<%BANNER%>

Concentration-Discharge Relationships for Streams and Rivers in Florida

Permanent Link: http://ufdc.ufl.edu/UFE0045545/00001

Material Information

Title: Concentration-Discharge Relationships for Streams and Rivers in Florida Patterns and Controls
Physical Description: 1 online resource (60 p.)
Language: english
Creator: Diamond, Jacob S
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2013

Subjects

Subjects / Keywords: catchment -- concentration -- discharge
Interdisciplinary Ecology -- Dissertations, Academic -- UF
Genre: Interdisciplinary Ecology thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Patterns and drivers of concentration-discharge relationships for 58 streams and rivers in Florida were analyzed using long-term datasets for geogenic solutes, nutrients, and organic solutes as well as land cover and geologic data. All solute concentrations exhibit much less variability than discharge, though each solute-type shows coherent patterning across watersheds. Geogenic solute concentrations tend to become slightly dilute with increasing discharge, nutrients tend to exhibit variable relationships with discharge, and organic solutes tend to become enriched with increasing discharge. Controls on concentration-discharge relationships include land cover, watershed size, and geology. Increasing agricultural cover converges mean concentrations of total inorganic nitrogen (TIN), total phosphorus (TP), and total organic carbon (TOC) to relatively constant values,the molar ratio of which is representative of biotic control. Wetlands reduce export of TIN and TP while increasing export of TOC. Watershed size exerts control on downstream solute fluxes by increasing the variability in source water contribution to streamflow. Geology is shown to have a significant control on geogenic, TP, TIN, and TOC solute concentrations. My results indicate that while concentration-discharge relationships in Florida are a complex function of source-water mixing, antecedent moisture conditions, season, and time-of-day, they can be generally and usefully characterized by a simple set of parameters.
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 Jacob S Diamond.
Thesis: Thesis (M.S.)--University of Florida, 2013.
Local: Adviser: Cohen, Matthew J.
Local: Co-adviser: Jawitz, James W.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2013
System ID: UFE0045545:00001

Permanent Link: http://ufdc.ufl.edu/UFE0045545/00001

Material Information

Title: Concentration-Discharge Relationships for Streams and Rivers in Florida Patterns and Controls
Physical Description: 1 online resource (60 p.)
Language: english
Creator: Diamond, Jacob S
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2013

Subjects

Subjects / Keywords: catchment -- concentration -- discharge
Interdisciplinary Ecology -- Dissertations, Academic -- UF
Genre: Interdisciplinary Ecology thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Patterns and drivers of concentration-discharge relationships for 58 streams and rivers in Florida were analyzed using long-term datasets for geogenic solutes, nutrients, and organic solutes as well as land cover and geologic data. All solute concentrations exhibit much less variability than discharge, though each solute-type shows coherent patterning across watersheds. Geogenic solute concentrations tend to become slightly dilute with increasing discharge, nutrients tend to exhibit variable relationships with discharge, and organic solutes tend to become enriched with increasing discharge. Controls on concentration-discharge relationships include land cover, watershed size, and geology. Increasing agricultural cover converges mean concentrations of total inorganic nitrogen (TIN), total phosphorus (TP), and total organic carbon (TOC) to relatively constant values,the molar ratio of which is representative of biotic control. Wetlands reduce export of TIN and TP while increasing export of TOC. Watershed size exerts control on downstream solute fluxes by increasing the variability in source water contribution to streamflow. Geology is shown to have a significant control on geogenic, TP, TIN, and TOC solute concentrations. My results indicate that while concentration-discharge relationships in Florida are a complex function of source-water mixing, antecedent moisture conditions, season, and time-of-day, they can be generally and usefully characterized by a simple set of parameters.
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 Jacob S Diamond.
Thesis: Thesis (M.S.)--University of Florida, 2013.
Local: Adviser: Cohen, Matthew J.
Local: Co-adviser: Jawitz, James W.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2013
System ID: UFE0045545:00001


This item has the following downloads:


Full Text

PAGE 1

1 CONCENTRATION DISCHARGE RELATIONSHIPS FOR STREAMS AND RIVERS IN FLORIDA: PATTERNS AND CONTROLS By JACOB S. DIAMOND A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUI REMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2013

