<%BANNER%>

Controls on the Generation of Secondary Porosity in Eogenetic Karst

Permanent Link: http://ufdc.ufl.edu/UFE0024273/00001

Material Information

Title: Controls on the Generation of Secondary Porosity in Eogenetic Karst Examples from San Salvador Island, Bahamas and North-Central Florida, USA
Physical Description: 1 online resource (140 p.)
Language: english
Creator: Moore, Paul
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: bahamas, calcite, carbonates, caves, diagenesis, dissolution, eogenetic, florida, geochemistry, hydrogeology, karst, limestone, porosity
Geological Sciences -- Dissertations, Academic -- UF
Genre: Geology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Carbonate rocks host large amounts of the world's water, petroleum, and natural gas reserves. Wise use and management of these resources require predictive models that characterize magnitudes and distributions of porosity and permeability. Predictive models are complicated by the formation of secondary porosity and coupled increases in permeability, which self-organize during dissolution and precipitation and facilitate the circulation of fluids. Most models describing porosity and permeability evolution have focused on telogenetic karst aquifers of dense, recrystallized limestone, where dissolution mainly results in conduits embedded in rocks with low matrix permeability. This study provides new insights into the generation of secondary porosity in eogenetic karst aquifers of San Salvador Island, Bahamas and north-central Florida, USA, where dissolution results in both conduits and isolated voids within rocks possessing high matrix permeability. Small carbonate islands, such as San Salvador, lack conduits, but develop secondary porosity as isolated voids in freshwater lenses. The location and size of these voids are functions of coupling dissolution and transport of reactions products, with small voids developed within the island's interior and large voids developed near the shoreline. Dissolution results primarily from CO2 generating carbonic acid at or near the water table rather than mixing of fresh and saline water as previously thought. Large carbonate platforms, such as Florida, contain conduits resulting from dissolution driven by carbonic acid, however, the high matrix permeability provides a significant component of flow that affects rates, magnitudes, and locations of dissolution. Dissolution along the conduit flow path is limited as water with elevated Ca++ concentrations flows from the matrix into the conduit and restricts dissolution at the conduit wall. In contrast, flow of water undersaturated with respect to calcite from conduits to matrix porosity creates a dissolution halo surrounding conduits, unlike telogenetic karst aquifers where dissolution is concentrated along the walls of fractures and conduits. These new concepts highlight stark contrasts in the evolution of porosity and permeability in eogenetic karst aquifers compared to their telogenetic counterparts. This information provides a better understanding to accurately predict how secondary porosity and permeability develop in carbonate rocks prior to burial diagenesis.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Paul Moore.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Martin, Jonathan B.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024273:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024273/00001

Material Information

Title: Controls on the Generation of Secondary Porosity in Eogenetic Karst Examples from San Salvador Island, Bahamas and North-Central Florida, USA
Physical Description: 1 online resource (140 p.)
Language: english
Creator: Moore, Paul
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: bahamas, calcite, carbonates, caves, diagenesis, dissolution, eogenetic, florida, geochemistry, hydrogeology, karst, limestone, porosity
Geological Sciences -- Dissertations, Academic -- UF
Genre: Geology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Carbonate rocks host large amounts of the world's water, petroleum, and natural gas reserves. Wise use and management of these resources require predictive models that characterize magnitudes and distributions of porosity and permeability. Predictive models are complicated by the formation of secondary porosity and coupled increases in permeability, which self-organize during dissolution and precipitation and facilitate the circulation of fluids. Most models describing porosity and permeability evolution have focused on telogenetic karst aquifers of dense, recrystallized limestone, where dissolution mainly results in conduits embedded in rocks with low matrix permeability. This study provides new insights into the generation of secondary porosity in eogenetic karst aquifers of San Salvador Island, Bahamas and north-central Florida, USA, where dissolution results in both conduits and isolated voids within rocks possessing high matrix permeability. Small carbonate islands, such as San Salvador, lack conduits, but develop secondary porosity as isolated voids in freshwater lenses. The location and size of these voids are functions of coupling dissolution and transport of reactions products, with small voids developed within the island's interior and large voids developed near the shoreline. Dissolution results primarily from CO2 generating carbonic acid at or near the water table rather than mixing of fresh and saline water as previously thought. Large carbonate platforms, such as Florida, contain conduits resulting from dissolution driven by carbonic acid, however, the high matrix permeability provides a significant component of flow that affects rates, magnitudes, and locations of dissolution. Dissolution along the conduit flow path is limited as water with elevated Ca++ concentrations flows from the matrix into the conduit and restricts dissolution at the conduit wall. In contrast, flow of water undersaturated with respect to calcite from conduits to matrix porosity creates a dissolution halo surrounding conduits, unlike telogenetic karst aquifers where dissolution is concentrated along the walls of fractures and conduits. These new concepts highlight stark contrasts in the evolution of porosity and permeability in eogenetic karst aquifers compared to their telogenetic counterparts. This information provides a better understanding to accurately predict how secondary porosity and permeability develop in carbonate rocks prior to burial diagenesis.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Paul Moore.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Martin, Jonathan B.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024273:00001


This item has the following downloads:


Full Text

PAGE 1

CONTROLS ON THE GENERATION OF SECONDARY POROSITY IN EOGENETIC KARST: EXAMPLES FROM SAN SALVADOR ISLAND, BAHAMAS AND NORTH-CENTRAL FLORIDA, USA By PAUL J. MOORE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009 1

PAGE 2

2009 Paul J. Moore 2

PAGE 3

The world didn't turn color until sometime in the 1930s, and it was pretty grainy color for a while, too. Bill Watterson To my Janie 3

PAGE 4

ACKNOWLEDGMENTS Completing a doctoral dissertation is not po ssible without the support of coworkers, friends, and family. First, I would like to acknow ledge the support of my dad, Joe Moore, who instilled in me a work ethic and integrity that defines the person I am today. Never one to remain idle, he constantly challenged me to push harder, become better, and refuse to settle for anything less than that warranted through hard work a nd perseverance. His ability to overcome any obstacle gives me the strength to press forward each day. While many teachers during my childhood and adolescence helped shape my intellect, none was more influential than my high school sc ience teacher, Bob Pursley. His knowledge of mathematics, physics, and chemistry was unparalleled and often presented in a humorous manner. Each class offered an opportunity to understand important pheno mena of the physical world, including calculating the trajectory of ar rows piercing classmates usually girls on hilltops, making soap from beef tallow, or characterizing Brownian motion with hip gyrations rivaled only by Chubby Checker. He exposed me to critical thinking, whic h I never appreciated during those immature years but now embrace. Needless to say, he had a profound influence on my desire to evaluate and understand the world around me. I would like to acknowledge Dr. John Mylroie, who not only became my de facto advisor during my formative years as an undergraduate at Mississippi State University, but also a dear friend. Whether jumping in and out of caves, hopp ing from island to island, or just enjoying some rum or bourbon, his infectious personality and passion for science redirected my journey through life. I am forever grateful for the mo ments we shared, and look forward to future discussions about time, eustasy, an d the absence of rock ESADMF. I thank my primary advisor, Dr. Jon Martin, for agreeing to mentor (and fund) me during my six-year tenure at the Univ ersity of Florida. Jon was not only instrumental in my 4

PAGE 5

development as a scientist, but also as a pers on. He showed me the importance of a work-life balance, where one can be equally dedicated to both career and family. Although our paths now diverge, I will never forget thos e trips to South America and Antarctica, watching Gollum collect samples in Majors Cave, and those exciting adve ntures to OLeno State Park I never thought I would miss those, but amazingly I do. In addition to these individuals, a host of ot hers provided much needed support through the years. I thank my committee members, Drs. Li z Screaton, Phil Neuhoff, Dave Hodell, and Joe Delfino, for agreeing to oversee my project. Spec ifically, I thank Liz for always keeping me on my toes, and Phil for stressing the importance of chemical potential. I thank Kelly Deuerling for providing invaluable assistance in the lab, and Mou Roy for always sharing a good laugh or some of her amazing cooking. I thank Jason Gulley for the lively convers ations, usually over scotch, about cave genesis and anything else th at came up, which was often. Thanks go out to Jonathan Hoffman, Derrick Newkirk, and Laura Ruhl, for being great roommates and not shooting me in the middle of the night during my visits to the kitchen for milk and cookies. I would also like to acknowledge the office staff, Nita Fahm, Su san Birungi, and Pam Haines, for always accommodating my requests and pleads. You guys kept me sane through the final months. Lastly, I would like to thank my family, w hose love and support are my foundation. My siblings, Hope Mosley and Kameren Gibson, have always been there to take my mind off school and because of them I am uncle PJ to Addis on, Jasmine, and Steven. My grandparents, Rita Hildick and Doris Moore, are my angels on Earth. I thank my aunt, Margaret Hildick-Pytte, for her encouragement, words of wisdom, and sharing of good wine. Finally, I wish to thank the love of my life, Jane Gustavson. She remained by my side during times of my own self doubt and 5

PAGE 6

irritable mood swings, always r eady to provide reassurance with a warm hug. She is my friend, my closest companion, and the reason this document reached completion. 6

PAGE 7

TABLE OF CONTENTS page ACKNOWLEDGMENTS ...............................................................................................................4 LIST OF TABLES ...........................................................................................................................9 LIST OF FIGURES .......................................................................................................................10 ABSTRACT ...................................................................................................................................11 CHAPTERS 1 INTRODUCTION..................................................................................................................13 Statement of Purpose ..............................................................................................................17 Dissertation Organization .......................................................................................................18 2 RAPID GENERATION OF MACROPOROSI TY ON CARBONATE ISLANDS: THE FLANK MARGIN CAVE HY POTHESIS REVISITED.......................................................20 Introduction .............................................................................................................................20 Location and Geologic Setting ...............................................................................................23 Water Chemistry .....................................................................................................................25 Methods ...........................................................................................................................25 Results .............................................................................................................................27 Bulk chemistry .........................................................................................................27 CO2 and saturation state ...........................................................................................28 Influence of mixing on saturation ............................................................................29 Generation of Macroporosity ..................................................................................................30 Mechanisms of Dissolution .............................................................................................30 Mixing dissolution ....................................................................................................30 Role of CO2 in dissolution and precipitation reactions ............................................32 Calculation of Dissolution Rates .....................................................................................35 Conclusions .............................................................................................................................37 3 GEOCHEMICAL AND STATISTICAL EVIDENCE FOR RECHARGE, MIXING, AND CONTROLS ON SPRING DISCHA RGE IN AN EOGENETIC KARST AQUIFER........................................................................................................................ .......47 Introduction .............................................................................................................................47 Study Area ..............................................................................................................................49 Methods ..................................................................................................................................51 River Stage and Potential Recharge ................................................................................51 Field Sampling and Laboratory Analysis ........................................................................52 Principal Component Analysis ........................................................................................53 Estimate of Vertical Flow Rate from Temperature Perturbations ...................................54 7

PAGE 8

Results .....................................................................................................................................55 River Stage and PE-T ......................................................................................................55 Water Temperature and Chemistry .................................................................................56 Principal Component Analysis ........................................................................................57 Discussion ...............................................................................................................................59 End-member Chemistry and Sources of Water ...............................................................60 Allogenic recharge ...................................................................................................60 Groundwater .............................................................................................................62 Influences of Vertical Flow on Shallow Water Chemistry .............................................64 Effects of Source Water and Flow Paths on Spring Discharge .......................................66 Conclusions .............................................................................................................................70 4 CONDUIT ENLARGMENT IN AN EOGENETIC KARST AQUIFER..............................85 Introduction .............................................................................................................................85 Background .............................................................................................................................87 Hydrologic and Geologic Conditions ..............................................................................87 River Conditions and Potential Recharge........................................................................89 Water Chemistry ..............................................................................................................90 Chemical Analysis and Geochemical Modeling .....................................................................91 Methods ...........................................................................................................................91 Results .............................................................................................................................92 Discussion ...............................................................................................................................95 Geochemical Evolution of Groundwater .........................................................................95 Estimates of Conduit Dissolu tion in the Sink-Rise System ............................................97 Implications of Conduit Enlargement in Eogenetic Karst Aquifers ..............................100 Conclusions ...........................................................................................................................103 5 SUMMARY AND CONCLUSIONS...................................................................................115 APPENDIX: FIELD DATA AND CHEMIC AL ANALYSIS OF WATER SAMPLES FROM SAN SALVADOR ISLAND, BAHAMAS..............................................................118 LIST OF REFERENCES .............................................................................................................124 BIOGRAPHICAL SKETCH .......................................................................................................140 8

PAGE 9

LIST OF TABLES Table page 2-1 Calculated dissolution of a freshwater lens at Sandy Point during the MIS 5e highstand. ...........................................................................................................................39 2-2 Calculation of dissolution at Altar Cave. ...........................................................................39 3-1 Summary of major ions, al kalinity, SpC, pH, and T of representative water samples. .....72 3-2 Variable loadings of PCA ..................................................................................................74 3-3 Fraction of water discharging from the River Rise originating from the River Sink and two groundwater end members. ..................................................................................75 4-1 Summary of major ions, alkalinity, SpC, pH, T, SI, and PCO2 of representative water samples. ............................................................................................................................105 4-2 Chemical reactions involving UFA minerals. ..................................................................107 4-3 River Rise composition estimated by mixing of River Sink, Well 2, and Well 4 endmember compositions and mass transfer of calcite and CO2...........................................108 4-4 Inverse modeling along a conduit flow path. ...................................................................110 A-1 Water chemistry from San Salvador Island, Bahamas.....................................................118 9

PAGE 10

LIST OF FIGURES Figure page 2-1 Maps of Bahamian archipelago and San Salvador Island..................................................40 2-2 Plots of major ion versus chlori de concentrations in all samples ......................................41 2-3 Comparison of DIC and total alkalinity. ............................................................................42 2-4 Relationship between PCO2 and aragonite and calcite saturation in all water samples ......43 2-5 Water samples collected from Ink Well blue hole in April 2007 ......................................44 2-6 Saturation indices for all San Salv ador waters versus percent seawater ...........................45 2-7 Plot of excess Ca2+, saturation index of aragonite, and log PCO2 of water from the Line Hole well field. ..........................................................................................................46 3-1 Lithostratigraphic and hydrostratigraph ic units of the Santa Fe River Basin. ...................76 3-2 Site location of the Santa Fe Sink-Rise system.................................................................77 3-3 Stage and discharge of the Santa Fe Ri ver at the River Rise and precipitation and potential recharge ...............................................................................................................78 3-4 Piper diagram of surface and groundwater in Santa Fe Sink-Rise system ........................79 3-5 Principal component loadings and scores for Santa Fe Sink-Rise water...........................80 3-6 Plots of ion concentrations versus Cl-................................................................................81 3-7 Diagrammatic sketch of boundary conditions for vertical steady-state flow and heat transfer at Well 2 ................................................................................................................82 3-8 Plot of Mg2+ versus SO4 2concentrations ..........................................................................83 3-9 Plots of source contributions vers us discharge at the River Rise ......................................84 4-1 Phase diagram showing relations among sampled water and aquifer minerals ...............111 4-2 Saturation indices of calcite, do lomite, and gypsum versus Log ..............112 2-2422SOCO/ aa 3 4-3 Bar graph showing Ca2+ concentrations at the River Sink and River Rise .....................113 4-4 Conceptual sketch showing the cross-sectional area of a conduit in an eogenetic karst aquifer ..............................................................................................................................114 10

PAGE 11

Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CONTROLS ON THE GENERATION OF SECONDARY POROSITY IN EOGENETIC KARST: EXAMPLES FROM SAN SALVADOR ISLAND, BAHAMAS AND NORTH-CENTRAL FLORIDA, USA. By Paul J. Moore May 2009 Chair: Jonathan B. Martin Major: Geology Carbonate rocks host large amounts of the wo rlds water, petroleum, and natural gas reserves. Wise use and management of these resources require pred ictive models that characterize magnitudes and distributions of porosity and permeability. Predictive models are complicated by the formation of secondary por osity and coupled increases in permeability, which self-organize during dissolu tion and precipitation a nd facilitate the circ ulation of fluids. Most models describing porosity and permeability evolution have focused on telogenetic karst aquifers of dense, recrystallized limestone where dissolution mainly results in conduits embedded in rocks with low matrix permeability This study provides new insights into the generation of secondary porosity in eogenetic karst aquifers of San Salvador Island, Bahamas and north-central Florida, USA, where dissolutio n results in both conduits and isolated voids within rocks possessing high matrix permeability. Small carbonate islands, such as San Salvador, lack conduits, but develop secondary porosity as isolated voids in freshwater lenses. The location and size of these voids are functions of coupli ng dissolution and transpor t of reactions products, with small voids developed within the islands interior and large voids developed near the shoreline. Dissolution results primarily from CO 2 generating carbonic acid at or near the water 11

PAGE 12

table rather than mixing of fresh and saline water as previously thought. Large carbonate platforms, such as Florida, contain conduits resulting from dissolution driven by carbonic acid, however, the high matrix permeability provides a significant component of flow that affects rates, magnitudes, and locations of dissolution. Di ssolution along the condu it flow path is limited as water with elevated Ca 2+ concentrations flows from the ma trix into the conduit and restricts dissolution at the conduit wall. In contrast, flow of water undersaturated w ith respect to calcite from conduits to matrix porosity creates a dissolution halo surrounding conduits, unlike telogenetic karst aquifers wher e dissolution is concentrated along the walls of fractures and conduits. These new concepts highlight stark contrasts in the evol ution of porosity and permeability in eogenetic karst aquifers compared to their telogenetic counterparts. This information provides a better understanding to accurately predict how secondary porosity and permeability develop in carbonate ro cks prior to burial diagenesis. 12

PAGE 13

CHAPTER 1 INTRODUCTION Carbonate rocks supply an estimated 25% of drinking water to the worlds population (Ford and Williams 2007) and host more than 60% and 40% of the worlds petroleum and natural gas reserves, respectiv ely (Schlumberger Market Analysis 2007). Wise use of these natural resources requires knowle dge about the various types of porosity, both primary and secondary, that tend to be heter ogeneously distributed in carbonate rocks. These various types of porosity include intergranular porosity within th e matrix rock, fractures, faults, bedding plane partings, and conduits enlarged through dissolution. This porosity and its distribution influence nearly all aspects of aquifer-reservoir characte ristics, including storage and distribution of permeability (e.g., Moore 2001; Ford and W illiams 2007). The range of porosity and permeability determines flow paths and allows highly variable flow rates, which can be both laminar and turbulent (Worthington 1994; Quinlan et al. 1996; White 1999; Halihan et al. 2000). While most flow in carbonate ro cks occurs through conduits and fr actures, storage is primarily in the matrix porosity (e.g., Worthington et al. 2000). The magnitude and distribution of porosity and permeability in car bonate rocks largely results from diagenetic and karst processes (e.g., Choquette and Pray 1970; Lucia 1995; White 1999; Worthington et al. 2000). While diagenesis refers to the sum of all chemical, physical, and biological processes affecting sediment following deposition (Morse a nd Mackenzie 1990), karst processes reflect the development of well-orga nized secondary porosity from dissolution, which facilitates the circulation of fluids (Huntoon 1995). Early work described the evolution of carbonate rocks based on time-porosity stages that reflect the rock cycle, with the terms eogenetic and telogenetic referring to before and after the reduction of primary depositional porosity by burial diagenesis (Choquette and Pr ay 1970). Recent work applied these terms to 13

PAGE 14

karst to designate carbonate rocks of varying age, diagenesis, and karstification (Vacher and Mylroie 2002). Eogenetic karst refers to young limestone that has only undergone meteoric diagenesis near its area of deposition, and main tains high matrix porosity as localized mineral dissolution and precipitation redist ributes porosity from interparticle to moldic pores (Vacher and Mylroie 2002). Examples of eogenetic karst include the Cenozoic rocks of Yucatan, Mexico, the Bahamas, and Florida, USA (Back and Hanshaw 1970; Carew and Mylroie 1997). In contrast, telogenetic karst has undergone burial diagenesis and uplift, resulting in a dense recrystallized limestone that develops conduits em bedded in a network of fractures of otherwise low matrix porosity and permeability (Vacher and Mylroie 2002). Examples of telogenetic karst include the Paleozoic rocks of Appalachia in the eastern United States and the Alps of Europe (White 1988; Ford and Williams 2007). The hallmark property that distinguishes eogenetic karst from telogenetic karst is the role of the matrix permeability in hydrogeology (e.g., Budd and Vacher 2004; Florea and Vacher 2007). Most conceptual and numerical models of the generation of secondary porosity in carbonate rocks have focused on conduit evolutio n in telogenetic karst aquifers (White 1988; Palmer et al. 1999; Ford and Williams 2007). Two critical processes affecting the evolution from fractures to conduits include the transition from la minar to turbulent flow and a kinetic shift in calcite dissolution rates (White 1977; Dreybrodt 1990; Palmer 1991). During early stages of conduit evolution, calcite dissolution rates drop from linear to fourth-order kinetics at about 90% saturation, thereby allowing slow widening of fractures under laminar flow. Low matrix permeability focuses flow and dissolution within the fractures, progressively enlarging competing flow paths until a preferential pathway captures most of the flow to become a protoconduit. Once water in the proto-conduit dr ops below 90% of calcite saturation, linear 14

PAGE 15

dissolution kinetics dominates re sulting in rapid enlarg ement of the conduits (e.g., Palmer 1991). Under typical groundwater gradients, the kinetic shift moves from slow, high-order kinetics to fast, low-order kinetics approximately at the tr ansition from laminar to turbulent flow during flow through an enlarging conduit (White 1977). These concepts have been used in deterministic and stochastic models to provide valuable insi ght on evaluating the earl y evolution of conduits (e.g., Groves and Howard 1994; Kaufmann and Braun 1999; Gabrovek and Dreybrodt 2001; Romanov et al. 2003a; Gabrovek et al. 2004), predicting the mo rphology of cave patterns (e.g., Palmer 1991; Howard and Groves 1995; Palmer 2001) modeling leakage rates beneath dam sites (Romanov et al. 2003b), and estimating the role of conduit growth on landscape evolution in a karst basin (Groves and Meiman 2005). Less understood, however, are processes controll ing the generation of secondary porosity in eogenetic karst, which produce both conduits as well as isolated meter-scale dissolution chambers (Mylroie and Vacher 1999; Florea et al. 2007). In Florida, con duits have been shown to occur primarily where streams flowing off siliciclastic sediments sink into underlying limestone and later reemerge downgradient (Katz et al. 1998; Screaton et al. 2004), and at springs where the convergence of groundwater flow lines concentrate diffuse flow (Beck 1986; White 2002). Although conduits develop in similar se ttings in telogenetic karst (e.g., Shuster and White 1971; Palmer 2001), flow in eogenetic karst may not remain focused within fractures or proto-conduits because large fractions of the flow occur in the high matrix permeability. The relationship between flow in conduits and the surrounding aquifer has been observed as mixing between conduit and matrix water depending on hydraulic head gradients between the conduit and surrounding aquifer (Katz et al. 1998; Screaton et al. 2004; L oper et al. 2005; Martin et al. 2006). Mixing of these chemically-distinct waters should reduce the magnitude and distribution 15

PAGE 16

of dissolution, thereby diminishing enlargement of the proto-conduit (cf. Gabrovek et al. 2004). These observations suggest that new concepts may be necessary to describe conduit development in eogenetic karst. In addition, not all caves found in eogeneti c karst reflect the evolution of conduit permeability (Mylroie and Carew 1990; Florea 2006). For example, many Florida caves are similar to caves found on small carbonate islands such as the Bahamas, which are characterized by laterally-extensive and vertically-restricte d chambers with blind pockets and dead-end passages (Mylroie and Carew 1990; Florea 2006). In both loca tions, geomorphic and glacioeustatic evidence suggests these isolated voids are phreatic in origin and develop along watertable horizons both within th e interior and along shorelines Cave morphology and the high matrix porosity and permeability of the surrounding a quifer suggests they likely formed under diffuse-flow conditions, and thus formed as dissolution chambers and not conduits that form from focused flow such as in telogenetic karst. Furthermore, they are believed to form with limited connection to the land surface, whereby di ffuse recharge to the water table occurs through the epikarst and vertical preferential flow paths, such as narrow fracture trends (Florea 2006) or along well-developed karstic fissures (W hitaker and Smart 1997a). Cave entrances form when the voids are breached by surface denudation, hillside erosion, or land alteration such as road construction and quarrying. One primary mechanism believed to drive cave development in these locations is the mixing of chemically-distinct water. In the Baha mas, voids known as banana holes form within the interior, reaching hundreds of m 3 in volume and are believed to have developed by mixing of vadose and phreatic water (Harri s et al. 1995). Conversely, larger voids, called flank margin caves, can reach thousands of m 3 in volume and are believed to have developed at the edge of 16

PAGE 17

freshwater lenses primarily from the convergen ce of two areas of mixi ng dissolution: mixing of vadose and phreatic water at the wa ter table and mixing of fresh and saline water at the halocline (Plummer 1975; Mylroie and Carew 1990). The pr imary difference between banana holes and flank margin caves is size, which is largely attr ibuted to more chemically-aggressive water along the lens edge (Mylroie and Carew 1990). However, in Florida, many large air-filled caves occur at water-table elevations consistent with presen t-day sea level or align with marine terraces of past sea-level highstands (Florea et al. 2007). Th ese caves reach volumes in excess of thousands of m 3 and likely formed in the abse nce of mixing of fresh and salin e water, since the thickness of fresh groundwater where these caves are located can exceed several hund red meters (Miller 1986). These observations suggest that processes in addition to dissolution mechanisms may influence the magnitude and distribution of cav e development in eogenetic karst aquifers. Statement of Purpose The objective of this investiga tion is to develop an understa nding of controls on porosity development in eogenetic karst aquifers. This study couples hydrogeological concepts with chemical compositions of water from San Salva dor Island, Bahamas and nor th-central Florida, USA with the known distributions of dissolution voids includ ing flank margin caves and conduits. Results contribute to the current und erstanding of secondary porosity generation in eogenetic karst in several ways, specifically, how chemical and physical factors control porosity density and distribution to provide new concepts to characterizing aquifers and reservoirs, and on fluid flow and recharge and their influen ce on water budgets and sp ring characterization. To meet this objective, this study addresse s several general questions pertaining to dissolution and porosity evolution in eogenetic karst aquifers: 1. Are flank margin caves caused primarily by th e mixing of fresh and saline water? Does cave development depend on additional proc ess to drive dissolution including the 17

PAGE 18

generation of carbonic acid? What role does fr eshwater lens hydrodynamics play in their enlargement? 2. Can matrix permeability affect spring discharg e and chemistry even in systems dominated by allogenic recharge and conduit flow? Does groundwater monitoring provide insight on sources of water to springs? 3. Does the presence of high matrix porosity and permeability affect the magnitude and distribution of conduit development in eoge netic karst aquifers? Are new concepts necessary for modeling conduit enlargem ent in eogenetic karst aquifers? Dissertation Organization This dissertation is presented in publishable paper format. Chapter 2 is titled Rapid generation of macroporosity on ca rbonate islands: the flank ma rgin cave hypothesis revisited and is in review at the Journal of Sedimentary Research This chapter contributes to the understanding of chemical and hydrologic proc esses most likely responsible for generating dissolution chambers within freshwater lens of small carbonate islands. The study was based on coupling bulk chemistry and geochemical modeling of water collected from San Salvador Island, Bahamas to the hydrodynamics of fr eshwater lenses. The major resu lt of this chapter is a single unifying quantitative model that describes how me ter-scale porosity can develop within diffuse flow fields. This chapter provi des insight on predicting the lo cation and magnitude of meterscale secondary porosity across carbonate platform s that have experienced meteoric diagenesis. Chapter 3 is titled Geochemical and statisti cal evidence for recharge, mixing, and controls on spring discharge in an eogenetic kars t aquifer and is in review at the Journal of Hydrology This chapter contributes to the understanding of matrix flow in eogenetic settings. The study was based on principal component analysis (PCA) of the bulk chemistry of surface and groundwater from a portion of the Upper Floridan aquifer in no rth-central Florida. The major result of this chapter suggests that matrix flow provides a significant component of flow even in eogenetic aquifers dominated by conduits that are sour ced by continuous allogenic recharge. The 18

PAGE 19

contributions of this chapter provide insight on how high matrix permeability affects aquifer recharge and spring discharge, a nd suggests that characterizing eoge netic karst aquifers requires attention to groundwater chemistry and additional flow paths rather than on the conduit water as is commonly the case for telogenetic settings Chapter 4 is titled Conduit enlargement in eo genetic karst aquifers and will be submitted to the Journal of Hydrology This chapter contributes to the understanding of the physical and chemical factors responsible for dissolution and formation of conduits in eogenetic karst setting and the differences between these factors in eoge netic and telogenetic karst. The study was based on the geochemical equilibria of water chemistr y from Chapter 3 and a mass-balance model that evaluates the mixing of water a nd mass transfer of mineral and gas phases within the conduit. This work found that dissolution of conduit walls is restricted by flow of Ca-rich water from the matrix and that dissolution is focused in matr ix porosity surrounding the conduits as well as at the conduit walls during high-flow events. This ch apter provides insight on how the evolution of eogenetic and telogenetic karst aquifers differ. Chapter 5 summarizes the principle findings of Chapters 2 through 4. 19

