2 6.7–7.5

1 1 2

6.7 6.7 6.3–6.4

5 2 7

0.02–0.09 0.04–0.05 0.03–0.25

11 22 9

0.20–0.71 0.25–0.81 0.19–0.44

1 1 2

0.16 0.29 1.7–1.9

4 2 6

1.1–2.4 3.3–5.1 20.0–49.2

22 22 9

27.7–82.6 38.5–131 28.6–58.6

1 1 2

26.1 65.2 264–318

4 2 6

0.7–1.5 1.6–2.3 1.1–1.8

22 22 9

3.7–41.1 6.7–29.5 2.2–27.5

1 1 2

4.0 10.2 54.9–55

# = the number of springs included in the range; multiple samples for the same spring were averaged; SC = specific conductance. Data sources are as follows: central Pennsylvania: Barna (2017); Bliss (2019); Downey (2017); Fink (2018); Gooch and Glazier (1991); Shuster and White (1971); eastern West Virginia: Corder (2008); Grand (2005); Vesper et al. (2009); southern West Virginia: Bausher (2018); Richards (2006).

are used as public water supplies, and at least two more headwater springs are currently used as bottled water sources.

Given the siliciclastic source rocks, the main control on ground-water flow in headwater springs is likely to be fractures or bedding planes (Trapp and Horn, 1997). Comparing well data, the USGS concluded that the siliciclastic units in the V&R produce a much less transmissive aquifer than do the carbonate units (Swain et al., 1991). Based on the importance of fractures for controlling flow, the size and number of springs, and the lack of solutional openings, it is likely that the systems feeding the headwater springs are fairly localized. Although many of the springs may be small, they have a cumulative impact on carbonate aquifers in the valleys. Jacobson and Langmuir (1970) illustrated the role of these headwater springs in recharging and dissolving the valley aquifers. The U.S. Environmental Protection Agency (EPA) also recognized the importance of overland flow from the ridges in recharging the valley carbonate aquifers (Ginsberg and Palmer, 1996); however, the EPA did not explicitly address ground-water and spring sources from the siliciclastic units.

Examples of headwater springs can be found in each region. In central Pennsylvania, headwater springs are located near the ridge and along the flanks of Tussey

Mountain. Sand Springs, on the southern side of Tussey Mountain, south of Boalsburg, can be readily found along the “Sand Spring Trail” in a popular hiking area (PA Department of Conservation and Natural Resources, 2015). Other springs in Rothrock State Forest are used as private water supplies for hunting cabins (Figure 8A). In eastern West Virginia, headwater springs have been identified in Cacapon State Park and near the Coolfont Resort; the latter was previously used for water at the resort (Corder, 2008).

In southern West Virginia, an extensive survey of springs along Peters Mountain identified more than 200 headwater springs in an approximately 40 km2 area (Richards, 2006). These springs are mapped primarily in the mixed shale strata of the Reedsville Formation, which is also a common source for headwater springs in Pennsylvania; most of the springs in Monroe County are small, although a few larger springs are actively being used as PSDs for the towns of Union and Gap Mills, PA. The structure and elevation of these springs are not consistent over the county, suggesting that they may emerge from discontinuous permeable layers that overlie less permeable ones. The interbedded lithologies in the Reedsville Formation could create highly localized flow paths for small springs. However, several of the springs in the Reedville Formation are highly temporally consistent in temperature and

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345–358

353

EEGS-26-03-01_1XO

July 8, 2020

8:14

Vesper and Herman

eral interaction with a low SC, regardless of location.

This lack of carbonate signal is borne out in measurements of major ions at these springs; Ca2+ and Mg2+ concentrations are typically low relative to the nearby karst springs. The headwater springs in central Pennsylvania and eastern West Virginia have pH values that are generally acidic (4.6–5.5), and the SC is low (0.02–0.09 mS/cm). In southern West Virginia, where the headwater springs flow from mixed lithologies, including some carbonate layers, the pH and SC values are higher (pH 5.3–7.7; SC 0.03–0.25 mS/cm) than the other headwater springs but still generally lower than nearby karst springs. Warm springs may overlap the SC for headwaters springs or be significantly higher.

Johnson et al. (2011) found that the wells in the siliciclastic units yielded water chemistries with a greater range than was typically found in the carbonate units; this was true for the compiled pH data in this work but not for the SC data (Figure 9).

Karst (Carbonate) Springs

Figure 9. Screening chemistry of springs: diamonds for central Pennsylvania, triangles for eastern West Virginia, and squares for southern West Virginia.

chemistry, suggesting that longer flow paths exist in some cases. Based on well data, Johnson et al. (2011) noted that most of the V&R siliciclastic aquifers were localized and had low yields, but that could change in areas with extensive fracturing.

The headwater spring chemistry (Figure 9 and Table 2) generally reflects the lack of carbonate min-

354

Karst springs have been the focus of most spring studies in the V&R, and a considerable body of literature exists on their locations, chemistry, and temporal variability. These springs are most commonly found in the valleys (Table 1 and Figure 7) where the carbonate rocks have been left exposed by differential erosion.

It is not uncommon for the headwater streams to be captured by sinkholes near the siliciclastic-carbonate contact. This water is an important allogenic recharge source for the valley springs (B1, Figure 7). The valley streams may disappear through sinkholes and reappear at springs (B2, Figure 7).

Although karst springs exist in a wide range of sizes, they are generally larger than the headwater springs.

Some of these springs are large enough, and sufficiently well known, to be included in numerous public records and are commonly identified on topographic maps. Their impact on development of the regions can be seen in the names of the towns near the springs, such as Spring Mills and Roaring Spring in Pennsylvania and Sweet Springs in West Virginia.

Karst springs may be in limestone or dolomite.

Limestone springs are more likely to have fast flow paths, while dolomite springs tend to have slow, fracture-controlled flow systems; hence, the limestone springs tend to be more variable chemically and in flow over time (Shuster and White, 1971; Hollyday and Hileman, 1996). Hollyday and Hileman (1996) reported that the dolomite wells were more likely to be controlled by joints and fractures, but they had a higher median specific capacity than wells in limestone.

The type of recharge feeding the carbonate aquifer impacts the variability of the springs. Karst springs

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345–358

EEGS-26-03-01_1XO

July 8, 2020

8:14

Spring Types in Valley and Ridge Province

closest to ridges are more likely to receive input from the headwater springs, and streams that flow from the ridges are aggressive in the dissolution of carbonates. In Pennsylvania, headwater streams flowing off of Tussey Mountain’s Schalls Gap are swallowed by a sinkhole near the shale-limestone contact; this water flows through a strike-parallel flow system that can be seen in the underlying Chisel Cave, before reappearing at the nearby Rock Spring (Shuster and White, 1971).

In the Tuscarora Creek watershed in the West Virginia portion of the Great Valley, the springs closest to the ridge and the North Mountain Thrust Fault were the most variable spatially and temporally (Vesper et al., 2009). In southern West Virginia, the springs close to sinkholes were the most variable. If the recharge to the carbonate aquifer is largely dispersed, then the springs tend to be less variable in flow and chemistry. In central Pennsylvania, the dolomite springs are located more centrally in the valleys due to the geologic structures, and they have a higher proportion of dispersed recharge. The combination of dispersed recharge and slower flow paths results in springs that tend to be more consistent in flow and chemistry.

The larger, more valley-central springs in the Great Valley of eastern West Virginia tend to have consistent flow and chemistry, although they do not have a dolomitic source signature. In this case, the more consistent chemistry may be due to the distance to the ridges, which would limit the rapid input of more aggressive waters.

The distinctions discussed here over-simplify a complex continuum of spring behavior, but in general, large dolomite-sourced springs like the Big Spring at Bellefonte, PA, have been consistent, high-quality water sources for significant populations, while flashier limestone springs have not played the same watersupply role in the region, but they are sometimes used for residential supplies. Both limestone- and dolomitecontrolled springs are also found in southern West Virginia. Rich Creek Spring in Monroe County, WV, is similar to Rock Spring in Centre County, PA. Both springs are chemically variable over time and fed by cave systems occurring mostly in limestones close to a formational contact (Shuster and White, 1971; Dasher, 2019). Big Spring and Thompson Spring in central Pennsylvania (Shuster and White, 1971) and Zenith Spring in southern West Virginia (Bausher, 2018) are highly consistent over time and have Ca/Mg molar ratios close to 1, indicating dolomite in the spring flow paths.

The chemistry of the karst springs generally shows evidence of significant dissolution of carbonate rocks: circumneutral pH (6.4–8.2), generally higher SC (0.19– 0.81 mS/cm), and higher Ca2+ and Mg2+ concentrations than those measured in the headwater springs

(Figure 9 and Table 2). The pH values of the central Pennsylvania springs are, as a group, slightly higher than the other regions, but the SC values are similar in all three regions. There is overlap with headwater springs in SC and pH values in southern West Virginia, but little to no overlap in central Pennsylvania and eastern West Virginia. Interestingly, the range of average temperatures in headwater and karst springs overlaps in all three regions despite a wide variety in sampling frequency and time of sampling.

Warm Springs Warm springs (∼20–40°C) exist in all three regions and are likely associated with deeper thrust faults in the V&R (Hobba et al., 1977; Trapp and Horn, 1997).

According to Trapp and Horn (1997), nearly all of the warm springs in the eastern United States are located in the V&R. Most warm springs have a high yield and are located in regions with vertical permeability due to bedding plane structures, fractures, or thrust faults (C, Figure 7). Although “hot springs” (∼>40°C) can be found in the V&R, our example areas only included thermal warm springs.

There are fewer warm springs recorded than either of the other two types of springs in each of the three regions. The four warm springs have higher temperatures than the headwater and karst springs, but the springs in the West Virginia regions are consistently warmer than Huntingdon Warm Spring in central Pennsylvania (Figure 9 and Table 2). The warm springs have comparable pH values that are within the range of the karst springs. The SC values of Huntingdon Warm and Ladies Springs (central Pennsylvania and eastern West Virginia, respectively) are comparable to the karst springs and within the range of the southern West Virginia headwater springs and considerably lower than the warm springs in southern West Virginia.

Summary of Flow Paths for the Spring Types Figure 7 presents a schematic diagram for the common types of springs in the V&R, starting with headwater springs on ridge flanks (A, Figure 7). These springs contribute to the recharge of the carbonate springs on the valley floors (B, Figure 7). The headwater springs and their chemically aggressive waters, in fact, help to develop the carbonate aquifer system by enlarging fractures and bedding plane partings through dissolution and often create sinkholes at the shale-carbonate contact. Karst springs (B, Figure 7) are recharged by the water from the ridges via both surface and subsurface contributions; they are also recharged from precipitation onto the valley and

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345–358

355

EEGS-26-03-01_1XO

July 8, 2020

8:14

Vesper and Herman

sinking streams in the valley. Warm springs discharge from deeper flow paths that are somewhat isolated from the other two systems (C, Figure 7). There is a vertical flow component to these springs commonly associated with faults in the V&R. Within each of these categories, there is significant variation. The scope of this variation in the V&R karst springs is much better studied, and many attempts have been made to classify springs (e.g., by flow and recharge) within this category. However, to provide a full picture of significant spring resources in the V&R, all three spring types should be examined.

SUMMARY The V&R springs are critical water resources for human and ecosystem use. Although there has been extensive research on springs in the V&R, the focus has been on chemically variable karst springs followed by the warm springs; minimal research has been completed on the siliciclastic headwater springs that contribute to recharge of the larger karst systems. These types of springs (headwater, karst, and warm) are common to the three regions in the central V&R. The temperature, chemistry, and locations of the springs are controlled by the structural geology as well as the formations and lithologies through which the recharge water travels. Although other types of springs are present throughout the V&R, these examples can provide a framework for comparing commonalities found throughout the region.

ACKNOWLEDGMENTS Data were obtained by the following students as part of a graduate or undergraduate research project: Joshua Barna, Emily Bausher, Benjamin Bliss, Lacoa Corder, Autum Downey, Madison Fink, Rachel Grand, Johnathan Moore, Jason Tarbert. Thanks go to Autum Downey and Jill Riddell for help with the figures.

This research was partially supported by grants from the following sources: (1) the National Energy Technology Laboratory (Regional University Alliance #DE-FE0004000; URS Activity #4.600.251.002); (2) West Virginia University Community Engagement and Program to Stimulate Competitive Research (PSCOR) grants; (3) USGS Water Research Institute 104(e); (4) West Virginia Conservation Agency, Eastern Panhandle Conservation District; (5) U.S. Department of Agriculture National Research Imitative Competitive grants program (#2003-35102-13537); (6) the Appalachian Freshwater Initiative (funded by National Science Foundation [NSF] EPSCOR grant #1417477); (7) the National Park Service (agreements P11AC60552 and J31001100007) with the Great

356

Lakes Northern Forest Cooperative Ecosystem Studies Unit (cooperative agreement H6000082000); (8) the West Virginia Department of Health and Human Services (agreement #G180409); and (9) National Science Foundation’s Hydrologic Sciences Program (NSF EAR-1417401).