PAGE 2

2 2013 Jacob S. Diamond

PAGE 3

3 To my family and mentors

PAGE 4

4 ACKNOWLEDGMENTS I would like to express deep gratitude to Dr. Matt Cohen for hi s guidance, critiques, encouragement, and inspiration. Furthermore, I would like to thank Dr. Daniel McLaughlin for his assistance and contributions to my thought process, and Drs. Jawitz and Martin for helping me structure my work and participating on my committee. In no particular order I would also like to acknowledge Ms. Erika Donahue for always being there for me, my parents for raising me as a curious young man, my friends for keeping me humble, my mentors for challenging me, and the giants for giving me shoulders to stand on.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 7 LIST OF FIGURES ................................ ................................ ................................ ......................... 8 LIST OF ABBREVIATIONS ................................ ................................ ................................ .......... 9 ABSTRACT ................................ ................................ ................................ ................................ ... 10 CHAPTER 1 INTRODUCTION AND BACKGROUND ................................ ................................ ........... 12 Patterns in Concentration Discharge Relationships ................................ ............................... 12 Controls on Concentration Discharge Relationships ................................ .............................. 15 2 METHODS ................................ ................................ ................................ ............................. 18 Patterns in Concentration Discharge Relationships ................................ ............................... 18 Controls on Concentration Discharge Relationships ................................ .............................. 20 3 RES ULTS ................................ ................................ ................................ ............................... 23 Patterns in Concentration Discharge Relationships ................................ ............................... 23 Log(C) Log(Q) ................................ ................................ ................................ ................ 23 Effect of Flow Weighted Concentration on C Q Relationships ................................ ...... 2 4 Controls on Concentration Discharge Relationships ................................ .............................. 32 Catchment Size ................................ ................................ ................................ ................ 32 Santa Fe River ................................ ................................ ................................ .......... 32 Suwannee River ................................ ................................ ................................ ........ 33 Withlacoochee River ................................ ................................ ................................ 34 ................................ ................................ ................................ ........ 35 Land Cover ................................ ................................ ................................ ...................... 35 Geology ................................ ................................ ................................ ........................... 42 4 DISCUSSION ................................ ................................ ................................ ......................... 43 Patterns in Concentration Discharge Relatio nships ................................ ............................... 43 Controls on Concentration Discharge Relationships ................................ .............................. 45 Catchment Size ................................ ................................ ................................ ................ 45 Land Cover ................................ ................................ ................................ ...................... 48 Conclusions ................................ ................................ ................................ ............................. 50

PAGE 6

6 APPENDIX ................................ ................................ ................................ ................................ .... 52 LIST OF REFERENCES ................................ ................................ ................................ ............... 57 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ......... 60

PAGE 7

7 LIST OF T ABLES Table page 3 1 Black and white color matrix for CV C /CV Q across site for all solutes .............................. 30 3 2 Geology results ................................ ................................ ................................ ................. 42

PAGE 8

8 LIST OF FIGURES Figure page 3 1 Concentration discharge relationships for selected sites and solut es plotted on logarithmic axes. ................................ ................................ ................................ ............... 26 3 2 Log log slopes of C Q relationships for 58 sites in Florida arranged alphabetica lly by site for select solutes ................................ ................................ ................................ .......... 27 3 3 Ranking of log(C) log(Q) slopes, averaged across sites ................................ .................... 28 3 4 Log(C) log(Q) slope as a function of mean daily flow weighted c oncentration for select solutes ................................ ................................ ................................ ..................... 30 3 5 CV C /CV Q as a function of mean flow weighted concentration (FWC) for select solutes ................................ ................................ ................................ ................................ 31 3 6 Mean daily flow weighted concentration (on log scale) as a function of % agricultural cover for select solutes ................................ ................................ ................... 37 3 7 CV C /CV Q as a function of % agricultural cover for nutrients TIN & TP .......................... 38 3 8 Mean flow weighted concentration as a function of % w etland cover for select solutes ................................ ................................ ................................ ................................ 39 3 9 CV C /CV Q as a function of % urban cover for select solutes ................................ ............. 40 3 10 Log(C) log(Q) slope as a function o f % urban cover for select solutes ........................... 41

PAGE 9

9 LIST OF ABBREVIATIONS C Concentration of a solute (in mg/L unless otherwise specified). C V Coefficient of variation, the standard deviation of a sample population normalized by its mean. F WC Flow weighted concentration. N AWQA Un assessment program. N WIS USGS national water information system. Q Discharge of a stream or river. S JRWMD S RWMD Suwannee River Water Management Distri ct. S TORET S WFWMD South West Florida Water Management District. T IN Total inorganic nitrogen. Sum of nitrate, nitrite, ammonium, and ammonia as nitrogen in a water sample. T KN Total Kj eldahl nitrogen. Sum of organic nitrogen, ammonia, and ammonium. T OC Total organic carbon. Sum of material derived from decaying vegetation, bacterial growth, and metabolic activities of living organisms or chemicals. T P Total phosphorus. Sum of all parti culate and dissolved forms of phosphorus in a water sample.

PAGE 10

10 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science CONCENTRATIO N DISCHARGE RELATIONSHIPS FOR STREAMS AND RIVERS IN FLORIDA: PATTERNS AND CONTROLS By Jacob S. Diamond May 2013 Chair: Matthew Cohen Cochair: James Jawitz Major: Interdisciplinary Ecology Patterns and drivers of concentration discharge relationships for 58 streams and rivers in Florida were analyzed using long term datasets for geogenic solutes, nutrients, and organic solutes as well as land cover and geologic data. All solute concentrations exhibit much less variability than discharge, though each solut e type shows coherent patterning across watersheds. Geogenic solute concentrations tend to become slightly dilute with increasing discharge, nutrients tend to exhibit variable relationships with discharge, and organic solutes tend to become enriched with i ncreasing discharge. Controls on concentration discharge relationships include land cover, watershed size, and geology. Increasing agricultural cover converges mean concentrations of total inorganic nitrogen (TIN), total phosphorus (TP), and total organic carbon (TOC) to relatively constant values, the molar ratio of which is representative of biotic control. Wetlands reduce export of TIN and TP while increasing export of TOC. Watershed size exerts control on downstream solute fluxes by increasing the varia bility in source water contribution to streamflow. Geology is shown to have a significant control on geogenic, TP, TIN, and TOC solute concentrations. My results indicate that while concentration discharge relationships in Florida are a complex function of source