PAGE 20

CHAPTER 2 RAPID GENERATION OF MACROPORO SITY ON CARBONATE ISLANDS: THE FLANK MARGIN CAVE HYPOTHESIS REVISITED Introduction On Holoceneand Pleistocene-aged carbonate islands, freshwater lenses are sites of extensive meteoric diagenesis including dissolu tion of carbonate minerals and development of secondary porosity. Carbonate mineral dissoluti on commonly results from formation of carbonic acid by the hydration of CO 2 (Kern 1960), acids produced fro m oxidation-reduction reactions (Smart et al. 1988), and mixing of waters with dissimilar chemical compositions, such as seawater and freshwater (Runnells 1969; Pl ummer 1975). Although dissolution is common in freshwater lenses, bulk values of porosity do not change greatly throughout most of the lens because the porosity is largely redistributed from interparticle to moldic pores through localized dissolution and reprecipitation (e.g., Vacher and Mylroie 2002). In sele cted locations of freshwater lenses, however, th e generation of secondary porosity can result in meter-scale macropores which have a significant impact on the accumulation and exploitation of freshwater and hydrocarbon resources (e.g., Craig 1988; Jaeggi 2006; Baceta et al. 2008; Labourdette and Mylroie 2008). Any development of macropores w ould require an environment that promotes dissolution while preventing reprecipitation of the di ssolution reaction products (e.g., Palmer 1991). In the Bahamas, geomorphic and glacio-eust atic evidence suggest s macropores develop along the water table and at the dist al edge of freshwater lenses in short time spans of thousands of years (Mylroie and Carew 1995). At the water table, macropores known as banana holes reach hundreds of m 3 in volume and are believed to have developed from mixing between vadose and phreatic water (Harris et al. 1995). Conversely, the largest macropores, called flank margin caves, can reach thousands of m 3 in volume and are believed to have developed at the edge of 20

PAGE 21

freshwater lenses primarily from the convergen ce of two areas of mixi ng dissolution: mixing of vadose and phreatic water at the wa ter table and mixing of fresh and saline water at the halocline (Mylroie and Carew 1990; Mylroie and Carew 1995). Although significant effort has gone into develo ping conceptual models of how and where macropores form, a single unifying quantitative model for their formation has never been developed, regardless of macropore importance for subsurface storage of freshwater and hydrocarbons (e.g., Romanov and Dreybrodt 2006). Porosity generation could accelerate at the lens edge due to increases in specific discharge, which s hould contribute to dissolution by enhancing the mixing of different water masses (Raeisi and Mylroie 1995; Rezaei et al. 2005). However, on small carbonate islands, where most flank margin caves are found (Labourdette et al. 2007), mixing processes appear to have litt le influence on dissolution (e.g., Plummer et al. 1976; Budd 1988; Anthony et al. 1989; McClain et al. 1992; Matsuda et al. 1995; Ng and Jones 1995; Moore et al. 2006; Whitaker and Smart 2007a ; Martin and Moore 2008). In addition to mixing, porosity generation may result from di ssolution by acids generated by aerobic and anaerobic oxidation-reduction reactions within the mixing zone where density interfaces trap organic matter and bacteria (Smart et al. 1988; Bottrell et al. 1991, 1993; Schwabe et al. 2008). Although mixing and oxidation-reduction reactions may drive some dissolution, most dissolution on these islands may be linked primarily to increased concentrations of CO 2 generating carbonic acid. Nonetheless, average dissolution rates fail to explain the location and magnitude of porosity generation required to develop mapped flank margin caves. For example, most Bahamian flank margin caves formed during the Marine Isotope Substage (MIS) 5e highstand in about 12 ky or less (Carew and Mylroie 1995a); however, this amount of time woul d be insufficient to form the 21

PAGE 22

caves given average estimates of dissolution rates for carbonate islands (e.g., Whitaker and Smart 2007a). I suggest the generation of macropores, such as banana holes and fla nk margin caves, may occur in select locations of fres hwater lenses from inputs of CO 2 coupled with the flow dynamics of freshwater lenses. Flow velo cities in freshwater lenses ar e greatest at the water table and increase towards the lens edge, where the highest degree of specific discha rge occurs (Vacher et al. 1990; Raeisi and Mylroie 1995) Consequently, the size of banana holes and flank margin caves may reflect the relative abil ity of the freshwater lens to transport reaction products away from the locus of porosity generation before th ey reprecipitate as ce ments. This hypothesis would predict flank margin caves are the la rgest macropores because they develop along shorelines at the edge of fres hwater lenses where specific disc harge is greatest and reaction products can be rapidly flushed to the ocean. C onversely, banana holes would be smaller than flank margin caves because they form along the water table within the islands interior, where flow velocities are less than at the lens edge. Th eir smaller size likely reflects these low flow velocities providing sufficient time for so me reprecipitation of reaction products. In this chapter, I evaluate the magnitude and rate of dissolution required for a flank margin cave on San Salvador Island, Bahamas to develop during the MIS 5e high stand, using chemical compositions of groundwater, freshwater-lens geom etry, and the volume of a flank margin cave. The water-chemistry data suggests that most dissolu tion occurs in low-salin ity water, is primarily governed by the influx of CO 2 and mixing has only a minor influence on dissolution. By coupling these estimates of dissolution with flow dynamics, I develop a mass balance model that estimates dissolution rates at the lens edge, and compare magnitudes of dissolution with the measured volume of a flank margin cave. The ab ility to predict the lo cation of macropores in 22

PAGE 23

carbonate islands will be important for understandi ng distributions of valuable natural resources such as freshwater and hydrocarbons. Location and Geologic Setting The Bahamian Archipelago is a NW-SE trendi ng, 1400-km long series of carbonate islands and shallow banks located along the eas tern margin of North America ( Figure 2-1A ). The archipelago is considered to be tectonically stab le, but subsiding isostatica lly at a rate of 1-2 m per 100 ky (Mullins and Lynts 1977; Carew and Mylr oie 1995a). Islands in the northwestern portion of the archipelago are isolated landmasses located on two large platforms, Little Bahama Bank and Great Bahama Bank. To the southeast, th e archipelago consists of small, isolated platforms that are capped by islands that cover most of the platform area. The islands are composed of variably-cemented Holocene and mi d-to-late Pleistocene ca rbonate sediments of subtidal, reef, beach, and dune facies that reach up to about 60 meters above present sea level (masl). Only dune facies exist a bove 8 masl, and all exposed mari ne facies in the Bahamas are believed to be deposited during the MIS 5e highstand (Carew and Mylroie 1997). Climate in the Bahamas includes a warm, rainy season from Ma y to October and a cool, dry season from November to April (Whitaker and Smart 1997a). Rain delivers freshwater to the islands during the passage of cold fronts in the winter, c onvective thunderstorms during summer, and tropical storms. San Salvador Island, located in the southeastern portion of the archipel ago, is about 20 km long and about 8 km wide ( Figure 2-1B ). Annual rainfall ranges from 1000 to 1250 mm/yr, with potential evapotranspiration (PET) estimated between 1250 and 1375 mm/yr (Sealey 1994). The negative water budget limits groundwater recharge to rain events that exceed PET. Surface streams are absent due to the high porosity ( 30%) of the limestone (Voge l et al. 1990; Vacher and Mylroie 2002), but about one third of the isla nd is covered by lakes ranging in salinity from 23

PAGE 24

marine to hypersaline. Lakes with marine salinity have tidal fluctuations and bulk chemical compositions similar to seawater, suggesting a di rect subsurface connectio n to the ocean (Davis and Johnson 1989; Martin and Moore 2008). Most of the land surface is covered by low-lying plains and eolian dunes that reach a maximum of 30 masl (Sealey 1994) and multiple freshwater lenses occur below the dunes ( Figure 2-1B ). These freshwater lenses drain into nearby interior lakes or along the coast line (Davis and Johnson 1989). San Salvador Island has numerous flank margin caves that formed about 125 ka during the 12-ky MIS 5e highstand when sea level was 4 to 6 masl (Mylroie and Carew 1990; Chen et al. 1991; Carew and Mylroie 1995a). Du ring this time, most of San Salvadors platform was flooded, and the bank was comprised of small, is olated strip-islands made up mostly of the eolian dunes observed today (Carew and Mylroie 1997). Flank margin caves are believed to have formed by freshwater and seawater mixing at the di stal edge of freshwater lenses that developed within the strip-islands (i.e., hypogenic caves, Palm er 1991). The caves initially formed as small, isolated dissolution chambers and increased in size in the absence of tu rbulent flow as voids coalesce along strike at the edge of the fres hwater lens (Labourdette et al. 2007). This mechanism has led to morphologies that are vert ically-restricted, horizontally-extensive caves with volumes varying from 10 to 15,000 m 3 (Mylroie et al. 1995). Cave entrances form when the caves are breached by surface denudation and hillside erosion. Altar Cave, a well-studied flank margin cave used in this study, is located in Grotto Beach Ridge on Sandy Point in the southwest portion of San Salvador Island ( Figure 2-1C ). The cave has an areal footprint of about 450 m 2 and a minimum volume of 1200 m 3 (Florea et al. 2004). Grotto Beach Ridge formed early in the MIS 5e transgression as a prograding sequence of subtidal, beach, back-beach, and eolian sediment s (Carew and Mylroie 1995b). This material 24

PAGE 25

would have been originally mostly aragonite (e.g., Budd 1988), but altera tion has converted the rocks to about 60% calcite (Vogel et al. 1990; Carew and Mylroie 1995b). Fossil corals from the base of Grotto Beach Ridge located above mode rn sea level about 500 m north of Altar Cave yielded a U/Th age of approximately 125 ka (Hat tin and Warren 1989). This timing indicates that Altar Cave formed concurrently with the deposit ion of its host rock (i .e., syndepositional karst; Labourdette and Mylroie 2008). While the MIS 5e highstand lasted for about 12 ky, Altar Cave must have formed in less time, perhaps at most 10 ky, to allow time for early development of both Grotto Beach Ridge and its freshwater lens following flooding of the bank during the transgression. Water Chemistry Methods Sixty-three water samples, ranging from fresh to hypersaline, were collected from inland lakes, cave pools, blue holes and wells on San Salvador Is land between April 2005 and April 2007 during three different sampling trips (Moore and Martin 2008) ( Table A-1 ). Most water bodies were sampled only once, including 12 we lls from the Line Hole well field, which penetrate a single freshwater lens ( Table A-1 ). The wells are evenly spaced approximately 30 m apart and are located about 200 to 500 m from the coast. Othe r water bodies, including cave pools and Ink Well Blue Hole, were sampled multip le times both spatially and temporally. The cave pools are at least several mete rs deep and were sampled in depth profiles to measure salinity gradients across the pycnocline (Moore et al. 20 06). At Ink Well, two sets of samples were collected near the center of the 15-m diameter blue hole from the surface to the sediment-water interface during a single day. Th e first set was collected in the morning under cloud cover and the other under direct sunlight about one hour after the initial set was collected. 25

PAGE 26

Water was collected from accessible locations by grab sampling, but wells, cave pools, and Ink Well Blue Hole were sampled using polyeth ylene tubing connected to a 12-V peristaltic pump (Moore et al. 2006). During sampling, I m easured pH and temperature using an Orion #250A portable pH-temperature meter, and condu ctivity and salinity using an Orion model #130 portable conductivity meter. Samples for major ion and Sr 2+ concentrations were collected in clean, dry HDPE bottles with no preservatives, and samples for dissolved inorganic carbon (DIC) were collected in glass bottles and poisoned with CuSO 4 No samples were collected for DIC during the April 2005 trip ( Table A-1 ). Total alkalinity was measured using the Gran titration within 24 hours of collect ion (e.g., Drever 1997). Major ion, Sr 2+ and DIC concentrations were analyzed at the Department of Geological Sciences at the University of Florida. Major ions were measured with an automated Dionex 500DX ion chromatograph. Strontium concentrations were measured w ith a Thermo Finnigan Element 2 Inductively Coupled Plasma Mass Spectrometer (ICP-MS). Disso lved inorganic carbon was measured with a Coulometrics CO 2 coulometer using a 3% AgNO 3 scrubber solution, N 2 as the carrier gas, and 2N HCl to evolve the CO 2 from the water. Precision of the measurements (1 ) was assessed by replicate measurements of internal standa rds, and was found to be about 1% for DIC concentration, 5% for the major ion concentrations, and about 8% for Sr 2+ concentrations. The geochemical code EQ3/6, Version 8 (Wolery 1992), was used to determine the partial pressure of CO 2 ion-activity products (Q), and carbonate mineral saturation indices [log (Q/K), where K is the equilibrium constant for a given mineral] for all water samples, as well as theoretical saturation states of mixtures of samples with compositions assumed to represent freshwater and seawater con centrations. All thermodynamic data were from the SUPCRT database (Johnson et al. 1992). Activ ity coefficients of aqueous species were calculated using the 26

PAGE 27

extended Debye-Hckle equation of Helgeson (1 969). Samples having a saturation index (SI) within .1 are assumed to be in equilibrium with respect to calcite and dolomite based on analytical errors in measurements of pH, alkalinity, and c oncentrations of Ca 2+ (e.g., Langmuir 1997). Charge balance errors for most samples are < 3%, but may be larger in freshwater samples with concentrations clos e to instrument detection limits. I use Cl as a conservative tracer to determine the amount of ions in the water that originate from carbonate mineral dissolution by assuming that Cl is chemically conservative on San Salvador Island and that there are no other so urces of reaction products to the water. These assumptions are reasonable since lithologies on San Salvador Isla nd, other than minor quartzitic paleosols, are primarily calcite and aragonite wi th no other Caor Cl-bearing minerals such as halite, gypsum, or anhydrite (Martin and Moore 2008). Excess ion concentrations originating from water-rock reactions (X EXCESS ) were determined by EXCESSSAMPLE SEAWATERSAMPLEX=X(X/Cl)(Cl) (2-1) where X and Cl are molar concentrat ions of any dissolved ion and Cl respectively. Results Bulk chemistry Water on San Salvador Island consists of mixtures of two end members, rain water and seawater. Some of the water samples have been concentrated through ev aporation as shown by their salinity, which ranges up to about 52 pr actical salinity unit (p su) or about 1.5 times seawater concentrations. Magnesium, Na + K + and SO 4 2concentrations correlate linearly with Cl ( Figures 2-2A, 2-2B, 2-2C, 2-2D ) indicating their concentrations vary only through seawater dilution and evaporation. In c ontrast, concentrations of Ca 2+ Sr 2+ and total alkalinity are enriched compared to values expected from conservative mi xing and evaporation, reflecting 27

PAGE 28

carbonate mineral dissolution ( Figures 2-2E, 2-2F, 2-2G ). The measured total alkalinity correlates with DIC suggesting it is mostly carbonate alkalinity ( Figure 2-3 ). The greatest enrichments of Ca 2+ Sr 2+ and alkalinity occur in samples with salinity less than 5 psu (e.g., Martin and Moore 2008). The con centration in excess of that expected from mixing of seawater and freshwater can be defi ned as the excess ion concentrations (i.e., Eq. 2-1 ). For example, excess Ca 2+ in samples with salinity less than 5 psu average about 2.0 mM, while brackish samples (7 to 30 psu) and sample s with marine salinity have excess Ca 2+ concentrations averaging about 0.90 mM and 0.32 mM, respectively ( Figure 2-2E ). The greatest enrichment of 48 times occurs in Sample LH-A (0.4 psu), which has an excess Ca 2+ concentration of about 3.5 mM. This sample also has an excess Sr 2+ concentration of 0.098 mM and excess alkalinity concentration of 7.4 mM, which refl ects a 155-times enrichment of Sr 2+ and 430-times enrichment of alkalinity ove r seawater contributions. CO 2 and saturation state Speciation-solubility modeling shows the SI of most samples are within about 0.5 SI units of saturation with respect to aragonite, although a few samples extend to as much as 1 and -1.5 SI units away from saturation ( Figure 2-4 ). This modeling also shows that log P CO2 values range over 2 orders of magnitude from -3.7 to -1.3 bars. The relationship between log P CO2 and saturation state follows two distinct trends : one is a linear corr elation between log P CO2 and SI (Trend 1 in Figure 2-4 ), while the other shows variable P CO2 with little change in SI (Trend 2 in Figure 2-4 ). Samples forming Trend 2 were collected from groundwater wells with salinities less than about 5.0 psu, and from two flank margin caves (Majors Cave and Crescent Top Cave) which have salinities rangi ng from about 2 to 35 psu. Trend 1 extends from water that is supersat urated with respect to aragonite and near equilibrium with atmospheric CO 2 to water undersaturated with re spect to aragoni te and calcite 28

PAGE 29

and P CO2 above atmospheric values ( Figure 2-4 ). Although samples from cave pools are near saturation with respect to ara gonite and calcite, they also sh ow a linear correlation between changes in SI and P CO2 ( Figure 2-4 ). The four samples with P CO2 near equilibrium with atmosphere were collected from surface water sites (Watlings Blue Hole, Six Pack Pond, and Crescent Pond) where salinities range from about 24 to 52 psu. These samples are the most highly supersaturated with re spect to carbonate minerals. A ll samples undersaturated with respect to calcite also come fr om a surface water site, Ink Well Bl ue Hole. In contrast with the other surface water sites, Ink Well Blue Hole exhibits a strong pycnocline with salinity ranging from 0.5 to 17.5 psu, but no thermocline with an average temperature of 24.2 0.4 C (1 ) ( Figure 2-5A ). Neither temperature nor salinity va ries with time of collection, but P CO2 and the saturation state of the water with respect to calcite changed from the early to late morning ( Figures 2-5B and C ). The early morning samples have pH values of 6.74 to 7.32 with lower and more erratic P CO2 concentrations and they are more undersat urated than samples collected later in the day ( Table A-1 and Figure 2-5C ). Comparison of SO 4 2versus Cl suggests sulfate reduction was not occurring at the time of sampling ( Figure 2-5D ). Influence of mixing on saturation Using the compositions of one freshwater samp le and two samples with seawater salinity but with different P CO2 values, I estimate how mixing of th ese end-member waters may control the saturation state with respec t to aragonite and calcite ( Figure 2-6 ). The freshwater end member (GRC2-06), with a salinity of 0.6 psu, has a log P CO2 of -1.83 bars and is slightly undersaturated with respect to aragonite (SI ARG = -0.03). The two saline samples (CP-05 and CT05) have salinity and chemical compositions si milar to seawater. Sample CP-05 was collected from Crescent Pond, an interior tidal lake, whil e CT-05 was collected from a pool in Crescent Top Cave about 15 m from the lake. Th e lake sample (CP-05) has a log P CO2 of -3.73 bars, near 29

PAGE 30

equilibrium with atmospheric CO 2 while the cave sample (CP-05) has a log P CO2 of -2.69 bars, which is about an order of magnitude hi gher than equilibrium with atmospheric CO 2 The lake water is supersaturated with respect to bot h aragonite and calcite with SI of 0.72 and 0.86, respectively, but the cave pool is slightly undersaturated with respect to calcite (SI CAL = -0.05) and aragonite (SI ARG = -0.19). Estimates based on models of mixed compositions of these endmember waters show how saturation states with respect to aragonite a nd calcite should vary depending on the fractions of fresh and sa line water, either with atmospheric P CO2 or elevated P CO2 in each sample ( Figure 2-6 ). These modeled mixing lines represent the expected saturation states of water on San Salvador Island following mixing of freshwater and seawater. None of the samples have saturation states that fall along th e modeled saturation state, regardless of their salinity. Generation of Macroporosity Mechanisms of Dissolution Mixing dissolution Comparison of the model of sa turation states expected from mixing of fresh and saline water to sampled water indicates that mixing processes have little affect on dissolution ( Figure 26 ). Dissolution by mixing processe s depends on mechanical mixing of water with different compositions and P CO2 Mixing of these distinct water types may be limited on small islands such as San Salvador from a combination of factors including magnitudes of recharge and the presence of well-defined transiti on zones. Transition zones form be tween the freshwater lens and underlying seawater from hydrodynamic dispersion a nd vertical mixing due to ocean tides, and their thickness depends on factors including transver se dispersivity and variations in vertical permeabilities (Bear and Todd 1960; Underwood et al. 1992; Griggs and Peterson 1993). On small carbonate islands, such factors are greatly influenced by rock deposition and diagenesis, 30

PAGE 31

and thus transition zones can vary from a few meters (Cant and Weech 1986) up to tens of meters thick (Matsuda et al. 1995; Peterson 1997), and are typically thicke st near the shoreline where tide fluctuations are greatest (e.g., Whitaker and Smart 1997b). Consequently, once a transition zone develops, its pr esence may inhibit mechanical mi xing of fresh and saline water and thus diminish the magnitude of dissolution Recent modeling suggests that flank margin caves could only form where heterogeneous hydr aulic conductivities prod uce thin transition zones, and conversely, that thick transition zones near the shoreline are unlikely to produce flank margin caves (Romanov and Dreybrodt 2006). In a ddition to a thin transition zone, this model also required an annual recharge rate of 1.11 m to produce groundwater fluxes sufficient enough to drive mixing dissolution that would produce flank margin caves. Most freshwater lenses on small islands, including San Salva dor Island, occur under narrow, inland eolianite ridges a nd back-beach dunes with limited catchment size and recharge (e.g., Davis and Johnson 1989). Thes e lenses receive recharge of about 0.3-0.4 m/yr based on the thickness of freshwater lenses (Budd and Vacher 1991). Although the amount of recharge is unknown for the Bahamas during the MIS 5e highs tand when flank margin caves formed, it seems unlikely to be as high as 1.11 m/yr requi red for dissolution accord ing to the Romanov and Dreybrodt (2006) model, considering the islands were forested during this time which would have increased evapotranspirati on (Carew and Mylroie 1997). Furt hermore, although recharge is nearly two times higher in the northern Bahama s than southern Bahamas (Cant and Weech 1986; Whitaker and Smart 1997a), flank margin caves are similar in size across the Bahamian archipelago (Labourdette et al. 2007), further suggesting flank margin caves may form where recharge is less than required by th e Romanov and Dreybrodt (2006) model. 31

PAGE 32

In contrast to small islands, large carbonate platforms, such as on the Yucatan Peninsula and South Andros Island, Bahamas, are characteri zed by mixed water that is undersaturated with respect to carbonate minerals, and could be responsible for the large caves found in these locations (e.g., Back et al. 1986; Stoessell et al. 1989; Whitaker and Smart 1997c). For example, dissolution has been observed on South Andros Island with in a bank-margin fracture characterized by vertically-extensive blue holes where vertical mixing is enhanced by strong circulation within the fracture system (Whitaker and Smart 1997c ). On the Yucatan Peninsula, mixing dissolution may lead to caves that extend about 12 km inland from the coast where an extensive catchment supplies large amounts of fres hwater, resulting in shar p transition zones that range from about 0.4 to 4 m thick in the c onduits (Beddows 2004; Smart et al. 2006). These hydrologic conditions, however, differ greatly fro m environments where most flank margin caves are found and the generated caves have di fferent morphologies from flank margin caves. Mixing of vadose and phreatic water has also been suggested as an important mechanism to dissolve carbonate minerals at the water table (Thrailkill 19 68; Bgli 1980), and possibly in the development of flank margin caves (Mylroie and Carew 1990) Recent results from North Andros Island, Bahamas suggest mixing at the top of the freshwater lens does not drive significant amounts of dissolution on carbonate islands (Whitaker and Smart 2007b). A more likely cause of the dissolution is from elevated CO 2 concentrations at the top of the water table from microbial oxidation of organic matter in soil and vadose zones (Whitaker and Smart 2007b). Role of CO 2 in dissolution and precipitation reactions All samples from San Salvador Island e ither follow a linear correlation between P CO2 and SI (Trend 1 in Figure 2-4 ) or remain near satu ration regardless of P CO2 (Trend 2 in Figure 2-4 ). The difference between these trends relate to the uptake and re lease of gaseous CO 2 from the 32

PAGE 33

water to the atmosphere and from buffering by water-rock interactions. Of the samples following Trend 1, those from Ink Well Blue Ho le exhibit the largest range in P CO2 but all remain undersaturated with respect to calcite regardless of the c oncentration of P CO2 These variations in P CO2 likely result from respiration and photosynthesis within the blue hole, which has fresh to brackish water and a large amount of sub-aqua tic vegetation, following a reaction such as: 222CO+HOCHO+O 2 (2-2) where CH 2 O represents solid organic matte r. Respiration would elevate P CO2 values, thereby lowering pH and carbonate mineral saturation stat e as shown by the samples collected in the early morning while cloud cover limited sunlig ht. Photosynthesis would become increasingly important during mid-morning as sun light increased, thereby consuming CO 2 increasing the pH and reducing undersaturation ( Figures 2-5B and C ). Water in the blue hole also probably remains undersaturated because limited contact between the water and the sides of the blue hole (about 7 m away) would reduce the buffering infl uence of the carbonate minerals. In contrast to Ink Well Blue Hole, other locations open to the atmosphere, including Watlings Blue Hole, Six Pack Pond, and Crescent Pond, are supersaturated with respect to aragonite and calcite ( Figure 2-4 ). All of these locations have near marine salinity and show diurnal tidal fluctuations with pe riodicities that are similar to, but out of phase with, ocean tides, suggesting they are connected to the oceans th rough preferential flow systems (Teeter 1995; Crump and Gamble 2006). Unlike the water chemistr y at Ink Well Blue Hole, carbonate mineral saturation at these locations re flect influence of mixing with open ocean water, rather than uptake and release of CO 2 during photosynthesis and respiration. Unlike surface-water samples, samples collected from groundwater wells and two flank margin caves have equilibrated with aragonite and calcite, regardless of the magnitude of the 33

PAGE 34

flux of CO 2 to the water. Sources of CO 2 at these sites could incl ude root respiration and oxidation of organic matter, and these processes are heterogeneously distributed in the vadose zone and along the water tabl e (Atkinson 1977; Wood and Petr aitis 1984; Whitaker and Smart 2007b). The range of P CO2 along Trend 2 reflects this he terogeneous distribution of CO 2 with the flank margin cave samples clustered around log P CO2 of -2.7 bars and groundwater samples ranging over an order of magnitude, from a value sim ilar to those in the flank margin caves to a log P CO2 of -1.3 bars ( Figure 2-4 ). The two samples collected from Crescent Top Cave have marine salinity, but those from Majors Cave were collected from a 5 m thic k transition zone with salinities ranging from about 2 to 30 psu, similar to the transition zone in Ink Well Blue Hole ( Table A-1 ). No photosynthesis occurs in this pool because of its location in a cave, and thus CO 2 concentrations do not have diurnal variations. Regardless of the lack of variations in CO 2 concentrations, their satu ration state is not controlled simply by mixing of fresh and saline water ( Figure 2-6 ). If elevated CO 2 concentrations control carbonate mine ral dissolution, then concentrations of excess Ca 2+ should correlate to changes in P CO2 reflecting the buffering capacity of carbonate mineral dissolution. This relationship occurs at th e Line Hole well field wh ere concentrations of excess Ca 2+ (2 to 4 mM) correlate with log P CO2 (-2.3 to -1.3 bars) over a distance of about 300 m extending across the freshwater lens ( Figure 2-7 ). Heterogeneity of the system is shown by the fact that the highest and lowest concentrations occur only 60 m apar t. Regardless of the variation in P CO2 and excess Ca 2+ concentration, all of the samples ar e within 0.2 SI units of saturation with respect to aragonite, reflecting fast equi libration between water and carbonate minerals as the CO 2 concentration of the water changes. Rapid variation in P CO2 could result from the local generation of CO 2 in the vadose zone (e.g., Whitaker and Smart 2007b). The high porosity of 34

PAGE 35

eogenetic rocks allows sufficient contact between the air mass in the vadose zone and the water table. If the fugacity of CO 2 in the vadose zone exceeds the P CO2 in the water, CO 2 will diffuse into the water, resulting in te mporary undersaturation with respect to carbonate minerals, which is then modified by mineral dissolution. Conversely, CO 2 degassing will occur once the P CO2 in the water exceeds CO 2 fugacity in the vadose zone, leading to carbonate mineral supersaturation and ultimately precipitation. Although these waters remain saturated with respect to carbonate minerals, the high range in excess Ca 2+ suggests heterogeneous disso lution and prec ipitation can occur across the water table. Di ssolution in these watertable settings has been suggested for the development of banana hol es (Harris et al. 1995). In summary, these results indicate disso lution is a result of variations in CO 2 concentrations in the water on San Salvador Island. Enrichment of Ca 2+ in low salinity water indicates that most of the disso lution has occurred since the freshwater recharged the island, and is a near-surface phenomenon (e.g., Vacher et al. 1990; Martin and Moore 2008). The correlation between P CO2 and excess Ca 2+ suggests that while groundwaters remain near saturation at the water table, increases in CO 2 in the vadose zone can lead to further dissolution. Although this mechanism may lead to a significant increase in dissolution at the wate r table, the following section explores if CO 2 -driven dissolution is sufficiently fast to form macroporosity at the edge of the freshwater lens where short resi dence times may restrict reprecipitation. Calculation of Dissolution Rates Under steady-state conditions, all recharge to a fr eshwater lens is lost to the ocean annually along with any CaCO 3 dissolved in the water (Plummer et al. 1976; Ant hony et al. 1989; Vacher et al. 1990). This dissolved CaCO 3 would represent the amount of secondary porosity formed, and when coupled with estimates of the volume of recharged water coul d provide information on the amount of secondary porosity forming in a particular region (e.g., Plummer et al. 1976; Budd 35