REFERENCES Barna, J. M., 2017, Mg/Ca and PCO2 Storm Hysteresis as an Indicator of Flowpaths and Recharge Sources at Two Karst Springs in Central Pennsylvania: Unpublished Senior Honor’s Thesis, Department of Geology & Environmental Geosciences, Bucknell University, Lewisburg, PA, 63 p.

Bausher, E., 2018, Qualitative and Quantitative Analysis of Carbonate Waters in the Peters Mountain Region of Monroe County, WV: Unpublished Master’s Thesis, Department of Geology and Geography, West Virginia University, Morgantown, WV, 123 p.

Berglund, J. L.; Toran, L.; and Herman, E. K., 2018, Using stable isotopes to distinguish sinkhole and diffuse storm infiltration in two adjacent springs. In Sasowsky, I. D., Byle, M.J., and Land, L. (Editors), Proceedings of the 15th Multidisciplinary Conference on Sinkholes and the Engineering and Environmental Impacts of Karst Sypmosium, April 2–6, Shepherdstown, West Virginia: National Cave and Karst Research Institute, Carlsbad, NM, pp. 53–63.

Berglund, J. L.; Toran, L.; and Herman, E. K., 2019, Deducing flow path mixing by storm-induced bulk chemistry and REE variations in two karst springs: With trends like these who needs anomalies?: Journal of Hydrology, Vol. 571, pp. 349–364.

Bliss, B. R., 2019, Tracking Annual, Seasonal, and Storm-toStorm Changes in Three Pennsylvania Karst Springs to Determine Flow Paths and Recharge Patterns: Unpublished Senior Honor’s Thesis, Department of Geology & Environmental Geosciences, Bucknell University, Lewisburg, PA, 63 p.

Butts, C. and Moore, E. S., 1936, Geology and Mineral Resources of the Bellefonte Quadrangle, Pennsylvania: U.S. Geological Survey Bulletin 855, 128 p.

Cohen, S., 1981, Historic Springs of the Virginias, A Pictorial History: Pictorial Histories Publishing Company, Charleston, WV, 218 p.

Corder, L. L., 2008, Hydrogeochemical Characterization of Springs and Wells in the Cacapon Mountain Aquifer: Unpublished Master’s Thesis, Department of Geology, West Virginia University, Morgantown, WV, 80 p.

Dasher, G. R., 2019, The Caves and Karst of Monroe County, West Virginia, 2nd ed.: West Virginia Speleological Survey, Barrackville, WV, 476 p.

Downey, A. R., 2017, Sandstone Springs as Contributors to Karst Aquifers in the Valley and Ridge Province: Unpublished Undergraduate Report (McNair Fellow Project), Department of Geology, West Virginia University, Morgantown, WV, 13 p.

Fink, M. S., 2018, Detecting Trends in Calculated and Directly Measured CO2 Concentrations in Karst Springs: Unpublished Senior Honor’s Thesis, Department of Geology & Environmental Geosciences, Bucknell University, Lewisburg, PA, 54 p.

Ginsberg, M. and Palmer, A. N., 1996, Guidelines for Wellhead and Springhead Protection Area Delineation in Carbonate Rocks: U.S. Environmental Protection Agency Report 904-B97-003, 66 p.

Glazier, D. S., 1991, The fauna of North America temperate cold springs: Patterns and hypotheses: Freshwater Biology, Vol. 26, pp. 527–542.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345–358

EEGS-26-03-01_1XO

July 8, 2020

8:14

Spring Types in Valley and Ridge Province Glazier, D. S., 1998, Springs as model systems for ecology and evolutionary biology: A case study of Gammarus minus Say (Amphipoda) in mid-Appalachian springs differing in pH and ionic content. In Botosaneanu, L. (Editor), Studies in Crenobiology: The Biology of Springs and Springbrooks: Backhuys, Leiden, Netherlands, pp. 49–62.

Glazier, D. S., 2014, Springs. Vol. 2014. Reference Module in Earth Systems and Environmental Sciences. Electronic document, available at http://www.sciencedirect.com/science/ article/pii/B9780124095489093222 Glazier, D. S. and Gooch, J. L., 1987, Macroinvertebrate assemblages in Pennsylvania (USA) springs: Hydrobiologia, Vol. 150, pp. 33–43.

Gooch, J. L. and Glazier, D. S., 1991, Temporal and spatial patterns in mid-Appalachian springs: Memoirs Entomological Society Canada, Vol. 123, No. S155, pp. 29–49.

Grand, R. V., 2005, Controls, Characterization, and Small Scale Chemical Variations of a Karst System: A Geochemical Assessment of Tuscarora Creek Watershed, West Virginia: Unpublished Master’s Thesis, Department of Geology, West Virginia University, Morgantown, WV, 114 p.

Grimsley, G. P. and White, I. C., 1916, Jefferson, Berkeley, and Morgan Counties: West Virginia Geological Survey, Wheeling, WV, 644 p.

Herman, E. K.; Toran, L.; and White, W. B., 2008, Threshold events in spring discharge: Evidence from sediment and continuous water level measurement: Journal Hydrology, Vol. 351, No. 1, pp. 98–106.

Herman, E. K.; Toran, L.; and White, W. B., 2009, Quantifying the place of karst aquifers in the groundwater to surface water continuum: A time series analysis study of storm behavior in Pennsylvania water resources: Journal Hydrology, Vol. 376, No. 1, pp. 307–317.

Hobba, W. A. J.; Chemerys, J. C.; Fisher, D. W.; and Pearson, F. J. J., 1977, Geochemical and Hydrologic Data for Wells and Springs in Thermal-Spring Areas of the Appalachians: U.S. Geological Survey Water-Resources Investigations Report 77-25, 36 p.

Hobba, W. A. J.; Fisher, D. W.; Pearson, F. J. J.; and Chemerys, J. C., 1979, Hydrology and Geochemistry of Thermal Springs of the Appalachians: U.S. Geological Survey Professional Paper 1044-E, 36 p.

Hollyday, E. F. and Hileman, G. E., 1996, Hydrogeologic Terranes and Potential Yield of Water to Wells in the Valley and Ridge Physiographic Province in the Eastern and Southeastern United States: U.S. Geological Survey Professional Paper 1422-C, 38 p.

Hurd, T. M.; Brookhart-Rebert, A.; Feeney, T. P.; Otz, M. H.; and Otz, I., 2010, Fast, regional conduit flow to an exceptional-value spring-fed creek: Implications for sourcewater protection in mantled karst of south-central Pennsylvania: Journal Cave Karst Studies, Vol. 72, No. 3, pp. 129– 136.

Jacobson, R. L. and Langmuir, D., 1970, The chemical history of some spring waters in carbonate rocks: Ground Water, Vol. 8, No. 3, pp. 5–9.

Jacobson, R. L. and Langmuir, D., 1974, Controls on the quality variations of some carbonate spring waters: Journal Hydrology, Vol. 23, pp. 247–265.

Johnson, G. C., 1997, Water Quality of Springs in the Valley and Ridge Physiographic Province in the Upper Tennessee River Basin: U.S. Geological Survey Water-Resources Investigations Report 02-4180, 22 p.

Johnson, G. C.; Zimmerman, T. M.; Lindsey, B. D.; and Gross, E. L., 2011, Factors Affecting Groundwater Quality in the Val-

ley and Ridge Aquifers, Eastern United States, 1993–2002: U.S. Geological Survey Scientific Investigations Report 2011-5115, 70 p.

Kresic, N. and Stevanovic, Z., 2009, Groundwater Hydrology of Springs: Engineering, Theory, Management and Sustainability: Butterworth-Heinemann, Burlington, MA, 592 p.

Kulander, B. R. and Dean, S. L., 1986, Structure and tectonics of central and southern Appalachian Valley and Ridge and Plateau Provinces, West Virginia and Virginia: AAPG Bulletin, Vol. 70, No. 11, pp. 1674–1684.

LaMoreaux, P. E. and Tanner, J. T., 2001, Springs and Bottled Waters of the World. Ancient History, Source, Occurrence, Quality and Use: Springer, Berlin, 315 p.

Langmuir, D., 1969, The chemistry of ground and surface waters in Nittany Valley: Earth Mineral Sciences, Vol. 38, No. 9, pp.

73–74.

Langmuir, D., 1971, The geochemistry of some carbonate ground waters in central Pennsylvania: Geochimica et Cosmochimica Acta, Vol. 35, pp. 1023–1045.

Lattman, L. H. and Parizek, R. R., 1964, Relationship between fracture traces and the occurrence of ground water in carbonate rocks: Journal Hydrology, Vol. 22, No. 1, pp. 1–26.

Long, A. J., 1960, Springs and springhouses: Pennsylvania Folklife, Vol. 11, No. 1, pp. 40–43.

Longenecker, J.; Bechtel, T.; Chen, Z.; Goldscheider, N.; Liesch, T.; and Walter, R., 2017, Correlating global precipitation measurement satellite data with karst spring hydrographs for rapid catchment delineation: Geophysical Research Letters, Vol. 44, No. 10, pp. 4926–4932.

McColloch, J. S., 1986, Springs of West Virginia: West Virginia Geological and Economic Survey, Charleston, WV, 493 p.

Moore, J. E., 2012, Potential Inhibitors and Sources of Error in the Measurement of Travertine Deposition Rates in a Karst Spring Influenced by Thermal Mineral Waters: Unpublished Master’s Thesis, Department of Geology, West Virginia University, Morgantown, WV, 117 p.

Myers, R. E., 1959, Some like it hot, some like it cold: Proceedings of the Pennsylvania Academy of Science, Vol. 33, pp. 152–155.

PA Department of Conservation and Natural Resources, 2015, Rothrock State Forest Public Use Map. Electronic document, available at http://www.docs.dcnr.pa.gov/ cs/groups/public/documents/document/dcnr_20032064.pdf PA Geological Survey, 2019, Pennsylvania Groundwater Information System (PaGWIS), 4th Ser. Electronic document, available at www.dcnr.state.pa.us/topogeo/groundwater/pagwis/ records/index.htm Pentacost, A., 2005, Travertine; Springer-Verlag, Berlin, 446p.

Perry, L. D.; Costain, J. K.; and Geiser, P. A., 1979, Heat flow in western Virginia and a model for the origin of thermal springs in the folded Appalachians: Journal Geophysical Research, Vol. 84, No. B12, pp. 6875–6883.

Reisch, C. E. and Toran, L., 2014, Characterizing snowmelt anomalies in hydrochemographs of a karst spring, Cumberland Valley, Pennsylvania (USA): Evidence for multiple recharge pathways: Environmental Earth Sciences, Vol. 72, No. 1, pp. 47–58.

Richards, B. G., 2006, Aqueous Geochemistry of Springs along Peters Mountain in Monroe County, WV: Unpublished Master’s Thesis, Department. of Geology, West Virginia University, Morgantown, WV, 73 p.

Saad, D. A. and Hippe, D. J., 1990, Large Springs in the Valley and Ridge Physiographic Province of Pennsylvania: U.S. Geological Survey Open-File Report 90-164, 17 p.

Scopus, 2019, Document details for Shuster and White, 1971, available at http://www.scopus.com, Accessed 8/11/19.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345–358

357

EEGS-26-03-01_1XO

July 8, 2020

8:14

Vesper and Herman Shultz, R. A.; Hobba, W. A. J.; and Kozar, M. D., 1995, Geohydrology, Ground-Water Availability, and Ground-Water Quality of Berkeley County, West Virginia, with Emphasis on the Carbonate-Rock Area: U.S. Geological Survey WaterResources Investigations Report 93-4073, 88 p.

Shuster, E. T. and White, W. B., 1971, Seasonal fluctuations in the chemistry of limestone springs: A possible means for characterizing carbonate aquifers: Journal Hydrology, Vol. 14, pp.

93–128.

Siddiqui, S. H. and Parizek, R. R., 1971, Hydrogeologic factors influencing well yields in folded and faulted carbonate rocks in central Pennsylvania: Water Resources Research, Vol. 7, No. 5, pp. 1295–1312.

Springer, A. E. and Stevens, L. E., 2009, Spheres of discharge of springs: Hydrogeology Journal, Vol. 17, pp. 83–93.

Swain, L. A.; Hollyday, E. F.; Daniel, C. C. I.; and Mesko, T. O., 1991, An overview of the Appalachian Valleys-Piedmont Regional Aquifer-System analysis. In Hotchkiss, W. R. and Johnson, A. I. (Editors), Regional Aquifer Systems of the United States, Aquifers of the Southern and Eastern States, Volume AWRA Monograph Series No. 17: American Water Resources Association, Bethesda, MD, pp. 43–57.