PAGE 11

11 water mixing, antecedent moisture conditions, season, and time of day, they can be generally and usefully characterized by a simple set of parameters.

PAGE 12

12 CHAPTER 1 INTRODUCTION AND BACKGROUND Streamflow and its associated chemical fluxes are the fi ltered integration of hydrologic and biogeochemical processes in a watershed. As such, the study of hydrochemical fluxes helps resolve questions regarding watershed scale structure and function (Feng et al. 2004; Kirchner et al. 2004). In addition to helpi ng reveal how earth surface processes work, an understanding of streamflow and solute dynamics in catchments is vital for water management decisions in which accurate predictions of chemical fluxes as well as controls on and ecosystem responses to those fl uxes are needed (e.g. nutrient flux into the Gulf of Mexico from the Mississippi River Basin). Unfortunately, the time scales at which watershed scale processes act can span decades or longer, creating a need for long term catchment monitoring programs tha t regularly collect both hydrologic and chemical data (Likens 2004 and references within). These monitoring programs are essential for determining ecosystem water and solute budgets, chemical weathering rates, and biologic uptake rates. In this study, long term monitoring data of solute and water fluxes are used to test and calibrate theoretical predictions of watershed processes. Patterns in Concentration Discharge Relationships There is a long legacy of efforts to understand what concentration discharge r elationships tell us about watershed form and function (e.g. Johnson et al. 1969; Walling and Webb 1986; Evans and Davies 1998; Godsey et al. 2009; Murphy et al. 2012). Early results from pristine catchments like the Hubbard Brook Experimental Forest revea led that concentrations of some geogenic solutes (e.g. Na + Si) in streamflow tend to decrease with increasing discharge, implying a dilution of solutes derived from weathering (Johnson et al. 1969; Clow and Drever 1996). However, some solutes become more concentrated with increasing discharge (e.g. Al 3+ NO 3 ) perhaps due to increasing source area contribution and other solutes (e.g. Mg 2+ Ca 2+

PAGE 13

13 Cl ) exhibit a buffered response (i.e., little concentration variation with flow; Johnson et al. 1969). Explanati ons for these stark differences among solutes were attributed to both watershed scale phenomena such as biologic uptake, and point scale reaction rate equilibria between soil water and weathered bedrock. Despite these generalities relating chemical weather ing, biotic action, and streamflow concentrations, there have been few studies that have satisfactorily linked these components (notable examples are Johnson et al. 1969 and Anderson et al. 2002). The coupling of solute fluxes to watershed flow generation and their relation to biota remains a core question in both hydrology and geochemistry (McGuire and McDonnell 2010; McDonnell et al. 2010). Many studies of concentration discharge tend to focus on fitting mixing models of different source waters to event s cale (i.e. short term) data (Evans and Davies 1998; Hornberger et al. 2001). Concentration discharge graphs from storm events frequently exhibit distinctive looped patterns, or hysteretic behaviors, that are dependent on both solute chemical behavior and a ntecedent moisture conditions (McGuire and McDonnell 2010). Evans and Davies (1998) showed that chemograph hysteresis could be theoretically represented by a 3 component mixing model of groundwater, surface event water, and soil water. Murphy et al. (2012) determined that a simple two component mixing model (direct runoff and baseflow) could adequately describe most observed hysteresis patterns implying that, at least for their study catchment, there are two main sources of water that contribute to streamfl ow. Further, these two sources evolve at different time scales: baseflow at monthly and inter annual scales, and shallow subsurface flow at daily and event time scales. While baseflow chemistry most likely represents deep soil equilibrium conditions, the c hemical signal during high flow conditions is a complex mixture of