PAGE 36

1988; Anthony et al. 1989; Vacher et al. 1990; McClain et al. 1992; Whitaker and Smart 2007a). I make these estimates for the Sandy Point region of San Salvador, where a well-studied flank margin cave has been mapped, depositional and di agenetic timing are know n, and the freshwater lens can be constrained by t opographic relief and glacio-eust atic sea-level variations. Average recharge across San Salvador Island is estimated to be 25% of annual rainfall or around 300 mm/yr (Cant and Weech 1986; Budd a nd Vacher 1991). Effective porosity is estimated at 30% (Vogel et al. 1990 ). The size of the freshwater lens at Sandy Point is estimated to be about 1.5 km 2 for the MIS 5e highstand by assuming the 6 m contour interval represents the boundary of the lens (Carew and Mylroie 1997) ( Figure 2-1C ). Recharge over this area would provide around 4.5 x 10 8 liters of water available to disc harge to the ocean each year. The average excess Ca 2+ concentration of 2.0 mM in San Salvador freshwater ( Figure 2-2E ) represents an average amount of Ca 2+ derived from dissolution of CaCO 3 similar to values estimated from other islands in the region (B udd 1988; Whitaker and Smart 2007a). These values of discharge and excess Ca 2+ concentration suggest 2.1 x 10 -4 kg/l of CaCO 3 would be flushed to the ocean each year. With a density of 2.93 g/cm 3 for aragonite, this magnitude of dissolution reflects 46 m 3 of CaCO 3 dissolved per year or 31 m 3 /km 2 yr when normalized to lens area ( Table 2-1 ). Given a dissolution rate of 31 m 3 /km 2 yr, approximately 138 m 3 of rock would have dissolved at the location of Altar Cave (areal footprint of 4.5 x 10 -4 km 2 ) during the estimated amount of time, about 10 ky, available fo r the cave to form. A volume of 138 m 3 is only about 12% of the cave volume, which has b een mapped to be a minimum of 1200 m 3 (Florea et al. 2004). This calculation indicates disso lution rate at the site of A ltar Cave was greater than the average value across the island. 36

PAGE 37

The amount of dissolution required to form A ltar Cave can also be estimated from the volume of water flowing through the space occupied by the cave. Flank margin caves, observed on this cliff and elsewhere throughout the Ba hamas, commonly have a semi-regular spacing parallel to the paleo-shoreline, suggesting all caves at any on e location formed concurrently, rather than sequentially (Vogel et al. 1990; Labourdette et al 2007). Although dissolution voids may alter the flow field, high porosity of island carbonates suggests flow would not have been diverted much beyond the boundary of the void (Las cu 2005). Consequently, I estimate the width of the catchment supplying water to form Alta r Cave was 50 m, and if it extended about 200 m upslope to the groundwater divi de to the east, the catchment area would be about 0.01 km 2 ( Figure 2-1C ). The width for the catchment area is pr obably a maximum given that the greatest width of the cave is about 15 m (Florea et al 2004). Assuming 300 mm/yr recharge, the annual flushing rate for this catchm ent area would be about 3.0 x 10 6 l/yr, which would need to remove only 8.2 x 10 -5 kg/l of dissolved CaCO 3 away from the locus of cav e development to form Altar Cave in 10 ky ( Table 2-2 ). This amount is about 40% of the average dissolved CaCO 3 concentration of 2.1 x 10 -4 kg/l measured in freshwater samples of San Salvador Island and other islands in the region (Budd 1988; Whitaker a nd Smart 2007a). Although the magnitude of dissolution required to form Altar Cave is only 40% of that estimat ed across the entire island, the calculation suggests the disso lution rate was about 267 m 3 /km 2 yr where Altar Cave formed, or about 9 times the average value of 31 m 3 /km 2 yr ( Table 2-2 ). Conclusions Diagenesis on carbonate islands usually dissolv es and reprecipitates carbonate sediments with little loss or gain in bul k porosity. The presence of flank margin caves and banana holes, however, indicates that secondary macroporosity forms along the wate r table and at the edge of freshwater lenses in relatively s hort time spans of thousands of years. Although the distal edge of 37

PAGE 38

freshwater lenses has been proposed to experience mixing dissolution, the models assume mechanical mixing between fresh and saline water end members. I suggest such mechanical mixing may be limited on small carbonate islands due to factors including limited recharge and thick transitions zones that inhibit mixing of the end members. Instead, most dissolution on San Salvador occurs in freshwater from elevated concentrations of CO 2 This dissolution is sufficient to produce flank margin caves but only when the hydrodynamics of the fr eshwater lens flush reaction products from the zone of dissolution. Within a freshwat er lens, specific discharge is elevated at the water table and increases to a maximum at the seaward edge of the lens. Where local sources of CO 2 drive dissolution near the water table at the lens edge, flank margin caves may develop as reactions products are flushe d to the ocean. Where local sources of CO 2 drive dissolution at the water table, banana holes ma y form. Banana holes are smaller than flank margin caves since lower flow velocities away fr om the lens edge may provide sufficient time for some reprecipitation. This hypot hesis predicts that cementa tion may be greater around the boundaries of banana holes than in areas that lack these caves. Mass balance calculations indicate that with focused dissolution, only about 40% of the carbonate found to be dissolved annually on San Salvador Island is required to form the observed flank margin caves. Because all excess Ca 2+ must be derived from carbonate dissolution, macroporosity through small ocean islands may be more widespread than is obser ved, possibly because not all caves have been breached or because some of the secondary porosity is not large enough for human exploration. Results of this study suggest th at distribution of resources, su ch as water and hydrocarbons, in ancient carbonate platforms may be concentrated in areas where local sources concentrate CO 2 and flow is sufficient to remove reaction pr oducts from the region wh ere dissolution occurs. 38

PAGE 39

Table 2-1. Calculated dissoluti on in a freshwater lens at Sandy Point during the MIS 5e highstand. Freshwater lens at Sandy Point Lens area (km 2 ) 1.5 Rainfall (mm/yr) 1200 Recharge (mm/yr) a 300 Flushing (l/yr) 4.5 x 10 8 Dissolved CaCO 3 (kg/l) b 2.1 x 10 -4 Dissolution (kg/yr) 9.5 x 10 4 Volume dissolved (m 3 /yr) c 46 (m 3 /km 2 yr) 31 a Based on estimated recharge of 25% of average total rainfall. b Based on average excess Ca 2+ in freshwaters from San Salvador Island. c Assuming aragonite density of 2.93 g/cm 3 and porosity of 30%. Table 2-2. Calculation of dissolution at Altar Cave. Lens section for cave development Catchment zone (km 2 ) 0.01 Rainfall (mm/yr) 1200 Recharge (mm/yr) 300 Flushing (l/yr) 3.0 x 10 6 Area of Alter Cave (km 2 ) a 4.5 x 10 -4 Required Dissolved CaCO 3 (kg/l) 8.2 x 10 -5 Dissolution (kg/yr) 246 Volume dissolved (m 3 /yr) 0.12 (m 3 /km 2 yr) b 267 a Areal footprint. b Dissolution rate for the formation of Altar Cave in 10 ky 39

PAGE 40

Figure 2-1. Maps of Bahamian archipelago and San Salvador Island. A) Map of Bahamian archipelago showing location of San Salva dor Island. B) Map of San Salvador Island showing location of sample sites and locat ion of Sandy Point. Gray area is land and white area is water. C) Map of Sandy Point in the southwest portion of San Salvador Island showing location of Altar Cave a nd recharge catchment for Altar Cave. Contour lines represent presen t-day land elevation in meters above sea level. Stippled area shows the area of the freshwater le ns at Sandy Point during the MIS 5e highstand, and dashed line approximates th e ground-water divide. Catchment area shows estimated maximum area (~0.01 km 2 ) of the freshwater lens that would contribute flow to Altar Cave location. 40

PAGE 41

Figure 2-2. Plots of major ion versus chloride co ncentrations in all samples. A) Plot of Mg 2+ B) Plot of Na + C) Plot of K + D) Plot of SO 4 2. E) Plot of Ca 2+ F) Plot of Sr 2+ G) Plot of Alkalinity. Dashed line in each plot re presents conservative mixing between freshwater and seawater end members. 41

PAGE 42

Figure 2-3. Comparison of DIC and total alkalinity in samples co llected in June 2006 and April 2007. Solid line is a linear regressi on DIC and total alkalinity. 42

PAGE 43

Figure 2-4. Relationship between PCO 2 and aragonite and calcite saturation in all water samples. Carbonate mineral saturation in sample s following Trend 1 is controlled by PCO 2 Conversely, samples following Trend 2 are in equilibrium with respect to aragonite regardless of PCO 2 43

PAGE 44

Figure 2-5. Water samples collected from Ink Well blue hole in April 2007. Black circles represent first series of samples collected in the morning during a one hour period with cloud cover, and white circles repres ent a second series of samples collected mid-morning with no cloud cover immediately following the first series of sample collection. A) Salinity versus depth. B) Comparison of PCO 2 versus depth. C) Comparison of calcite undersaturat ion versus depth. D) Plot of SO 4 2versus Cl Solid line represents expected SO 4 2concentration due to conservative mixing. Three white circles have similar chemistry, and thus ove rlay each other near the graphs origin. 44

PAGE 45

Figure 2-6. Saturation indices (SI) for all San Salvador waters re lative to aragoni te and calcite versus percent seawater. Curved lines show SI for mixing between freshwater sample GRC2-06 and seawater sample CP-05 (solid line) and saline water sample CT-05 (dashed line). 45

PAGE 46

Figure 2-7. Plot of excess Ca 2+ saturation index of aragonite, and log P CO2 of water from the Line Hole well field, showing a strong correlation between P CO2 and excess Ca 2+ The wells penetrate the top of a single fres hwater lens, and are aligned almost perpendicular from the shore from about 200 to 500 m. 46

PAGE 47

CHAPTER 3 GEOCHEMICAL AND STATISTICAL EVIDENCE FOR RECHARGE, MIXING, AND CONTROLS ON SPRING DISCHARGE IN AN EOGENETIC KARST AQUIFER Introduction Karst aquifers are characterized by heterogeneou s distributions of various types of porosity including intergranular porosity within the matr ix rock, fractures along jo ints, faults and bedding planes, and conduits enlarged thr ough dissolution. This porosity dist ribution influences nearly all aspects of aquifer characteristic s, including aquifer storage and di stribution of permeability. The range of porosity and permeability determines flow paths and allows extreme flow rates including both laminar and tu rbulent flow (Worthington 1994; Quinlan et al. 1996; White 1999; Halihan et al. 2000). While most flow in karst aquifers occurs through conduits, storage is primarily in the matrix porosity (e.g., Worthingt on et al. 2000). Matrix porosity and permeability also affect recharge to the aquifer, which can vary on seasonal and individual storm time scales. Karst aquifer recharge commonly occurs as point s ource (allogenic) recharge into swallets or as diffuse recharge through the va dose zone (e.g., White 1988; Ford and Williams 2007; Ritorto et al. in press). Upward flow of water from deep within an aquifer may also contribute to an aquifers shallow water budget and chemistr y, depending on the distribution of porosity, permeability, and hydraulic head (Kohout et al. 1977; Smith and Fuller 1977; Hughes et al. 2007). Flow paths and sources of recharge to kars t aquifers have long been assessed through physical and chemical variations in springs (e.g., Shuster and Wh ite 1971; Dreiss 1989). In most well-studied cases, springs are in regions with dense, recrystallized rocks (i.e., telogenetic karst, Vacher and Mylroie 2002), where low matrix perm eability restricts most of the flow to conduits and fracture networks. Individual springs exhibiting large variati ons in discharge and chemical composition through time have been inferred to be dominated by allogenic recharge and conduit 47

PAGE 48

flow. In contrast, springs with smaller amounts of chemical variability a nd discharge have been inferred to be dominated by diffuse recharge and diffuse flow through frac ture networks (Shuster and White 1971; Ternan 1972; Smart and H obbs 1986; Hess and White 1988; Dreiss 1989). These studies assume that spring variability resu lts largely from variation in recharge and the flow paths of that recharge. Considering only these few parameters limits the understanding of the karst system that can be derived from variations in spring-water chemistry. For example, physical and chemical variations in springs issu ing from the karstic In ner Bluegrass region of Kentucky fail to reflect the geometry of the a quifers conduit system because differences in lengths of flow paths mask variations in c onduit sizes that source the springs (Scanlon and Thrailkill 1987). Consequently, a qu estion I explore in this paper is what additional insight can be gained from physical and chemical monitori ng of spring flow and chemical composition. Large springs also discharge from carbonate rocks that retain high matrix porosity and permeability (i.e., eogenetic karst, Vacher a nd Mylroie 2002). In these rocks, high matrix permeability allows access to aquifer storage and di ffuse recharge, which constitute a substantial component of spring discharge (e.g., Florea and Vacher 2006; Ritort o et al. in pr ess). Numerous springs that discharge from the eogenetic Upper Floridan aqui fer (UFA) appear to be fed primarily from diffuse recharge transmitted thro ugh the rock matrix (e.g., Martin and Gordon 2000; Florea and Vacher 2006). Othe r springs discharging from the UFA are directly connected by conduits to allogenic inputs so that the so urce of water to these springs depend on the hydraulic head gradient between the conduit a nd surrounding aquifer matr ix (Katz et al. 1998; Screaton et al. 2004; Loper et al. 2005; Martin et al. 2006). When allogenic inputs allow conduit hydraulic head to exceed head in the surrounding matrix, allogenic recharge accounts for most to all of the spring discharge, with an additional fraction of the recharge stored temporarily in the 48

PAGE 49

matrix until the flood pulse recedes and hydraulic head gradients reverse (e.g., Screaton et al. 2004; Martin et al. 2006). Following head reversal spring discharge is a mi xture of water stored temporarily in the matrix, allogenic recharge, and water recharged diffusely to the matrix from the surface. This interaction between allogenic re charge and diffusely recharged water can lead to high variability in discharge and spring-water chemistry (e.g., Katz et al. 1998; Crandall et al. 1999; Katz et al. 2001; Martin and Dean 2001; Katz 2004; Katz et al. 2004; Screaton et al. 2004; Martin et al. 2006). Assessing origins of water, wh ich is needed to unders tand susceptibility of karst areas to contamination, requires a clear und erstanding of processes causing variations in groundwater chemistry and connectivity betwee n conduit and matrix porosity (e.g., McConnell and Hacke 1993; Plummer et al. 1998; Katz 2004). In this chapter I use major element chemis try, physical conditions including river stage, precipitation, evapotranspiration (ET), temperature gradients of groundwater, and a multivariate statistical method (principal component analysis ; PCA) to evaluate how multiple sources of water and variations in aquifer flow paths infl uence a first magnitude sp ring draining a portion of the eogenetic UFA. I suggest that knowledge of the spatia l and temporal variation of groundwater chemistry is necessary to separate sources of water and co mponents of flow, which cannot be resolved by only monitoring spring disc harge, and that mixing of these water sources plays an important role in temporal variations of spring chemistry. Because of the importance of groundwater sources to spring-water chemistry, matr ix porosity in eogenetic aquifers appears to be significant to spring discharge even where dominated by conduits. Study Area The Santa Fe River is a tributary of the Su wannee River, with a watershed covering about 3600 km 2 in north-central Florida (Hunn and Slack 1983). Land use in the watershed is mainly agricultural, primarily as improved pastures and row and field crops (Kautz et al. 2007). In the 49

PAGE 50

watershed, Oligocene and Eocene carbonate rocks make up the UFA ( Figure 3-1 ). The aquifer is confined by the Hawthorn Group to the northeast, comprised in part of Miocene and younger siliciclastic-dominated rocks (Scott, 1988; Gros zos et al., 1992), and is unconfined in the southwest where the confining unit has been removed by erosion ( Figure 3-2 ). The erosional edge of the Hawthorn Group is referred to as the Cody Scarp (Puri and Vernon 1964). To the northeast of the scarp, surface water is common on the confining unit, but is limited to the southwest where streams crossing the scarp eith er become losing streams, sink underground and reemerge, or disappear underground with no clear point of reemergence. The Santa Fe River flows westward from La ke Santa Fe for about 40 km until it reaches the Cody Scarp, where it sinks in to a 36-m deep sinkhole at the Ri ver Sink in OLeno State Park ( Figure 3-2 ). The river flows underground through a ne twork of conduits until it reemerges about 6 km from the River Sink as a first magnitude spring, called the River Rise, marking the headwaters of the lower Santa Fe River (Martin and Dean 2001). The conduits rise to the surface intermittently between the River Sink and River Rise at several karst windows ( Figure 3-2 ). At the Santa Fe Sink-Rise system, the UF A is about 430 m thick, unconfined at the surface, and is covered by a thin veneer (about 4 m, depending on land-surface elevation) of unconsolidated sands and sediments (Miller 1986). In this area, Oligo cene carbonate rocks are absent and no middle confining unit exists, resulting in the UFA extending from the Upper Eocene Ocala Limestone to the lower confining unit of the Lower Eocene Cedar Key Formation (Miller 1986) ( Figure 3-1 ). Potable water extracted from the aquifer is estimated to come from the upper 100 m of the Ocala Limest one, with more mineralized wate r in deeper po rtions of the aquifer (Hunn and Slack 1983; Mi ller 1986). Porosity and matrix permeability of the Ocala Limestone average about 30% and 10 -13 m 2 respectively (Budd and Vacher 2004; Florea 2006). 50

PAGE 51

Exploration of the submerged c onduits upstream of the River Rise has resulted in over 15 km of surveyed passage (Poucher 2007). Average dimensi ons of the conduits range from 18 to 24 m wide and 12 to 18 m high with an average depth of about 30 m below the ground surface (mbgs) (Screaton et al. 2004; Poucher 2007). The conduit system has not been completely mapped from the River Sink to River Rise, but high flow rates detected by natural and ar tificial tracers show the two locations are linked by conduits (Hiser t 1994; Martin and Dean 1999; Moore and Martin 2005). Previous work has shown that water discha rging from the River Rise varies between sources from the River Sink and from groundwater, defined here as water stored in the aquifer surrounding the conduits (e.g., Martin and Dean 1999; Martin and Sc reaton 2001; Screaton et al. 2004; Martin et al. 2006). During high flow, discharg e at the River Rise is mostly derived from water entering the conduit system at the River Si nk. As river stage and in put into the River Sink decreases, increasingly larger percentages of groundwater drain from the surrounding aquifer into the conduit system to discharge at the River Rise (Martin and Screaton 2001). Methods River Stage and Potential Recharge Stage of the Santa Fe River was monitore d about 200 m downstream of the River Rise with an automatic pressure tran sducer with an accuracy of .03 m. A separate barometric data logger (.0045 m) was used for barometric co mpensation of the non-vented transducer. The water levels were recorded at 10-minute intervals, and the data were downloaded from the recorder at fourto five-week intervals. When the data were downloaded, the river stage was measured from a staff gauge, and the recorded water level was referenced to the gauge for each download period to correct for dr ift. The relationship between st age and discharge at the River Rise was calculated based on the rating curve de veloped by Screaton et al. (2004), using data 51

PAGE 52

collected by the Suwannee River Water Management District (SRWMD). Potential recharge was estimated as precipitation minus evapotranspiration (P-ET) during the study period by Ritorto et al. (in press). Briefly, daily valu es of P-ET are estimated using daily precipitation data collected in OLeno State Park using an automated rain gauge maintained by SRWMD ( http://www.srwmd.state.fl.us/index.asp?NID=99 ), and the Penman-Monteith model for estimating daily ET, which estimates water loss to the atmosphere from a vegetative surface (Dingman 2002). Field Sampling and Laboratory Analysis Sixteen sampling trips were conducted from January 2003 to April 2007 to collect water from eight groundwater monitori ng wells, one sinking stream (Riv er Sink), one first magnitude spring (River Rise), and four intermediate karst windows ( Figure 3-2 ). Monitoring wells were drilled to depths of about 30 mbgs, approximately at the depth of the conduits, and screened over 6-m (20 foot) depth intervals using 250 m PVC screening material attached to 51 mm (2 inch) diameter PVC linear. Groundwater samples were collected from monitoring wells using a Grundfos II submersible pump. Surf ace-water samples were collected on shore with a peristaltic pump attached to tubing that was pushed close to spring boils when vi sible, or in the deepest part of the sinkhole if no boil was pres ent. Field measurements of temp erature (T), pH, and specific conductivity (SpC) were recorded with a YS I multiprobe model 556 prior to sampling. The probe was calibrated at the star t of each sampling day, and calib ration was checked several times while in the field. All samples were collected unfiltered in high density polyethylene (HDPE) bottles. Samples collected for cations were preserved with either sulfuric (Na + and K + ) or nitric acid (Ca 2+ and Mg 2+ ) to a pH < 2.0, while samples for anions and alkalinity were collected with no preservatives. Samples were stored on wet ice until they were delivered to the laboratory for analysis. 52

PAGE 53

Concentrations of major ions (Na + K + Ca 2+ Mg 2+ Cl and SO 4 2) and alkalinity were analyzed by a NELAC-certified laboratory, Advanced Environm ent Laboratories, Inc., in Gainesville, FL. Analyses were determined in accordance with Environment Protection Agency (EPA) protocols for each analyte (EPA 1983). Data from quality-assurance samples indicate no contamination resulted from sampling procedur es and equipment, and that good analytical reproducibility occurred in the laboratory. Charge balance er rors for most samples were % except for samples whose concentrations we re near instrument detection limits. Principal Component Analysis Principal component analysis (PCA) is a multiv ariate statistical technique used to reduce the complexity of and decipher pa tterns within large data sets by determining a small number of variables that account for the grea test variance in all of the original variab les (Wold et al. 1987; Jolliffe 2002). For this study, PCA was applied to a normalized data matrix of 9 variables (river stage, pH, Cl SO 4 2, Ca 2+ Na + Mg 2+ K + and alkalinity concentrati ons) from 211 water samples using the princomp function in Matlab (S tatistics toolbox 5.0, Mathworks, Waltham, MA). Because the data have large ranges and diffe rent units of measurement (e.g., stage and concentration), data were normalized by centeri ng the data set about zero by subtracting the means of each variable set from the measured value for individual samples and dividing each value within the variable set by its standard devi ation (Stetzenbach et al 1999; Chen et al. 2007). Consequently, each variable was normalized to unit variance and thus contributes equally to the analysis. Principal components (PC) are eigenvectors of the correlation matr ix of the normalized data set, and represent correlation coefficients, ca lled loadings, between each variable and each PC. Since the correlation matrix is symmetrical, the eigenvector s are orthogonal and thus each PC is projected as an uncorrelated axis in a new space that helps explain the relationship among 53

PAGE 54

data points or variables along each PC. Positive loadings show a direct relationship, and those with the strongest absolute magnitude exert the greatest influence on the PC. The first PC accounts for the greatest fraction of variance of the co rrelation matrix, followed by subsequent components reflecting less variance. Principal co mponent scores are transformed data points projected into PC space by axis rotation and correla ting the weight of each loading variable to the original, normalized data. By plotting PC scor es, similarities and disparities can be observed between the samples. For example, PC scores that cluster show thei r variance results from similar variable loadings, and thus suggest simila r processes influence the samples. Furthermore, PC scores that vary al ong linear trends suggest variable loadings that affect these samples may exhibit some systematic variations such as, in th is case, through time or with changing stage. In contrast, dissimilar PC scores show samples th at are unrelated, likely suggesting these samples are influenced by independent processes. Estimate of Vertical Flow Rate from Temperature Perturbations Upwelling of deep water can be estimated assuming vertical flow within the UFA drives heat transfer following one-dimensiona l steady-state flow described by: 2 wwZ ZZ 2c v TT = 0 zkz (3-1) where T z is temperature at depth z, w is density of water, c w is heat capacity of water, v z is vertical Darcy velocity of water (positive for downward flow and negative for upward flow), k is thermal conductivity of the porous material (Bredehoeft and Papadopul os 1965). With boundary conditions of T O as the uppermost T at z = 0 and T L as the lowermost T at z = L yields a solution to Equation 3-1 for T Z of: 54

PAGE 55

ZOLOz exp -1 L T = T + T-T exp -1 (3-2) (Bredehoeft and Papadopulos 1965), where L is the thickness of the vertical section and is the dimensionless Peclet number for heat transfer: wwZ cvL = k (3-3) Rearranging Equation 3-3 for v z yields an expression for th e vertical flow rate of: Z wwk v = cL (3-4) Results River Stage and PE-T Average river stage during the entire study peri od was 10.2 masl with an average discharge of about 16 m 3 /s ( Figure 3-3 ). Samples collected during trips S-2, S-9, and S-11 occurred during high flow events when the river was above aver age stage. All other sa mples were collected during average or low flow. Within the study area, changes in river stage appear to correlate po sitively over long time periods with P-ET, but this relationship seems to breakdown for individual events suggesting that antecedent conditions are important to river stage and discharge ( Figure 3-3 ). Between January 2003 and April 2007, average annual P-ET was about 400 mm (Ritorto et al. in press). The maximum annual P-ET of about 990 mm occurred in 2004 due to an active hurricane season (see Florea and Vacher 2007), which resulted in the highest stage of 14.1 ma sl. This high stage occurred immediately after Hu rricane Frances delivered a to tal of 400 mm of P-ET to OLeno State Park over a 6-day period in September 2004 ( Figure 3-3 ). The lowest stage of 9.6 masl 55

PAGE 56

occurred in April 2007 following a year-long drought that resulted in the ar ea receiving a total of 83 mm of P-ET. In addition to long-term events affecting stag e, short-duration storms that exceed ET also cause variable responses in rive r stage, but these events did no t cause a systematic response in the river ( Figure 3-3 ). Six rain events that produced a total of 232 mm of P-ET over a 39-day period in February and March 2003 caused a 3m rise in river stage on March 13, 2003. In February 2004, about half of the P-ET in Fe bruary and March 2003 (144 mm) produced only a tenth of the increase in river st age (0.33 m) seen the previous year. Conversely, only about 82 mm of P-ET over an 11-day period in March 2005 resulted in a 1.8-m rise in river stage on March 30, 2005. Water Temperature and Chemistry Temperature of the water at the surface wate r sites vary depending on the air temperature (e.g., Martin and Dean 1999), but temperature of the groundwater is more consistent, although variable among the wells. Temperatures at all wells, except Well 2, aver aged around 21C with small variations between sampling times ( Table 3-1 ). These measured temperatures are similar to average air temperature in the region as well as the typical temperatur e of water discharging from the regional springs (Hunn and Slack 1983). In contrast, water temper atures are higher and more variable at Well 2 than all other wells, ra nging from about 22 to 26C, with the highest temperature measured follo wing a one-year drought. The chemical variations from two surface-w ater sites (River Si nk and River Rise) and three groundwater wells (Wells 2, 4, and 7) are shown in a Piper diagram ( Figure 3-4 ). A statistical summary of the major chemistry is shown in Table 3-1 These five sites show the greatest variation in water chemistr y, and all of the other sites that were sampled during the study (data not reported here) have chemical compositions similar to one of these five sites. These five 56

PAGE 57

sites are thus used to represent the con tinuum of water chemistry across the region ( Figure 3-4 ). The variation in water chemistry reflects thr ee end-member sources that develop two mixing trends. One trend extends from one end member characterized by a Ca-HCO 3 composition ( in Figure 3-4 ) to another with Ca 2+ and Mg 2+ as the primary cations, but with more SO 4 2and less HCO 3 as the charge-balancing anion ( in Figure 3-4 ). The composition of Well 4 reflects the Ca-HCO 3 -type end member. Water from Well 2 has a chemical composition reflecting a strong influence from the Ca-Mg-SO 4 -type end member, although the high CV of major element concentrations, SpC, and T suggests contributions of this end member are variable at this site ( Table 3-1 ). For example, SpC and T at Well 2 range from 488 to 1315 S/cm and 22 to 26 C, respectively. Well 7 falls along the mixing trend between Well 4 and 2, suggesting it may be influenced by both sources of water ( Figure 3-4 ). The third end member is characterize d by elevated concentrations of Na + and Cl and occurs at the River Sink at high flow ( in Figure 3-4 ). This end member develops a second mixing trend that is confined to water collecte d from the surface-water sites, but this trend reflects extensive mixing between all three end members. During certain sample trips (e.g., S-3, S-4, S-6, S-10, and S-12 through S-16), water from the River Sink and River Rise fall along the mixing line between Wells 4 and 2, reflecting little influence from the Na-Cl-type end member ( Figure 3-4 ). Principal Component Analysis Principal component analysis identifies wh ich of the measured components provide the greatest variation in the compos ition of the water (e.g., Stetzenbach et al. 1999, 2001; Chen et al. 2007; Fournier et al. 2007). The first three PCs (eigenvalues of 4.87, 2.13, and 1.16, respectively) explain a total of 91% of the variance, or 54%, 24%, and 13%, respectively ( Table 3-2 ). When PC 1 and 2 are considered together, differences in the loadings are represented by 57