Toran, L.; Herman, E. K.; and Berglund, J. L., 2019, Advances in monitoring to understand flow paths in karst:

358

Comparison of historic and recent data from the Valley and Ridge of Pennsylvania. In Younos, T.; Schreiber, M.; and Kosič Ficco, K. (Editors), Karst Water Environment: Advances in Research, Management and Policy: Cham, Springer International Publishing, pp. 65–89.

Toran, L. and Reisch, C. E., 2013, Using stormwater hysteresis to characterize karst spring discharge: Groundwater, Vol. 51, No. 4, pp. 575–587.

Trapp, H. J. and Horn, M. A., 1997, Ground Water Atlas of the United States. Segment 11 Delaware, Maryland, New Jersey, North Carolina, Pennsylvania, Virginia, West Virginia: U.S. Geological Survey Hydrologic Atlas 730-L, 24 p.

U.S. Geological Survey (USGS), 2016, National Water Information System data available on the World Wide Web (USGS Data for the Nation). Electronic document, available at http://waterdata.usgs.gov/nwis/ Vesper, D. J.; Grand, R. V.; Ward, K.; and Donovan, J. J., 2009, Geochemistry and implications for structural control in a spring-dense karst watershed, the Appalachian Great Valley, West Virginia, USA: Environmental Geology, Vol. 53, No. 3, pp. 667–678.

White, I. C., 1885, The Geology of Huntingdon County: Pennsylvania Geological Survey, Harrisburg, PA, 528 p.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 345–358

EEGS-26-02-07_1XO

July 8, 2020

8:35

Hydrology of a Southern Appalachian Hypocrene Spring-Fed Fen JEFFREY WILCOX* EMILY BRADSHAW MARINO University of North Carolina Asheville, One University Heights, CPO #2330, Asheville, NC 28804

ADAM WARWICK MEGAN SUTTON The Nature Conservancy Southern Blue Ridge Office, 5 Barbetta Drive, Asheville, NC 28806

Key Terms: Wetland Hydrology, Fen, Seep, Springs, Sarracenia, Prescribed Fire, Vegetation Management, Blue Ridge Province ABSTRACT Garland Seep is a Southern Appalachian fen that supports a population of federally endangered green pitcher plants (Sarracenia oreophila). The wetland is underlain by clayey stream deposits above fractured bedrock, is located at the base of a mountain slope, and is fed by groundwater that originates as recharge on the adjacent hillslope. Groundwater wells were installed following a hydrologic restoration in the mid-1990s and have been monitored at varying frequencies since that time. The 20+ year record provides evidence that Garland Seep can be classified as a “hypocrene fen,” in which spring flow rarely reaches the ground surface because of low discharge rates and high evapotranspiration (ET). In general, water-level fluctuations followed seasonal ET patterns, with higher water levels in the winter and early spring (when ET is low) and lower levels in the summer and fall. During wetter years, the water table remained near the ground surface for much of the year, with the clay layer underlying the site retaining moisture even after water levels had dropped. The “clay wetting” period was shorter during dryer years and corresponded with a reduction in the number of pitcher plant clumps observed at the site. In addition to the geologic and climatic controls on hydrology, previous landowners used fire to maintain open space for grazing, and The Nature Conservancy has continued the practice to combat woody vegetation and to open the canopy. Prescribed burns reduce ET (at least initially), cause a rise in water levels, and have helped maintain a thriving Sarracenia population.

*Corresponding author email: jwilcox@unca.edu

INTRODUCTION Garland Seep is a Southern Appalachian mountain wetland in western North Carolina (United States) that supports a vibrant population of federally endangered green pitcher plants (Sarracenia oreophila) (Figure 1). Sarracenia seedlings require high soil moisture and light intensity, and as a result, land-use conversion, hydrologic alterations, and absence of fire have likely contributed toward relegating this carnivorous plant to Garland and a few dozen other known locations in northern Georgia and Alabama (United States) (USFWS, 2011). The presence of Sarracenia at Garland Seep is curious because it grows in clumps in an atypical “wetland” where the ground surface is mostly solid and dry (Figure 2). The goal of this study was to identify and describe the hydrologic conditions supporting the wetland and its rare flora. Garland Seep is presented in this special issue as a classic example of a “hypocrene spring,” a classification that helps explain the recent success of wetland management to preserve—and expand—the Sarracenia population.

SITE DESCRIPTION Garland Seep is a 5-acre (2.02-ha) Southern Appalachian wetland that has been owned and managed

by The Nature Conservancy (TNC) since 1988. It is also designated as a Conservation Partnership Area for the Mountain Bogs National Wildlife Refuge, which was officially established in April 2015 (USFWS, 2019). The seep supports a variety of grasses, sedges, and moisture-tolerant forbs in addition to the federally endangered green pitcher plants (Sarracenia oreophila). Rapidly growing trees and shrubs within and surrounding the herbaceous wetland core include tulip poplar (Liriodendron tulipifera), red maple (Acer rubrum), red chokeberry (Aronia arbutifolia), swamp rose (Rosa palustris), and sweet white azalea

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 359–366

359

EEGS-26-02-07_1XO

July 8, 2020

8:35

Wilcox, Marino, Warwick, and Sutton

Figure 1. Federally endangered green pitcher plants (Sarracenia oreophila). Photos by Karen Vaughn.

(Rhododendron arborescens) (Schwartzman, 2016; Weakley and Schafale, 1994).

The wetland is located about 2,000 ft (610 m) above sea level, between the western base of a mountain slope and the eastern shore of an artificial valley reservoir (Figure 3). The site is underlain by Hemphill loam and Nikwasi fine sandy loam (Warwick, 2019): gray, clayrich soils with rounded but irregularly shaped pebbles and cobbles (Richardson and Huang, 1996; Braun, 1997). Well-construction reports for nearby domestic wells indicate the unconsolidated clay unit is 10–12 ft (3.0 to 3.7 m) thick and rests directly above fractured bedrock (Wilcox, 2012). Prior to TNC’s acquisition of the property in 1988, several 6-in. (15-cm) drainage tiles had been installed within the clay layer

to drain the site. The outlets of these drainage tiles were plugged in the early 1990s (TNC, 2010), but the gravel beds surrounding the drainage tiles continued to provide preferential flow zones beneath the wetland (Braun, 1994). In 1995, TNC excavated the drain tiles at 50-ft (15-m) intervals and permanently blocked them in an effort to restore the natural hydrology of the site.

Garland Seep has been subject to controlled burns since the early 1900s, and wildfires before then likely played a role in preventing rapid forest succession (Warwick, 2019). Historical records indicate that a single landowner burned the site every winter from 1908 to maintain open land for grazing cattle. Following a 20-year absence of fire, TNC resumed the practice in

Figure 2. Sarracenia oreophila grow in clumps at Garland Seep, where the ground surface is typically dry enough to wear normal shoes and even kneel in jeans. Photos by Jeff Wilcox (left) and Karen Vaughn (right).

360

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 359–366

EEGS-26-02-07_1XO

July 8, 2020

8:35

Hydrology of a Hypocrene Spring-Fed Fen

Figure 3. Site location and cross section, Garland Seep, Clay County, NC (Unied States).

April 1992 and conducted their ninth prescribed burn in March 2019.

TNC has monitored the Sarracenia oreophila population since 1990, counting flowers, leaves (pitchers), and clumps of pitchers along eight transects. As the Sarracenia population grew—purportedly in response to the aforementioned management activities—it became infeasible to count flowers and pitchers along all eight transects or to differentiate between neighboring clumps that had grown together. Nevertheless, TNC continued counting pitchers and flowers along three of the original transects, and it became clear that the Sarracenia were thriving. Whereas 157 pitchers and seven flowers were counted along the three transects in 1990, there were 5,614 pitchers and 390 flowers in the 2018 survey (Figure 4). Only 43 flowers were counted across the entire site during the first survey in 1990, while over 2,000 were counted in 2018 (Warwick, 2019).

METHODS Hydrologic Monitoring A network of 16 groundwater monitoring wells was installed in the mid-1990s following hydrologic

restoration (drain tile blockage and removal) to provide a baseline prior to residential development on the hillslope above Garland Seep (Richardson and Huang, 1996). The wells were constructed with 1.25-in.

(3.2-cm) diameter steel or 2-in. (5.1-cm) polyvinyl chloride (PVC) pipe with 5-ft (30.5-cm) or 10-ft (61cm) screens. Well depths ranged from 2.0 to 13.0 ft (1.6 to 4.0 m) below ground surface and included three sets of well nests to evaluate the vertical component of groundwater flow. Richardson and Huang (1996) also conducted slug tests and collected water samples for water-quality analyses. Measured hydraulic conductivities ranged from 1.3E-4 to 0.49 ft/d (4.0E-5 to 0.15 m/d). Following the conclusion of this initial study, a resident living near Garland Seep continued to measure water levels on a weekly basis from 1996 to 2013 (unless he was on vacation), after which we deployed pressure transducers (Onset HOBO U-20, Onset Corporation, Bourne, MA) in six of the wells to record water levels at hourly intervals.

Reliable weather data were not available dating back to 1996 in the vicinity of Garland Seep or even the nearest towns in Clay County. We installed a tipping-bucket rain gauge onsite in 2013, and records from the closest weather station at Andrews-Murphy

Figure 4. The Nature Conservancy has documented an increase in the number of Sarracenia oreophila leaves (pitchers) and flowers along three survey transects (TNC, 2010).

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 359–366

361

EEGS-26-02-07_1XO

July 8, 2020

8:35

Wilcox, Marino, Warwick, and Sutton

Figure 5. Mean monthly precipitation (a), evapotranspiration (b), and net precipitation (c), 1996–2018.

airport about 15 miles (24 km) away only go back to June 2001. The closest weather stations with complete precipitation records (1996–2018) were located over 16 miles (26 km) away, to the west in Murphy, NC, and to the east at Coweeta Hydrologic Laboratory in Otto, NC. Given precipitation variability from individual storms in a mountainous region, we downloaded only monthly precipitation totals from the State Climate Office of North Carolina (SCONC, 2019). SCONC also provided monthly evapotranspiration (ET) estimates using the FAO Penman-Monteith method (Allen et al., 1998) and net radiation, ground heat flux, temperature, humidity, and wind speed data from the Andrews-Murphy airport.

2013 and a minimum of 33.9 in. (86 cm) in 2016. Mean monthly precipitation ranged from 3.20 ± 2.21 in. (8.1 ± 5.6 cm) in October to 6.13 ± 3.05 in. (15.6 ± 7.7 cm) in December (Figure 5a). Mean annual reference ET at the Andrews Airport was 43.4 ± 2.3 in. (110 ± 6 cm), with monthly ET ranging from 1.17 ± 0.11 in. (3.0 ± 0.3 cm) in December to 5.92 ± 0.57 in. (15.0 ± 1.5 cm) in June (Figure 5b). Mean annual net precipitation (PET) was +20.8 ± 12.5 in. (+53 ± 32 cm) and ranged from +42.5 in. (108 cm) in 2018 to −0.6 in. (−2 cm) in 2016. Monthly net precipitation ranged from +4.96 ± 2.92 in. (12.6 ± 7.4 cm) in December to −0.97 ± 2.41 in. (−2.5 ± 6.1 cm) in August (Figure 5c).

Water Levels and Groundwater Flow

RESULTS Precipitation and Evapotranspiration Precipitation measurements collected onsite at Garland Seep from 2013 to 2018 were generally consistent with those collected in Murphy, NC (west of Garland), but about 20 percent lower than those collected at the Coweeta Hydrologic Laboratory (east of Garland Seep). For this reason, we relied on the historical precipitation record from Murphy, NC, from 1996 through 2018; mean precipitation was 58.3 ± 11.7 in.

(148 ± 30 cm), with a maximum of 77.7 in. (197 cm) in 362

Groundwater flows from east to west beneath Garland Seep, with a horizontal gradient of approximately 0.05 (Figure 6). The vertical gradient ranges from −0.36 to +0.24, with both extremes recorded at the well nest located near the upper edge of the wetland boundary and Sarracenia habitat (wells 1 and 14 in Figure 6). The vertical component of groundwater flow at this well nest is generally upward in the late winter and spring months and downward in the late summer and fall months (Figure 7). During dryer years (2000, 2001, 2007, 2008, 2016) there is a longer period of downward flow (“draining” conditions), while

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 359–366

EEGS-26-02-07_1XO

July 8, 2020

8:35

Hydrology of a Hypocrene Spring-Fed Fen

Figure 6. Garland Seep water-table map.