PAGE 14

14 many different source waters. Hence, the degree of hydrochemical variation in a river is itself a metric indicative of source water mixing. While most of the previous work on concentration discharge relationships in streams has been performed on the event scale, recent studies that draw on long term data collections have started to highlight the importance of inter annual patterns in solute transport. Despite earlier findings of dilution of major geogenic solutes with increasing discharge, there appears to be pervasive chemostasis defined as small variation in concentration despite large variation in discharge in catchments of varying magnitude (10 1 10 6 km 2 ) for both geogenic and anthrop ogenic (i.e. those under significant human control) solutes at annual time scales (e.g. Godsey et al. 2009; Basu et al. 2010; Basu et al. 2011a; Guan et al. 2011). Chemostatic behavior appears to be a result of chemical mass legacies that result in flow li mited systems (Basu et al. 2010; Godsey et al. 2010; Murphy et al. 2012). In other words, catchments with a large source mass of solute buffer natural hydrologic variations to yield concentrations that are effectively constant with respect to flow. The imm ediate consequence of catchment chemostasis is that the power function that commonly characterizes the relationship between flow and concentration, C=aQ b collapses (b=0, C=a), and load becomes directly proportional to flow (L=aQ). In this situation, load variation can be predicted by flow variation alone. This linear relationship between flow and load has important implications for predicting fluxes of chemical constituents to receiving water bodies due to variable environmental conditions (Godsey et al. 2 009; Jawitz and Mitchell 2011). For chemostasis to occur at inter annual time scales, a large storage is required in the flow generating zone of the subsurface. This raises the questions of where exactly storage occurs for specific solutes, and what the pr ocesses and reactions are that lead to their fate and transport.

PAGE 15

15 One possibility is that the storage zone is a well mixed reservoir with respect to solutes and ages solute concentrations. However, this paradigm conflicts with observations of time variance of event chemistry (Botter et al. 2010; McDonnell et al. 2010). I ndeed, Godsey et al. (2009), using a linear well mixed model, found that the storage zone required to produce observed concentration discharge relationships would be many times larger than is physically possible. On the contrary, evidence suggests that the storage zone is not a well mixed reservoir, but rather is a complex, non linear filter for stochastic rainfall inputs, which generally creates power law (or gamma) distributions of residence times (as opposed to short tailed exponential distributions gene rated by well mixed models; Kirchner et al. 2004; Zhang and Schilling 2005). Furthermore, since the travel time distribution of water through the storage zone will significantly affect its chemical composition, one would expect a similar power law distribu tion in the chemistry of the stream. This has been evidenced by work on fractal stream chemistry, where streamwater concentrations of chloride exhibit 1/f fractal scaling over variable time scales (Kirchner et al. 2000; Zhang and Schilling 2005). Since the magnitude of a storm event impacts discharge analyses can reveal information about what controls solute delivery, water storage, and flow generation. Controls on Concentra tion Discharge Relationships Solute flux from a watershed is a result of many interacting processes including equilibrium reactions with bedrock and the subsurface, formation of secondary minerals, biologic action, and availability of solute mass. On one e nd of the spectrum, solute flux is

PAGE 16

16 controlled by flow generation and on the other, by equilibrium processes. If simple dilution (i.e. constant flux of solute despite variable flux of water) controlled streamflow concentrations, the expected behavior would be a marked decline in solute concentrations with increasing flow. In chemostatic systems, in contrast, solute production and/or mobilization must be nearly proportional to water fluxes, both on storm and inter annual timescales (Godsey et al. 2009). Likew ise, in systems where solutes are enriched with flow, event water mobilizes solutes not available during lower flow periods; this may arise by connecting new flow generating locations to the stream network or by enabling solute generating processes. It is therefore important to note the specific mechanisms that govern chemostatic behavior. Chemostasis could arise from very long water residence times with respect to solute equilibrium times. However, Godsey et al. (2009) point out that, at least for some geo genic solutes (e.g. Si), equilibrium reaction rates are very slow, even compared to water transit times of a year, implying that streamwater is always far from equilibrium with respect to weathered material. In other words, the downstream delivery of silic a from the subsurface is a reaction limited process, suggesting that dilution would be a dominant signal. Since Si dilution with increasing flows is not observed, the mechanism controlling chemostasis in this case is likely due to a ubiquitous source mass which is homogeneously distributed throughout the subsurface. Chemostasis observed for more reactive solutes (e.g. Mg 2+ Ca 2+ Na + ), whose flux is transport limited, may instead be a function of underlying geology and its connection with the water table. Mass accumulation in the subsurface and biogeochemical cycling also can potentially contribute to observed chemostatic behavior. Indeed, a long term study of intensively managed catchments concluded that nutrient (total phosphorus and total nitrogen) chemostasi

PAGE 17

17 between 1950 and 2005, the diffuse TP and TN inputs to the soil surface increased by 185%, but only 1% and 16% of TP and TN mass inputs at the soil surface reached the nearest surface water body within the 5 years, a response reflective of biogeochemical retention within the basin. Clearly, land surface processes are strongly coupled to solute and flow dynamics in watersheds. In this work, I build upon these studies by comparing long term concentration discharge relationships for several solutes across an array of variably sized streams and rivers in Florida and a nalyze the potential driving forces of these relationships. Specifically, I seek to address two main questions. First, what are the concentration discharge (C Q) relationships for different solutes in Florida streams and rivers and how do they vary across sites and solutes? Second, what controls variation in these solute discharge relationships? For the latter, I focus on watershed size, land cover, and geologic controls. I hypothesize that solutes will exhibit coherent concentration discharge relationships across watersheds which will vary according to solute characteristics. In particular I predict that geogenic solutes will exhibit chemostasis, nutrients will exhibit large C Q variability, and organic solutes will become enriched with increasing flows. Fu rthermore, I predict that agricultural cover will exert control on C Q relationships for nitrogen and phosphorus by increasing total concentrations of each and by reducing relative variation in concentrations. Likewise, I predict that increasing wetland co ver will control export of total organic carbon by increasing downstream concentrations and decreasing relative variability in that export. I also expect that geogenic solutes will be similarly controlled by increasing presence of characteristic geologic f ormations. More generally, I predict that increasing solute supply be it through increasing watershed area, geologic supply, or terrestrial loading will increase downstream solute concentrations and decrease relative solute variability.