PAGE 58

two clusters and one outlier ( Figure 3-5A ). One cluster shows a str ong positive loading (loading > 0.3: italics font Table 3-2 ) of Na + Mg 2+ K + Cl and SO 4 2on PC 1. While these components carry similar weights on PC 1, only K + also has a strong positive loading on PC 2 followed by weaker positive loadings of Cl and Na + Although SO 4 2and Mg 2+ are heavily loaded on PC 1, they show no loading on PC 2. The ot her cluster of pH, alkalinity, and Ca 2+ has a weak positive loading on PC 1 and a strong negative loading on PC 2. The single variable that plots as an outlier in Figure 3-5A is river stage, which has a weak ne gative loading on PC 1 and a strong positive loading on PC 2. When PC 2 and PC 3 are considered together, a strong inverse relationship exists between the pH and river stage, suggesting that these two components are responsible for most of the 13% variance on PC 3, since Ca 2+ has similar loadings on both PC 2 and PC 3 and alkalinity remains negativ ely loaded on PC 3 although less on PC 2 ( Table 3-2 ). The PC scores for each sample are calculated as the sum of the PC loading times the normalized values for that sample, e.g., 4PC 1 score = 0.08(pH) + 0.42(Cl ) + 0.44(SO) + 0.27(Ca) + 0.44(Na) + 0.44( Mg) + 0.39(K) + 0.05(alkalinity) 0.10(stage). (3-5) These values thus represent the relative influence each loading has on the water sample for a given PC. While all surfaceand groundwater site s were included in the PCA, only sites that reflect the greatest variation in water chemis try and most closely define the end-member compositions (i.e., those sites shown on Figure 3-4 ) are plotted in Figure 3-5B The advantage of plotting PC scores in this fashion over using Piper diagrams is that the variation in samples can be observed at a hi gher resolution, thereby revealing additional information and relationships previously unrecognized (e.g., Melloul and Collin 1992; Laaksoharju et al. 1999; Olofsson et al. 2006). Fo r example, the strong positive loading of K + on both PC 1 and 2 suggests multiple sources of K + such as dissolution of K-bearing minerals, 58

PAGE 59

application of fertilizers to the land surface, and seawater. This information is masked in the Piper diagram because Na + and K + are grouped together during ion balancing. Although samples from Wells 2, 4, and 7 lie along th e mixing trend between the Ca-HCO 3 and Ca-Mg-SO 4 -type end members in the Piper diagram ( Figure 3-4 ), their projection in PC space allows observations of additional relationshi ps and disparities ( Figure 3-5B ). For example, water from Wells 4 and 7 have slightly negative PC 1 scores with minimal variability, but show greater variability on PC 2. Conversely, water from Well 2 is highly variab le on both PC 1 and PC 2 scores with the strongest positive PC 1 scores of any water sa mpled. In addition to the groundwater samples, surface-water samples from the River Sink and Ri ver Rise show some variance on PC 1, which are scattered and overlap each othe r on the negative side, but separa te into two distinct groups on the positive side. Most of the variance in thes e samples occurs on the positive side of PC 2, which relates directly with st age and inversely with loadi ngs of pH, alkalinity and Ca 2+ Discussion Temporal variations in spring discharge and chemistry have often been used to understand groundwater flow paths and sources of recharge in both telogenetic and eogenetic aquifers because springs are commonly assumed to reflect pr ocesses that occur over large scales and may be the only point of access to the groundwater (e.g., Shuster a nd White 1971; Dreiss 1989; Katz 2004; Vesper and White 2004; Toth and Katz 2006). Recent studies, however, suggest monitoring the spatial and tempor al variations in groundwater ma y elucidate additional aquifer parameters unrecognized by only monitoring kars t springs (Scanlon 1989; Martin and Dean 2001; Toran et al. 2007). In the following section, I use representative end-member water types to describe the sources of water to the sink-ri se system, followed by a mass-balance calculation to estimate the relative contribution each source provides to spring discharge at the River Rise. Comparison of these results to physical conditions, including river stage, precipitation, and ET, 59

PAGE 60

provides insight to the complex nature of the aquifer that could be overlooked if aquifer characteristics were determined only by monitori ng the spring. This analysis illustrates the importance of coupling groundwater monitoring, physical conditions, and spring discharge and chemistry when interpreting the physical and ch emical characteristics of karst aquifers. End-member Chemistry and Sources of Water Allogenic recharge When PC loadings and scores are considered together, the source of water entering the River Sink has a statistical association with stage ( Figure 3-5 ). Positive loadings of stage, K + Na + and Cl and negative loading of pH, Ca 2+ and alkalinity on PC 2 s uggest allogenic recharge at the River Sink delivers incr easing concentrations of K + Na + and Cl but dilutes pH, Ca 2+ and alkalinity as stage increases ( Figure 3-5A ). These relationships indica te that water entering the River Sink during high flow is evolved rain wate r flowing overland or in the shallow subsurface during storm events with minimal groundwater contribution (cf. Skla sh and Farvolden 1979). The evolved rain water accounts for the Na-Cl-type end member ( in Figure 3-4 ), which has an average Na + /Cl ratio of 0.81 0.19 (1 ), close to the 0.86 ratio of seawater. Seawater could be an important contributi on to major element chemistry with positive loading on PC 2 (Na + Cl and K + ), although other factors such as introduction of contaminants and reactions with siliciclastic minerals in the confining Hawthorn Group also could be important. Seawater would be the primary source of Na + and Cl to the region as sea spray becomes entrained in precipitation when tropical storms and summertime convective thunderstorms move inland from the coast. Some of the water has Na + /Cl ratios in excess of seawater values, which may reflect excess Na + due to leaching of soil particulates in the atmosphere (Junge and Werby 1958) or due to cation exchange in the siliciclastic Hawthorn Group (Rose 1989). Cation exchan ge could also remove Na + from the water, which would 60

PAGE 61

explain the Na + /Cl ratios that are below seawater value. Potassium is unlikely to be derived only from sea spray since the average K + /Cl ratio of 0.08 0.02 (1 ) exceeds by a factor of 4 the 0.02 ratio of seawater. Although K + could result from dissolution of K-bearing minerals in the Hawthorn Group (Edwards et al. 1998), these minerals occur in trace amounts that would unlikely provide the observed concentra tions. The elevated concentration of K + in allogenic water, as reflected by its str ong positive loading on PC 2 ( Figure 3-5A ), probably results from leaching of artificial fertilizers used for agriculture (Katz et al. 2001; Chelette et al. 2003). The negative loadings of pH, Ca 2+ and alkalinity on PC 2 refl ect dilute rainwater entering the River Sink during high flow. In these conditions pH values are lower than what would be expected for water buffered by dissolution of car bonate minerals, and mineral sources of Ca 2+ and alkalinity (e.g., HCO 3 ) are scarce in upper sections of the Hawthorn Group (Scott 1988). Although middle portions of the Hawthorn Group contain limestone and dolostone units (Groszos et al. 1992), the negative loadings of Ca 2+ and alkalinity and no loading of Mg 2+ on PC 2 ( Figure 3-5A ) suggest allogenic recharge has not inte racted with these carbonate minerals. Sulfate also shows no statistical association with stage on PC 2 ( Figure 3-5A ), suggesting this water has not dissolved mineral s ources of S, such as gypsum, anhydrite, or pyrite, which exist in minor amounts throughout the Hawthorn Group (Lazareva and Pichler 2007). During times of little precipitation, river stage drops as lesser amount s of runoff from the confined area contribute to river flow, and water at the River Sink trends towards an intermediate composition between the two groundwater end members ( and in Figure 3-4 ). This mixing between the three end members is observed in the PCA where River Sink scores on PC 2 show a strong positive association with stage during high flow, but become negative during low flow as loadings of pH, Ca 2+ and alkalinity exert a stronger in fluence on the composition of allogenic 61

PAGE 62

water ( Figure 3-5 ). These relationships suggest that, dur ing low flow conditions, water at the River Sink is a mixture of all ogenic runoff and groundwater from the UFA, which has a different composition than water entering the River Sink during high flow. Consequently, water entering the UFA through swallets may be time-dependent mixtures of water that originates from the surface or the surrounding aquifer depending on condi tions such as river stage, precipitation, and ET. Groundwater The differences in chemical compositions be tween water from Wells 2 and 4 reflect two distinct sources. Well 4 has Ca-HCO 3 -type water similar to most shallow groundwater of the UFA and results from rain water equilibrati ng with the Ocala Limestone (Sprinkle 1989). Although Well 4 is located only about 100 m from the conduit, its variation on PC 2 scores shows no statistical association with stage ( Figure 3-5B ). Most of the variation of Well 4 on PC 2 scores likely results from subtle changes in pH, Ca 2+ and alkalinity, whose loadings exert the greatest influence on the Ca-HCO 3 -type water ( Figure 3-5B ). Water at Well 4 is likely to originate from diffuse recharge as indicat ed by the small variations in solute/Cl ratios ( Figure 36A ). The magnitude of diffuse recharge has been shown to exceed allogenic recharge at the River Sink depending on conditions including ET soil saturation, and precipitation (Ritorto 2007). The Ca-Mg-SO 4 -type water from Well 2 results from pr ocesses other than, or in addition to, simple limestone dissolution. Although all wells are screened at similar depths below the land surface, water collected from Well 2 is the most mi neralized in the region with the highest major element concentrations and SpC. Well 2 also has the highest T of all water collected ( Table 3-1 ). Consequently, the positive loadings of K + Cl Na + Mg 2+ and SO 4 2on PC 1, coupled with the strong positive PC 1 scores of Well 2, suggest this water source deliv ers most of these ions to the 62

PAGE 63

sink-rise system ( Figure 3-5 ). Water with similar SO 4 2concentrations (in excess of 400 mg/l, i.e. about 4.2 mmol/kg H 2 O) was previously observed from a municipal well in High Springs, FL (less than 5 km away from We ll 2) that was open to the UFA from about 105 to 150 mbgs (Hunn and Slack 1983). The nearby presence of deep, mine ralized water could reflect a source of water that would give Well 2 its unique chemical comp osition. While the source of mineralized water deep within the Floridan aquifer system ha s not been determined (e.g., Phelps 2001), the increased salt contents cannot result from only mixing with seawater. Comparing ratios of dissolved components to Cl concentrations to their seawater values suggests the mineralized water at Well 2 has concentrations of Mg 2+ and SO 4 2that exceed values expected from seawater fractions by a factor of 11 and 49 tim es, respectively, and the average Na + /Cl ratio of 0.96 0.06 (1 ) ( Table 3-1 ) at Well 2 is about 10% higher than the seawater value. Nonetheless, elevated concentrations of K + at Well 2, as reflected by it s strong positive loading on PC 1 ( Figure 3-5A ), suggests dilute seawater deep within the aquifer may account for some of the mineralized water since this is the likely source of K + in the UFA (Sprinkle 1989). Other than seawater as a source of salts at Well 2, water-rock reactions could provide its elevated ion concentrations. Elevated concentr ations could result from water reacting with minerals in leaky portions of the Hawthorn Group which then moves along deep flow paths due to regional head gradients (Lawrence and Upc hurch 1982). An alternate explanation for the elevated concentrations could result from evaporite dissolution and dedolomitization occurring deep within the aquifer (e.g., Plummer 1977; Hans haw and Back 1979; Jones et al. 1993). In the lower portions of the UFA, evaporite minerals and dolomite are known to occur (Miller 1986) ( Figure 3-1 ). Dissolution of gypsum or anhydrite releases Ca 2+ and SO 4 2, which initiates calcite 63

PAGE 64

precipitation and subsequently promotes add itional dissolution of gypsum or anhydrite and dolomite if present (Plummer and Back 1980). Although near-surface reactions in the Hawthorn Group could elevate ion concentrations in the UFA, dissolution of evaporite minerals and do lomite in deeper portions of the aquifer are likely responsible for the observed concentrations at Well 2. These processes would elevate concentrations of SO 4 2, Mg 2+ and Ca 2+ but would not increase the concentration of K + ( Figure 3-6B ). Dissolution of Ca-bearing minerals, however, would not explain the linearity between Na + and Cl or the value of Na + /Cl molar ratio of 0.96 0.06 (1 ), which is similar to the Na/Cl molar ratio of halite and suggest halite dissolution although no halite has been reported in the Floridan Aquifer system (Miller 1986). Influences of Vertical Flow on Shallow Water Chemistry Most work on groundwater flow at the study site and other karst systems has focused on horizontal flow through conduits and surrounding aquifer following rapid recharge through swallets and discharge from springs (Martin and Dean 2001; Screaton et al 2004; Martin et al. 2006; Ritorto et al. in press). Few studies have considered vertical flow through karst aquifers or the geographic distributions and controls of wh ere vertical flow could occur (e.g., Sprouse 2004). The chemical variations at Well 2, where measur ed temperatures are significantly higher than surrounding wells, indicate that upward flow is important in the region, which I estimate below using Equations 3-2 and 3-4 ( Figure 3-7 ). For T Z I use a measured T of 26 C at Well 2, which represents the highest T observe d at Well 2 and occurred followi ng a one-year drought (S-15 and S-16, Figure 3-3 ). The drought may have increased hydrauli c head differences between the deep and shallow portions of the aquifer as drought co nditions have greater effect on the shallower portions of the aquifer. In addition to head differences, the drop in river stage during the drought 64

PAGE 65

minimizes flow through conduits, thereby reducing horizontal flow which may also alter the T at the well (e.g., Lu and Ge 1996). Considering this conceptualization of vertical flow at Well 2, I estimate z and L to be 23 and 423 m, respectively during this time ( Figure 3-7 ). Although there is no water-table well near Well 2, I estimate T O to be about 21 C based on the av erage groundwater T (e.g., Wells 4 and 7, Table 3-1 ) and average air T for the area (Hunn and Sl ack 1983). I estimate a temperature of 28.6 C for T L at the base of the UFA, assuming an average geothermal gradient of about 1.8 x 10 -2 C/m across the region (Reel and Griffi n 1971; Smith and Lord 1997). Solution to Equation 3-2 using these T values suggests at Well 2 is about -19.6. Using a heat capacity of 4184 J/kg C, density of 1000 kg/m 3 for water, and thermal conductivity of limestone of 3 W/m C (Deming 2002), Equation 3-4 yields an upward Darcy velocity at Well 2 of about 1 m/yr. Although I observe the temperature anomaly re sulting from vertical flow only at Well 2, the deep-water source appears to have a significant impact on the regional shallow-water chemistry as shown by the chemical compositions at Wells 2 and 7, River Sink, and River Rise. During low flow conditions, water at the River Si nk and River Rise appear to be intermediate mixtures of the groundwater end members ( and in Figure 3-4 ), although water from the River Sink lies closer to the Ca-HCO 3 -type end member while the River Rise lies closer to the Ca-Mg-SO 4 -type end member ( Figure 3-4 ). This difference in water chemistry at low flow suggests the River Rise receives a greater contribution from the deep-water source than the River Sink. Dilution of the deep-water source at Well 2 is shown by the variation in Well 2 scores on PC 1, which changes with SpC ( Figure 3-5B ). As dilute allogenic water reaches Well 2, the concentrations of K + Cl Na + Mg 2+ and SO 4 2decrease, resulting in PC 1 scores plotting 65

PAGE 66

towards the graphs origin ( Figure 3-5B ). These changes suggest that Well 2 is more closely linked to surface water than the other we lls, possibly through unmapped conduits ( Figure 3-2 ). The deep-water source at Well 2 requires greater vertical permeability than the other wells. Higher permeability could result from vertical fract ures that would provide a flow path for deep water, and if these fractures ar e linked to the conduit sourcing the River Rise, could explain the greater influence of deep water there than at the River Sink ( Figure 3-4 ). The only other location with a signal from the deep-water source is Well 7 ( Figure 3-4 ), but its location is about 1 km away from the closest known conduit ( Figure 3-2 ). Although simulations of regional groundwater flow suggest water upwelling from d eep flow paths exert little influence on firstmagnitude springs draining the UFA (Bush and Johnston 1988), deep water at the sink-rise system suggests heterogeneous permeability ca n greatly alter groundwat er flow fields and reflects the importance of multiple flow paths in karst aquifers (e.g., Knochemus and Robinson 1996). Effects of Source Water and Flow Paths on Spring Discharge Volumes of allogenic and diffuse recharge have been estimated for the River Rise (e.g., Martin and Dean 2001; Screaton et al. 2004; Ritorto et al. in pre ss), but contributions from the deep source have not yet been included in water mass-balance estima tes although the chemical composition of the Rive Rise water indicates th e deep source contribut es to its discharge. Estimating the volume of deep water sourcing the Ri ve Rise is difficult because of uncertainty in the chemical composition of the deep-water e nd member. While chemical compositions of end members represented by allogenic recharge and shallow sources can be measured directly at the River Sink and Well 4, respectively (e.g. Figure 3-4 ), the composition of the end member reflecting the deep-water source can not be directly sampled. Instead, water chemistry at Well 2 is a mixture of both deep and shallow water, and consequently mass-ba lance calculations can 66

PAGE 67

only approximate the relative fractions of wate r sourcing the River Rise While dissolution and precipitation reactions within the conduit may affect spring composition to some degree, I assume the mixing of the three representative end members largely accounts for most of the chemical variation at the River Rise ( Figures 3-4 and 3-5B ). I use concentrations of Mg 2+ and SO 4 2to estimate the relative fractions of the three sources of water discharging from th e River Rise. Concentrations of Mg 2+ and SO 4 2show strong linear correlations at the Rive r Sink, River Rise, and Well 2 ( Figure 3-8 ). The linear relationship suggests that concentrations are controlled by di lution, which is most likely to occur from mixing of allogenic recharge and the concentrated deep-water source as shown by the PCA ( Figure 3-5 ). In contrast to the deep-water sour ce at Well 2, concentrations of Mg 2+ and SO 4 2of diffuse recharge at Well 4 are low, rema in relatively constant, and have nearly the same ratio through time (see Figures 3-6A and 3-8 ) suggesting this water is not a ffected by inputs of allogenic or deep water. Although Well 4 does exhibit a linear trend on PC 2 scores ( Figure 3-5B ), no systematic cause for the variation exists. In order to observe how temporal variations in the magnitudes of sources affect spring discharge, water fractions were calculated using Mg 2+ and SO 4 2concentrations from each sample trip. Assuming contributi ons only from the three identifi ed end members, water at the River Rise consists of volumetric fractions of each end member, X, RSW2W4X= X+ X+ X (3-6) where the subscripts represent all ogenic recharge at the River Sink (S), the deep source at Well 2 (W2), diffuse recharge at Well 4 (W4), and disc harge at the River Rise (R), which equals 1. Individual equations were written for Mg 2+ and SO 4 2concentrations where: RRSSW2W2W4W4XMg= XMg+ XMg+ XMg (3-7) 67

PAGE 68

R4RS4SW24W2W44W4XSO= XSO+ XSO+ XSO (3-8) Rearranging Equation 3-6 for X S and substituting into Equation 3-7 and solving for X W4 gives RRSW2W2S W4 W4SX(Mg-Mg)-X(Mg-Mg) X= Mg-Mg (3-9) and rearranging Equation 3-6 for X W4 and substituting into Equation 3-8 and solving for X S gives R4R4W4W24W24W4 S 4S4W4X(SO-SO)-X(SO-SO) X= SO-SO (3-10) Substituting Equations 3-9 and 3-10 into Equation 3-6 and solving for X W2 yields 4R4W4 RS 4S4W4 W4S W2 4W24W4 W2S 4S4W4 W4SSO-SOMg-Mg 1SO-SOMg-Mg X SO-SOMg-Mg 1SO-SOMg-Mg (3-11) Variables X W4 and X S are found using back-substitution of solutions to Equation 3-11 into Equations 3-9 and 3-10 respectively. Equations 3-9 3-10 and 3-11 provide the mixing fractions of source water contributing to discharge at the River Rise fo r all the sampling times except January 2003 (S-1, Figure 3-3 ) prior to the installation of Well 4 ( Table 3-3 ). Results of the mixing calculations show that flow through the sink-rise system is quite complex. Nonetheless, discharge at the Rive Ri se correlates positively with allogenic recharge (River Sink), inversely with the deep-water source (Well 2), but lacks a correlation with diffuse recharge (Well 4) ( Figure 3-9 ). These results agree with th e PCA, which suggests that as allogenic recharge increases with stage the magnitude of the deep-water source decreases. This decrease in deep water may reflect elevated he ad in the conduit limiting upward flow. The lack of correlation between di scharge and diffuse recharge as represented by the Well 4 fraction ( Figure 3-9C ) suggests that hydraulic head between th e conduit and surroundi ng aquifer, and the related exchange of water betw een the conduit and matrix, do not change systematically with 68

PAGE 69

river stage. During times when diffuse recharge exceeds allogeni c recharge, hydraulic head in the surrounding aquifer could exceed conduit head as diffuse recharge elevates the water table and causes flow from the matrix to the conduit a nd ultimately to discharge from the River Rise (Martin and Dean 2001; Screaton et al. 2004; Martin et al. 2006; Ritorto et al. in press). Certain sampling times provide information on how differences in hydraulic head between the conduit and surrounding aquife r may affect the chemical co mposition of water discharging from the River Rise. Prior to sampling on Ap ril 30, 2003 and January 17, 2006, river stage dropped rapidly, which would re sult in rapidly decreasing head in the conduit (indicated as in Figure 3-9 ). If head in the conduit dropped more quick ly than head in th e surrounding aquifer, pressure gradients would drive flow toward the conduit (Screaton et al. 2004; Martin et al. 2006), decreasing the fraction of alloge nic water to the River Rise a nd simultaneously increasing the fractions of matrix water. Consequently, these two sample times show the elevated fraction of water from Well 4 (diffuse recharge) relative to the River Sink fracti ons (allogenic recharge) ( Figure 3-9 ). During times of low flow, the conduit acts as a low-resistance drain th at allows water to converge on it (e.g., Freeze and Cherry 1979; Ford and Williams 2007). This process is observed during a drought from July 2006 to April 2007 (S-13 S-16, Figure 3-3 ), when river stage constantly fell from about 10 to 9.7 masl, far be low the average stage of 10.2 masl. During this time, discharge from the River Rise was close to an even mixture of allogenic water (River Sink) and groundwater (Wells 2 and 4) (see in Figure 3-9 ). The fraction of deep water (Well 2) was at a maximum, averaging around 20% of the tota l discharge, suggesting that first-magnitude springs draining the UFA may recei ve significant contributions of flow from upward movement from deep flow paths (Katz, 2004). The fraction of diffuse water (Well 4) is more variable than 69

PAGE 70

the fraction of deep water, ranging from about 20 to 40%. This variability likely reflects changes in head gradient between the conduit and surrounding aquifer due to differences in antecedent conditions such as prior precip itation and ET. Variations in these factors w ould alter the elevation of the water table so that different amounts of matrix water woul d flow to the conduit for similar river stages ( Figure 3-3 ). Although matrix flow in unconf ined eogenetic aquifers can provide significant amounts of sp ring discharge, its contributi on through time at any one spring must be sensitive to processes affecting hydr aulic head gradients between conduits and surrounding aquifer. Conclusions Spatial and temporal monitoring of surf aceand groundwater chemistry along with observations of physical parameters including river stage, precipit ation, and ET in the Santa Fe River Sink-Rise system of the eogenetic UFA provide insight on how multiple sources of water and several different flow paths may affect sp ring discharge in karst aquifers. Chemical monitoring and PCA suggest that mixing of two shallow sources (d iffuse and allogenic recharge) and one upwelling deep-water source explains 91 % of the chemical variation in the sink-rise system. The previously unrecognized deep-wat er source is the primary influence on majorelement chemistry by providing most of Na + Mg 2+ K + Cl and SO 4 2to the system. Estimates of vertical flow, based on maxi mum observed temperatures, are on the order of 1 m/yr, and this flow appears to contribute up to nearly 20% of the di scharge at the River Rise. The contribution from the deep source depends inversely on flow conditions. The presence of a deep source suggests that care must be taken in the evalua tion of karst aquifers based on the chemical composition of spring water, which may not be re stricted to shallow portions of the aquifer. Water flowing through karst aquifers from all ogenic inputs to springs should reflect an evolution of the recharged wate r by water-rock reactions along conduit flow paths. Comparison 70

PAGE 71

71 of relative fractions of source water, however, suggest the deep-water source and local diffuse recharge cause significant changes in the chemi cal composition of discharge even in a system dominated by allogenic recharge and conduit flow. While variati ons in spring chemistry likely reflect water-rock reac tions along conduit flow paths betw een sinks and springs, mixing of different sources may play a more dominate role in the temporal variability of spring chemistry. Consequently, any characterization of karst a quifers using spring-wate r chemistry requires understanding the variety of sources of waters and their chemical compositions.

PAGE 72

Table 3-1. Summary of major ions, alkalinity, SpC, pH, and T of representative water samples. Location Cl SO4 Ca Na range x CV range x CV range x CV range x CV River Sink 0.23 0.64 0.38 27 0.02 0.48 0.24 69 0.19 1.39 0.78 58 0.19 0.46 0.30 19 River Rise 0.32 0.64 0.47 17 0.02 1.15 0.63 61 0.20 2.02 1.24 55 0.24 0.52 0.39 23 Well 2 0.42 1.66 1.24 28 1.19 4.47 3.43 28 1.97 4.77 3.92 18 0.61 1.61 1.26 23 Well 4 0.22 0.28 0.25 6 0.02 0.05 0.04 18 2.07 2.36 2.21 3 0.18 0.22 0.20 6 Well 7 0.28 0.54 0.42 18 0.02 0.29 0.16 38 1.54 2.77 2.19 19 0.19 0.34 0.27 17 Table 3-1. Continued 72 Location Mg K alkalinity pH range x CV range x CV range x CV range x CV River Sink 0.09 0.64 0.35 58 0.020 0.047 0.028 27 0.16 3.04 1.56 71 5.40 7.79 6.94 9 River Rise 0.09 0.72 0.44 54 0.024 0.041 0.027 15 0.16 3.16 1.90 58 4.70 7.37 6.90 9 Well 2 0.57 2.04 1.48 32 0.035 0.082 0.06 28 2.04 4.28 3.89 13 6.48 7.10 6.84 3 Well 4 0.05 0.09 0.06 16 0.005 0.010 0.009 14 2.80 4.30 4.03 9 6.48 7.19 6.87 3 Well 7 0.15 0.24 0.18 13 0.015 0.023 0.019 13 2.16 5.12 4.12 19 6.50 7.40 6.95 4

PAGE 73

73 Table 3-1. Continued Location SpC T range x CV range x CV River Sink 73 412 256 48 10.0 27.7 19 27 River Rise 73 560 371 46 11.0 26.4 20 20 Well 2 488 1315 1058 20 22.0 26.3 25 4 Well 4 390 449 428 3 20.9 21.7 21 1 Well 7 306 -550 434 20 20.3 20.9 21 1 Range and mean (x) of concentrations in mmol/kg H 2 O, coefficient of variation (CV) in percent, pH is unitless, SpC in S/cm and T in C.