Figure 7. Net precipitation (top) and water levels measured at Garland Seep (1996–2018). The dates of prescribed fire are indicated with triangles above the figure.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 359–366

363

EEGS-26-02-07_1XO

July 8, 2020

8:35

Wilcox, Marino, Warwick, and Sutton

in wetter years, upward flow dominates. Artesian conditions are often so great that the deeper well 14 is freeflowing. During these wetter periods—and years— with extended artesian conditions, water levels approach the ground surface and occasionally exceed it (see hydrographs for water-table wells 1 and 2 in Figure 7). We estimated a maximum upward seepage velocity of 0.8 to 3.0 in./mo (2.0 to 7.6 cm/mo) based on a vertical gradient of +0.24 and hydraulic conductivity values measured by Richardson and Huang (1996).

The well nest located just upgradient of the core wetland complex (wells 9 and 12, Figure 6) showed a similar seasonal pattern, with higher water levels in winter and spring than in summer and fall, but artesian conditions were rare, and the potentiometric surface only occasionally approached the ground surface (Figure 7, bottom).

DISCUSSION A Hypocrene Spring Water levels measured within the pitcher plant habitat at Garland Seep frequently rose within inches (centimeters) of the ground surface over the 20-year record, particularly during wetter seasons and years. When this happened, the clay layer immediately underlying the site retained moisture even after water levels had dropped. The repeated wetting of this shallow clay layer from below (as indicated by artesian conditions in wells 9 and 14, Figure 7) is probably the most important hydrologic factor supporting the Sarracenia. For example, the numbers of pitcher plant clumps were lower in 2002 (compared to previous monitoring, 1990–1994), then increased for 5 years until dropping again in 2007–2008 (Roe and Croll, 2009). This pattern corresponds with precipitation and water-level data also collected at the site—the water table remained near the ground surface for much of the year during wetter years (e.g., 2003–2006), while the “clay wetting period” was shorter during dryer years (2000–2001, 2007–2008). Groundwater discharge through the clay layer was mostly offset by evapotranspiration, even in wetter seasons and years, allowing wetland flora to flourish in a wetland that is not wet.

This conceptual model is consistent with a “hypocrene spring,” which Springer and Stevens (2009) define as follows: …springs in which groundwater levels come near, but do not reach the surface … discharge from the springs is low enough that evaporation or transpiration consumes all discharge and there is no surface expression of the water.

364

Wetland Management TNC has actively managed Garland Seep since the early 1990s and has documented a dramatic increase in Sarracenia clumps, pitchers, and flowers (Figure 4).

While there may be several contributing factors to the success of TNC management, the recognition of Garland Seep as a hypocrene spring provides a hydrologic mechanism with which to explain the thriving Sarracenia population. First, TNC recognized that drain tiles running below the site were likely drying the wetland. The manner in which the drain tiles functioned, though, turned out to be unusual: rather than draining excess water downward from the soil surface, as in a typical wetland, the tiles instead intercepted artesian flow before it could reach the shallow soil or ground surface above. Plugging the drain tile outlets and excavating the gravel beds around the tiles had the effect of allowing more water to reach the root zone.

TNC also recognized the historic importance of fire at Garland Seep and has burned the site every 2 to 5 years since re-introducing controlled fire in 1992 (see markers at the top of Figure 7). The greatest observed hydrologic impact of these fires followed a prescribed burn on 29 March 2015. When we returned to the site on 1 May 2015, we observed standing water at the ground surface and thousands of new pitcher leaves (Figure 8). Precipitation had been below average every month so far that year, and the rise in water levels following the burn was unmistakable (see wells 1 and 14 in Figure 7). The obvious explanation is that the burn reduced ET (at least initially) and allowed upwelling groundwater to reach the surface. The reasons for which the 2015 prescribed fire had such clear effects on ET and water levels compared to those conducted previously are not as obvious. Looking at the historical record, standing water would have also been observed in the vicinity of well 1 following burns in 1997, 2005, 2009, and 2011. These burns coincided with relatively wet years—years in which P-ET exceeded 20 in.

(50 cm)—but none resulted in such a significant or long-lasting rise in water levels as observed in 2015.

TNC had recently hired a new stewardship manager prior to the 2015 burn, and the firing techniques may have been different than when TNC relied on partners to lead burns. The new manager also emphasized aggressive removal of competing hardwoods, which would have reduced overall ET and possibly intensified the impact of prescribed fire on the undergrowth.

Perhaps the weather conditions simply allowed for a particularly effective burn. In any case, periodic prescribed fires at Garland Seep may help preserve the Sarracenia population by both keeping an open canopy and reducing ET to keep water levels near the surface.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 359–366

EEGS-26-02-07_1XO

July 8, 2020

8:35

Hydrology of a Hypocrene Spring-Fed Fen

Figure 8. New Sarracenia oreophila leaves on May 1 (left) and May 22 (right) following 29 March 2015 prescribed fire.

CONCLUSIONS Garland Seep is a classic example of a hypocrene spring. Discharge through the clay layer was low and was typically exceeded by ET. The water table rarely reached the ground surface, but the clay layer was able to retain moisture and support federally endangered Sarracenia oreophila and other wetland flora. Wetland management by TNC has included prescribed fire and woody vegetation removal. These activities were not only used to combat succession and open the canopy but they have preserved—and even enhanced—site hydrology by reducing ET above the seep. The result has been a wetland with frequent standing water and a thriving Sarracenia population.

ACKNOWLEDGMENTS Funding for this project was provided by the U.S. Fish and Wildlife Service and The Nature Conservancy. Both organizations also contributed countless personnel and volunteer hours to conduct the prescribed burns, pitcher plant counts, and other wetland management activities described in this article. We thank Margit Bucher, Philip Croll, Mamie Colburn, Jennifer Lamb, Andrew Roe, Robert Sutter, and Karen Vaughn (TNC–North Carolina); Malcom Hodges and Erick Brown (TNC–Georgia); Mara Alexander, Anita Goetz, Gary Peeples, Rebekah Reid, and Carolyn Wells Swed (USFWS); Bill Champion and Chad Cook (U.S. Forest Service); and Carrie Radcliff (Atlanta Botanical Garden). We particularly want to recognize the efforts of a neighbor caretaker who not only watched over the site but who diligently recorded water

levels (almost) every week from 1996 through 2012. Finally, we thank Douglas Wilcox and three anonymous reviewers for their comments and suggestions for improving the manuscript.

REFERENCES Allen, R. G.; Pereira, L. S.; Raes, D.; and Smith, M., 1998, Crop Evapotranspiration—Guidelines for Computing Crop Water Requirements: FAO irrigation and drainage Paper 56.

Rome, Italy: Food and Agriculture Organization of the United Nations.

Braun, D., 1994, TNC Memorandum to Margit Bucher on “Riverbend” and “Garland” Preserve Reports.

Braun, D., 1997, TNC Memorandum to Margit Bucher, Comments on “Garland” Seep report (1996).

Richardson, C. J. and Huang, Y. P., 1996, An Ecological Analysis of “Garland” Seep, Clay County, North Carolina Before and After Restoration of Hydrologic Conditions: Final report submitted to the Plant Conservation Program, North Carolina Department of Agriculture, 66 p.

Roe, A. and Croll, P., 2009, Monitoring Summary of Green Pitcher Plant (Sarracenia oreophilia) at “Garland” Seep: Report prepared for The Nature Conservancy, 7 p.

Schwartzman, E., 2016, Botanical Inventory and Invasive Plant Assessment of “Garland” Seep: Report prepared for The Nature Conservancy, 44 p.

Springer, A. E. and Stevens, L. E., 2009, Spheres of discharge of springs: Hydrogeology Journal, Vol. 17, No. 83, https://doi.org/10.1007/s10040-008-0341-y.

State Climate Office of North Carolina (SCONC), NC State University, 2019, CRONOS: Electronic document, available at http://climate.ncsu.edu/cronos/ The Nature Conservancy (TNC), 2010, Hydrologic Summary of “Garland” Seep, Clay County, 17 p.

U.S. Fish and Wildlife Service (USFWS), 2011, Green Pitcher Plant: Sarracenia oreophila: Electronic document,

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 359–366

365

EEGS-26-02-07_1XO

July 8, 2020

8:35

Wilcox, Marino, Warwick, and Sutton available at https://www.fws.gov/southeast/pdf/fact-sheet/ green-pitcher-plant.pdf USFWS, 2019, Mountain Bogs National Wildlife Refuge: Electronic document, available at https://www.fws.gov/refuge/ mountain_bogs/ Warwick, A., 2019, “Garland” Seep Preserve Draft Management Plan: The Nature Conservancy, 46 p.

366

Weakley, A. S. and Schafale, M. P., 1994, Non-alluvial wetlands of the southern Blue Ridge—Diversity in a threatened ecosystem: Water, Air, and Soil Pollution, Vol. 77, pp. 359– 383.

Wilcox, J. D., 2012, Hydrologic controls on the presence and distribution of Sarracenia oreophilia at “Garland” Seep: Final report to The Nature Conservancy, 27 p.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 359–366

EEGS-26-03-02_1XO

July 8, 2020

14:20

Geochemical Variability in Karst-Siliciclastic Aquifer Spring Discharge, Kaibab Plateau, Grand Canyon ALEXANDER J. WOOD ABRAHAM E. SPRINGER* BENJAMIN W. TOBIN School of Earth and Sustainability, Northern Arizona University, Box 4099, Flagstaff, AZ 86011

Key Terms: Geochemistry, Modeling/Statistics, Hydrogeology, Geological Process, Infiltration ABSTRACT The source area of groundwater for springs discharging from lithologically variably perched aquifers is essential to understand when establishing baseline aquifer characteristics. Stratigraphic data from hydrostratigraphic outcrops and geochemical data from springs were used to characterize the hydrogeology of a remote, data-poor aquifer. This study focuses on the hydrogeological variability within the shallow karst-siliciclastic Coconino (C) aquifer on the Kaibab Plateau, north of Grand Canyon National Park. Stratigraphic data were collected from 8 locations, and 22 C aquifer springs were sampled for 18 months. Stable isotope analyses indicate that groundwater is biased to winter recharge in the form of snow and shows similar isotopic signature for groundwater storage areas for all C aquifer springs. Stratigraphic analyses show that the primary water-bearing unit in the C aquifer thins dramatically from south to north and has evaporite lithofacies directly above the unit. Principal component analysis (PCA) indicates that the hydrogeochemistry is influenced by SO4 2− , Cl− , Mg2+ , Ca+ , specific conductivity, alkalinity, and δD variability. The stratigraphic variability influences geochemistry at multiple locations and has geochemical variabilities that correlate with changing lithology. Based on the PCA results, groundwater subbasins were delineated based on geochemical variability. This study provides new analytical tools for land managers and karst hydrogeologists to evaluate lithologically complex aquifers by evaluating the stratigraphy and with high-resolution data. Cost-effective stratigraphic analyses and high-resolution spring sampling can and should be used to evaluate lithologically complex aquifers in remote, data-poor regions.

*Corresponding author email: abe.springer@nau.edu

INTRODUCTION When evaluating a mixed karst-siliciclastic aquifer, it is important to understand how spatially variable hydrostratigraphic units impact the physical and chemical behavior of the aquifer system (Kohlhepp et al., 2017). Understanding groundwater flow paths are especially important when establishing connectivity between recharge areas and springs within laterally extensive, shallow (<200 m thick) aquifers. If soil cover is minimal, then geochemical variability may be a attributed to dissolution of soluble hydrostratigraphic units (Goldscheider and Drew, 2007; Gleeson et al., 2009; and Kohlhepp et al., 2017). The presence of distinct, regionally extensive hydrostratigraphic units such as evaporite and/or carbonate karst units render solute analyses a simple and cost-effective tool to evaluate how groundwater flows through these types of lithologically complex aquifer systems (Worthington and Ford, 1995; Goldscheider et al., 2008). Lithological variability can be used as a natural geochemical marker for complex groundwater flow systems and can efficiently delineate groundwater catchment areas in a shallow aquifer system (Goldscheider et al., 2008) The aquifer properties of karst and siliciclastic units are a result of hydrostratigraphic unit thickness, but aquifer heterogeneity is heavily dependent on the presence of soluble units (Schwartz and Zhang, 2003; Hartmann et al., 2019). Consequently, aquifer properties such as transmissivity and hydraulic conductivity can be orders of magnitude different between siliciclastic and karst hydrostratigraphic units (Cook et al., 2003; Ford and Williams, 2007). If regional groundwater flow to springs interacts with unique karst lithologies, then the resulting spring geochemistry can be used to delineate relative groundwater catchments and flow paths (Goldscheider et al., 2008). For the purpose of this study, karst-siliciclastic aquifers are defined as having unique karst and siliciclastic hydrostratigraphic units that are greater than 5 m thick and are laterally extensive. This study used springs to determine the behavior of a mixed karst-siliciclastic aquifer by determining how (1) recharge varies spatially and (2)

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

367

EEGS-26-03-02_1XO

July 8, 2020

14:20

Wood, Springer, and Tobin

how lithological variability locally impacts groundwater geochemistry. This study contributes to the global understanding of karst-siliciclastic aquifers through this case study of the classic Coconino (C) aquifer of the Kaibab Plateau of the Grand Canyon.