PAGE 18

18 CHAPTER 2 METH ODS Patterns in Concentration Discharge Relationships My first objective was to establish concentration discharge relationships for nutrients (N and P), geogenic solutes (Si, Ca, Mg, Na, Cl), and total organic carbon for an array of variably sized streams and rivers located throughout Florida, USA. I focused on three water management districts (WMDs) in Florida of similar topographic relief and underlying geology in order to simplify inter s River (SJRWMD), and Southwest Florida (SWFWMD) were chosen for analysis. Most solute concentration data were obtained from the STORET ( Stor age and Ret rieval) database provided by the Florida Department of Environmental Protection (FDEP), which archives p hysical, data were obtained from the USGS National Water Quality Assessment Program (NAWQA). It should be noted that these water quality data are coarse resoluti on (monthly in most cases, but sometimes as infrequently as quarterly). Furthermore, sampling time of day varies greatly, potentially complicating analysis of solutes that exhibit diel signals (Heffernan and Cohen, 2010; DeMontety et al. 2011). Solute conc entrations were not corrected by precipitation chemistry since concentrations for all solutes are generally orders of magnitude higher in streamflow (Godsey et al. 2009). Solute concentration measurements were not indexed by their position on the hydrograp h or by their season and antecedent moisture conditions. Sites were chosen according to the following criteria: (1) on a stream or river, and (2) at least 5 continuous years of concentration data records. These reduced the number of study sites from ~20,00 0 in all of STORET to ~200, most of which were at least second order rivers. The period of record ranged from 5 to 31 years (median = 12 years). Discharge data, which are

PAGE 19

19 unavailable in STORET, were obtained from the USGS National Water Information System (NWIS). NWIS was queried for all daily flow records for streams and rivers within the study domain. Average annual runoff from selected sites varied from 2.8 to 75.6 cm/yr (median = 26.3 cm/yr). To consolidate concentration and flow records, I developed a matching method based on site GPS coordinates. Stations were paired if the scalar distance between them was less than or equal to 100 meters, stations without a match were discarded. This process reduced the number of sites to ~100. Flow data from NWIS sta tions were then matched in time to water quality data from STORET. I selected the following solutes based on data availability and management relevance: total inorganic nitogen (TIN), total Kjeldahl nitrogen (TKN), total phosphorus (TP), phosphate (PO4), s pecific conductance (SpC), silica (Si), sodium (Na), chloride (Cl), calcium (Ca), magnesium (Mg), turbidity, and total organic carbon (TOC). At all sites there were at least a few sampling dates where streamflow chemistry was missing for solutes of interes t and some sites may not have had any streamflow chemistry at all for certain solutes. Final site selection eliminated sites for which fewer than 50 dates with combined flow and chemistry data were available. The final number of sites used for analysis (n= 58) included 16 from SRWMD, 22 from SJRWMD, and 20 from SWFWMD. Finally, water quality data with the following comments were eliminated from the database: off scale low (K), off scale high (L), sample held beyond accepted holding time (Q), value reported l ess than laboratory detection limit (T), value reported above detection limit (V), sample was improperly preserved (Y), and other comments that deemed the data point either inaccurate or unusable. Data losses from this screening were substantial (up to 30% of the data at some sites). I additionally compared flow distributions on dates with water quality samples with the longer term flow regime to ensure no sampling biases.

PAGE 20

20 For almost every site, flows sampled by this study were representative of the overall flow regime. Stations that preferentially sampled base flows with water quality measurements were discarded. I plotted solute concentrations against average daily. I fit a power function to each relationship and extracted the power function exponents. I a lso extracted the coefficients of determination (R 2 ), residual plots, and 95% confidence intervals. Residual plots and R 2 values indicated that 15% of the fitted power functions were poor representations of observed concentration discharge relationships; h owever relationships for all sites and solutes are reported. The most frequently sampled constituent was specific conductance. Dissolved oxygen (DO) and temperature were also frequently sampled, but were not included in this analysis because they are highl y seasonal, vary significantly over the course of a day, and generally independent of other solute measurements. I then calculated the coefficient of variation and flow weighted concentration for each solute for all 58 sites. Normally, the coefficient of variation (CV) is calculated by normalizing the standard deviation of a dataset by its mean. Because I used power functions to describe these hydrochemical data, I used the following equation was used to calculate the CV: Controls on Concentration Discha rge Relationships This synthesis also examined catchment controls on observed concentration discharge relationships. Contributing watersheds were initially characterized by their hydrologic unit code (HUC), a standardized watershed classification system de veloped by USGS. Depending on the location of a station, different HUC levels were chosen to represent the contributing watershed area; some sites had contributing areas that included multiple HUCs. Relevant information