PAGE 74

Table 3-2. Variable loadings of PCA. Loadings Variables PC 1 PC 2 PC 3 pH 0.08 a -0.42 0.53 Cl 0.42 0.15 0.00 SO 4 0.44 0.00 -0.07 Ca 0.27 -0.46 -0.39 Na 0.44 0.07 -0.02 Mg 0.44 0.02 0.07 K 0.39 0.31 0.09 Alkalinity 0.05 -0.62 -0.36 Stage -0.10 0.32 -0.65 eigenvalues 4.87 2.13 1.16 % Variance 54 24 13 % Cumulative 54 78 91 a Loadings greater than |0.3| are in italics. 74

PAGE 75

Table 3-3. Fraction of water discharging from th e River Rise originating from the River Sink and two groundwater end members. Sample Date Sample Period Rise Discharge (m 3 /s) River Sink (%) Well 2 (%) Well 4 (%) 3/2/03 3/5/03 3/19/03 S-2 57.9 81 a 0 19 4/30/03 S-3 12.0 40 24 36 1/23/04 S-4 5.2 83 18 -1 3/8/04 S-5 9.6 74 5 21 5/5/04 S-6 6.1 57 20 23 1/19/05 S-7 18.0 87 19 -6 3/18/05 S-8 20.2 76 13 11 7/18/05 S-9 49.5 93 3% 3 10/27/05 S-10 15.7 76 11 13 1/17/06 S-11 30.4 65 4 31 4/12/06 S-12 10.3 74 20 6 7/13/06 S-13 7.5 55 16 29 10/10/06 S-14 5.2 46 20 34 01/17/07 S-15 3.9 67 17 17 04/10/07 S-16 3.6 42 21 37 a Percentages calculated based on solutions to Equations 3-9, 3-10, and 3-11. 75

PAGE 76

Figure 3-1. Lithostratigraphic a nd hydrostratigraphic units of the Santa Fe River Basin. Thickness of units not implied in the diag ram. Modified from Miller (1986), Scott (1988), and Martin and Dean (2001). 76

PAGE 77

Figure 3-2. Site location of the Santa Fe SinkRise system showing locations of surface water and ground water sampling sites. Insert ma p shows location of Santa Fe Sink-Rise system in relation to north-central Florida. Dotted line represents erosional edge of the Hawthorn Group to the northeast (gray ar ea) marking the conf ined portion of the Upper Floridan Aquifer, with the white area representing the unconfined portion of the UFA where the Hawthorn is absent. 77

PAGE 78

Figure 3-3. Stage and discharge of the Santa Fe River at the Ri ver Rise, and precipitation and potential recharge (precipitation minus ev apotranspiration; P-ET) amounts estimated within OLeno State Park. Gray dots repres ent times of sample collection. Dashed line represent average stage ( 10.2 masl) during study period. 78

PAGE 79

Figure 3-4. Piper diagram showi ng the hydrochemical facies of surface and groundwater in Santa Fe Sink-Rise system. Representative end members are 1) Ca-HCO 3 type water, 2) CaMg-SO 4 type water, and 3) Na-Cl type water. Inset in diamond shows mixing trends between the three end members. 79

PAGE 80

Figure 3-5. Principal component lo ading and scores for Santa Fe Si nk-Rise water. A) Plot of PC loadings for major ions, alkalinity, pH, and ri ver stage. B) Plot of PC scores for River Sink, River Rise, and Wells 2, 4, and 7. 80

PAGE 81

Figure 3-6. Plots of ion concentrations versus Cl A) Water from Well 4. B) Water from Well 2. 81

PAGE 82

Figure 3-7. Diagrammatic sketch of boundary cond itions for vertical stea dy-state flow and heat transfer at Well 2. 82

PAGE 83

Figure 3-8. Plot of Mg 2+ versus SO 4 2concentrations showing the te mporal, linear variation at the River Sink, River Rise, and We ll 2. Well 4 shows little change near the origin of the graph. 83

PAGE 84

Figure 3-9. Plots of source contri butions versus discharge at the River Rise. A) Percentage of River Sink source. B) Percentage of Well 2 source. C) Percentage of Well 4 source. 84

PAGE 85

CHAPTER 4 CONDUIT ENLARGMENT IN AN EOGENETIC KARST AQUIFER Introduction Most models of karst aquife rs are commonly based on studies of dense, recrystallized limestone (i.e., telogenetic karst, Vacher and Mylroie 2002). Within these aquifers, conduits are embedded in a network of joints, fractures, and fissures in a groundmass of otherwise low matrix porosity and permeability. Conduits develop along th ese initially narrow flow paths as a function of flow-path geometry, water chemistry, and fl ow rates (e.g., Ford and Ewers 1978; Palmer 1991). As conduits enlarge, they capture more of the flow, th ereby enhancing dissolution and further enlarging the conduits until only a few routes carry the majority of flow (e.g., White 1988; Ford and Williams 2007). Low matrix permeability of telogenetic aquifers keep the flow focused within the developing conduit and cause enlargement through dissolution along the conduit wall. Flow and any dissolved CaCO 3 from the regional gr oundwater is considered negligible (e.g., Halihan and Wicks 1998). The concepts of conduit flow in telogenetic ka rst aquifers allow disso lution rates to be estimated based on the magnitude of disequilib rium between the conduit water and carbonate minerals. Early calculations used kinetics of cal cite dissolution to deri ve expressions for the retreat of conduit walls (Dreybrodt 1990; Palmer 1991). These kinetic expressions have been used to model the early development of conduits (e.g., Groves and Howard 1994; Kaufmann and Braun 1999; Gabrovek and Dr eybrodt 2001; Romanov et al. 2003a), morphologies of cave patterns (e.g., Palmer 1991; Howard and Groves 1995; Palmer 2001), leakage rates beneath dam sites (Romanov et al. 2003b), and to estimate the role of conduit growth on landscape evolution in a karst basin (Groves and Me iman 2005). Recent studies have suggested that dissolution rates may be expressed as a function of flow velo city through the conduit, whereby dissolved CaCO 3 85

PAGE 86

at a spring reflects the amount of dissolution along the conduit fl ow path (Grasso and Jeannin 2002; Grasso et al. 2003). This understanding of how conduits develop appears to work for telogenetic limestone, especially where alloge nic recharge to the conduit occurs through sinkholes and swallets (e .g., Palmer 2001), but may not fit fo r eogenetic karst aquifers, which have orders of magnitude greater matrix permeab ility than their telogene tic counterparts (Vacher and Mylroie 2002; Budd and Vacher 2004). The limitations of telogenetic models of conduit development for explaining eogenetic karst conduits stems from high matr ix permeability, which allows recharge to and discharge from aquifer storage. This exchange of water vari es depending on hydraulic head between the conduit and surrounding aquifer, affecting both regional groundwater and spring chem istry (e.g., Katz et al. 1998; Crandall et al. 1999; Mart in and Screaton 2001; Moore et al. submitted). For example, water that drains from the surr ounding aquifer can cons titute a substantial component of water flowing in the conduits that would be absent in telogenetic aquifers (e.g., Martin and Dean 2001; Florea and Vacher 2006). Conversel y, water lost from the condu it to surrounding aquifer mostly occurs during high flow when water undersaturated with respect to calci te may drive dissolution within the aquifer matrix (Katz et al. 1998; Crandall et al. 1999; Screaton et al. 2004; Ritorto et al. in press). This interaction between the conduit and surrounding aquifer should affect how conduits enlarge in eogenetic karst, whereby dissolution occurs at the conduit wall as well as within the aquifer matrix. These observations su ggest that new concepts may be necessary to describe conduit developm ent in eogenetic karst. The central question I address in this paper is how high matrix porosity and permeability of eogenetic karst aquifers affect the magnitude and distribution of conduit development and dissolution within the surrounding matrix porosity. I use gro undwater chemistry, geochemical 86

PAGE 87

reactions, and the physical and ch emical variations of a first magnitude spring draining a portion of the Upper Floridan aquife r (UFA) to estimate magnitude s and locations of dissolution occurring within a 6-km long conduit networ k. Although recharge to the conduit occurs primarily by allogenic runoff, diffuse recharge through the rock matrix and deep-water upwelling also source the conduit (Ritorto et al. in press, Moore et al. submitted). I use compositions of water from each source and the mass transfer of calcite and CO 2 to estimate the extent of dissolution within c onduit. These results suggest that concepts of conduit enlargement used in telogenetic karst may fail to properly explain conduit development in eo genetic karst. Although this study focuses on the eogenetic karst of northcentral Florida, the results discussed here are relevant to other karst aquifers that re tain high matrix porosity and permeability. Background Hydrologic and Geologic Conditions The Santa Fe River is a tributary of the Su wannee River, with a wa tershed covering about 3600 km 2 in north-central Florida (Hunn and Slack 1983). In the watershed, the UFA is confined by the Hawthorn Group to the northeast, and is unconfined in the southwest due to erosion ( Figure 3-2 ). The erosional edge of the Hawthorn Group, referred to as the Cody Scarp, forms the boundary between the confined and unconfined UFA (Puri and Vernon 1964). The Santa Fe River flows westward from Lake Santa Fe fo r about 40 km until it reaches the Cody Scarp, where it sinks into a 36-m deep sinkhole at the River Sink in OLeno State Park ( Figure 3-2 ). The river flows underground through a network of conduits until it reemerges about 6 km from the River Sink as a first magnitude spring, called the River Rise, marking the headwaters of the lower Santa Fe River (Martin and Dean 2001). Th e conduits rise to th e surface intermittently between the Sink and Rise at several karst windows ( Figure 3-2 ). 87

PAGE 88

At the Santa Fe Sink-Rise system, the UFA is about 430 m thick, unconfined at the surface, and is covered by a thin veneer (about 4 m, depending on land-surface elevation) of unconsolidated sands and sediments (Miller 1986). No middle confining unit exists in this area and thus the UFA extends to th e lower confining unit of the Cedar Key Formation (Miller 1986) ( Figure 3-1 ). Potable water extracted from the aquife r is estimated to come from the upper 100 m of the Ocala Limestone, and the water becomes increasingly mineralized with depth into the aquifer (Hunn and Slack 1983; Miller 1986). In this area, the water table averages about 4 meters below ground surface (mbgs) (Ritorto et al. in pr ess). Porosity and matrix permeability of the Ocala Limestone average about 30% and 10 -13 m 2 respectively (Budd and Vacher 2004; Florea 2006). Exploration of the conduits upstream of th e River Rise has resulted in over 15 km of surveyed passage, and represents one of the longest conduit systems in Florida (Poucher 2007). Average dimensions of the cross section of the conduits range from 18 to 24 m wide and 12 to 18 m high and they occur at an average depth of about 30 mbgs (Poucher 2007). Conduits have not been mapped to connect the River Sink to the River Rise, but high flow rates detected by natural and artificial tracers show the two locations mu st be linked by conduits (Hisert 1994; Martin and Dean 1999; Moore and Martin 2005). Previous work has shown that water discha rging from the River Rise varies between sources from the River Sink and from groundwater, defined here as water stored in the aquifer surrounding the conduits (e.g., Mar tin and Screaton 2001; Screaton et al. 2004; Martin et al. 2006). During high flow in the River Sink, discharg e at the River Rise is mostly derived from water entering the conduit system at the River Si nk. As river stage and in put into the River Sink decrease, increasingly la rger volumes of groundwater drain from the matrix into the conduit system to discharge at the River Rise (Martin and Screaton 2001). 88

PAGE 89

River Conditions and Potential Recharge River Rise stage data were recorded a bout 200 m downstream of the spring with an automatic pressure transducer with an accuracy of .03 m (Moore et al. submitted). The relationship between stage and discharge at th e River Rise was calculated based on the rating curve developed by Screaton et al. (2004), usin g data collected by the Suwannee River Water Management District (SRWMD). Potential r echarge was estimated as precipitation minus evapotranspiration (P-ET) during the study period by Ritorto et al (in press). Briefly, daily values of P-ET were calculated using daily precipitation data collected in OLeno State Park with an automated rain gauge maintained by SRWMD ( http://www.srwmd.state.fl.us/index.asp?NID =99 ), and the Penman-Monteith model for estimating daily ET, which estimates water loss to the atmosphere from a vegetative surface (Dingman 2002). Over the study period from January 2003 to Ap ril 2007, the amount of recharge, estimated as P-ET, and river stage and disc harge varied greatly, but not alwa ys in concert. Average annual P-ET was about 400 mm (Ritorto et al in press), which resulted in an average river stage and discharge of 10.2 masl and 16 m 3 /s, respectively ( Figure 3-3 ). The maximum annual P-ET of about 990 mm occurred in 2004 when Hurricane Frances delivered a total of 400 mm, similar to the average annual P-ET, at the OLeno State Park rain gauge over a 6-da y period in September. This rainfall corresponds to the maximum stage and discharge of 14.1 masl and 200 m 3 /s respectively, which occurred on September 10, 2004 ( Figure 3-3 ). From April 2006 to April 2007 the area received a total of 83 mm of P-ET or less than a fifth of the average annual amount. This low rainfall resulted in the lowest stage and discharge during the study of 9.6 masl and 2.2 m 3 /s, respectively. Sampling trips S-2, S-9, and S-11 occurred during high flow events when the river was above average stage. All other samples were collected during average or low flow. 89

PAGE 90

Water Chemistry Sixteen sampling trips were conducted from January 2003 to April 2007 to collect water from eight groundwater monitori ng wells, one sinking stream (Riv er Sink), one first magnitude spring (River Rise), and four intermediate karst windows ( Figure 3-2 ) (Moore et al. submitted). Of the 14 sites sampled, two surface water sites (River Sink and River Rise) and three groundwater sites (Wells 2, 4, and 7) show the gr eatest variation in wate r chemistry (Moore et al. submitted). All of the other site s that were sampled during th e study (data not reported here) have chemical compositions similar to one of these fi ve sites, and thus these five sites are used to represent the continuum of wate r chemistry across the region ( Figure 3-4 ). A statistical summary of the major chemistry is shown in Table 4-1 The variation in water chemistry reflects th ree representative end-member water types identified by Moore et al. (submitted). These thre e end members include a Na-Cl-type water, CaHCO 3 -type water, and Ca-Mg-SO 4 -type water that result from thr ee distinct sources and develop two mixing trends ( Figure 3-4 ). One trend extends from the Ca-HCO 3 -type water at Well 4 ( in Figure 3-4 ) to the Ca-Mg-SO 4 -type water at Well 2 ( in Figure 3-4 ). The composition of Well 4 is similar to shallow groundwater of the UFA and reflects rain water equilibrating with the Ocala Limestone (Moore et al. submitted). Water found at Well 2 is the most mineralized in the region and likely reflects water upwelling from deep within the aquifer, where elevated temperatures reflect the geothe rmal gradient and elevated SO 4 2and Mg 2+ concentrations may reflect gypsum dissolution, calcite precipitation, and dedolomitization (Moore et al. submitted). Well 7 falls along the mixing trend between Wells 4 and 2, suggesting it may be influenced by both sources of water ( Figure 3-4 ). The third end member is characterize d by elevated concentrations of Na + and Cl and occurs at the River Sink at high flow ( in Figure 3-4 ). This end member develops a second 90

PAGE 91

mixing trend that is confined to water collected from surface-water sites and reflects evolved rain water flowing overland with litt le contact with carbonate rock s (Moore et al. submitted). When the river is below average stage (e.g., S3, S-4, S-6, S-10, and S-12 through S-16: Figure 3-3 ), water from the River Sink and River Rise fall along the mixing line be tween Wells 4 and 2, reflecting little influence from the Na-Cl-type end member. Chemical Analysis and Geochemical Modeling Methods The geochemical code, PHREEQC, Version 2.15.0, (Parkhurst and Appelo 1999), was used to determine the distribution of a queous species, partial pressure of CO 2 ion-activity products (Q), and mineral saturati on index (SI) for all water samples. The SI for a given mineral is found by Q SI = log K (4-1) where K is the equilibrium constant for a given mi neral. Samples within 0.1 SI are assumed to be in equilibrium with respect to calcite and dolo mite based on analytical errors in measurements of pH, alkalinity, and concentrations of Ca 2+ (e.g., Langmuir 1997). Thermodynamic data for calculations were from the phreeqc.dat database. Ionic strength of water samples was < 0.1 molal and thus activity coefficients of aqueous sp ecies were calculated using an extended version of the Debye-Hckle equation (T ruesdell and Jones 1974). Charge balance errors for most samples were less than 5% except for samples whose concentrations were near instrument detection limits. The inverse modeling code in PHREEQC was used to assess changes in major dissolved constituents between input at the River Sink a nd discharge at the River Rise as a means to estimate the extent of calcite dissolution along the conduit flow path. The mass-balance model 91

PAGE 92

attempts to match the water composition at the River Rise based on mixing proportions of water with compositions of water found at the River Sink and Wells 2 and 4, and the phase transfer of calcite and CO 2 gas. In modeling mass transfer, I assu me a quasi-steady state for each sample day and thus hydrodynamic dispersion and diffu sion are assumed to be minor processes compared to the chemical evolution along the conduit flow path for that day (Wigley et al. 1978; Gandolfi et al. 2001). In this study, mass-balance calculations determined in PHREEQC are constrained by the concentration of Ca 2+ Mg 2+ Na + Cl SO 4 2, and alkalinity in the initial and final waters. Concentrations of K + were not included in modeling because their low concentrations resulted in failed convergence of iterations in the mass-balance calculations. Prior to each simulation, each solution is charge balanc ed by adjusting the ion concentrations within predetermined uncertainty limits, which were set based on the charge imbalance for each water analysis. These limits allow the model to charge balance the initial and final solutions given the range of uncertainty within the analyzed water ch emistry. If a solution can not be adjusted to charge balance using the given uncertainty limits, no models will be found. Results The UFA is mostly comprised of calcite with lesser amounts of dolomite and gypsum ( Figure 3-1 ), and thus dissolution of the aquifer mine rals can be described largely by reactions given in Table 4-2 Deviations from equilibrium of th ese reactions can be evaluated through calculations of SI and an activ ity-phase diagram that represents the major aquifer-bearing minerals in the basin (cf. Helgeson 1969) ( Figure 4-1 ). In phase diagrams, boundaries between mineral phases are based on values of the equilib rium constants for the reactions; for example, the equilibrium between calcite and dolomite is derived from the expression for the equilibrium constant of Reaction 1 ( Table 4-2 ), which in logarithmic form is 92

PAGE 93

2+ 2+Ca Mglog = log K a a (4-2) where a represents ion activity and log K is 0. 38 (Plummer and Busenberg 1982). All other phase boundaries in Figure 4-1 were calculated using the logar ithmic form of the K expression for each reaction in Table 4-2 Figure 4-1 depicts the relative stabil ity of calcite, dolomite, and gypsum at 25 C and 1 atm as a function of the activities of Ca 2+ Mg 2+ SO 4 2, and CO 3 2of the water collected during the study period. Although activities of Ca 2+ Mg 2+ and SO 4 2are primarily controlled by mineral dissolution reactions, the activity of CO 3 2is a function of pH and magnitude of the CO 2 dissolved in the wate r in the form of +2 23 3HCO = 2H + CO -2 (4-3) which can be represented by a K expression in logarithmic form +2 23 23 3HCO HCO HC Olog K = 2 log + log log aaa (4-4) and has a log of -16.68 (Plummer and Busenberg 1982). Considering can be represented by (Drever 1997), 23HCOK 23HCOa 2COCOKP Equation 4-4 becomes +2 23 2 2 3HCO COCO HC Olog K = 2 log + log log Klog Paa (4-5) where log has a value of -1.47 at 25 C (P lummer and Busenberg 1982). Rearranging 2COK Equation 4-5 for log and substituting values for log and log gives 23COa 2COK 23HCOK 22 3CO COlog = log P+ 2 pH 18.15a (4-6) Equation 4-6 shows how relates to the pH and P 23COa CO2 for each sample. 93

PAGE 94

Samples from the five sites representing end-member water compositions fall into three groups on the phase diagram ( Figure 4-1 ). Each group has ratios that vary little compared to their ratios, which suggests most variation in water chemistry results from changes in SO 2+2+CaMg/ aa 2-2422SOCO/ aa 3 4 2concentration, P CO2 and pH. The Ca-HCO 3 -type water, as represented by samples from Well 4, plots in the calcite field with a constant log ratio of about 1.58 ( 2+2+CaMg/ aa Figure 4-1 ). This ratio suggests the wate r reacts primarily with calcite and has little contact with dolomite. In contrast, the Ca-Mg-SO 4 -type water from Well 2, as well as surface water from the River Sink and River Rise, plot on the phase boundary between calcite and dolomite, suggesting this water reacts with bot h minerals (Reaction 1 in Table 4-2 ). This water also plots closer to the phase boundaries of gypsum than water from the other wells, reflecting reactions with this mineral as well (e.g. Reactions 2 and 3, Table 4-2 ). Although some surface-water samples plot into the gypsum field, these samples were collect ed during high flow and thus are dilute with minor amounts of SO 4 2, low pH and high P CO2 ( Table 4-1 ). Water from Well 7 plots in the calcite field, intermediate between samples from Wells 4 and 2, but clos er to the phase boundary between calcite and dolomite than Well 4. Th ese relationships correspond to the mixing relationships shown in Figure 3-4 but reflect which minerals the water has reacted with. While the phase diagram shows which minerals may have reacted with the water, the SI indicates whether the water can theoretically dissolve or precipitate a give n mineral. Most of the water is undersaturated with respect to calcite, dolomite and gypsum ( Figure 4-2 ). Some groundwater samples at all the well s are saturated with respect to calcite and some samples from Well 2 are saturated with respect to dolomite ( Figures 4-2A and B ). The SI of calcite and dolomite in the groundwater samples show a linear variation through time, although not systematically with river stage. Although all gr oundwater remains undersatur ated with respect to 94

PAGE 95

gypsum, a strong trend exists among the wells, wi th Well 2 water plotting closest to saturation ( Figure 4-2C ). The SI of surface-water sites decreases with river stage indicating that allogenic runoff becomes more dilute as stage increases ( Figures 4-2D-F ). Most surface water is undersaturated with respect to calcite, dolomite and gypsum, although some samples are at saturation with calcite and dolomite at low flow. Discussion Conduits in telogenetic karst result from di ssolution along fractures, joints, and fissures depending on disequilibrium between the conduit wa ter and wall rock (e.g., Palmer 1991; Grasso and Jeannin 2002). In eogenetic karst aquifers, however, high matrix pe rmeability provides a significant component of flow (Budd and Vacher 2004; Florea and Vacher 2007), allowing multiple sources of water to converge on the co nduits (e.g., Moore et al. submitted). Although fractures and partings have a strong influence on many conduits in eogenetic karst (Florea et al. 2007), the interaction of water with different chemistry and high matrix permeability should influence how these conduits enlarge over time. In the following sections, I use mass-balance modeling of groundwater and surface -water chemistry to evaluate how matrix permeability may influence the magnitude and distribution of cond uit development in eogenetic karst aquifers. Geochemical Evolution of Groundwater Shallow groundwater in the sink-rise region, reflected in compositions of water from Well 4, is the product of CO 2 -rich diffuse recharge dissolvi ng Ocala Limestone to elevate concentrations of Ca 2+ and HCO 3 (e.g., Ritorto et al. in press; Moore et al. submitted). This CO 2 derives from root respiration and oxidation of or ganic matter in the soil and vadose zone (e.g., Atkinson 1977; Wood and Petraitis 1984; Ritorto et al. in press), which decreases pH and drives the water to undersaturation with respect to calcite and dolomite ( Figures 4-2A and B ). Although the Ocala Limestone has up to 2 mole % Mg and trace amounts of dolomite (Plummer 1977; 95

PAGE 96

Miller 1986), insufficient amounts of the Mg is dissolved to cause the shallow groundwater to reach saturation with respect to dolomite ( Figure 4-2B ). The limited amount of Mg 2+ in the shallow groundwater results in Well 4 water proj ecting well into the calcite field on the phase diagram ( Figure 4-1 ). Deep water in the region, as reflected in wa ter at Well 2, is more mineralized than the shallow water. Although this mineralization could result from leaching of ions during recharge through leaky portions of the Hawthorn Group that flow along deep flow paths due to regional head gradients (Lawrence and Upchurch 1982; Wicks and Herman 1994; Katz et al. 2004), recent work suggests elevated conc entrations reflect dissolution of calcium sulfate minerals and dedolomitization occurring deep within the aquifer (Moore et al. submitted). This observation is supported by the saturation indices with re spect to calcite, dolomite and gypsum ( Figures 4-2AC ) and the projection of Well 2 water on the pha se boundary between calcite and dolomite and close to the phase boundary of gypsum ( Figure 4-1 ), which suggests the deep water dissolved calcium sulfate minerals and may be at equilibrium with respect to both calcite and dolomite (cf. Plummer 1977; Hanshaw and Back 1979; Jones et al. 1993). Surface water at the River Sink and River Rise is a mixture of allogenic recharge and groundwater, with proportions th at vary with flow conditions (Moore et al. submitted) ( Figure 34 ). Water budget calculations indicate that while input at the River Sink accounts for the largest amount of recharge to the sink-ri se system, most of this wate r flows through the conduits and discharges at the River Rise w ithout any loss to the surrounding a quifer (Ritorto et al. in press). Consequently, water from the River Sink and River Rise have si milar chemistry and fall on the phase boundary between calcite and dolomite, but at various distances from the field for gypsum ( Figure 4-1 ). The projection of surface water on th e phase boundary between calcite and 96

PAGE 97

dolomite reflects reaction with these minerals, although there is little dolomite in the stratigraphic section at the depths of the conduits as reflected in water compositions at Well 4 ( Figure 4-1 ). Chemical compositions of River Sink and River Rise waters indicate some portion of their source was deeper within the stratigra phic section than where they were sampled, reflecting mixing with deep mineralized water as characterized by water compositions at Well 2 (cf. Jones et al. 1993). Contributions of deep water to the sink-rise system correlate inversel y with river stage and are greatest during droughts as surface water chem istry trends towards the Well 2 end member (Figures 3-4 and 4-1 ) (Moore et al. submitted). The projec tion of River Sink and River Rise water towards and into the gypsum field occurs only during high discharge conditions when runoff has dilute concentrations, low pH, and high P CO2 ( Table 4-1 and Figure 4-1 ). This water is greatly undersaturated with respect to all minerals ( Figures 4-2D-F ). Consequently, although this water projects into the gypsum field ( Figure 4-1 ), it most likely reflects decreasing from low pH and high P 23COa CO2 rather than an increase in the from reaction with calcium sulfate minerals. The only other location with a signal from the deep-water source is Well 7 (Figures 24SOa 3-4 and 4-1 ), but its location is about 1 km aw ay from the closest known conduit ( Figure 3-2 ). The presence of mineralized water supports earlie r observations that deep-water upwelling may be responsible for high SO 4 2concentrations in shallow portions of the UFA throughout the region, although delivery of this water may be restricted to vertic al flow paths such as fractures and faults (e.g., Hunn and Slack 1983; Moore et al. submitted). Estimates of Conduit Dissolution in the Sink-Rise System I estimate dissolution along the conduit flow path between the River Sink and River Rise using the inverse modeling code in PHREEQC. Results of th e model are not unique, but do 97

PAGE 98

provide insight to possible water-rock reactions that account for spring chemistry at the River Rise. In these simulations, the groundwater flow path is th e conduit network and the final solution is the River Rise water, which is a ssumed to obtain its composition from mixing with the representative end member water type s, i.e., River Sink and Wells 2 and 4 ( Figure 3-4 ), and reactions with calcite and CO 2 gas. I allow the model to react only with stoichiometric calcite, since the conduits are confined to the Ocala Li mestone, which contain only trace amounts of magnesium-bearing minerals (Miller 1986). Given that water from the River Rise is either at equilibrium or undersaturated with respect to calcite ( Figure 4-2D ), I assume calcite only dissolves, whereas CO 2 is allowed to be taken up or released from solution during the simulations. Each simulation resulted in one inverse model that accounted for the chemical variation at the River Rise for all sampling times excep t January 2003 and March 2003 (S-1 and S-2, Figure 3-3 ). January 2003 was prior to the installation of Well 4 so data critical for the model were lacking for that sample time ( Table 4-3 ). River stage was elevat ed during the March 2003 sampling, and thus the River Rise and River Sink are dilute and have similar compositions, reflecting minimal gain of groundwater by the conduits ( Table 4-1 ). To estimate dissolution during this time, water at the River Ri se was modeled by reacting calcite and CO 2 with River Sink water (SI CAL = -4.2 and log P CO2 = -1.4) until ach ieving the SI of calcite and P CO2 determined in the River Rise water (SI CAL = -4.8 and log P CO2 = -0.69). Modeling results show that most of the time water chemistry at th e River Rise can be e xplained by mixing of representative end-member water types and mass transfer of calcite and CO 2 or in the case of March 2003, water from the River Sink alone reacting with calcite and CO 2 While dissolution along the conduit flow path only occurred five times, there was uptak e and release of CO 2 during 98

PAGE 99

each simulation suggesting that dissolution has a minor affect on the mass transfer of CO 2 ( Table 4-3 ). Uptake and release of CO 2 between the River Sink and River Rise may result from processes including mixing of different source wa ters, aerobic decay of organic carbon, and interaction with the atmosphere. Release of CO 2 could occur when the conduit water emerges to the surface at karst window s along the flow path ( Figure 3-2 ). Uptake of CO 2 along the flow path reflects CO 2 being generated within the conduit and/or additional CO 2 entering the conduit from the surrounding aquifer. One source of CO 2 may result from aerobic d ecay of organic carbon that is flushed into the conduit. Dissolved organic carbon has been shown to exceed 30 mg/l at the River Sink during high flow (Zimmerm an et al. 2006). In March 2003, CO 2 concentration is about 20 times more than occurred during the remaining 14 sample trips and the excess CO 2 may be derived from oxidation of organic carbon ( Table 4-3 ). During this sampling time, hydraulic head in conduit exceeded hydrau lic head in the surrounding aquife r (Martin et al. 2006) (S-3, Figure 3-3 ), restricting the amount of CO 2 that could enter the conduit from the matrix. The CO 2 charged water undersaturated with respect to cal cite was lost from the conduit to the surrounding aquifer, and would have dissolv ed some of the aquifer matrix during its residence time there (Martin and Dean 2001; Screaton et al. 2004; Ritorto et al. in press). This loss of CO 2 -rich water to the surrounding aquifer may also e xplain the strong linear variation in SI CAL and SI DOL at in the wells ( Figures 4-2A and B ), which would vary due to changes in pH and P CO2 of the groundwater ( Eq. 4-6 ); however, no systematic cause for th e variation exists (Moore et al. submitted). During average and low flow when hydr aulic head gradients reverse, increases in CO 2 along the conduit flow path could result from a combination of aerobic decay and input of CO 2 -rich water from the matrix ( Table 4-1 ). The relative concentrations of these two sources 99