BACKGROUND In the Grand Canyon physiographic region, there is a shallow karst-siliciclastic aquifer overlying a deeper karst aquifer (Tobin et al., 2017). The upper aquifer unit is the C aquifer, an aquifer system with complex lithology comprised of sandstone, shales, evaporite, and carbonate units. The deeper karst aquifer unit is the Redwall-Muav (R) aquifer, which is deeply buried below the C aquifer by 300 m of variable siliciclastic units. Throughout northern Arizona, the R and C aquifers are used for municipal water supply and are ubiquitous throughout the region (Monroe et al., 2004; Bills et al., 2007; and Pool et al., 2011). However, localized hydrological studies often have limited applicability when addressing the spatial heterogeneity in northern Arizona aquifers because of dissimilarities in regional structure, the incision of the Colorado River, and lateral variations in the R and C aquifers’ hydrostratigraphic units (Bills et al., 2007; Pool et al., 2011; and Beisner et al., 2017). For example, in areas adjacent to the Grand Canyon, the C aquifer is generally unsaturated to partially saturated, while the R aquifer is fully saturated along deformation zones (Huntoon, 1970, 2000; Monroe et al., 2004; and Pool et al., 2011).

The majority of the Grand Canyon’s perennial tributaries to the Colorado River are spring-fed streams originating from R aquifer springs (Tobin et al., 2017).

KAIBAB PLATEAU STUDY AREA The Kaibab Plateau is located to the north of the Grand Canyon in Arizona, USA. The Kaibab Plateau is a high-elevation, semi-arid to sub-humid montane to sub-alpine environment with an aerial extent of 2,560 km2 (Figure 1). The Kaibab Plateau has bimodal seasonal precipitation distribution and receives approximately 652 mm of precipitation annually during summer monsoons and winter snow (PRISM Climate Group, 2004). The Kaibab Plateau is dominated by mixed-conifer forests and had widespread logging in the early 1900s (O’Donnell et al., 2018). The Kaibab Plateau is dually managed by the Kaibab National Forest and Grand Canyon National Park. The southern portion of the Kaibab Plateau by the North Rim Lodge has an abrupt escarpment of 1,500 m from the incision of the Colorado River. The western (Rainbow Rim) and eastern (East Rim) boundaries are marked by prominent monoclines and low plateaus that are

368

bound by the incision of Kanab Canyon and Marble Canyon, respectively (Figure 1). Directly to south and across the Grand Canyon lies the South Rim and the Coconino Plateau.

Bedrock and Structural Geology of the Kaibab Plateau The Kaibab Plateau’s C aquifer hydrostratigraphic units are the Hermit Formation (lower confining unit), the Coconino Sandstone, Toroweap Formation, and the Kaibab Formation (Figure 2). The thickness of the C aquifer’s hydrostratigraphic units (Permian, 298.9– 251.9 Ma) have lateral variations that were produced during deposition by tectonic instability and the encroaching Kaibab Sea (Blakey and Ranney, 2008). The Hermit Formation is a mudstone to very fine-grained sandstone and is an aquiclude in all localities except where it is fractured (Huntoon, 1970). The Coconino Sandstone is a very fine- to medium-grained quartzose sandstone and has primary, intergranular porosity with some secondary porosity from dolomite cement dissolution (Middleton et al., 1990). The Toroweap Formation is comprised of repeating packages of mudstone, sandstone, carbonate, and evaporite mudstone beds (Turner, 2003). Although some intergranular porosity exists in siliciclastic lithologies of the Toroweap Formation, the majority of porosity exists from the dissolution of carbonates and evaporite karst lithologies (Huntoon, 1970; Turner, 2003). The Kaibab Formation is comprised of packages of fossiliferous limestone and dolomite and has secondary porosity from carbonate dissolution (Sorauf and Billingsley, 1991). Because the Kaibab Formation has experienced extensive mechanical and chemical erosion, it crops out only along the rim of the Grand Canyon and along high ridges in the interior of the Kaibab Plateau (Huntoon, 1970).

The Kaibab Plateau is a Cretaceous aged, north– south- to northwest–southwest-oriented uplift that is bound by monoclines and is typical of Laramide orogeny deformation (Billingsley, 2000; Davis and Bump, 2009). These two monoclines, the East Kaibab and Crazy Jug monoclines, form the eastern and western margins of the Kaibab Plateau, respectively, and have between 540 and 910 m of structural relief from adjacent plateaus (Figure 1). Between the two monoclines, there are five major north–south to northwest– southwest Neogene-aged extensional normal faults that are tectonically related to the eastern transition zone of the Basin and Range Province (Figure 1) (Huntoon, 1970; Billingsley, 2000). Both Cretaceousand Neogene-aged deformation events created structural geometries that orient groundwater flow and have produced the Grand Canyon region’s most rapid karst

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

EEGS-26-03-02_1XO

July 8, 2020

14:20

C Aquifer

Figure 1. Overview of the Kaibab Plateau. (A) Map showing the location of the Kaibab Plateau in the Grand Canyon physiographic region.

(B) Digital elevation model of the Kaibab Plateau with the perimeter of the Kaibab Plateau marked in a black-dashed line. Similar spring discharge areas are denoted by colored ellipses on the map, and these were determined later in the study by statistical analyses using stable isotope, water quality, and cation/anion data. Structural geology, spring locations, and digital elevation models are modified from their original versions (Billingsley, 2000; Ryan et al., 2009; and Ledbetter et al., 2014).

groundwater flow regime (Huntoon, 2000; Hill and Polyak, 2010; and Jones et al., 2017).

C Aquifer Properties and Kaibab Plateau Groundwater Flow The Kaibab and Toroweap formations are heavily karstified with nearly 6,000 identified collapse features (Jones et al., 2017). Except for regional spring characterizations (Beisner et al., 2017; Tobin et al., 2017) and karst aquifer vulnerability (Jones et al., 2019), almost all previous hydrogeological research on the Kaibab Plateau has focused on the deeper R aquifer.

Elsewhere in the northern Arizona physiographic region, C aquifer studies have noted highly variable aquifer properties attributed to porosity differences that arise from variability in (1) C aquifer hydrostratigraphic unit, (2) degree of fracturing, and/or (3)

secondary mineralization (Hart et al., 2002; Monroe et al., 2004; Bills et al., 2007; and Pool et al., 2011).

Traditional aquifer property tests have highlighted the C aquifer’s variability with porosity varying from 0.01 to 0.50 percent, specific yields ranging from 0.00001 to 3, storage coefficients between 0.005 and 0.01, transmissivities between 0.9 and 845 m2 /d (10–9,100 ft2 /d), and hydraulic conductivity values from 0.06 to 2.1 m/d (0.019–6.88 ft/d) (Hart et al., 2002; Bills et al., 2007; and Pool et al., 2011). Locally, aquifer properties can be one or two orders of magnitude higher, such as on the Defiance Plateau in eastern Arizona, which is a Laramide-style uplift and is structurally similar to the Kaibab Plateau (Cooley et al., 1969). Lower specific yield, storage coefficient, transmissivity, and hydraulic conductivity values are generally in less fractured portions of the aquifer, while higher values are associated with more fractured aquifers (Hart et al., 2002; Pool et al., 2011).

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

369

EEGS-26-03-02_1XO

July 8, 2020

14:20

Wood, Springer, and Tobin

METHODS To evaluate the heterogeneity of a karst-siliciclastic aquifer, quasi-monthly water sampling was collected at 22 spatially and vertically representative C aquifer springs over a period of 19 months (Figure 1). Because of difficult winter access, 11 of the springs were sampled only during spring, summer, and fall. Over 19 months, 16 unique field site visits occurred.

Field Methods Water quality data (temperature, pH, alkalinity, and specific conductivity), stable isotopes, dissolved ions, and discharge data were collected during 175 unique site visits to C aquifer springs. To compare the variable emergent conditions for C aquifer springs, each spring was qualitatively characterized by its sphere of discharge (Springer and Stevens, 2009). Discharge magnitude, Q{90} /Q{90} , was determined for springs with multiple measurements according to Springer et al.

(2008), where Q is the 90 and 10 percent quartile of measured discharge.

Geochemistry and Water Quality YSI Professional Plus multiparameter water quality probes (Yellow Springs, OH) were used to measure pH, temperature, and specific conductivity and were calibrated daily throughout the field sampling process using pH buffers and specific conductivity calibration fluids from YSI and HACH (Loveland, CO). Discharge measurements were recorded using volumetric or cutthroat Baski 1- or 4-in. flumes (Englewood, CO) following established methods (±3) (Buchanan and Somers, 1969). Water samples were collected in 30-mL polypropylene bottles and were analyzed for Mg2+ , Ca+ , Na+ , K+ , F− , Cl− , NO2 − , Br− , NO3 , PO4 3− , and SO4 2− in accordance with U.S. Geological Survey (USGS) preservation and storage standards (Wilde et al., 2004). Water sample analyses were conducted using ion chromatography and inductively coupled plasma mass spectrometry by the Arizona Laboratory for Emerging Contaminants at the University of Arizona, Tucson. Alkalinity values were collected as equivalent CaCO3 through direct titrator methods (error ±2 mg/L) in the field using LaMotte Field Alkalinity Test Kits (Chestertown, MA).

Additional water samples were collected for stable isotope analyses of δ18 O and δD. Samples were collected following established USGS stable isotope preservation and storage procedures (Hem, 1985). Water samples were processed at the Colorado Plateau Stable Isotopes Laboratory in Flagstaff, Arizona, and were analyzed for O and H isotopes using Los Gatos

370

Research’s Off-Axis Integrated Cavity Output coupled to a CTC LC-PAL liquid autosampler for simultaneous D/H and 18 O/16 O ratio measurements of H2 O. Isotope values were reported relative to Vienna Standard Mean Ocean Water.

Physical Hydrology To assess the lateral variability and thickness of karst and siliciclastic units, nine stratigraphic columns were produced near springs with adequate outcrops that spatially highlighted the depositional variability of the Coconino Sandstone and Toroweap Formation.

Stratigraphic columns included the Hermit Formation, the Coconino Sandstone, and the Toroweap Formation. Stratigraphic columns were produced beginning with the Hermit Formation to midway through the Toroweap Formation, and the vertical resolution ranged from 12 to 62 m. Stratigraphic columns were prepared by hand according to modern stratigraphic techniques and were later digitized (Northern American Commission on Stratigraphic Nomenclature, 2005). A modified Folk–Dunham classification was used to evaluate siliciclastic and carbonate lithofacies (Folk, 1959; Dunham, 1962; and Leeder, 1999).

Field stratigraphic columns were digitized and combined into a correlation plate to compare C aquifer unit changes from south to north on the Kaibab Plateau.

Stable Isotopes δD and δ18 O Because summer and winter meteoric water are isotopically unique, a local meteoric waterline (LMWL) was produced to understand precipitation seasonality. This was accomplished using linear regression of δ18 O versus δD from meteoric water data from past Kaibab Plateau studies (Ross, 2005; Brown, 2010) and from this study. For this study, meteoric water was collected by all-weather, plastic rain gauges with mineral oil to prevent evaporation at three different elevation locations through the duration of spring sampling. To understand how winter meteoric water evolves during snowmelt, sinkhole runoff samples were opportunistically collected at two different locations within DeMotte Park. To evaluate seasonal recharge and regional groundwater storage, plots of δ18 O versus δD using C aquifer groundwater from springs, meteoric water, and sinkhole surface water runoff were used to highlight the evolution of source to sink water isotope.

To distinguish between Pacific Ocean, winter-derived, and tropical, summer monsoonal moisture sources, the deuterium excess value (d-excess) was calculated using

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

EEGS-26-03-02_1XO

July 8, 2020

14:20

C Aquifer

δ18 O and δD (Dansgaard, 1964): d-excess = δD8 × δ O 18

(1)

The d-excess equation is also the equation for the global meteoric waterline (GMWL) and is more sensitive than analyses of δD versus δ18 O when evaluating moisture source areas (Sharp, 2006).

are invariably correlated, it is essential to maintain a diverse suite of ions to drive stronger diversity in the geochemical data set variation in order to understand how spatially unique lithology impacts groundwater geochemistry.

RESULTS

Geochemical Analysis

Overview

Because geochemical speciation techniques were not used in this study to analyze dissolved ions, only samples with pH values near neutral (6.7–8.7) were included in this study so that we could assume alkalinity as CaCO3 = HCO3 (Hem, 1985). Of the original 159 samples, only 4 samples had pHs less than 6.7, and these samples were removed. After removing low pH sites from the analysis, the remaining sites had pH ranges of 6.75 to 8.75 and an average of 7.44. A charge balance error (CBE) was used to evaluate the quality of the laboratory analyses, field collection, and ion sampling for the remaining samples:

( cations − anions ) × 100 (2) ( cations + anions )

Springs in this study discharge from either the Coconino Sandstone (10 springs) or the Toroweap Formation (12 springs) (Table 1). Springs located near each other all discharged from the same geologic unit.