PAGE 21

21 included total watershed area, lengt h of river/stream, and location of the sampling station with respect to the outlet of the watershed. Roughly half of the sites were best described by HUC 12s (10 1 10 2 km 2 ) and the other half were represented HUC 10s (10 2 10 3 km 2 ). Some sites were described by an even smaller watershed delineation termed a WBID (water body ID) that ranges on the order of 10 0 km 2 In all, drainage areas ranged from 17.4 to 24,320 km 2 (median = 385 km 2 ). Land use data was derived from FLUCCS codes (Florida Land Use, Cover and Forms Classification System), which are publicly available through the Florida Department of Transportation. Using ArcGIS, FLUCCS codes were intersected with Hydrologic Unit Codes (HUCs) to get land cover data for each watershed. Land covers of interest i ncluded wetlands, urban, and agricultural lands and these were quantified with a percent coverage of the total watershed area. It is important to note that land cover is a dynamic variable and the values used in this study reflect a one time measurement of land cover performed in either 2008 or 2010 depending on the site Generally, HUCs were comprised of 10 50% forests and wetlands, with a select few having upwards of 40 50% agricultural cover (mainly in southwest Florida in the Myakka River basin). Strati graphy/geology maps for the state of Florida were obtained from the Florida Department of Environmental Protection website and were intersected with HUC codes to get at the underlying geology contributing to each site. Formations of interest included the H awthorn Group, Ocala limestone, and major rock types. Ocala limestone, from the upper Eocene, consists of nearly pure limestones and occasional dolostones with many foraminifera fossils throughout. It is at or near the surface within west central and north central Florida and in these places it exhibits extensive karstification. This karst network plays a significant role in the hydrology of

PAGE 22

22 the area due to its high permeability and transmissivity. The Hawthorn Group varies in thickness and composition thro ughout the north, panhandle, and central parts of Florida. Generally it consists of low permeability siliciclastic clays that can contain large amounts of phosphatic material, though it is often 100 feet below land surface. Geology for a site was entered a s a binary value of either a formation being present (1) or not present (0) in the contributing watershed. A geological formation was present if it covered at least a continuous 20% of the drainage area. Site characteristics are compiled and described in t he appendix.

PAGE 23

23 CHAPTER 3 RESULTS Patterns in Concentration Discharge Relationships Log(C) Log(Q) Log log concentration discharge plots suggest characteristic behavior for solutes across the study catchments. Most geogenic solutes (i.e. Ca, Mg, Na, Si) exhi bit a very shallow decrease in concentration with increasing discharge, but silica shows no variation in concentration with changing discharge. Nutrient concentrations exhibited more variable relationships with discharge, but total phosphorus concentration s varied 65% less than nitrogen concentrations. In general, TOC and TKN concentrations increased with increasing discharge. A sample of these plots for select sites ( Figure 3 1 ) highlights two sites from each water management district and compares nutrient s, base cations, and total organic carbon within each site. The SWFWMD data was incomplete or insufficient for Mg, Na, and Ca so for these sites (Figure 3 1, e f) Cl and SpC serve as proxies. The best fit log(C) log(Q) slopes for solutes tend to be sim i lar across watersheds (Figure 3 2 ). While there are fewer data points for silica than for other solutes because of site screening criteria, it clearly exhibits chemostatic behavior across all sites (Fig. 3 2a). Other geogenic solutes, (Ca, Mg, and Na; the latter are not shown) tend to exhibit dilution behavior with increasing discharge as evidenced by the persistent, shallow negative log log slopes across all sites (Figure 3 2b). Similar to Si, total phosphorus (TP) averages a log log slope of 0 across all sites (Figure 3 2d), though variation between sites is clearly greater. In contrast, total inorganic nitrogen (TIN) has both a much more variable signal in its concentration discharge relationship among sites (Figure 3 2c), and evidence of predominantly di lution behavior. Total organic carbon (TOC) appears to have a consistent enrichment effect with increasing discharge, though