PAGE 100

cannot be determined from these data, but could be assessed by measured changes in organic carbon concentrations and terminal electron acceptors (e.g., O 2 NO 3 and metal oxides) along the flow path. Conduit wall retreat can be estimated dur ing the five times dissolution was found ( Table 43 ) assuming the conduit is about 6000 m long with an average cross-sect ional area of about 386 m 2 (Screaton et al. 2004). These dimensions indicate a surface area of about 4 x 10 5 m 2 Using a measured value for discharge at the River Rise, a molar volume of calcite of 3.69 x 10 -5 m 3 and 30% porosity for the Ocala Limestone, an estimated dissolution rate ranges from about 2 x 10 -6 to 6 x 10 -6 m/day ( Table 4-4 ). Although the occasional sample trips make it difficult to estimate yearly averages, these daily values of dissolu tion agree well with estimates of maximum wall retreat in telogenetic conduits (3 x 10 -6 to 3 x 10 -7 m/day; Palmer 1991). Implications of Conduit Enlargemen t in Eogenetic Karst Aquifers The conduit network between the River Sink an d River Rise repres ents a dynamic flow path that transmits water through the subsurface and increases Ca 2+ concentrations at all sampling times ( Figure 4-3 ). This increase in Ca 2+ would be consistent with dissolution of the conduit wall rock in a telogenetic aquifer (Pal mer 1991; Howard and Groves 1995; Grasso and Jeannin 2002), but mo st increases in Ca 2+ concentrations at the River Rise can be explained by flow of allogenic and diffuse recharge to th e conduit, coupled with upwelling of mineralized deep water based on a two component mixing model using only Mg 2+ and SO 4 2(Moore et al. submitted). Inverse modeling used here relies on concentrations of Mg 2+ SO 4 2, Ca 2+ Na + Cl and alkalinity and thus may be more sensitive to estimates of mixing because of uncertainty limits set for each component ( Table 4-3 ). Consequently, the increase in Ca 2+ concentrations along the conduit flow path like ly reflects dissoluti on within the surrounding aquifer with 100

PAGE 101

subsequent flow to the conduit, rather than di ssolution at the conduit wall (Screaton et al. 2004; Ritorto et al. in press). Although water flowing from the River Sink to River Rise remains undersaturated with respect to calcite most of the time ( Figure 4-2D ), modeling results suggest dissolution is limited along the conduit flow path ( Table 4-4 ). Rates of dissolution are controlled by the combination of a thin boundary layer of static water at the contact between wall rock and water, and reaction processes including the hydration of CO 2(aq) (Kern 1960) and dissolution at the mineral surface by reaction with H + and H 2 CO 3 o (the PWP Equation; Plumme r et al. 1978). Thickness of the boundary layer depends on flow conditions through th e conduit, but are generally on the order of a fraction of a millimeter (Dreybrodt and Gabrov ek 2002). This layer acts as a weak barrier between the undersaturated water and conduit wa ll rock, whereby reactants and dissolution products move through the layer by molecular di ffusion (Liu and Dreybrodt 1997). However, as water from the matrix flows into the conduit, di ssolution may be limited as reactants such as H + H 2 CO 3 o and CO 2(aq) are restricted from contacting the conduit wall. These observations suggest that conduit enlargement in eogenetic karst may result from processes in addition to strict dissolution of the conduit wall. Conduits may increase in size in eogenetic karst when allogenic recharge flows to the surrounding aquifer causing dissolution within the matrix porosity (S creaton et al. 2004; Ritorto et al. in press). Although the boundary layer also controls diss olution rates within these pore spaces, slower flow and longer residence time allo ws sufficient time for matrix water to reach saturation (e.g., Figure 4-2A ). Water losses from the conduit typically occur during high flow when water has a SI CAL of less than -3 SI (S-2, S-9, and S-11: Figure 3-3 ). Recent estimates of dissolution from allogenic water equilibrating with the matrix rock in the sink-rise system 101

PAGE 102

suggested up to 500 m 3 of limestone could have dissolved ov er a 63 month period (Ritorto et al. in press). Averaging this amount over th e surface area of the conduit wall (4 x 10 5 m 2 ), suggests a dissolution rate of about 2 x 10 -4 m/yr or about 7 x 10 -7 m/day. This value is about an order of magnitude lower than daily rates es timated along the conduit flow path ( Table 4-4 ), but still within range of maximum wall retreat values determined for telogenetic karst (3 x 10 -6 to 3 x 10 -7 m/day; Palmer 1991). Although the loss of conduit wa ter to the surr ounding aquifer is episodic, this process suggests significant amounts of di ssolution may occur adjacent to the conduit or high permeability zones where allogenic water can fl ow into the matrix (Ritorto et al. in press). Because of the high matrix porosity in eogenetic karst, dissolution by this process would result in a friable halo surrounding the conduit ( Figure 4-4 ). Similar results were found from core samples collected from the walls of air-f illed caves in west-central Flor ida, which show an order of magnitude increase in matrix permeability near the wall surface compared to matrix permeability within the wall inte rior (Florea 2006). This increase in matrix permeability would not be expected in telogenetic karst aquifers; instead, water lost to the surr ounding aquifer in these settings will enlarge fracture flow paths within an otherwise impenetr able matrix (Ford and Williams 2007). The development of a friable halo will larg ely depend on the movement of water between the conduit and surrounding aquifer, which varies due to differenc es in hydraulic conductivity (Martin et al. 2006) ( Figure 4-4 ). Particle tracking simulations during the March 2003 flood (S-2, Figure 3-3 ) suggested the movement of water into the matrix ranges from a bout 0.4 to 8 m based on differences in hydraulic head between river stag e at the River Sink and water levels in Wells 1 and 4 (Martin 2003) ( Figure 3-2 ). Although these simulations a ssumed flow paths within a homogeneous and isotropic aquifer, preferential flow paths will a llow water to move farther into 102

PAGE 103

the matrix (Martin and Dean 2001) (K 2 in Figure 4-4 ). Following reversal of the hydraulic gradient between the wells and conduit, the si mulated water was estimated to return to the conduit in about two weeks, indi cating sufficient time for calcite dissolution in the matrix to drive the water to about 80 to 90% calcite sa turation (Plummer et al. 1978; White 1988). This dissolution should also increase matrix permeability over time a nd allow more undersaturated water to flow into to the surrounding aquifer in a feed-back loop that would preferentially enlarge conduits along the flow path that already started. In addition, as matrix rock becomes less consolidated within the friable halo, phys ical erosion of the conduit wall rock should increase as suspended sediments and particles are transported through th e conduit (e.g., Dogwiler and Wicks 2004). This development of a friable ha lo and removal of the wall rock by dissolution and/or sediment transport may explain the irre gular solution pockets and spongework textures common to many Florida conduits th at receive allogenic recharge (Kincaid 1999; am Ende 2000; Palmer 2007). Conclusions In the Santa Fe Sink-Rise area, water origin ates from three sources including allogenic water at the River Sink, diffuse recharge through the vadose zone that equilibrates with Ocala Limestone, and upwelling of water from deep in th e aquifer, which is more mineralized than the shallow water following reactions with calcite, dolomite, and calcium sulfate minerals. Although allogenic recharge provides signifi cant amounts of undersaturated wa ter to the UFA, most of the recharge flows through the conduits discharging at the River Rise w ith little intera ction with the surrounding aquifer. This limited in teraction results in model estim ates of dissolution occurring only during 5 of the 14 sampling times. When hydraulic head within the conduit exceeds hydraulic head in the surrounding aqui fer, allogenic water undersaturated with respect to calcite is lost to the surrounding aquifer and dissolves matrix rock in th e vicinity of the conduit. This 103

PAGE 104

104 process generates a friable halo, suggesti ng conduit enlargement may result from both dissolution and physical erosion of less consolidated matrix rock. Extrapolating these results to eogenetic karst in general suggest that conduits develop in eogenetic kars t aquifers by mixing of multiple water sources and flow into and out of the high matrix porosity. Models describing conduit enla rgement in telogenetic karst assume an impenetrable boundary at the conduit wall, whereby increases in Ca 2+ concentration along th e flow path reflect dissolution of the wall. In eogenetic karst, Ca 2+ concentrations could increase in the conduit water from inputs of Ca-rich water from the ma trix when hydraulic head s within the surrounding aquifer exceed hydraulic head in the conduit, a c ondition that is common in eogenetic karst. This flow of water from the matrix to the conduit would also limit dissolution of the conduit water by limiting contact of undersaturated water with th e conduit wall. These processes may explain why dissolution within th e conduit occurred only during 30% of the sampling times. Although conduit wall retreat in the sink-rise system has daily rates comparable to those from telogenetic settings, this dissolution is no t continuous. The long-term average dissolution of the conduit walls results from a combination of di ssolution at the conduit wall, as well as within the matrix surrounding the conduit, which is a proce ss not expected to occu r in telogenetic karst aquifers with low matrix permeability. These results provide insight on the difficulty in applying concepts of conduit enlargement de veloped for telogenetic karst in eogenetic karst aquifers, and illustrate the influence of matrix permeability on groundwater flow and the conduit enlargement in eogenetic settings.

PAGE 105

Table 4-1. Summary of major ions, alkalinity, SpC, pH, T, SI, and P CO2 of representative water samples. Location Cl SO4 Ca Na range x CV range x CV range x CV range x CV River Sink 0.23 0.64 0.38 27 0.02 a 0.48 0.24 69 0.19 1.39 0.78 58 0.19 0.46 0.30 19 River Rise 0.32 0.64 0.47 17 0.02 1.15 0.63 61 0.20 2.02 1.24 55 0.24 0.52 0.39 23 Well 2 0.42 1.66 1.24 28 1.19 4.47 3.43 28 1.97 4.77 3.92 18 0.61 1.61 1.26 23 Well 4 0.22 0.28 0.25 6 0.02 0.05 0.04 18 2.07 2.36 2.21 3 0.18 0.22 0.20 6 Well 7 0.28 0.54 0.42 18 0.02 0.29 0.16 38 1.54 2.77 2.19 19 0.19 0.34 0.27 17 Table 4-1. Continued 105 Location Mg K alkalinity pH range x CV range x CV range x CV range x CV River Sink 0.09 0.64 0.35 58 0.020 0.047 0.028 27 0.16 3.04 1.56 71 5.40 7.79 6.94 9 River Rise 0.09 0.72 0.44 54 0.024 0.041 0.027 15 0.16 3.16 1.90 58 4.70 7.37 6.90 9 Well 2 0.57 2.04 1.48 32 0.035 0.082 0.06 28 2.04 4.28 3.89 13 6.48 7.10 6.84 3 Well 4 0.05 0.09 0.06 16 0.005 0.010 0.009 14 2.80 4.30 4.03 9 6.48 7.19 6.87 3 Well 7 0.15 0.24 0.18 13 0.015 0.023 0.019 13 2.16 5.12 4.12 19 6.50 7.40 6.95 4

PAGE 106

106 Table 4-1. Continued Location SpC T SI CAL SI DOL range x CV range x CV range x CV range x CV River Sink 73 412 256 48 10.0 27.7 19 27 -4.5 0.32 -1.40 n/a b -9.0 0.4 -3.1 n/a River Rise 73 560 371 46 11.0 26.4 20 20 -4.8 0.02 -1.10 n/a -9.5 -0.4 -2.5 -99 Well 2 488 1315 1058 20 22.0 26.3 25 4 -0.7 0.06 -0.19 n/a -1.4 -0.1 -0.7 63 Well 4 390 449 428 3 20.9 21.7 21 1 -0.7 0.04 -0.27 n/a -1.5 -1.4 -2 -22 Well 7 306 -550 434 20 20.3 20.9 21 1 -0.6 0.08 -0.23 n/a -1.3 -0.8 -1.5 -27 Table 4-1. Continued Location SI GYP log P CO2 range x CV range x CV River Sink -4 -2.0 -2.7 -24 -2.7 -1.4 -2.1 -14 River Rise -4 -1.6 -2.1 -32 -2.8 -0.7 -2 -22 Well 2 -1.5 -0.8 -0.9 -19 -2.0 -1.1 -1.5 -15 Well 4 -3.1 -2.8 -2.9 -3 -1.8 -1.1 -1.5 -12 Well 7 -3.2 -2.0 -2.3 -11 -2.1 -1.0 -1.6 -22 Range and mean (x) of concentrations in mmol/kg H 2 O, coefficient of variation (CV) defined as mean/standa rd deviation in percent, pH and SI values are unitless, SpC in S/cm, T in C, log P CO2 in atm, a minimum detectable value of SO 4 2, b n/a is not applicable due to both positive and negative numbers.

PAGE 107

107 Table 4-2. Chemical reacti ons involving UFA minerals. Chemical Reactions log K 1 2 CaCO 3 + Mg 2+ CaMg(CO 3 ) 2 + Ca 2+ 0.38 2 2 CaSO 4 2 H 2 O + 2 CO 3 22 CaCO 3 + 2 SO 4 2+ 2 H 2 O 7.99 3 2 CaSO 4 2 H 2 O + Mg 2+ + 2 CO 3 2CaMg(CO 3 ) 2 + Ca 2+ + 2 SO 4 22 H 2 O 8.37 Log K is the equilibrium constant for each reaction at 25 C. Gibbs energy and enthalpy of formation for calcite from Plummer and Buse nberg (1982), dolomite from Ball and Nordstrom (1991), and gypsum from Anderson and Crerar (1993).

PAGE 108

Table 4-3. River Rise composition estimated by mixing of River Sink, Well 2, and Well 4 end-member compositions and mass transfer of calcite and CO 2 Date River Rise pH Alkalinity Cl SO4 Ca Na Mg Calcite CO 2 (g) Uncertainty Limits (mmolal) (mmolal) (mmolal) (mmolal) (mmolal) (mmolal) (10 -5 moles/l) River Sink Well 2 Well 4 (%) (%) (%) 3/5/03 Modeled composition a 4.70 0.165 0.358 0.021 0.193 0.274 0.099 0.3 716 b n/a c n/a n/a (%) difference 0 3 12 0 -5 10 4 4/30/03 Modeled composition 6.80 2.240 0.539 0.622 1.397 0.422 0.404 -29 7.5 7.5 7.5 (%) difference 0 4 5 -5 -2 4 0 1/23/04 Modeled composition 7.37 2.959 0.522 0.929 1.788 0.466 0.649 10 15 2.5 2.5 2.5 (%) difference 0 5 3 -2 -5 5 -5 3/8/04 Modeled composition 6.87 0.846 0.511 0.342 0.603 0.335 0.250 --2 2.5 2.5 2.5 (%) difference 0 3 4 -4 0 4 3 5/5/04 Modeled composition 7.36 3.022 0.516 1.035 1.833 0.509 0.717 -6 2.5 2.5 2.5 (%) difference 0 5 0 0 -4 -2 5 1/19/05 Modeled composition 6.91 1.581 0.539 0.423 1.035 0.353 0.272 --27 5 2.5 15 (%) difference 0 4 13 -6 11 4 -13 3/18/05 Modeled composition 7.06 1.532 0.514 0.467 0.959 0.375 0.343 8.6 -27 6 2.5 2.5 (%) difference 0 -7 0 -6 6 6 -2 7/18/05 Modeled composition 6.59 0.533 0.374 0.145 0.357 0.219 0.133 --27 30 10 10 (%) difference 0 3 7 10 -10 -10 0 108

PAGE 109

109 Table 4-3. Continued Date River Rise pH Alkalinity Cl SO4 Ca Na Mg Calcite CO 2 (g) Uncertainty Limits (mmolal) (mmolal) (mmolal) (mmolal) (mmolal) (mmolal) (10 -5 moles/l) River Sink Well 2 Well 4 (%) (%) (%) 10/27/05 Modeled composition 7.19 2.096 0.433 0.586 1.233 0.367 0.434 --47 6 6 6 1/17/06 Modeled composition 7.27 0.560 0.424 0.203 0.398 0.283 0.155 --33 5 5 5 (%) difference 0 4 5 -5 -5 0 1 4/12/06 Modeled composition 6.96 2.560 0.432 0.685 1.483 0.356 0.519 3.2 -11 2.5 2.5 15 (%) difference 0 2 2 -4 -4 5 0 7/13/06 Modeled composition 6.98 3.148 0.422 0.997 1.868 0.444 0.691 -34 5 5 5 (%) difference 0 5 0 4 -5 -1 5 10/10/06 Modeled composition 7.15 3.259 0.502 1.024 1.925 0.483 0.739 --11 4.5 4.5 4.5 (%) difference 0 5 -1 -5 -4 0 5 1/17/07 Modeled composition 7.26 3.196 0.492 1.052 1.974 0.466 0.689 --37 4 4 6 (%) difference 0 2 3 0 -2 0 -4 4/10/07 Modeled composition 7.24 3.158 0.483 1.088 1.912 0.498 0.748 7.2 -8 7.5 7.5 7.5 (%) difference 0 0 -5 -5 1 5 5 a March 2003 composition was not determined by inverse modeling, but by reacting water from Ri ver Sink with calcite and CO 2 until achieving the SI of calcite and P CO2 determined in River Rise water. b negative sign denotes uptake and pos itive sign denotes release of CO 2 c n/a is not applicable for March 2003 due to different modeling approach.

PAGE 110

Table 4-4. Inverse modeling of source water mixing and mass transfer of calcite and CO 2 along a conduit flow path. Date Rise Discharge Source Contribution Calcite CO 2 (g) (m 3 /s) Well 2 Well 4 Sink (10 -6 m/day) (moles/s) 3/5/03 a 58.0 0.00 0.00 1.00 0.3 b 415 c 4/30/03 11.9 0.18 d 0.07 0.74 -e 3 1/23/04 5.1 0.18 0.08 0.75 6 0.8 3/8/04 9.8 0.03 0.02 0.96 --0.2 5/5/04 6.1 0.17 0.04 0.78 -0.4 1/19/05 18.0 0.17 0.14 0.69 --5 3/18/05 20.5 0.10 0.04 0.86 2 -5 7/18/05 49.5 0.04 0.04 0.92 --13 10/27/05 15.7 0.09 0.17 0.74 --7 1/17/06 30.4 0.02 0.01 0.97 --10 4/12/06 10.2 0.18 0.00 0.82 4 -1 7/13/06 7.4 0.14 0.10 0.76 -3 10/10/06 5.1 0.16 0.11 0.73 --0.6 1/17/07 3.9 0.16 0.18 0.66 --1 4/10/07 3.7 0.15 0.07 0.78 3 -0.3 a Modeled by allowing River Sink water to achieve SI CAL and P CO2 values determined at River Rise. b Dissolution rate calculated from flux of calc ite at spring (Rise discharge times calcite concentration in moles/l (Table 4-3 )), molar volume of calcite (3.69 x 10 -5 m 3 ), 30% for Ocala Limestone, and surface area along the flow path of 4.2 x 10 5 m 2 (Screaton et al. 2004). c Positive signs indicates uptake of CO 2 and negative sign indicates release of CO 2 d Mixing fractions calculated by inverse modeling in PHREEQC. e -indicates days when no dissolution occurred. 110

PAGE 111

Figure 4-1. Phase diagram s howing relations among sampled water and the aquifer-bearing minerals calcite, dolomite, gypsum. Lines on diagram represent calculated phase boundaries between minerals using reactions in Table 4-2. Solid lines represent equilibrium between mineral phases and dotte d line represents metastable extension of calcite-dolomite phase boundary into th e gypsum field. Oval indicates dilute surface samples collected during high flow and arrow points to all other surface samples collected during average or low flow. 111

PAGE 112

Figure 4-2. Saturation indices of calc ite, dolomite, and gypsum versus Log A) SI 2-24322SOCO/ aa CAL at groundwater wells. B) SI DOL at groundwater wells. C) SI GYP at groundwater wells. D) SI CAL at River Sink and River Rise. E) SI DOL at River Sink and River Rise. F) SI GYP at River Sink and River Rise. Dashed lines represents sa turation (SI = 0). 112

PAGE 113

Figure 4-3. Bar graph showing Ca 2+ concentrations at the River Sink (black bar) and River Rise (gray bar) for each sample trip highlighted on Figure 3-3 113

PAGE 114

Figure 4-4. Conceptual sketch s howing the cross-sectional area of a conduit in an eogenetic karst aquifer. Gray area represents the halo of dissolution in the matrix. Dashed line represents conduit developing along fr acture with hydraulic conductivity (K 1 ), and dotted line represents secondary flow path with hydraulic conductivity (K 2 ) in a matrix of lower hydraulic conductivity (K 3 ). 114

PAGE 115

CHAPTER 5 SUMMARY AND CONCLUSIONS The evolution of porosity in carbonate rocks exerts a str ong control on fluid flow and storage. Most previous studies on porosity evolution have focused on conduit development in telogenetic karst aquifers. A lthough conduits exist in eogeneti c karst aquifers, high matrix porosity and permeability provide a major compone nt of flow, and this flow can generate localized areas of large-scale secondary porosit y where dissolution occurs In this study, I evaluated how matrix flow a ffects the distribution and magn itude of secondary porosity development by coupling water chemistry and geochemical modeling with hydrodynamics from eogenetic karst aquifers of Sa n Salvador Island, Bahamas and north-central Florida, USA. On San Salvador Island, most dissolution occu rs in low-salinity water from elevated concentrations of CO 2 rather than from the mixing of wate r with different chemical compositions as previously thought. Flow remains diffuse from recharge to discharge along the shoreline, and the generation of meter-s cale secondary porosity within freshw ater lenses stays focused along the water table, where flow velocities are greatest and increase to a maximum at the seaward edge of a lens. This dissolution coupled w ith flow dynamics near the wate r table at the lens edge are sufficient to produce flank margin caves and bana na holes because reactions products are flushed from the zone of dissolution. Where local sources of CO 2 drive dissolution near the water table at the lens edge, flank margin caves may develop as reaction products are flushed to the ocean. Banana holes may form wh ere local sources of CO 2 drive dissolution at the water table within the islands interior, but are smaller than flank ma rgin caves because lower flow velocities away from the lens edge provide sufficient time for so me reprecipitation. This hypothesis predicts that banana holes may be surrounded by more tightly cemented rock if reaction products are not move far beyond the developing void. 115

PAGE 116

On the Florida platform, chemical and hydr ologic observations of groundwater and spring discharge suggest that high matrix permeability provi des a significant component of flow even in portions of an aquifer dominated by conduits. In the Santa Fe River Sink-Rise system, spring discharge, and thus water within the conduit, is composed of th ree chemically-distinct sources including allogenic recharge at a swallet, diffuse recharge through the vadose zone, and upwelling of mineralized water fr om deep within the aquifer. The relative amounts of these sources are controlled by vari ations in hydraulic head betw een the conduit and surrounding aquifer, which is a function of rainfall, evapotra nspiration, and river stag e. Dissolution along the conduit flow path is limited in eogenetic karst aq uifers because the flow of Ca-rich water from the matrix to the conduit, whic h occurs most of the time, lim its conduit water undersaturated with respect to calcite from c ontacting the conduit wall. Instead, most dissolution occurs when hydraulic head in the conduit ex ceeds hydraulic head in the su rrounding aquifer during high-flow events and water undersaturated with respect to calcite en ters the aquifer matrix in the vicinity of the conduit. This process predicts the generation of a friable halo of less consolidated matrix rock around the conduit, whereby enlargem ent results from both dissolution within the matrix as well as along the conduit wall. This conceptualizat ion is distinct from that used for conduit enlargement in telogenetic karst aquifers. Results of this dissertation show how matrix porosity and permeability allow meter-scale pores to develop in eogenetic karst aquifers. Is olated voids may form along water-table horizons lacking discrete inputs or outputs within a diffuse flow field, and conduits may enlarge, in part, from dissolution of the matrix rock surrounding a conduit. This work provides new concepts for better understanding how secondary por osity evolves in eogenetic ka rst aquifers, which is vital for long-term exploitation of natural resources including freshwater and hydrocarbons. One 116

PAGE 117

117 important finding was that voids developing on small carbonate is lands only require recharge catchments slightly larger than the size of the vo id itself, suggesting that catchment size may not limit cave development. In addition, the genera tion of these voids do not require mixing of freshwater and seawater but only inputs of CO 2 which does not restrict porosity development only to areas where mixing occurs. Although this work addresses the enlargement of conduits that develop in sink-rise system s receiving allogenic recharge, many conduits draining eogenetic karst aquifers lack discrete inputs. An importa nt question, that remains unanswered, is how do conduits that receive only calcite saturated diffuse flow from the a quifer matrix form. Part of the answer may lie in processes similar to thos e observed on small carbonate islands, whereby dissolution and flow at the water table generate isolated voids that later become submerged and breached at the surface.