Springs had an average elevation of 2,348 m, and springs had an average of seven unique site visits. C aquifer springs exhibit a range of spheres of discharge, including 13 hillslope, three gushet, two helocrene, and one rheocrene spring (Springer et al., 2008). Hillslope springs are the most common sphere of discharge because erosional processes have exposed the lower confining layers in the C aquifer’s hydrostratigraphic units.

On completion of CBE analysis, 24 of the remaining 155 samples had CBEs exceeding the ±10 CBE error quality control threshold for low-ionic-strength waters (Fritz, 1994). Once these high-CBE samples were removed, CBE had an average absolute value of 3.22 for the remaining 131 samples (Wood, 2019).

To describe groundwater flow in the C aquifer, geochemistry and water isotope data were evaluated by geologic discharge unit and Kaibab Plateau discharge area (Figure 1). Using RStudio (RStudio Team, 2015), analysis of variance (ANOVA) was used to identify significant differences, and, when present, the Tukey honest significant difference (HSD) test was run to determine which groupings were statistically significant (Wickham and RStudio, 2017). To assess in what areas of C aquifer groundwater storage and flow occur, the data set was evaluated with principal component analysis (PCA). PCA was used to reduce the complexity of the large C aquifer data set while maintaining relevant trends and patterns (Lever et al., 2017). Input variables used in this PCA were normally distributed and then z-scored prior to analysis. To fully capture the inherent geochemical variability of a mixed-lithology aquifer with evaporite (gypsum, halite, and anhydrite) and carbonate (dolomite and limestone) karst units, 16 of the original 19 variables were used in the PCA. These variables included discharge, pH, specific conductivity, temperature, Mg2+ , Ca+ , Na+ , K+ , F− , Cl− , Br− , NO3 , SO4 2− , δ18 O, δD, and alkalinity as CaCO3 .

While some of these analytes (such as Na+ and Cl− )

A synthesis of past and current Kaibab Plateau δ18 O and δD data indicate that C aquifer groundwater isotope data lie between the LMWL and GMWL (Figure 3a). Meteoric water samples range from δ18 O −17.46‰ to −3.36‰ and δD −123.49‰ to −46.71‰.

C aquifer groundwater and sinkhole runoff isotope values overlap with heavier (less negative) values. C aquifer groundwater ranges from δ18 O −14.56‰ to −12.16‰ and δD −99.75‰ to −84.31‰. Sinkhole runoff isotope values are between meteoric precipitation and groundwater and vary from δ18 O −12.9‰ to −10.45‰ and δD −92.87‰ to −73.62‰. Plots of averaged C aquifer δ18 O versus δD values grouped by area and unit indicate groundwater mixing processes within C aquifer groundwater storage (Figure 3).

To evaluate spatial and seasonal variability of C aquifer groundwater, ANOVA was tested on all δ18 O, δD, and d-excess values (Table 2). Results indicate that there are significant differences between Kaibab Plateau areas for δ18 O (F4,154 = 14.38, p < 0.01), δD (F4,154 = 48.22, p < 0.01), and d-excess (F4,154 = 6.24, p < 0.01). Grouping by season (winter: December–February; spring: March–May; summer: June–August; autumn: September–November) yielded significant differences for δ18 O (F3,155 = 3.65, p = 0.01) and d-excess (F3,155 = 3.67, p < 0.05) and insignificant differences for δD (F3,155 = 2.39, p > 0.05). Tukey HSD indicates that Mangum and Grand Canyon National Park areas are statistically different from other discharge areas (Table 2).

Stable Isotopes δD and δ18 O

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

371

EEGS-26-03-02_1XO

July 8, 2020

14:20

Wood, Springer, and Tobin Table 1. Twenty-two Coconino aquifer springs including unit of discharge, discharge area, sphere of discharge, number of site visits, elevation, and Meinzer classification. See Figure 1 for field map area and for the locations of springs. Note that a higher discharge magnitude indicates a more variable discharge pattern, while a lower discharge magnitude represents a less variable discharge pattern (Meinzer, 1923; Springer and Stevens, 2009).

Spring Name

Unit of Discharge

Mangum Nail Oak Castle Riggs North Canyon Big Moquitch Mourning Dove South Canyon Parissawampitts Rainbow Timp Lower Thompson Upper Thompson Bright Angel Robbers Roost Watts Quaking Aspen Oquer Stina Two Pasture

Coconino Coconino Coconino Coconino Coconino Coconino Coconino Coconino Coconino Coconino Toroweap Toroweap Toroweap Toroweap Toroweap Toroweap Toroweap Toroweap Toroweap Toroweap Toroweap Toroweap Toroweap

Discharge Area

Sphere of Discharge

Big Big Big Big Big East Rim Big Mangum Mangum East Rim Rainbow Rainbow Rainbow GRCA GRCA GRCA GRCA Rainbow Rainbow Rainbow Rainbow Rainbow Rainbow

Gushet/rheocrene Gushet Hillslope Hillslope Helocrene Gushet Gushet Rheocrene Hillslope Gushet Helocrene Hillslope Hillslope Hillslope Hillslope Gushet Hillslope Hillslope Hillslope Hillslope Hillslope Rheocrene Hillslope

Discharge Magnitude

Site Visits

Elevation (m)

4th 3rd 2nd 3rd 3rd 4th 5th 2nd 2nd 2nd 2nd 2nd 3rd 2nd 2nd No flow 4th 2nd 3rd No flow 2nd 2nd 2nd

13 10 9 15 8 11 27 3 2 2 7 7 7 8 8 4 9 7 7 4 7 2 2

2,183 2,163 2,054 2,195 2,210 2,485 2,150 2,160 2,208 2,577 2,366 2,455 2,416 2,547 2,556 2,471 2,521 2,442 2,398 2,550 2,414 2,315 2,412

GRCA = Grand Canyon National Park.

Physical Hydrogeology Stratigraphic analyses indicate immense variations of C aquifer hydrostratigraphic units from north to south (Figure 4). Dramatic thinning was observed in the Coconino Sandstone in areas where evaporite lithofacies of the Toroweap Formation became more widespread. The Hermit Formation’s contact

with the overlying Coconino Sandstone is marked by 1- to 5-cm-thick laminated carbonate mudstone beds (Huntoon, 1970).

The Coconino Sandstone varies in thickness from 5 m by Warm Spring to 40 m by North Canyon Wash. The Coconino Sandstone has planar tabular cross-beds and grain sizes that vary from very fine to medium (Figure 4). The basal Coconino Sandstone

Figure 2. The hydrostratigraphy of the Coconino aquifer.

372

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

EEGS-26-03-02_1XO

July 8, 2020

14:20

C Aquifer

Figure 3. δ18O and δD of Coconino aquifer groundwater and Kaibab Plateau precipitation. (A) δ18O and δD groundwater (blue), sinkhole runoff (green), and meteoric (yellow, rain and snow). Data are from this study, Ross (2005), Brown (2010), and Schindel (2015). Averaged δ18O versus δD by (B) area and (C) unit. Error bars are the standard mean error of each site’s averaged data δ18O versus δD.

Table 2. Tukey honest significant difference test results showing significant differences (p < 0.05) between δ18 O, δD, and d-excess by area and season. This comparison between areas with p < 0.05 are marked as “Different,” while areas with p>0.05 are marked as “Similar.”

By area East Rim–Big GRCA–Big Mangum–Big Rainbow–Big GRCA–East Rim Mangum–East Rim Rainbow–East Rim Mangum–GRCA Rainbow–GRCA Rainbow–Mangum By season Spring–autumn Summer–autumn Winter–autumn Summer–spring Winter–spring Winter–summer

δ18 O

δD

d-Excess

Similar Similar Different Similar Different Different Similar Similar Different Different

Similar Different Different Similar Similar Different Similar Different Similar Different

Similar Similar Similar Similar Different Different Similar Similar Similar Similar

Similar Different Similar Similar Similar Different

Similar Similar Similar Similar Similar Similar

Similar Different Similar Similar Similar Different

GRCA = Grand Canyon National Park.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

373

EEGS-26-03-02_1XO

July 8, 2020

14:20

Wood, Springer, and Tobin

Figure 4. A correlation plate showing the stratigraphic sections from eight locations on the Kaibab Plateau. These locations were chosen for their proximity to springs and for lithologically representative lithofacies of Coconino aquifer hydrostratigraphic units. The Coconino Sandstone thins from 40 to 5 m from south to north. The Toroweap Formation has numerous packages of mudstone, siliciclastic, and carbonate lithofacies, resulting in a perched, groundwater flow setting.

is generally marked with open fractures, while the top varies from open to calcite-filled fractures. The Toroweap Formation is composed of repeating packages of siliciclastic and carbonate lithofacies, occasionally having evaporite lithofacies (Figures 4 and 5).

Evaporite facies are prevalent in the northern portion of the Kaibab Plateau but not in the western or southern portions. When present, evaporite lithofacies are found directly above the Coconino Sandstone and are part of the Toroweap Formation (Figure 4). Evaporite facies were identified as having weak bedding, being non-carbonate, and having variable crystal (halite) intergrowth (Figure 5). Within the Toroweap Formation, carbonate, mudstone, and siliciclastic facies are present in repeating packages and vary in size from centimeters to meters thick. Karst occurs along pri-

374

mary fracture sets in packages of carbonate with underlying and confining mudstone lithofacies (Figure 5). In these locations, mudstone facies have calcitesealed fractures, while carbonate facies have open fractures.

Geochemical Results Principal component (PC) 1 and PC 2 represent 57.1 percent of the variation in the data (Figure 6 and Table 3). Therefore, all data were re-projected on PC 1 and PC 2 modes of variability (Figure 6). Because PC 1 and PC 2 do not account for all data set variability, PC 3 is also compared in our results. Including PC 3, PC 1, PC 2, and PC 3 account for 66.1 percent of variability from this data set (Table 3). PC 1 is dominated

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

EEGS-26-03-02_1XO

July 8, 2020

14:20

C Aquifer

Using PC 1, PC 2, and PC 3, ANOVA determined statistically significant differences when grouped by area along PC 1 (F4,93 = 29.5, p < 0.01), PC 2 (F4,93 = 117.3, p < 0.01), and PC 3 (F4,93 = 12.3, p < 0.01).

For PC 1 and PC 2, Tukey HSD indicates that the Mangum and Rainbow areas are different from other areas (Table 5). Geochemistry is different by unit in PC 1 (F1,96 = 12.3, p < 0.01) and PC 2 (F1,96 = 53.1, p < 0.01) and is insignificant in PC 3 (F1,96 = 1.8, p > 0.05).

Seasonal groupings showed insignificant results for PC 1 (F3,94 = 0. 8, p > 0.05), PC 2 (F3,94 = 0.8, p > 0.05), and PC 3 (F1,94 = 0.6, p < 0.01).

DISCUSSION

Figure 5. An annotated photo of a Toroweap Formation stratigraphic section by Mangum Camp. Note the repeating packages of carbonate (limestone: micrite and wackestone) and siliciclastic (mudstone evaporate) lithologies. Toroweap Formation fractures are open in karst layers and are closed in siliciclastic layers. This creates miniature perched groundwater horizons in the Toroweap Formation.

by the positive loadings of SO4 2 (0.41), Cl− (0.40), and Na+ (0.41), while PC 2 is dominated by the negative loadings of alkalinity as CaCO3 (−0.42), specific conductivity (−0.39), and Ca+ (−0.38) (Figure 6 and Table 4). PC 3 (not shown in Figure 6) is dominated by negative loading of δ18 O (−0.61) (Table 4).

These results show that dissolution in spatially variable evaporites and carbonate karst units are driving the 57 percent of the variability in the water chemistry data set (Table 3 and 4). Evaporite dissolution, represented by SO4 2+ , Cl− , and Na+ in PC 1, is a primary driver in data set variability, while carbonate dissolution, represented by Ca2+ , alkalinity, and specific conductivity in PC 2, is a secondary driver of data set variability. δ18 O, represented by PC 3, accounts for 9 percent of data set variability and can be related to heterogeneity of recharge to unique Kaibab Plateau discharge areas.