PAGE 24

24 log log slopes tend to be relatively shallow (<= 0.2; Figure 3 2e) other solutes that exhibit similar consistent enrichment behavio r across sites are TKN and turbidity. Average solute discharge slopes across sites illustrate the marked variation across solutes from dilution through chemostasis to enrichment dynamics (Figure 3 3 ). While there is some deviation from average behavior, general patterns of solute behaviors emerge. Geogenic solutes tend to exhibit dilution or chemostatic effects with increasing discharge, nutrients are far more variable across catchments (though TP is generally chemostatic), and organic solutes (TOC, TKN, turbidity) are enriched with increasing discharge. Notably, magnesium and calcium have the same distribution of log(C) log(Q) slopes across catchments. Also worth noting is the large difference between TKN (0.21) and TIN ( 0.13) log(C) log(Q) slopes sugges ting organic N is strongly enriched with flow. Contrasting solute and discharge variation (using CV C /CV Q ) across sites reveals that all solutes generally vary far less than discharge (CV C /CV Q between 10 and 50%; Table 3 1 ), though there was considerable v ariation across sites and solutes. Reported CV ranges indicate that despite the fact that concentration discharge relationships can be characterized as enriching, dilute, or chemostatic most catchments exhibit much less variability in concentration (CV C = 0.08 4.5) compared to variability in flow (CV Q = 0.4 631) for all solutes. Despite similar behaviors across solutes the ranking parallels what was observed for log(C) log(Q) slopes, where geogenic solutes tend to exhibit far less variation with respect to discharge compa red to organic solutes (Table 3 1 ). Effect of Flow Weighted Concentration on C Q Relationships The absolute magnitude of solute concentration evidently has no influence on log log slopes for geogenic and organic solutes like calcium, silic a, and TOC, but may have a marginal influence on the slopes for total phosphorus (Figure 3 4 ). As mean concentration increases, total

PAGE 25

25 phosphorus converges to chemostatic concentration discharge relationships. This result is reinforced by the observation th at TP asymptotically approaches low CV C /CV Q values as mean concentration increases (R 2 =0.14, p = 0.07; Figure 3 5 ). TOC exhibits a similar asymptotic approach to chemostasis (R 2 =0.16, p = 0.0008). TIN, Mg, and Ca (not shown) have an opposite response, with significant linear increases in CV C /CV Q with increasing mean concentration (R 2 = 0.55, 0.38, and p = <0.0001, 0.0003, respectively).

PAGE 26

26 Figure 3 1 Concentration discharge relationships for selected sites and solutes plotted on logarithmic axes. By row, s ites are from Suwannee River Water Management District (a d), and South West Florida Water Management District (e f) (top to bottom). Notice general small variations in concentration despite large variation s in discharge.

PAGE 27

27 Figure 3 2. Log log slopes of C Q relationships for 58 sites in Florida arranged alphabetically by site for select solutes. Geogenic solutes (a b), nutrients (c d), and carbon (e) each exhibit unique behavior, though for each sol ute category behavior is similar across sites. Sites with tidal influence or point source discharge are omitted.

PAGE 28

28 Figure 3 3. Ranking of log(C) log(Q) slopes, averaged across sites, error bars are 95% confidence intervals. Geogenic solutes have the most n egative slopes, organic solutes the most positive, and nutrients the most variable.

PAGE 29

29 Table 3 1. Black and white color matrix for CV C /CV Q across site for all solutes. Sites are ranked from top to bottom by lowest to highest CV C /CV Q averaged across all so lutes. Solutes are ranked left to right by lowest to highest CV C /CV Q values, averaged across all sites. Values are color ranked from low high by black to white. Empty spaces indicate lack of sufficient data. Sites with significant influence from tides and point source discharge are not included.

PAGE 30

30 Figure 3 4 Log(C) log(Q) slope as a function of mean daily flow weighted concentration for select solutes. There is little effect of increasing mean concentration on the shape of the C Q relationship for mo st solutes, but TP converges on chemostasis with increasing concentration. The two fitted lines on the TP graph are for negative and positive b values converging to 0.

PAGE 31

31 Figure 3 5 CV C /CV Q as a function of mean flow weighted concentration (F WC) for select solutes. Best fit lines, R 2 and 95% confidence intervals (dashed lines surrounding solid lines) are also provided. TOC and TP both exhibit similar declines in CV C /CV Q with increased FWC, but Mg and TIN increase their CV C /CV Q with increasing concentrations. P values for coefficients in clockwise order, starting from top left: 0.0008, <0.0001, 0.0003, and 0.0747.

PAGE 32

32 Controls on Concentration Discharge Relationships Catchment Size This dataset enables investigation of changes in streamflow hydroc hemical fluxes associated with increasing drainage area and varying source inputs. Results from four major rivers illustrate the effects of increasing drainage area on concentration discharge relationships. Santa Fe River The first station (near Graham: 0 2320700) sampled along the river drains a large swamp and lake system and has a base flow on the order of 0.85 m 3 s 1 (drainage area = 246 km 2 ). The second station (Worthington Springs: 02321500) is after the confluence of New River with the Santa Fe and h as base flow on the order of 8.5 m 3 s 1 (drainage area = 1,490 km 2 ), the following station (USHWY441: 02321975) is located a few miles after the river routes underground through a three mile karst conduit (drainage area = 2,225 km 2 ), and the final station (near Hildreth: 02322800) is located just before its confluence with the Suwannee River, after receiving input from the Ichetucknee River and many artesian springs whose total discharge is on the order of 17 m 3 s 1 of aquifer water to the Santa Fe (drainag e area = 3,559 km 2 ). From the first station to the last, calcium and magnesium average flow weighted concentrations increase by 12.5 and 3.2 times, respectively (increase in drainage area by an order of magnitude). Nitrate followed a similar exponential in crease in concentration with distance along the river, increasing its concentration an order of magnitude from ~0.05 mg/L to ~0.5 mg/L. Interestingly, phosphorus flow weighted concentrations exhibit an inverse parabolic behavior, increasing to a peak after the confluence of New River (~0.2 mg/L) before dropping back down (~0.1 mg/L) near the confluence with the Suwannee. TOC concentrations decrease linearly by a factor of 5 (~50 mg/L to 10 mg/L) from the beginning to the end of the river run. Variation in c oncentration with respect to variation in discharge increased for every measured solute with downstream