PAGE 118

APPENDIX FIELD DATA AND CHEMICAL ANALYSIS OF WATER SAMPLES FROM SAN SALVADOR ISLAND, BAHAMAS Table A-1. Water chemistry from San Salvador Island, Bahamas. Location Samples pH Temp. Sal. Na + K + Mg 2+ Ca 2+ Sr 2+ Cl SO 4 2Alk. (C) (psu) (mM) (mM) (mM) (mM) (mM) (mM) (mM) (mM) Majors Cave MC0-05 7.68 23.8 27.8 358 7.33 41.9 8.82 0.095 424 21.5 2.40 Majors Cave MC1-05 7.68 23.7 27.8 362 7.39 42.2 8.86 0.095 423 21.4 2.42 Majors Cave MC2-05 7.67 23.7 27.8 362 7.38 42.4 9.04 0.095 425 21.5 2.44 Majors Cave MC3-05 7.67 23.7 27.8 367 7.57 42.6 8.79 0.094 425 21.5 2.50 Majors Cave MC4-05 7.59 23.8 28.0 370 7.74 43.1 9.23 0.097 429 21.8 2.48 Majors Cave MC5-05 7.55 23.9 28.3 379 7.59 44.7 9.45 0.096 436 22.1 2.56 Majors Cave MC6-05 7.68 23.5 27.8 362 7.42 42.3 8.97 0.094 426 21.6 2.37 Bat Nook BN0-05 7.63 23.3 25.3 328 6.68 38.4 8.21 0.088 385 19.3 2.43 Bat Nook BN1-05 7.55 23.8 27.1 363 7.30 42.0 8.48 0.096 413 20.8 2.46 Bat Nook BN2-05 7.59 23.8 27.5 363 7.25 42.2 8.59 0.097 422 21.3 2.40 Watling's B.H. WB-05 8.51 30.6 27.8 369 7.39 43.2 9.26 0.098 422 21.3 2.22 Ink Well IW-05 7.70 24.2 5.2 66.5 1.29 7.1 3.32 0.039 77.8 3.71 2.85 Crescent Top Cave CT-05 7.44 26.5 35.5 483 19.10 51.8 9.80 0.111 545 28.3 2.09 Crescent Pond CP-05 8.44 28.4 37.8 492 10.30 57.6 11.20 0.115 581 30.4 2.08 Over Bridge Well 1 OBW1-05 7.87 26.1 0.3 3.1 0.00 1.8 1.56 0.049 2.9 0.31 4.02 Over Bridge Well 2 OBW2-05 7.87 27.8 0.5 4.4 0.00 2.4 1.66 0.048 4.3 0.4 5.05 North Point Well NP-05 7.60 26.1 1.9 18.8 0.19 2.2 4.89 0.256 30.2 0.2 1.80 Six Pack Pond SPP-05 8.37 25.3 51.7 679 13.70 78.3 15.30 0.194 781 38.7 2.52 Majors Cave MC0-06 8.00 24.3 1.8 23 0.71 2.2 1.17 0.012 23.1 0.9 3.03 Majors Cave MC1-06 7.84 24.2 7.6 93.2 2.32 9.5 2.77 0.028 104 4.8 2.71 Majors Cave MC2-06 7.78 24.1 20.3 243 4.94 27.6 6.21 0.067 269 12.6 2.78 Majors Cave MC3-06 7.76 24.1 27.8 330 6.99 38.6 8.19 0.089 390 19.8 2.82 Majors Cave MC4-06 7.78 24.1 30.0 348 7.09 40.8 8.52 0.093 421 21.5 2.80 Majors Cave MC5-06 7.72 24.1 31.0 356 7.48 41.3 8.48 0.094 424 21.6 2.81 Majors Cave MC6-06 7.59 24.1 31.1 357 7.54 41.7 8.60 0.096 414 20.9 2.88 Bat Nook BN0-06 7.64 24.1 27.4 324 6.79 37.8 7.96 0.088 366 18.3 2.94 118

PAGE 119

Table A-1. Continued Location Samples pH Temp. Sal. Na + K + Mg 2+ Ca 2+ Sr 2+ Cl SO 4 2Alk. (C) (psu) (mM) (mM) (mM) (mM) (mM) (mM) (mM) (mM) Bat Nook BN1-06 7.65 24.1 29.2 337 7.26 39.5 8.11 0.092 387 19.3 2.93 Bat Nook BN2-06 7.61 24.1 30.3 355 7.46 41.1 8.39 0.091 403 20.2 2.81 Bat Nook BN3-06 7.57 24.2 31.1 362 7.50 41.9 8.62 0.092 412 20.7 2.88 Bat Nook BN4-06 7.54 24.2 31.5 368 7.77 42.7 8.71 0.096 416 20.9 2.86 Crescent Top Cave CT-06 7.55 26.6 40.1 456 9.87 53.8 10.30 0.107 542 28.4 2.57 Field Station Wells GRC1-06 7.40 27.3 0.5 4.6 0.15 2.1 1.98 0.148 7.4 0.4 4.84 Field Station Wells GRC2-06 7.27 26.7 0.6 5.8 0.16 2.4 2.03 0.155 9.2 0.5 4.79 Museum Well MW-06 7.37 27.3 0.9 10.6 0.26 1.6 1.76 0.161 15.8 0.6 4.37 North Victoria Well NVH-06 7.48 26.5 10.0 113 1.97 13.0 3.85 0.066 128 6.4 3.47 Ink Well B.H. IW0-07 6.58 24.2 3.0 36.4 0.91 3.8 1.72 0.021 39.2 2.1 2.87 Ink Well B.H. IW1-07 6.89 23.4 0.5 6.5 0.34 1.1 1.21 0.005 9.2 0.5 3.01 Ink Well B.H. IW2-07 6.72 23.9 2.7 32.3 0.82 3.4 1.61 0.017 33.6 1.9 2.69 Ink Well B.H. IW3-07 7.19 24.3 12.2 145 2.98 15.8 2.58 0.044 156 8.5 3.88 Ink Well B.H. IW4-07 6.74 24.5 17.5 205 4.38 23.5 5.22 0.058 225 11.6 3.88 Ink Well B.H. IW5-07 7.27 24.6 8.1 107 2.39 11.7 3.46 0.037 123 6.0 3.61 Ink Well B.H. IW6-07 7.32 24.1 1.4 18.1 0.51 2.0 1.25 0.014 19.2 1.3 2.52 Ink Well B.H. IW7-07 7.47 23.9 0.5 6.5 0.34 1.0 1.22 0.013 9.2 0.5 3.17 Ink Well B.H. IW8-07 7.50 24.6 0.5 6.5 0.35 1.1 1.22 0.013 9.2 0.5 3.12 Ink Well B.H. IW9-07 7.42 23.8 0.5 6.5 0.33 1.0 1.17 0.013 9.3 0.5 3.15 Ink Well B.H. IW10-07 7.22 24.6 3.4 52.7 1.28 5.7 2.34 0.025 59.2 2.9 3.22 Ink Well B.H. IW11-07 7.13 24.8 14.4 170 3.66 19.4 4.48 0.050 183 9.6 3.97 Line Hole Well LH1-07 7.29 27.1 0.3 3.9 0.07 0.4 2.32 0.027 6.2 0.2 4.56 Line Hole Well LH2-07 7.53 27.1 0.3 4.2 0.09 0.5 1.84 0.019 5.7 0.3 3.95 Line Hole Well LH3-07 7.27 27.1 0.2 2.8 0.05 0.3 2.30 0.079 3.9 0.2 4.52 Line Hole Well LH4-07 6.94 27.1 0.8 7.9 0.09 0.6 4.19 0.079 13.6 0.4 7.41 Line Hole Well LH5-07 7.04 26.8 0.5 4.36 0.04 0.4 3.44 0.039 7.11 0.2 6.15 Line Hole Well LH6-07 6.95 26.2 2.0 20.8 0.31 1.9 4.28 0.088 22.7 1.0 6.96 Line Hole Well LH7-07 7.04 26.2 1.3 14.5 0.17 1.2 3.87 0.065 15.3 0.8 6.14 119

PAGE 120

Table A-1. Continued Location Samples pH Temp. Sal. Na + K + Mg 2+ Ca 2+ Sr 2+ Cl SO 4 2Alk. (C) (psu) (mM) (mM) (mM) (mM) (mM) (mM) (mM) (mM) Line Hole Well LH8-07 7.06 26.4 0.90 9.6 0.13 0.8 3.00 0.055 15.4 0.5 6.25 Line Hole Well LH9-07 6.98 26.3 0.90 7.5 0.09 0.7 4.11 0.117 15.6 0.3 6.58 Line Hole Well LH10-07 7.17 26.4 0.80 6.9 0.20 0.7 3.56 0.029 14.8 0.2 6.58 Line Hole Well LHA-07 7.05 26.9 0.40 2.7 0.03 0.3 3.54 0.098 3.9 0.2 7.38 Line Hole Well LHB-07 7.22 26.6 2.40 25.2 0.49 2.9 3.70 0.052 27.2 1.3 7.12 Museum Well MW-07 7.36 26.6 3.50 38.5 0.81 4.0 3.17 0.278 44.0 1.2 5.34 North Point Well NP-0 7.79 26.0 0.70 11.9 0.46 0.6 0.90 0.046 8.6 1.2 6.60 Short Stop Well SS-07 7.99 25.5 0.50 5.1 0.18 1.4 1.66 0.076 6.7 0.5 4.44 Watling's B.H. WB-07 8.47 23.7 24 281 5.91 32.6 6.27 0.075 305 16.5 3.11 120

PAGE 121

Table A-1. Continued Location Samples DIC log PCO 2 SI CAL SI ARG Charge Bal. (mM HCO 3 ) (bars) (log Q/K) (mM HCO 3 ) Majors Cave MC0-05 a -2.80 0.16 0.02 -0.3 Majors Cave MC1-05 -2.80 0.17 0.02 0.3 Majors Cave MC2-05 -2.79 0.17 0.02 0.2 Majors Cave MC3-05 -2.78 0.17 0.02 0.8 Majors Cave MC4-05 -2.70 0.11 -0.04 0.8 Majors Cave MC5-05 -2.64 0.09 -0.06 1.4 Majors Cave MC6-05 -2.81 0.16 0.01 -1.1 Bat Nook BN0-05 -2.74 0.11 -0.04 0.3 Bat Nook BN1-05 -2.66 0.03 -0.11 1.6 Bat Nook BN2-05 -2.71 0.06 -0.08 0.6 Watling's B.H. WB-05 -3.73 0.96 0.82 1.6 Ink Well IW-05 -2.59 0.22 0.08 0.4 Crescent Top Cave CT-05 -2.69 -0.05 -0.19 1.9 Crescent Pond CP-05 -3.73 0.90 0.72 -0.4 Over Bridge Well 1 OBW1-05 -2.51 0.44 0.30 10.7 Over Bridge Well 2 OBW2-05 -2.42 0.66 0.51 10.8 North Point Well NP-05 -2.63 0.29 0.15 1.1 Six Pack Pond SPP-05 -3.63 0.90 0.76 1.2 Majors Cave MC0-06 2.31 -2.82 0.27 0.13 4.3 Majors Cave MC1-06 2.67 -2.78 0.19 0.05 1.6 Majors Cave MC2-06 2.87 -2.79 0.28 0.13 3.1 Majors Cave MC3-06 2.88 -2.80 0.30 0.16 -0.3 Majors Cave MC4-06 2.83 -2.83 0.32 0.18 -1.6 Majors Cave MC5-06 2.84 -2.77 0.26 0.11 -0.9 Majors Cave MC6-06 -2.63 0.15 0.01 0.7 Bat Nook BN0-06 3.09 -2.66 0.20 0.06 2.1 Bat Nook BN1-06 -2.67 0.21 0.06 1.3 121

PAGE 122

Table A-1. Continued Location Samples DIC log PCO 2 SI CAL SI ARG Charge Bal. (mM HCO 3 ) (bars) (log Q/K) (mM HCO 3 ) Bat Nook BN2-06 -2.66 0.15 0.01 1.8 Bat Nook BN3-06 -2.60 0.13 -0.01 1.7 Bat Nook BN4-06 3.83 -2.58 0.10 -0.04 2.1 Crescent Top Cave CT-06 2.62 -2.65 0.12 -0.03 -0.6 Field Station Wells GRC1-06 5.30 -1.96 0.26 0.11 -0.4 Field Station Wells GRC2-06 5.06 -1.84 0.11 -0.03 -0.8 Museum Well MW-06 4.33 -1.98 0.10 -0.05 -10.3 North Victoria Well NVH-06 4.34 -2.30 0.09 -0.06 1.4 Ink Well B.H. IW0-07 2.78 -1.60 -1.23 -1.37 3.3 Ink Well B.H. IW1-07 2.90 -1.78 -0.81 -0.96 -4.7 Ink Well B.H. IW2-07 2.66 -1.72 -1.08 -1.23 4.8 Ink Well B.H. IW3-07 3.82 -2.03 -0.44 -0.58 2.4 Ink Well B.H. IW4-07 3.95 -1.64 -0.70 -0.84 3.1 Ink Well B.H. IW5-07 3.68 -2.10 -0.46 -0.33 0.4 Ink Well B.H. IW6-07 2.48 -2.24 -0.46 -0.60 2.3 Ink Well B.H. IW7-07 2.92 -2.29 -0.15 -0.30 -6.6 Ink Well B.H. IW8-07 -2.66 0.15 0.01 1.8 Ink Well B.H. IW9-07 -2.60 0.13 -0.01 1.7 Ink Well B.H. IW10-07 3.83 -2.58 0.10 -0.04 2.1 Ink Well B.H. IW11-07 2.62 -2.65 0.12 -0.03 -0.6 Line Hole Well LH1-07 5.30 -1.96 0.26 0.11 -0.4 Line Hole Well LH2-07 5.06 -1.84 0.11 -0.03 -0.8 Line Hole Well LH3-07 4.33 -1.98 0.10 -0.05 -10.3 Line Hole Well LH4-07 4.34 -2.30 0.09 -0.06 1.5 Line Hole Well LH5-07 2.78 -1.60 -1.23 -1.37 3.3 Line Hole Well LH6-07 2.90 -1.78 -0.81 -0.96 -4.7 Line Hole Well LH7-07 2.66 -1.72 -1.08 -1.23 4.8 122

PAGE 123

123 Table A-1. Continued Location Samples DIC log PCO 2 SI CAL SI ARG Charge Bal. (mM HCO 3 ) (bars) (log Q/K) (mM HCO 3 ) Line Hole Well LH8-07 6.37 -1.57 0.10 -0.04 -11.8 Line Hole Well LH9-07 7.99 -1.41 0.24 0.10 -14.8 Line Hole Well LH10-07 -1.61 0.37 0.22 -16.9 Line Hole Well LHA-07 7.76 -1.46 0.31 0.17 -2.7 Line Hole Well LHB-07 7.13 -1.69 0.33 0.18 3.4 Museum Well MW-07 -1.93 0.27 0.13 1.9 North Point Well NP-0 5.96 -2.28 0.34 0.19 -4.6 Short Stop Well SS-07 3.76 -2.68 0.64 0.49 0.2 Watling's B.H. WB-07 2.68 -3.54 0.90 0.75 3.6 a Sample not collected for analysis

PAGE 124

LIST OF REFERENCES AM ENDE, B., 2000, Wakulla 2 building the first fully 3-D cave map: National Speleological Society News, v. 58, p. 131-140. ANDERSON, G.M., and CRERAR, D.A., 1993, Thermodynamics in Geochemistry: New York, Oxford University Press, 588 p. ANTHONY, S.S., PETERSON, F.L., MACKENZIE, F.T., and HAMLIN, S.N., 1989, Geohydrology of the Laura fresh-water lens, Majuro Ato ll: a hydrogeochemical approach: Geological Society of America Bulletin, v. 101, p. 1066-1075. ATKINSON, T.C., 1977, Carbon dioxide in the atmosphere of the unsaturated zone: An important control of groundwater hardness in limestones: Journal of Hydrology, v. 35, p. 111-123. BACETA, J.I., PUJALTE, V., WRIGHT, V.P., and LAPOINTE, P., 2008, Eogenetic karstification and palaeohydrology of a Danian land-attached carbonate shelf (North Spain), in Sasowsky, I.D., Freazel, C.T., Mylroie, J.E., Palmer, A. N., and Palmer, M.V., eds., Karst from Recent to Reservoirs, Karst Waters Institut e Special Publication 14, p. 21-22. BACK, W., and HANSHAW, B.B., 1970, Comparison of chemi cal hydrogeology of the carbonate peninsulas of Florida and Yucatan: J ournal of Hydrology, v. 10, p. 330-368. BACK, W., HANSHAW, B.B., HERMAN, J.S., and VAN DRIEL, J.N., 1986, Differential dissolution of a Pleistocene reef in the ground-water mixing zone of coasta l Yucatan, Mexico: Geology, v. 14, p. 137-140. BALL, J.W., and NORDSTROM, D.K., 1991, User's Manual for WATEQ4F, with Revised Thermodynamic Data Base and Test Cases for Ca lculating Speciation of Major, Trace, and Redox Elements in Natural Wa ters: Open-File Report 91-183, United States Geological Survey, 192 p. BEAR, J., and TODD, D.K., 1960, The transition zone betwee n fresh and salt waters in coastal aquifers: Water Resources Centre Contributi on, v. 29: Berkeley, Hydraulic Laboratory, University of California, 156 p. BECK, B.F., 1986, Ground water monitoring considerat ions in karst on young limestones, In: Symposium on Environmental Problems in Kars t Terranes and Their Solutions (October, 1986): Bowling Green, Kentucky, National Water Well Association, p. 229-248. BEDDOWS, P.A., 2004, Groundwater hydrology of a coasta l conduit carbonate aquifer: Caribbean coast of the Yucatan Peninsula [unpublished Ph .D. Diss.]: University of Bristol, 303 p. BGLI, A., 1980, Karst hydrology and physical speleology: New York, Springer-Verlag, 284 p. 124

PAGE 125

BOTTRELL, S.H., CAREW, J.L., and MYLROIE, J., 1993, Inorganic and bacteriogenic origins for sulfate crusts in flank margin cave s, San Salvador Island, Bahamas, in White, B., ed., Proceedings of the Sixth Symposium on th e Geology of the Bahamas: San Salvador, Bahamas, Bahamian Field Station, p. 17-21. BOTTRELL, S.H., SMART, P.L., WHITAKER, F., and RAISWELL, R., 1991, Geochemistry and isotope systematics of sulphur in the mixi ng zone of Bahamian blue holes: Applied Geochemistry, v. 6, p. 97-103. BREDEHOEFT, J.D., and PAPADOPULOS, I.S., 1965, Rates of Vertical Groundwater Movement Estimated from the Earth's Thermal Profile: Water Resources Research, v. 1, p. 325-328. BUDD, D.A., 1988, Aragonite-to-calcite transforma tion during fresh-water diagenesis of carbonates: Insights from porewater chemistry: Geological Society of America Bulletin, v. 100, p. 1260-1270. BUDD, D.A., and VACHER, H.L., 1991, Predicting the thickness of fresh-water lenses in carbonate paleo-islands: Journal of Sedime ntary Petrology, v. 61, p. 43-53. BUDD, D.A., and VACHER, H.L., 2004, Matrix permeability of the confined Floridan Aquifer, Florida, USA: Hydrogeology J ournal, v. 12, p. 531-549. BUSH, P.W., and JOHNSTON, R.H., 1988, Ground-waer hydraulic s, regional flow, and groundwater development of the Flor idan Aquifer system in Florid a and in parts of Georgia, South Carolina and Alabama: Professional Pa per 1403-C, U.S. Geological Survey, 80 p. CANT, R.V., and WEECH, P.S., 1986, A review of the factor s affecting the development of Ghyben-Hertzberg lenses in the Bahamas: Journal of Hydrology, v. 84, p. 333-343. CAREW, J.L., and MYLROIE, J.E., 1995a, Quaternary tectoni c stability of the Bahamian Archipelago; evidence from fossil coral reefs and flank margin caves: Quaternary Science Reviews, v. 14, p. 145-153. CAREW, J.L., and MYLROIE, J.E., 1995b, Depositional model a nd stratigraphy for the Quaternary geology of the Bahama Islands, in Curran, H.A., and White, B., eds., Terrestrial and Shallow Marine Geology of the Bahamas a nd Bermuda: Geological Society of America, Special Paper 300, p. 5-32. CAREW, J.L., and MYLROIE, J.E., 1997, Geology of the Bahamas, in Vacher, H.L., and Quinn, T.M., eds., Geology and Hydrogeology of Ca rbonate Islands, Developments in Sedimentology 54: Amsterdam, Elsevier, p. 91-139. CHELETTE, A.R., PRATT, T.R., and KATZ, B.G., 2003, Nitrate loading as an indicator of nonpoint source pollution in the lower St Marks-Wakulla Rivers watershed: Northwest Florida Water Management District, Water Re sources Special Report 02-1, 138 p. 125

PAGE 126

CHEN, J.H., CURRAN, H.A., WHITE, B., and WASSERBURG, G.J., 1991, Precise chronology of the last interglacial period: 234 U230 Th data from fossil coral reefs in the Bahamas: Geological Society of America Bulletin, v. 103, p. 82-97. CHEN, K., JIAO, J.J., HUANG, J., and HUANG, R., 2007, Multivariate stat istical evaluation of trace elements in groundwater in a coastal area in Shenzhen China: Environmental Pollution, v. 147, p. 771-780. CHOQUETTE, P.W., and PRAY, L.C., 1970, Geologic nomenclature and classification of porosity in sedimentary carbonates: American Association of Petroleum Geologi sts Bulletin, v. 54, p. 207-244. CRAIG, D.H., 1988, Caves and Other Feat ures of Permian Karst in San Andres Dolomite, Yates Field Reservoir, West Texas, in James, N.P., and Choquette, P.W., eds., Paleokarst: New York, Springer-Verlag, p. 342-363. CRANDALL, C.A., KATZ, B.G., and HIRTEN, J.J., 1999, Hydrochemical evidence for mixing of river water and groundwater during high-flow conditions, lower Suwannee River basin, Florida, USA: Hydrogeology J ournal, v. 7, p. 454-467. CRUMP, M.A., and GAMBLE, D.W., 2006, Hydroclimatic Analysis of a Carbonate Island Pond Through the Development of a Hydrologic Land scape Unit Model: Phys ical Geography, v. 27, p. 554-570. DAVIS, R.L., and JOHNSON, J., C.R., 1989, Karst hydrology of San Salvador, in Mylroie, J.E., ed., Proceedings on the 4th Symposium on the Geology of the Bahamas: San Salvador, Bahamas, Bahamian Field Station, p. 118-135. DEMING, D., 2002, Introduction to Hydrogeology: New York, McGraw-Hill, 468 p. DINGMAN, S.L., 2002, Physical Hydrology: Upper Saddle River, Prentice-Hall, 646 p. DOGWILER, T., and WICKS, C.M., 2004, Sediment entrainment and transport in fluviokarst systems: Journal of Hydrology, v. 295, p. 163-172. DREISS, S.J., 1989, Regional scale transport in a karst aquifer, 1. Component separation of spring flow hydrographs: Water Resources Research, v. 25, p. 117-125. DREVER, J.I., 1997, The Geochemistry of Natural Waters: Englewood Cliffs, Prentice-Hall, 437 p. DREYBRODT, W., 1990, The role of dissolution kinetics in the development of karst aquifers in limestone: Journal of Geology, v. 98, p. 639-655. DREYBRODT, W., and GABROVSEK, F., 2002, Basic processes and mechanisms governing the evolution of karst, in Gabrovsek, F., ed., Evolution of Ka rst: From Prekarst to Cessation: Postojna-Ljubljana, Zalozba ZRC, p. 115-154. 126

PAGE 127

EDWARDS, L.E., WEEDMAN, S.D., SIMMONS, K.R., SCOTT, T.M., BREWSTER-WINGARD, G.L., ISHMAN, S.E., and CARLIN, N.M., 1998, Lithostratigraphy, Petrography, Biostratigraphy, and Strontium-Isotope Stratigra phy of the Surfical Aquifer System of Western Collier County, Florida: Open-File Report 98-205, United States Geological Survey, 79 p. EPA, 1983, Methods for the chemical analysis of water and wastes: Cinc innati, United States Environmental Protection Agency, 552 p. FLOREA, L.J., 2006, The Karst of West-Central Florid a [unpublished Ph.D. diss.]: University of South Florida, 564 p. FLOREA, L.J., and VACHER, H.L., 2006, Springflow hydrographs: E ogenetic vs. telogenetic karst: Ground Water, v. 44, p. 352-361. FLOREA, L.J., and VACHER, H.L., 2007, Eogenetic karst hydrology: Insights from the 2004 hurricanes, peninsular Florida: Ground Water, v. 45, p. 439-446. FLOREA, L.J., MYLROIE, J.E., and PRICE, A., 2004, Sedimentation and porosity enhancement in a breached flank margin cave: Carbonates and Evaporites, v. 19, p. 75-85. FLOREA, L.J., VACHER, H.L., DONAHUE, B., and NAAR, D., 2007, Quaternary cave levels in peninsular Florida: Quaternary Sc ience Reviews, v. 26, p. 1344-1361. FORD, D.C., and EWERS, R.O., 1978, The development of limestone cave systems in the dimensions of length and depth: Canadian J ournal of Earth Science, v. 15, p. 1783-1798. FORD, D.C., and WILLIAMS, P.W., 2007, Karst Hydrogeology and Geomorphology: Chichester, Wiley, 562 p. FOURNIER, M., MASSEI, N., BAKALOWICZ, M., DUSSART-BAPTISTA, L., RODET, J., and DUPONT, J.P., 2007, Using turbidity dynamics and ge ochemical variability as a tool for understanding the behavior and vulnerability of a karst aquifer: Hydrogeology Journal, v. 15, p. 689-704. FREEZE, R.A., and CHERRY, J.A., 1979, Groundwater: Englewood Cliffs, Prentice-Hall, 290 p. GABROVEK, F., and DREYBRODT, W., 2001, A model of the early e volution of karst aquifers in limestone in the dimensions of length and de pth: Journal of Hydrology, v. 240, p. 206-224. GABROVEK, F., ROMANOV, D., and DREYBRODT, W., 2004, Early karstification in a dualfracture aquifer; the role of exchange flow between prominen t fractures and a dense net of fissures: Journal of Hydr ology, v. 299, p. 45-66. GANDOLFI, C., FACCHI, A., and WHELAN, M.J., 2001, On the Relative Role of Hydrodynamic Dispersion for River Water Quality: Water Resources Research, v. 37, p. 2365-2375. GRASSO, D.A., and JEANNIN, P.-Y., 2002, A global experimental system approach of karst springs hydrographs and chemographs : Ground Water, v. 40, p. 608-617. 127

PAGE 128

GRASSO, D.A., JEANNIN, P.Y., and ZWAHLEN, F., 2003, A deterministic approach to the coupled analysis of karst springs' hydrographs and ch emographs: Journal of Hydrology, v. 271, p. 65-76. GRIGGS, J.E., and PETERSON, F.L., 1993, Ground-water flow dynamics and development strategies at the at oll scale: Ground Water, v. 31, p. 209-220. GROSZOS, M., CERYAK, R., ALLISON, D., COOPER, R., WEINBERG, M., MACESICH, M., ENRIGHT, M.M., and RUPERT, F., 1992, Carbonate units of the Inte rmediate Aquifer system in the Suwannee River Water Management District, Florida: Open-file report 54, Florida Geological Survey, 22 p. GROVES, C., and MEIMAN, J., 2005, Weathering, geomorphic work, and karst landscape evolution in the Cave City groundwat er basin, Mammoth Cave, Kentucky: Geomorphology, v. 67, p. 115-126. GROVES, C.G., and HOWARD, A.D., 1994, Early development of karst systems 1. Preferential flow path enlargement under laminar flow: Water Resources Research, v. 30, p. 28372846. HALIHAN, T., and WICKS, C., 1998, Modeling of storm response in conduit flow aquifers with reservoirs: Journal of Hydr ology, v. 208, p. 82-91. HALIHAN, T., SHARP, J.M., JR., and MACE, R.E., 2000, Flow in the San Antonio segment of the Edwards aquifer: Matrix, fractures, or conduits?, in Sasowski, I.D., and Wicks, C.M., eds., Groundwater Flow and Contaminant Transport in Carbonate Aquifers: Rotterdam, A.A. Balkema, p. 129-146. HANSHAW, B.B., and BACK, W., 1979, Major geochemical pr ocesses in the evolution of carbonate-aquifer systems: Journal of Hydrology, v. 43, p. 287-312. HARRIS, J.G., MYLROIE, J.E., and CAREW, J.L., 1995, Banana holes: uniq ue karst features of the Bahamas: Carbonates and Evaporites, v. 10, p. 215-224. HATTIN, D.E., and WARREN, V.L., 1989, Stratigraphic anal ysis of a fossil Neogoniolithoncapped patch reef and associated facies, San Salvador, Bahamas: Coral Reefs, v. 8, p. 1930. HELGESON, H.C., 1969, Thermodynamics of hydrothermal systems at elevated temperatures and pressures: American Journal of Science, v. 267, p. 729-804. HESS, J.W., and WHITE, W.B., 1988, Storm response of the ka rstic carbonate aquifer of south central Kentucky: Journal of Hydrology, v. 99, p. 235-252. HISERT, R.A., 1994, A multiple tracer approach to determine the ground and surface water relationships in the western Santa Fe River, Columbia C ounty, Florida [unpublished Ph.D. diss.]: University of Florida, 211 p. 128

PAGE 129

HOWARD, A.D., and GROVES, C.G., 1995, Early development of ka rst systems 2. Turbulent flow: Water Resources Research, v. 31, p. 19-26. HUGHES, J.D., VACHER, H.L., and SANFORD, W.E., 2007, Three-dimensional flow in the Florida platform: Theoretical analysis of Kohout convect ion at its type locality: Geology, v. 35, p. 663-666. HUNN, J.D., and SLACK, L.J., 1983, Water resources of the Santa Fe River Basin, Florida: WaterResources Investigations Report 83-4075, United States Geological Survey, 105 p. HUNTOON, P.W., 1995, Is it appropriate to apply porous media groundwater circulation models to karstic aquifers?, in El-Kadi, A.I., ed., Groundwater M odels for Resource Analysis and Management: Boca Raton, Lewis Publishers, p. 339-358. JAEGGI, D., 2006, Multiscalar porosity structure of a Miocene reefal carbonate complex [unpublished PhD. diss.]: Swiss Federal Institute of Technology (ETH), 196 p. JOHNSON, J.W., OELKERS, E.H., and HELGESON, H.C., 1992, SUPCRT92; a software package for calculating the standard molal thermodynamic properties of minerals, gases, aqueous species, and reactions from 1 to 5000 bar and 0 to 1000 degrees C: Computers and Geosciences, v. 18, p. 899-947. JOLLIFFE, I.T., 2002, Principle component anal ysis: New York, Springer, 487 p. JONES, I.C., VACHER, H.L., and BUDD, D.A., 1993, Transport of calcium, magnesium and SO 4 in the Floridan aquifer, west-central Florida: implications to cementation rates: Journal of Hydrology, v. 143, p. 455-480. JUNGE, C.E., and WERBY, R.T., 1958, The concentration of chloride, sodium, potassium, calcium, and sulfate in rain water of over the United States: Journal of the Atmospheric Sciences, v. 15, p. 417-425. KATZ, B.G., 2004, Sources of nitrate contamination and ag e of water in large karstic springs of Florida: Environmental Ge ology, v. 46, p. 689-706. KATZ, B.G., BHLKE, J.K., and HORNSBY, H.D., 2001, Timescales for nitrate contamination of spring waters, northern Florida, USA: Chemical Geology, v. 179, p. 167-186. KATZ, B.G., CHELETTE, A.R., and PRATT, T.R., 2004, Use of chemical and isotopic tracers to assess sources of nitrate and age of ground wate r, Woodville Karst Plain, USA: Journal of Hydrology, v. 289, p. 36-61. KATZ, B.G., CATCHES, J.S., BULLEN, T.D., and MICHEL, R.L., 1998, Changes in the isotopic and chemical composition of ground water resulti ng from a recharge pulse from a sinking stream: Journal of Hydrology, v. 211, p. 178-207. KAUFMANN, G., and BRAUN, J., 1999, Karst aquifer evolution in fractured rocks: Water Resources Research, v. 35, p. 3223-3238. 129