PC 1 and PC 2 indicated that the primary dissimilarity between C aquifer springs occurs because of variability in SO4 2+ , Cl− , Na+ , Ca2+ , alkalinity as CaCO3 , and specific conductivity values (Figure 6, Table 5 and Table 6). These regional geochemical differences are corroborated by regional changes in evaporite and carbonate C aquifer hydrostratigraphic units (Figures 4 and 6). C aquifer spring geochemistry is indicative of unique water–rock interactions along structurally controlled flow paths from primary recharge areas (Figure 6). However, not all C aquifer springs receive recharge homogeneously, as seen by the variability in δ18 O in the PC 3. For all PCs, variability is statistically insignificant when evaluated by season and is significant when evaluated by discharge area.

Plots of δ18 O versus δD indicate that C aquifer groundwater (δ18 O −14.56‰ to −12.16‰ and δD −99.75‰ to −84.31.11‰) is similar to winter precipitation and high-elevation snowmelt runoff into sinkholes (δ18 O −10.45‰ to −12.9‰ and δD −92.87‰ to −73.62‰) (Figure 3). This suggests that C aquifer recharge is biased to winter-snow recharge, which is consistent with past isotopic studies from the Grand Canyon springs aquifer (Beisner et al., 2017; Jones et al., 2017; and Tobin et al., 2017). Considering the C aquifer’s large lateral extent and extensive karstification, spring water isotopes are seasonally homogenized and exhibit ranges of isotopic values that are not consistent with meteoric precipitation samples (δ18 O −17.46‰ to −3.36‰ and δD −123.49‰ to −46.71‰) (Table 2). Based on previous structural geology research (Huntoon, 1970; Billingsley, 2000), the middle and highest portions of the Kaibab Plateau contain large structural catchments that could buffer seasonal, isotopically unique precipitation events and maintain winter-biased groundwater storage. One such structural catchment is proximal to DeMotte Park (Figure 1), which is situated at 2,667 m and is higher than any spring (2,545–2,577 m) included in this study.

Isotopic differences do exist at some C aquifer springs

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

375

EEGS-26-03-02_1XO

July 8, 2020

14:20

Wood, Springer, and Tobin

Figure 6. Principal component analysis (PCA) with all data by Kaibab Plateau Area. (A) PCA grouped by area. The Rainbow area groups along the Principal component (PC) 2 axis (defined by Ca2+ , alkalinity as CaCO3 , and specific conductivity variability), while the Mangum area groups along the PC 1 axis (defined by SO4 2− , Na+ , and Cl− variability. (B) Plan view of the locations of the springs delineated by discharge unit (points) and different discharge areas (ellipses). Note that all discharge areas have one discharge unit. The PCA plot (A) and plan view map (B) are identically colored to view how the Coconino aquifer geochemistry varies across the Kaibab Plateau.

and are theorized to be from secondary storage areas and flow paths (Table 2 and Figure 3).

The greatest spatial change in C aquifer hydrostratigraphy is seen in its primary water-bearing unit, the Coconino Sandstone, as it thins from 45 m in the south to 5 m in the north (Figure 4). This regional change in lithology is corroborated by a profound hydrogeochemical change that is statistically significant in areas with less Coconino Sandstone and more evaporitic deposits (Table 4). Middle, eastern, and southern springs have minimal dissolved ion concentrations and are likely connected to recharge areas via flow paths that are entirely contained within the Coconino Sandstone (Figures 1 and 2). Northern C aquifer groundwater has higher SO4 2+ , Cl− , and Na+ and are likely connected to recharge areas with flow paths through Coconino Sandstone and overlying evaporite lithologies of the Toroweap Formation

(Figures 1, 2, and Table 6). Western C aquifer springs discharging from the Toroweap Formation have higher Ca2+ , alkalinity as CaCO3 , and specific conductivity, which indicates horizontal flow through carbonate lithologies (Table 6). Because the Kaibab Formation is heavily eroded and largely absent in these areas, the carbonate geochemistry of western C aquifer springs is likely derived solely from groundwater flow in the Toroweap Formation. Within the Toroweap Formation, groundwater can remain perched between repeating packages of mudstone and carbonate units (Figures 5 and 7). Groundwater flow infiltrates the perched aquifer setting within the Toroweap Formation as a result of offset along normal faults along primary Coconino Sandstone groundwater flow paths (Figure 8).

It is likely that groundwater remains perched within the Toroweap Formation until deformation zones allow groundwater to transfer back into the underly-

Table 3. Fraction of variance (%) for 12 of the 16 principal components (PCs). PC 1 (32%) and PC 2 (25%) account for 57% of the total data set variance. PCs 13–16 were omitted from this table.

PC

% Variance

376

1

2

3

4

5

6

7

8

9

10

11

12

0.32

0.25

0.09

0.08

0.07

0.04

0.04

0.03

0.03

0.01

0.01

0.01

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

EEGS-26-03-02_1XO

July 8, 2020

14:20

C Aquifer Table 4. Eigenvalues for principal component analysis (PCA) by area with eigenvalues for principal component (PC) 1, PC 2, and PC 3.

Note the high positive loadings (>0.40) for PC 1 (bold), the high negative loadings (<−0.37) for PC 2 (italic), and the high negative loading (−0.62) for PC 3 (bold italic).

PC 1

PC 2

PC 3

Variables

−0.08 0.09 −0.04 0.10 −0.03 −0.04 0.41 0.16 0.29 −0.02 0.38 0.40 0.38 0.27 0.41 −0.04

0.33 0.11 0.31 − 0.40 − 0.21 − 0.37 0.10 − 0.30 0.05 − 0.39 − 0.06 0.08 0.00 − 0.01 0.03 − 0.42

0.24 − 0.22 0.16 0.25 − 0.62 − 0.33 − 0.05 0.41 − 0.02 0.01 − 0.05 − 0.11 0.08 0.18 − 0.14 0.28

Discharge Temperature pH Specific conductivity δ18 O δD Na− Mg2+ K+ Ca2+ Fv− Cl− Br− NO3 − SO4 2− Alkalinity

ing Coconino Sandstone (Figure 7). Toroweap Formation springs, especially those by the Rainbow Rim, are all interpreted to be a result of this type of perched flow within carbonate lithologies because they have

Table 5. Tukey honest significant difference test results showing significant differences (p < 0.05) between areas principal components (PCs). Areas with p < 0.05 are marked as “Different,” while areas with p > 0.05 are marked as “Similar.”

East Rim–Big GRCA–Big Mangum–Big Rainbow–Big GRCA–East Rim Mangum–East Rim Rainbow–East Rim Mangum–GRCA Rainbow–GRCA Rainbow–Mangum

PC 1

PC 2

PC 3

Similar Different Different Different Similar Different Similar Different Similar Different

Similar Different Different Different Similar Different Different Different Different Similar

Similar Different Similar Similar Different Similar Similar Different Different Similar

GRCA = Grand Canyon National Park.

increased Ca2+ , alkalinity, and specific conductivity (Figure 5).

Results from this study suggest that groundwater flow originates in the highest and central portions of the Kaibab Plateau and that groundwater geochemistry is altered by contact with regionally unique lithologies. These results corroborate previous studies and suggest extensive lateral flow through C aquifer

Table 6. Averaged geochemistry data from all 22 sites. Springs are grouped by unit and area (Figure 1). The Mangum discharge area has statistically significant differences in sodium (F1,130 = 59.97, p < 0.01), chloride (F1,130 = 41.64, p < 0.01), and sulfate (F1,130 = 13.20, p < 0.01) (bold). The Rainbow discharge area has statistically significant differences in (F1,130 = 51.01, p < 0.01) and alkalinity as CaCO3 (F1,130 = 72.74, p < 0.01) (italic).

Spring Name Big Castle Riggs North Canyon South Canyon Bright Angel Lower Thompson Robbers Roost Upper Thompson Mangum Moquitch Mourning Dove Nail Oak Parissaw-ampitts Pasture Quaking Aspen Rainbow Stina Timp Two Watts

Discharge Unit Coconino Coconino Coconino Coconino Coconino Toroweap Toroweap Toroweap Toroweap Coconino Coconino Coconino Coconino Coconino Toroweap Toroweap Toroweap Toroweap Toroweap Toroweap Toroweap Toroweap

Discharge Area Big Big Big East Rim East Rim GRCA GRCA GRCA GRCA Mangum Mangum Mangum Mangum Mangum Rainbow Rainbow Rainbow Rainbow Rainbow Rainbow Rainbow Rainbow

Ca2+ (mg/L)

Mg2+ (mg/L)

Alkalinity as CaCO3 (mg/L)

SO4 2− (mg/L)

Cl− (mg/L)

Na− (mg/L)

42.60 42.98 47.21 39.98 46.22 54.37 43.43 52.28 51.42 54.91 61.20 54.65 48.92 52.14 64.92 62.75 66.61 68.60 67.95 59.60 66.51 81.46

22.48 23.62 20.51 23.14 24.29 32.20 19.45 22.74 24.49 31.34 29.40 27.60 27.87 33.15 32.71 35.51 26.36 35.95 32.58 34.02 23.68 27.29

208.29 221.99 210.00 202.96 215.55 244.00 191.57 224.98 214.75 255.32 238.00 233.75 254.68 216.29 297.14 275.76 279.71 317.00 283.75 297.94 239.67 280.29

3.13 4.06 8.21 2.16 2.02 1.78 2.30 4.19 2.69 14.68 24.74 19.85 11.04 67.27 2.70 2.56 2.13 2.56 2.88 2.72 2.63 2.60

2.06 2.35 2.38 1.01 1.61 1.61 1.12 1.19 2.06 5.30 4.90 4.00 4.71 10.71 1.47 2.11 0.96 1.19 1.43 1.25 1.23 1.12

2.45 2.59 2.62 1.41 2.26 1.93 1.34 1.13 1.91 4.13 3.91 3.60 3.59 7.20 1.78 1.99 1.47 1.67 1.61 1.55 1.50 1.65

GRCA = Grand Canyon National Park.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

377

EEGS-26-03-02_1XO

July 8, 2020

14:20

Wood, Springer, and Tobin

Figure 7. Conceptual diagram of groundwater flow on the Kaibab Plateau. A cross-sectional view of the Kaibab Plateau with (A) an east-towest section and (B) a north-to-south section. Many normal faults are present in the east-to-west section, which perches groundwater in the Toroweap Formation, forcing for carbonate–groundwater interactions. In the north-to-south section, the Coconino Sandstone thins to the north, and the Toroweap Formation (especially evaporite facies) thickens to the north.

378

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

EEGS-26-03-02_1XO

July 8, 2020

14:20

C Aquifer

Figure 8. Comparison of hypothesized Coconino aquifer groundwater flow paths developed by this study and previous studies. (A and B) Huntoon (2000). (C) Jones et al. (2017). (D) This study. The blue area is the theorized extent of the structural catchment around DeMotte Park. Data are modified from Huntoon (1970) and Billingsley (2000).

hydrostratigraphic units (Huntoon, 1970, 2000; Beisner et al. 2017; and Jones et al., 2017). Overall, the variability of C aquifer spring variability can be separated into three groundwater sub-basins: (1) higher-Cl− -

SO4 2+ type of groundwater to the north, (2) low-ionic groundwater in the east-central-southern areas, and (3) higher-Ca2+ -alkalinity as CaCO3 groundwater to the west.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

379

EEGS-26-03-02_1XO

July 8, 2020

14:20

Wood, Springer, and Tobin

CONCLUSIONS Isotopic signatures of C aquifer groundwater suggest that groundwater recharge is biased toward higher-elevation, structural catchments on the Kaibab Plateau. Recharge enters the C aquifer through karst features in the heavily eroded Kaibab and Toroweap formations, then moves vertically through deformation structures deeper into the Toroweap Formation and then laterally through the Coconino Sandstone (Figure 8). In the higher-elevation structural catchments of the Kaibab Plateau, it is estimated that the Coconino Sandstone is 30–50 m thick, while the Toroweap Formation is 35–90 m thick (Figures 4 and 8), providing substantial groundwater storage. All springs have baseline carbonate geochemistry that indicates water–carbonate interactions as recharge moves through the Kaibab Formation and is stored in carbonate portions of the Toroweap Formation. The active flow paths of all springs then move recharge laterally through the siliciclastic Coconino Sandstone and occasionally through perched groundwater horizons of the Toroweap Formation. Ion geochemistry suggests that differences in spring geochemistry are inherited from lithological variations along structurally and lithologically controlled flow paths.

Both western and northern groundwater flow are significantly impacted by the water–rock interaction in the Toroweap Formation; however, there are little to no evaporite facies present in the western portions, resulting in the observed differences in SO4 2+ , Na+ , and Cl− ions between these areas. These results indicate that karst groundwater flow paths in a mixed-lithology aquifer system result in regional geochemical variability can be used as a natural tracer.

In lithologically complex aquifer systems, the dominant lithology of groundwater flow paths across the region can be identified using spring geochemical variability coupled with stratigraphic mapping. These analyses show the importance of combining lithology and hydrogeochemistry to better understand flow path dynamics of complex aquifer systems. Future use of these techniques will increase the ability to assess hydraulic connectivity between the perched C aquifer and underlying R aquifer as well as provide insight into other systems with perched and confined/semiconfined groundwater flow paths.

ACKNOWLEDGMENTS Grand Canyon National Park through Cooperative Agreement P17AC00244 (study no. GRCA-00746, permit no. GRCA-204-SCI-0057) of the Colorado Plateau Cooperative Ecosystems Studies Unit provided all of the funding for most of the fieldwork travel

380

and most of the resources for equipment and water quality analyses. Thank you, Dr. Michael Smith and Dr. Rod Parnell, for your assistance and insight into this project and manuscript. Lastly, thank you to the volunteers who helped collect these data over 16 trips and 160 total volunteer days.

REFERENCES Beisner, K. R.; Paretti, N. V.; Tillman, F. D.; Naftz, D. L.; Bills, D. J.; Walton-Day, K.; and Gallegos, T.J., 2017, Geochemistry and hydrology of perched groundwater springs: Assessing elevated uranium concentrations at Pigeon Spring relative to nearby Pigeon Mine, Arizona (USA): Hydrogeology Journal, Vol. 25, p. 539–556, doi:10.1007/s10040-016-1494-8.

Billingsley, G. H., 2000, Geologic Map of the Grand Canyon 30 × 60 Quadrangle, Coconino and Mohave Counties, Northwestern Arizona: U.S. Geological Survey Geologic Investigations Series I-2688.

Bills, D.; Flynn, M.; and Monroe, S., 2007, Hydrogeology of the Coconino Plateau and Adjacent Areas, Coconino and Yavapai Counties, Arizona: U.S. Geological Survey Professional Paper, 116 p.

Blakey, R. C. and Ranney, W., 2008, Ancient Landscapes of the Colorado Plateau: Grand Canyon Association, Grand Canyon, AZ. 155 p. doi:97819346560307.

Brown, C. R., 2010, Physical, Geochemical, and Isotopic Analysis of R-Aquifer Springs, North Rim, Grand Canyon, Arizona: Unpublished Master’s Thesis, Northern Arizona University, 146 p.

Buchanan, T. J. and Somers, W. P., 1969, Discharge Measurements at Gaging Stations: Techniques of Water-Resources Investigations of the United States Geological Survey, Book 3— Applications of Hydraulics, 171 p.

Cook, P. G.; Land, C.; and Osmond, G., 2003, A Guide to Regional Flow in Fractured Rock Aquifers: CSIRO, Canberra, Australia, 151 p.

Cooley, M. E.; Harshbarger, J. W.; Akers, J. P.; and Hardt, W. F., 1969, Regional Hydrogeology of the Navajo and Hopi Indian Reservations, Arizona, New Mexico, and Utah: U.S Geological Survery Professional Paper 521-A. Dansgaard, W., 1964, Stable isotopes in precipitation: Tellus, Vol. 16, pp. 436–468, doi:10.3402/tellusa.v16i4.8993.

Davis, G. H. and Bump, A. P., 2009, Structural geologic evolution of the Colorado Plateau: Geological Society of America, Vol. 1204, pp. 99–124, doi:10.1130/2009.1204(05).

Dunham, R. J., 1962, Classification of carbonate rocks according to depositional textures. In Ham W. E. (Editor), Classification of Carbonate Rocks, Vol. 1: American Association of Petroleum Geologists, Tulsa, OK, 117 p.

Folk, R., 1959, Practical petrographic classification of limestone: American Association of Petroleum Geologists Bulletin, Vol. 43, pp. 1–38.

Ford, D. and Williams, P., 2007, Karst Hydrogeology and Geomorphology: John Wiley & Sons, West Sussex, U.K. Fritz, S. J., 1994, A survey of charge-balance errors on published analyses of potable ground and surface waters: Groundwater, Vol. 32, pp. 539–546, doi:10.1111/j.1745-6584.1994.tb00888.x.

Gleeson, T.; Novakowski, K.; and Kurt Kyser, T., 2009, Extremely rapid and localized recharge to a fractured rock aquifer: Journal of Hydrology, Vol. 376, pp. 496–509, doi:10.1016/j.jhydrol.2009.07.056.

Goldscheider, N. and Drew, D., 2007, Methods in Karst Hydrogeology: Taylor and Francis Group, London, U.K., 279 p.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

EEGS-26-03-02_1XO

July 8, 2020

14:20

C Aquifer Goldscheider, N.; Meiman, J.; Pronk, M.; and Smart, C., 2008, Tracer tests in karst hydrogeology and speleology: International Journal of Speleology, Vol. 37, pp. 27–40, doi:10.5038/1827-806X.37.1.3.

Hart, R. J.; Ward, J. J.; Bills, D. J.; and Flynn, M. E., 2002, Generalized Hydrogeology and Ground-Water Budget for the C Aquifer, Little Colorado River Basin and Parts of the Verde and Salt River Basins, Arizona and New Mexico: U.S. Geological Survey Water-Resources Investigations Report 02-4026, 54 p.

Hartmann, A.; Goldscheider, N.; Wagener, T.; Lange, J.; and Weiler, M., 2019, Karst water resources in a changing world: Review of hydrological modeling approaches: Reviews of Geophysics, pp. 672–697, doi:10.1029/2018RG000611.

Hem, D., 1985, Study and Interpretation of the Chemical Characteristics of Natural Water: U.S. Geological Survey Professional Paper, doi:10.3133/wsp2254.

Hill, C. A. and Polyak, V. J., 2010, Karst hydrology of Grand Canyon, Arizona, USA: Journal of Hydrology, Vol. 390, pp.

169–181, doi:10.1016/j.jhydrol.2010.06.040.

Huntoon, P. W., 1970, The Hydro-Mechanics of the Ground Water System in the Southern Portion of the Kaibab Plateau, Arizona: Unpublished PhD Dissertation, University of Arizona, 273 p.

Huntoon, P., 2000, Variability of Karstic Permeability Between Unconfined and Confined Aquifers, Grand Canyon Region, Arizona: Environmental and Engineering Geoscience, Vol. 6, pp. 155–170.

Jones, C. J. R.; Springer, A. E.; Tobin, B. W.; Zappitello, S. J.; and Jones, N. A., 2017, Characterization and hydraulic behaviour of the complex karst of the Kaibab Plateau and Grand Canyon National Park, USA. In Parise, M.; Gabrovsek, F.; Kaufmann, G.; and Ravbar, N. (Editors), Advances in Karst Research: Theory, Fieldwork, and Applications: Special Publications 466, Geological Society of London, London, U.K., https://doi.org/10.1144/SP466.5.

Jones, N. A.; Hansen, J.; Springer, A. E.; Valle, C.; and Tobin, B. W., 2019, Modeling intrinsic vulnerability of complex karst aquifers: Modifying the COP method to account for sinkhole density and fault location: Hydrogeology Journal, Vol. 27, pp.

2857–2868, https://doi.org/10.1007/s10040-019-02056-2.

Kohlhepp, B.; Lehmann, R.; Seeber, P.; Küsel, K.; Trumbore, S. E.; and Totsche, K. U., 2017, Aquifer configuration and geostructural links control the groundwater quality in thin-bedded carbonate–siliciclastic alternations of the Hainich CZE, central Germany: Hydrology and Earth Systems, Vol. 21, pp. 6091–6116.

Ledbetter, J. D.; MGIS; Stevens, L. E.; Springer, A. E.; and Brandt, B., 2014, Springs Inventory Database: Online Database. Springs and Springs-Dependent Species: Electronic document, available at http://www.springsdata.org Leeder, M., 1999, Sedimentology and Sedimentary Basins: Blackwell Science, Malden, MA. Lever, J.; Krzywinski, M.; and Altman, N., 2017, Points of significance: Principal component analysis: Nature Methods, Vol. 14, pp. 641–642, doi:10.1038/nmeth.4346.

Middleton, L. T.; Elliot, D. K.; and Morales, M., 1990, Coconino Sandstone (S. S. Beus and M. Morales, Editors): Oxford University Press, New York, pp. 183–202.

Monroe, S. A.; Antweiler, R. C.; Hart, R. J.; Taylor, H. E.; Truini, M.; Rihs, J. R.; and Felger, T. J., 2004, Chemical Characteristics of Ground-Water Discharge Along the South Rim of Grand Canyon in Grand Canyon National Park, Arizona, 2000–2001: U.S. Geological Survey Professional Paper, http://pubs.water.usgs.gov/sir2004-5146.

North American Commission on Stratigraphic Nomenclature, 2005, North American Stratigraphic Code: AAPG Bulletin, Vol. 89, pp. 1547–1591, doi:10.1306/07050504129.

O’Donnell, F. C.; Flatley, W. T.; Springer, A. E.; and Fule, P. Z., 2018, Forest restoration as a strategy to mitigate climate impacts on wildfire, vegetation, and water in semiarid forests: Ecological Applications, Vol. 28, pp. 1459–1472, doi:10/1002/eap.1746.

Pool, D. R.; Blasch, K. W.; Callegary, J. B.; Leake, S. A.; and Graser, L. F., 2011, Regional Groundwater-Flow Model of the Redwall-Muav, Coconino, and Alluvial Basin Aquifer Systems of Northern and Central Arizona: U.S. Geological Survey Scientific Investigations Report 2010-5180, Vol. 1.1, 116 p.

PRISM Climate Group, 2004, PRISM Climate Data: Electronic document, available at http://www.prism.oregonstate.edu Ross, L. E., 2005, Interpretive Three-Dimensional Numerical Groundater Flow Modeling, Roaring Springs, Grand Canyon, Arizona: Unpublished Master’s Thesis, Northern Arizona University, 148 p.

RStudio, 2015, RStudio: Electronic document, available at https://rstudio.com/products/rstudio/download Ryan, W. B. F.; Carbotte, S. M.; Coplan, J. O.; O’Hara, S.; Melkonian, R.; Arko, R.; Weissel, R. A.; Ferrini, V.; Goodwillie, A.; Nitsche, F.; Bonczkowski, J.; and Zemsky, R., 2009, Global Multi-Resolution Topography Synthesis: Electronic document, available at http://geomapapp.org Schwartz, B. F. and Zhang, H., 2003, Groundwater: Hermitage Publishing Services, Ossining, NY. Schindel, G. M., 2015, Determining Groundwater Residence Times of the Kaibab Plateau, R-Aquifer Using Temperature, Grand Canyon National Park, Arizona: Unpublished Master’s Thesis, Northern Arizona University, 125 p.

Sharp, Z., 2006, Principles of Stable Isotope Geochemistry: Prentice Hall, Englewood Cliffs, NJ. 360 p.

Sorauf, J.E., and Billingsley, G.H., 1991, Members of the Toroweap and Kaibab Formations, Lower Permian, northern Arizona and southwestern Utah: Mountain Geologists, v.

1991, p. 9–24.

Springer, A. E. and Stevens, L. E., 2009, Spheres of discharge of springs: Hydrogeology Journal, Vol. 17, pp. 83–93, doi:10.1007/s10040-008-0341-y.

Springer, A. E.; Stevens, L. E.; Anderson, D. E.; Parnell, R. A.; and Kreamer, D. K., 2008, A comprehenisve springs classification styem: Integrating geomorephic, hydrogeochemical and ecological criteria. In Stevens, L. E. and Meretsky, V. J. (Editors), Aridland Springs in North America: University of Arizona Press, Tucson, AZ, pp. 1–21.

Tobin, B. W.; Springer, A. E.; Kreamer, D. K.; and Schenk, E., 2017, Reviw: Distribution, flow, and quality of Grand Canyon Springs, Arizona, (USA): Hydrogeology Journal, p. 1– 12, doi:10.1007/s10040-017-1688-8.

Turner, C. E., 2003, Toroweap Formation. In Beus, S. S. and Morales, M., (Editors), Grand Canyon Geology: Oxford University Press, Oxford, U.K., pp. 180–195.

Wickham, H. and RStudio, 2017, Tidyverse: Electronic document, available at https://www.tidyverse.org Wilde, F. D.; Radtke, D. B.; Gibs, J.; and Iwatsub, R. T., 2004, Processing of water samples, version 2.2. In National Field Manual for the Collection of Water-Quality Data: U.S. Geological Survey, Washington, DC, pp. 449–674.

Wood, A. W., 2019, Hydrogeology of the Coconino Aquifer, Kaibab Plateau, Grand Canyon: Arizona, Unpublished Master’s Thesis, Northern Arizona University, 132 p.

Worthington, S. R. H. and Ford, D. C., 1995, High sulfate concentrations in limestone springs: An important factor in conduit initiation? Environmental Geology, Vol. 25, p. 9–15.

Environmental & Engineering Geoscience, Vol. XXVI, No. 3, August 2020, pp. 367–381

381

EEGS-26-03-02_1XO

July 8, 2020

14:20

0