PAGE 130

KAUTZ, R., STYS, B., and KAWULA, R., 2007, Florida Vegetation 2003 and Land Use Change Between 1985-89 and 2003: Florida Scientist, v. 70, p. 12-23. KERN, D.M., 1960, The hydration of carbon dioxide: Jour nal of Chemical Education, v. 37, p. 14-23. KINCAID, T., 1999, Morphologic and Fract al Characterization of Saturated Karstic Caves [unpublished Ph.D. diss.]: Univ ersity of Wyoming, 174 p. KNOCHEMUS, L.A., and ROBINSON, J.L., 1996, Descriptions of an isotropy and heterogeneity and their effect on ground-water flow and areas of contribution to public supply wells in a karst carbonate aquifer system: Water-Supply Paper 24 75, United States Geological Survey, 47 p. KOHOUT, F.A., HENRY, H.R., and BANKS, J.E., 1977, Hydrogeology related to geothermal conditions of the Floridan Plateau, in Smith, K.L, and Griffin, G.M., The geothermal nature of the Floridan Plateau: Florida Department of Natural Resources, Bureau of Geology Special Publication 21, 1-41 p. LAAKSOHARJU, M., TULLBORG, E.-L., WIKBERG, P., WALLIN, B., and SMELLIE, J., 1999, Hydrogeochemical conditions and evoluti on at the sp HRL, Sweden: Applied Geochemistry, v. 14, p. 835-859. LABOURDETTE, R., and MYLROIE, J.E., 2008, Flank margin caves: Preservation probability and burial outcomes, in Sasowsky, I.D., Feazel, C.T., Mylroie, J.E., Palmer, A.N., and Palmer, M.V., eds., Karst from Recent to Reservoirs, Ka rst Waters Institute Special Publication 14, p. 107-114. LABOURDETTE, R., LASCU, I., MYLROIE, J.E., and ROTH, M.J., 2007, Process-like modeling of flank margin caves: From genesis to burial evol ution: Journal of Sedi mentary Research, v. 77, p. 965-979. LANGMUIR, D., 1997, Aqueous Environmental Geochemistry: Upper Saddle River, Prentice-Hall, 600 p. LASCU, I., 2005, Speleogenesis of large flank marg in caves of the Bahamas [unpublished M.S. thesis]: Mississippi St ate University, 218 p. LAWRENCE, F.W., and UPCHURCH, S.B., 1982, Identification of rech arge areas using geochemical factor analysis: Ground Water, v. 20, p. 680-687. LAZAREVA, O., and PICHLER, T., 2007, Naturally occurring ar senic in the Miocene Hawthorn Group, southwestern Florida: Potential implication for phosphate mining: Applied Geochemistry, v. 22, p. 953-973. 130

PAGE 131

LIU, Z., and DREYBRODT, W., 1997, Dissolution kinetics of ca lcium carbonate minerals in H 2 OCO 2 solutions in turbulent flow; the role of the diffusion boundary layer and the slow reaction H 2 O + CO 2 = H + + HCO 3 : Geochimica et Cosmochimica Acta, v. 61, p. 28792889. LOPER, D.E., WERNER, C.L., CHICKEN, E., DAVIES, G., and KINCAID, T., 2005, Coastal Carbonate Aquifer Sensitivity to Tides: Eos, Transact ions, American Geophysical Union, v. 86, doi:10.1029/2005EO390001. LU, N., and GE, S., 1996, Effect of horizontal heat and fluid flow on the ve rtical temperature distribution in a semi-confi ning layer: Water Resources Research, v. 32, p. 1449-1453. LUCIA, F.J., 1995, Rock-fabric petrophys ical classification of carbona te pore space for reservoir characterization: American Association of Pe teroleum Geologists Bulletin, v. 79, p. 12751300. MARTIN, J.M., 2003, Quantification of the matrix hydr aulic conductivity in the Santa Fe River Sink/Rise system with implications on the exchange of water between the matrix and conduits [unpublished MS thesis]: University of Florida, 64 p. MARTIN, J.B., and DEAN, R.W., 1999, Temperature as a natura l tracer of short residence times for ground water in karst aquifers, in Palmer, A.N., Palmer, M.V., and Sasowsky, I.D., eds., Karst Modeling: Karst Waters Inst itute Special Publication 5, p. 236-242. MARTIN, J.B., and GORDON, S.L., 2000, Surface and ground water mixing, flow paths, and temporal variations in chemical compositions of karst springs, in Sasowsky, I.D., and Wicks, C., eds., Groundwater Flow and Cont aminant Transport in Carbonate Aquifers: Rotterdam, A.A. Balkema, p. 65-92. MARTIN, J.B., and DEAN, R.W., 2001, Exchange of water between conduits and matrix in the Floridan Aquifer: Chemical Geology, v. 179, p. 145-165. MARTIN, J.B., and SCREATON, E.J., 2001, Exchange of Matrix and Conduit Water with Examples from the Floridan Aquifer, in Kuniansky, E.L., ed., United St ates Geological Survey Karst Interest Group: St. Petersburg, Florida, United States Geological Survey, p. 38-44. MARTIN, J.B., and MOORE, P.J., 2008, Sr concentrations and isotope ratios as tracers of groundwater circulation in carbonate platforms: Examples from San Salvador Island and Long Island, Bahamas: Chemical Geology, v. 249, p. 52-65. MARTIN, J.M., SCREATON, E.J., and MARTIN, J.B., 2006, Monitoring well responses to karst conduit head fluctuations: Implications for flui d exchange and matrix transmissivity in the Floridan aquifer, in Wicks, C.M., and Harmon, R.S., eds., Perspectives on Karst Geomorphology, Hydrology, and Geochemistry: A Tribute Volume to Derek C. Ford and William B. White: Geological Society of Amer ica, Special Publication 404, p. 209-217. 131

PAGE 132

MATSUDA, H., TSUJI, Y., HONDA, N., and SAOTOME, J.-I., 1995, Early diagenesis of Pleistocene carbonates from a hydrogeochemical point of view, Irabu Island, Ryukyu Islands: porosity changes related to early carbonate diagenesis, in Budd, D.A., Harries, P.M., and Saller, A., eds., Unconformities and Porosity in Carbonate Strata: American Association of Petroleum Geologists Memoir, v. 63, p. 35-54. MCCLAIN, M.E., SWART, P.K., and VACHER, H.L., 1992, The hydrogeochemistry of early meteoric diagenesis in a Holocene deposit of biogenic carbonates: Jo urnal of Sedimentary Research, v. 62, p. 1008-1022. MCCONNELL, J., and HACKE, C.M., 1993, Hydrogeology, water quality, and water resources development potential of the U pper Floridan aquifer in the Valdosta area, south-central Georgia: Water-Resources Investigations Report 93-4044, United States Geological Survey, 44 p. MELLOUL, A., and COLLIN, M., 1992, The 'principal components' statistical method as a complementary approach to geochemical methods in water quality f actor identification; application to the Coastal Plai n aquifer of Israel: Journal of Hydrology, v. 140, p. 49-73. MILLER, J.A., 1986, Hydrogeologic framework of the Floridan aquifer system in Florida and parts of Georgia, Alabama, and South Carolina: Professional Paper 1403-B, United States Geological Survey, 91 p. MOORE, C.H., 2001, Carbonate reservoirs porosity evolution and diagenesis in a sequence stratigraphic framework, Developments in se dimentology 55: Amsterdam, Elsevier, 444 p. MOORE, P.J., and MARTIN, J.B., 2005, Exchange of water between conduit and matrix porosity in eogenetic karst? Evidence from a quantitativ e dye trace: Geological Society of America Abstracts with Programs, v. 37, p. 435. MOORE, P.J., and MARTIN, J.B., 2008, Water chemistry on San Salvador Island, Bahamas, in Parks, L., and Freile, D., eds., Proceedings of the 13th Symposium on the Geology of the Bahamas and Other Carbonate Regions: San Salv ador, Bahamas, Gerace Research Centre, p. 1-10. MOORE, P.J., MARTIN, J.B., and GAMBLE, D.W., 2006, Carbonate Water Mixing in a Flank Margin Cave, in Davis, R.L., and D.W., G., eds., Proceedings of the 12th Symposium on the Geology of the Bahamas and Other Ca rbonate Regions: San Salvador, Bahamas, Gerace Research Center, p. 123-129. MOORE, P.J., MARTIN, J.B., and SCREATON, E.J., submitted, Geochemical and statistical evidence of recharge, mixing, and contro ls on spring discharge in an eo genetic karst aquifer: Journal of Hydrology. MORSE, J.W., and MACKENZIE, F.T., 1990, Geochemistry of sedimentary carbonates, Developments in Sedimentology 48: Amsterdam, Elsevier, 707 p. 132

PAGE 133

MULLINS, H.T., and LYNTS, G.W., 1977, Origin of the northwe stern Bahama Platform; review and reinterpretation: Geologi cal Society of America Bulletin, v. 88, p. 1447-1461. MYLROIE, J.E., and CAREW, J.L., 1990, The flank margin model for dissolution cave development in carbonate platforms: Earth Surface Processes and Landforms, v. 15, p. 413-424. MYLROIE, J.E., and CAREW, J.L., 1995, Karst devel opment on carbonate islands, in Budd, D.A., Harris, P.M., and Saller, A., eds., Unconformities and Porosity in Carbonate Strata: American Association of Petroleum Geologists Memoir, v. 63, p. 55-76. MYLROIE, J.E., and VACHER, H.L., 1999, A conceptual view of carbonate island karst, in Palmer, A.N., Palmer, M.V., and Sasowsky, I.D., eds., Karst Modelling: Karst Waters Institute Special Publication 5, p. 48-57. MYLROIE, J.E., CAREW, J.L., and VACHER, H.L., 1995, Karst development in the Bahamas and Bermuda, in Curran, H.A., and White, W.B., eds., Terrestrial and Shallow Marine Geology of the Bahamas and Bermuda: Geological Soci ety of America, Special Paper 300, p. 251267. NG, K.-C., and JONES, B., 1995, Hydrogeochemistry of Grand Cayman, British West Indies; implications for carbonate diagenetic studies : Journal of Hydrology, v. 164, p. 193-216. OLOFSSON, B., JERNBERG, H., and ROSENQVIST, A., 2006, Tracing leachates at waste sites using geophysical and geochemical modelling: Environmental Geology, v. 49, p. 720-732. PALMER, A.N., 1991, Origin and morphology of limestone caves: Geological Society of America Bulletin, v. 103, p. 1-21. PALMER, A.N., 2001, Dynamics of cave development by allogenic water: Acta Carsologica, v. 30, p. 14-32. PALMER, A.N., 2007, Cave Geology: Dayton, Cave Books, 454 p. PALMER, A.N., PALMER, M.V., and SASOWSKY, I.D. (Editors), 1999, Karst Modeling: Charles Town, Karst Waters Institute, 265 p. PARKHURST, D.L., and APPELO, C.A.J., 1999, User's guide to PHREEQC (Version 2): WaterResources Investigations Report 99-4259, United States Geological Survey, 326 p. PETERSON, F.L., 1997, Hydrogeology of the Marshall Islands, in Vacher, H.L., and Quinn, T.M., eds., Geology and Hydrogeology of Carbonate Islands, Developments in Sedimentology 54: Amsterdam, Elsevier, p. 611-636. PHELPS, G.G., 2001, Geochemistry and Origins of Mine ralized Waters in the Floridan Aquifer System, Northeastern Florida: Water-Res ources Investigations Report 01-4112, United States Geological Society, 70 p. 133

PAGE 134

PLUMMER, L.N., 1975, Mixing of sea water with calcium carbonate ground water, in Whitten, E.H.T., ed., Quantitative Studies in the Ge ological Sciences: Geological Society of America Memoir, v. 142, p. 219-238. PLUMMER, L.N., 1977, Defining reactions and mass transf er in part of the Floridan Aquifer: Water Resources Research, v. 13, p. 801-812. PLUMMER, L.N., and BACK, W., 1980, The mass balance approach ; application to interpreting the chemical evolution of hydrologic systems: Am erican Journal of Science, v. 280, p. 130142. PLUMMER, L.N., and BUSENBERG, E., 1982, The solubilities of calc ite, aragonite and vaterite in CO 2 -H 2 O solutions between 0 and 90C, and an evaluation of the aqueous model for the system CaCO 3 -CO 2 -H 2 O: Geochimica et Cosomochim ica Acta, v. 46, p. 1011-1040. PLUMMER, L.N., WIGLEY, T.M.L., and PARKHURST, D.L., 1978, The kinetics of calcite dissolution in CO 2 -water systems at 5 to 60 C and 0.0 to 1.0 ATM CO 2 : American Journal of Science, v. 278, p. 179-216. PLUMMER, L.N., VACHER, H.L., MACKENZIE, F.T., BRICKER, O.P., and LAND, L.S., 1976, Hydrogeochemistry of Bermuda; a case hi story of ground-water diagenesis of biocalcarenites: Geological Society of America Bulletin, v. 87, p. 1301-1316. PLUMMER, L.N., BUSENBERG, E., MCCONNELL, J.B., DRENKARD, S., SCHLOSSER, P., and MICHEL, R.L., 1998, Flow of river water into a Ka rstic limestone aquifer. 1. Tracing the young fraction in groundwater mixtures in the Upper Floridan Aquifer near Valdosta, Georgia: Applied Geochemistry, v. 13, p. 995-1015. POUCHER, M., 2007, O'Leno/Old Bellamy cave exploration, http://cavesurvey.com/ob_cave.htm accessed on: February 20, 2009. PURI, H.S., and VERNON, R.O., 1964, Summary of the Geolog y of Florida and a Guidebook to the Classic Exposures: Special Publicati on 5, Florida Geological Survey, 312 p. QUINLAN, J.F., DAVIES, G.J., JONES, S.W., and HUNTOON, P.W., 1996, The applicability of numerical models to adequately characterize ground-water flow in karstic and other tripleporosity aquifers, in Ritchey, J.D., and Rumbaugh, J.O., eds., Subsurface fluid-flow (ground-water and vadose zone) modeling: West Conshohocken, American Society for Testing and Materials, Special Technical Paper 1288, p. 114-133. RAEISI, E., and MYLROIE, J.E., 1995, Hydrodynamic behavior of caves formed in the fresh-water lens of carbonate islands: Carbonate s and Evaporites, v. 10, p. 207-214. REEL, D.A., and GRIFFIN, G.M., 1971, Potentially petroliferous trends in Florida as defined by geothermal gradients: Transacti ons Gulf Coast Association of Geological Societies, v. 21, p. 31-36. 134

PAGE 135

REZAEI, M., SANZ, E., RAEISI, E., AYORA, C., VAZQUEZ-SUNE, E., and CARRERA, J., 2005, Reactive transport modeling of calcite dissolution in the fresh-salt water mixing zone: Journal of Hydrology, v. 311, p. 282-298. RITORTO, M., 2007, Impacts of diffuse recharge on tr ansmissivity and water budget calculations in the unconfined karst aquifer of the Santa Fe River basin [unpublished Ms. thesis]: University of Florida, 145 p. RITORTO, M., SCREATON, E.J., MARTIN, J.B., and MOORE, P.J., in press, Magnitudes and chemical effects of diffuse and focused recharge in an eogenetic karst aquifer: An example from the unconfined Floridan aquifer: Hydrogeology Journal. ROMANOV, D., and DREYBRODT, W., 2006, Evolution of porosity in the saltwater-freshwater mixing zone of coastal carbonate aquifers: An alternative modelling approach: Journal of Hydrology, v. 329, p. 661-673. ROMANOV, D., GABROVEK, F., and DREYBRODT, W., 2003a, The impact of hydrochemical boundary conditions on the evolution of limestone karst aquifers: Journal of Hydrology, v. 276, p. 240-253. ROMANOV, D., GABROVEK, F., and DREYBRODT, W., 2003b, Dam sites in soluble rocks: a model of increasing leakage by disso lutional widening of fractures beneath a dam: Engineering Geology, v. 70, p. 17-35. ROSE, S., 1989, The heavy-metal adsorption character istics of Hawthorne Formation (Florida, U.S.A.) sediments: Chemical Geology, v. 74, p. 365-370. RUNNELLS, D.D., 1969, Diagenesis, chemical sediments, and the mixing of natural waters: Journal of Sedimentary Research, v. 39, p. 1188-1201. SCANLON, B.R., and THRAILKILL, J., 1987, Chemical Similarities among Physically Distinct Spring Types in a Karst Terrain: Journal of Hydrology, v. 89, p. 259-279. SCANLON, B.R., 1989, Physical Controls on Hydrochemi cal Variability in the Inner Bluegrass Karst Region of Central Kentucky: Ground Water, v. 27, p. 639-646. SCHLUMBERGER MARKET ANALYSIS, 2007, http://www.slb.com/content/services/ solutions/ reservoir/carbonates.asp accessed on February 20, 2009. SCHWABE, S.J., CAREW, J.L., and HERBERT, R.A., 2008, Making caves in The Bahamas: Different recipes, same ingredients, in Park, L.E., and Freile, D., eds., Proceedings of the 13th Symposium on the Geology of the Baha mas and Other Carbonate Regions: San Salvador, Bahamas, Gerace Re search Centre, p. 153-168. SCOTT, T.M., 1988, The lithostratigraphy of the Hawt horn Group (Miocene of Florida), Florida Geological Survey Bulletin 59, 147 p. 135

PAGE 136

SCREATON, E., MARTIN, J.B., GINN, B., and SMITH, L., 2004, Conduit properties and karstification in the Santa Fe River Sink-Rise System of th e Floridan Aquifer: Ground Water, v. 42, p. 338-346. SEALEY, N.E., 1994, Bahamian landscapes; an introduc tion to the physical geography of the Bahamas: Nassau, Media Publishing, 128 p. SHUSTER, E.T., and WHITE, W.B., 1971, Seasonal fluctuations in the chemistry of limestone springs: A possible means for characterizing car bonate aquifers: Journal of Hydrology, v. 14, p. 93-128. SKLASH, M.G., and FARVOLDEN, R.N., 1979, The role of groundwater in storm runoff: Journal of Hydrology, v. 43, p. 45-65. SMART, P.L., and HOBBS, S.L., 1986, Characterisation of carbona te aquifers: a conceptual base, In: Symposium on the Environmental Problems in Karst Terranes and their Solutions (October, 1986): Bowling Green, Kentucky, National Water Well Association, p. 1-14. SMART, P.L., DAWANS, J.M., and WHITAKER, F., 1988, Carbonate dissolution in a modern mixing zone: Nature, v. 335, p. 811-813. SMART, P.L., BEDDOWS, P.A., COKE, J., DOERR, S., SMITH, S., and WHITAKER, F.F., 2006, Cave development on the Caribbean coast of the Yu catan Peninsula, Quintana Roo, Mexico, in Harmon, R.S., and Wicks, C.M., eds., Pers pectives on Karst Geomorphology, Hydrology, and Geochemistry A Tribute Volume to Derek C. Ford and William B. White: Geological Society of America, Special Paper 404, p. 105-128. SMITH, D.L., and FULLER, W.R., 1977, Terrestrial heat flow values in Florida and the effects of the aquifer system, in Smith, K.L., and Griffen, G.M., ed s., The Geothermal Nature of the Floridan Plateau: Florida Department of Na tural Resources, Bureau of Geology Special Publication 21, p. 91-130. SMITH, D.L., and LORD, K.M., 1997, Tectonic evolution and ge ophysics of the Florida basement, in Randazzo, A.F., and Jones, D.S., eds., The geology of Florida: Gainesville, University Press of Florida, p. 13-26. SPRINKLE, C.L., 1989, Geochemistry of the Floirdan A quifer System in Florida and in Parts of Georgia, South Carolina, a nd Alabama: Professional Paper 1403-I, United States Geological Survey, 105 p. SPROUSE, B.E., 2004, Chemical and isotopic evidence for exchange of water between conduit and matrix in a karst aquifer: An example from the Santa Fe River Sink/Rise system [unpublished M. Sc. thesis]: Un iversity of Florida, 84 p. STETZENBACH, K.J., FARNHAM, I.M., HODGE, V.F., and JOHANNESSON, K.H., 1999, Using multivariate statistical analysis of gr oundwater major cation and trace element concentrations to evaluate groundwater flow in a regional aquifer: Hydrological Processes, v. 13, p. 2655-2673. 136

PAGE 137

STETZENBACH, K.J., HODGE, V.F., GUO, C., FARNHAM, I.M., and JOHANNESSON, K.H., 2001, Geochemical and statistical evidence of d eep carbonate groundwater within overlying volcanic rock aquifers/aquitards of southern Nevada, USA: Journal of Hydrology, v. 243, p. 254-271. STOESSELL, R.K., WARD, W.C., FORD, B.H., and SCHUFFERT, J.D., 1989, Water chemistry and CaCO 3 dissolution in the saline part of an open-flow mixing zone, coastal Yucatan Peninsula, Mexico: Geologi cal Society of America Bulletin, v. 101, p. 159-169. TEETER, J.W., 1995, Holocene saline lake history, San Salvador Island, Bahamas, in Curran, H.A., and White, W.B., eds., Terrestrial and Shallow marine Geology of the Bahamas and Bermuda: Geological Society of Amer ica, Special Paper 300, p. 117-124. TERNAN, J.L., 1972, Comments on the use of a calcium hardness variability index in the study of carbonate aquifers, with referen ce to the central Pennines, En gland: Journal of Hydrology, v. 129, p. 87-108. THRAILKILL, J., 1968, Chemical and hydrologic factors in the excavation of limestone caves: Geological Society of America Bulletin, v. 79, p. 19-45. TORAN, L., HERMAN, E.K., and WHITE, W.B., 2007, Comparison of flowpaths to a well and spring in a karst aquifer: Gr ound Water, v. 45, p. 281-287. TOTH, D., and KATZ, B., 2006, Mixing of shallow and deep groundwater as indicated by the chemistry and age of karstic springs: Hydrogeology Journal, v. 14, p. 1060-1080. TRUESDELL, A.H., and JONES, B.F., 1974, WATEQ, a computer pr ogram for calculating chemical equilibria of natural waters: Journal Resear ch 2, United State Ge ological Survey, p. 233248. UNDERWOOD, M.R., PETERSON, F.L., and VOSS, C., 1992, Groundwater lens dynamics of atoll islands: Water Resources Research, v. 28, p. 2889-2902. UPCHURCH, S.B., 1992, Quality of water in Florida' s aquifer systems, Florida's Ground Water Quality Monitoring Program: Tallahassee, Florida Geological Survey Special Publication 34, p. 12-51. VACHER, H.L., and MYLROIE, J.E., 2002, Eogenetic karst from the perspective of an equivalent porous medium: Carbonates and Evaporates, v. 17, p. 182-196. VACHER, H.L., BENGTSSON, T.O., and PLUMMER, L.N., 1990, Hydrology of meteoric diagenesis; residence time of meteoric ground water in island fresh-water lenses with application to aragonite-calcite stabilization rate in Bermuda : Geological Society of America Bulletin, v. 102, p. 223-232. 137

PAGE 138

VESPER, D.J., and WHITE, W.B., 2004, Storm pulse chemogra phs of saturation index and carbon dioxide pressure: implications for shifting recharge sources during storm events in the karst aquifer at Fort Campbell, Kentucky/Te nnessee, USA: Hydrogeol ogy Journal, v. 12, p. 135-143. VOGEL, P.N., MYLROIE, J.E., and CAREW, J.L., 1990, Limestone petrology and cave morphology on San Salvador Island, Bahamas: Cave Science, v. 17, p. 19-30. WHITAKER, F.F., and SMART, P.L., 1997a, Climatic contro l of hydraulic conductivity of Bahamian limestones: Ground Water, v. 35, p. 859-868. WHITAKER, F.F., and SMART, P.L., 1997b, Hydrogeology and hydrology of the Bahamian archipelago, in Vacher, H.L., and Quinn, T.M., eds., Geology and Hydrogeology of Carbonate Islands, Developments in Sedime ntology 54: Amsterdam, Elsevier, p. 183-216. WHITAKER, F.F., and SMART, P.L., 1997c, Groundwater circulat ion and geochemistry of a karstified bank-marginal fracture system, S outh Andros Island, Bahamas: Journal of Hydrology, v. 197, p. 293-315. WHITAKER, F.F., and SMART, P.L., 2007a, Geochemistry of mete oric diagenesis in carbonate islands of the northern Bahamas; 2, Geochemical modelling and budgeting of diagenesis: Hydrological Processes, v. 21, p. 967-982. WHITAKER, F.F., and SMART, P.L., 2007b, Geochemistry of mete oric diagenesis in carbonate islands of the northern Bahamas; 1, Evidence from field studies: H ydrological Processes, v. 21, p. 949-966. WHITE, W.B., 1977, Role of solution kinetics in the development of karst aquifers, in Tolson, J.S., and Doyle, F.L., eds., Karst Hydrology: Huntsville, UAH Press, p. 503-517. WHITE, W.B., 1988, Geomorphology and Hydrology of Karst Terrains: New York, Oxford University Press, 464 p. WHITE, W.B., 1999, Conceptual models for karstic aquifers, in Palmer, A.N., Palmer, M.V., and Sasowsky, I.D., eds., Karst Modeling: Karst Wa ters Institute Special Publication 5, p. 1116. WHITE, W.B., 2002, Ground water flow in karst: matrix flow and conduit flow with implications for the Floridan Aquifer, in Martin, J.B., Wicks, C.M., and Sasowski, I.D., eds., Hydrologeology and Biology of Post-Paleozoic Carbonate Aquifers: Charles Town, Karst Waters Institute Special Publication 7, p. 9-13. WICKS, C.M., and HERMAN, J.S., 1994, The effect of a confining unit on the geochemical evolution of ground water in th e Upper Floridan aquifer system: Journal of Hydrology, v. 153, p. 139-155. WIGLEY, T.M.L., and PLUMMER, L.N., 1976, Mixing of carbonate waters: Geochimica et Cosomochimica Acta, v. 40, p. 989-995. 138

PAGE 139

WIGLEY, T.M.L., PLUMMER, L.N., and PEARSON JR, F.J., 1978, Mass transfer and carbon isotope evolution in natural water systems: Geoc himica et Cosmochemica Acta, v. 42, p. 11171139. WOLD, S., ESBENSEN, K., and GELADI, P., 1987, Principal component analysis: Chemometrics and Intelligent Laboratory Systems, v. 2, p. 37-52. WOLERY, T.J., 1992, EQ3NR, A computer progr am for geochemical aqueous speciation solubility calculations: Theoretical manua l, users guide, and related documentation (Version 7.0), Lawrence Live rmore National Laboratory WOOD, W.W., and PETRAITIS, M.J., 1984, Origin and Distribu tion of Carbon-Dioxide in the Unsaturated Zone of the Southern High-Plains of Texas: Water Resour ces Research, v. 20, p. 1193-1208. WORTHINGTON, S.R.H., 1994, Flow velocitie s in unconfined carbonate aq uifers: Cave and Karst Science, v. 21, p. 21-22. WORTHINGTON, S.R.H., FORD, D.C., and CAVIES, G.J., 2000, Matrix, fracture and channel components of storage and flow in a Paleozoic limestone aquifer, in Sasowsky, I.D., and Wicks, C.M., eds., Groundwater Flow and cont aminant Transport in Carbonate Aquifers: Rotterdam, A.A. Balkema, p. 113-128. ZIMMERMAN, A.R., MOORE, P.J., and MARTIN, J.B., 2006, Dissolved organic matter-mineralmicrobe interaction in a karst aquifer: The Santa Fe River Sink/R ise system, Floridan Aquifer, Florida: Eos Trans. AGU, v. 87, Fall Meeting Abstract B33A-1146. 139

PAGE 140

BIOGRAPHICAL SKETCH Paul Joseph Moore III (PJ) was born south of Interstate 10 in Pas cagoula, Mississippi on a crisp November morning in 1975. After receivi ng his diploma from Pum pkin Patch Kindergarten in Moss Point, Mississippi, he continued his education at Escatawpa Elementary School in Escatawpa, Mississippi, where he completed years one through six. Followi ng an incendiary year in 1988, PJ transferred from Magnolia Junior High School in Moss Point, Mississippi to Vancleave Middle School in Vancl eave, Mississippi to complete y ear seven. He graduated from Vancleave High School in May 1994. Following graduation, PJ worked as a marine electrician at Ingalls Shipbuilding (now Northup Grumman) in Pascagoula, MS for about four years before continuing his education at Missis sippi State University in 1998. He graduated from Mississippi State University with a Bachelor of Science in geology in May 2002. After an unexpected oneyear hiatus, PJ matriculated into the doctoral pr ogram in the Department of Geological Sciences at the University of Florida under the tutelage of Dr. Jonathan B. Martin. He received his Ph.D. in May 2009 and soon after stated, what a long, strange trip its been. 140