PAGE 33

33 distance, with every solute exhibiting low CV C /CV Q (<0.1) at the first station and 1 2 orders of magnitude larger CV C /CV Q at the final station. The most dramatic changes in CV C /CV Q appear at the last station, after the confluence with the Ichetucknee, with the largest changes occurring with TOC, TP, and TIN. This is related to changes in log(C) log(Q) slopes for TIN and TP which change sign with increasin g distance downstream. Total inorganic nitrogen initially has a slight positive slope (0.2), which becomes very slightly negative ( 0.05) for the middle reaches and then steeply negative after the Ichetucknee ( 0.35). Phosphorus exhibits nearly the opposit e behavior, moving from a slight negative slope ( 0.2) to near 0 slope in the middle reaches and finally a very large positive slope at the downstream station (0.73). Suwannee River Despite the order of magnitude difference in drainage areas, very similar behavior can be observed with increasing drainage are and distance downstream for the Suwannee River. Stations sampled along the Suwannee draw on the many different source water contributions to the river: the first site (White Springs: 02315500; drainage area = 6,294 km 2 ) is located ~115 km below the drain of the Okefenokee Swamp (a 1,770 km 2 peat accreting wetland that makes up the headwaters of the Suwannee), the second site (Dowling Park: 02319000) is located after the major confluences of the Alapaha and Withlacoochee Rivers with the Suwannee (drainage area = 18,622 km 2 ), a third site (Branford: 02320500) is located further along the river with several contributions from first (>2.8 m 3 s 1 ) and second magnitude (0.28 2.8 m 3 s 1 ) springs (drainage area = 20,409 km 2 ), and a final site (Bell: 02323000) located after the confluence with the Santa Fe River (drainage area =24,430 km 2 ). Both Ca and Mg exhibit massive concentration increases (15 and 5 fold, respectively), with a drainage area increase of 18,0 00 km 2 (385%), almost identical to the increases seen for the Santa Fe. Even more remarkable are the similarities observed for increases in TIN concentrations (0.01 0.47 mg/L) and the peak observed in TP

PAGE 34

34 concentrations (0.12 up to 0.17 back down to 0.15 mg /L). TOC flow weighted concentrations were reduced by a factor of 3 (54 20 mg/L), very close to the factor of 5 decrease observed for the Santa Fe system. Patterns in log(C) log(Q) slopes for total phosphorus and total inorganic nitrogen again show similar ity to the Santa Fe catchment, where TIN becomes more negative ( 0.16 0.61) with increasing drainage area and TP changes from a negative ( 0.18) slope to a slightly positive (0.06) slope, though the absolute values of the slope differ. TOC slope increas es monotonically from 0.08 at the swamp drain to 0.93 after the confluence with the Santa Fe. Ca, Mg, and Si all tend towards more negative log(C) log(Q) slopes with increasing drainage area. In line with results from the Santa Fe, CV C /CV Q values for all s olutes increase by 1 2 orders of magnitude with increasing drainage area, and tend to move towards a value of 1, except for phosphorus which hovers at a value of ~0.4 at the final 3 stations. Withlacoochee River Four sites along a forested section of the Withlacoochee River (note: not the tributary of the Suwannee) in south central Florida appear to be characterized by similar concentration discharge trends. The first site is located near Pineola, FL (02312598) right after the river flows through Nelson La ke (drainage area = 2,408 km 2 ); the second site (Floral City: 023126000) is just 3 miles downstream after the river flows through Bonnet Lake (drainage area = 2,577 km 2 ); the third site is located after the river combines with the drain from Lake Panasoffk ee (02312720; drainage area = 3,781 km 2 ); and the final site (Rutland 02312722) is four miles downstream (drainage area = 3,937 km 2 ). Total inorganic nitrogen concentrations increase from 0.03 0.05 mg/L, total phosphorus concentrations peak at 0.10 mg/L at the second site and decrease to 0.06 mg/L at the last site, and total organic carbon concentrations decrease from 29 19 mg/L. Coefficients of variation increased by close to an order of magnitude for all solutes along the river reach. Log(C) log(Q) slopes monotonically increased for total phosphorus and

PAGE 35

35 total organic carbon (0.04 0.27 and 0.04 0.4, respectively), whereas the slope for total inorganic nitrogen decreased from slightly positive to near zero (0.17 0.01). Sites sampled along t drainage area = 5,292 km 2 ) and after (Osceola: 02234010; drainage area = 5,763 km 2 ) Lake Harney, a 24.5 km 2 lake to the northeast of the Orlando metropolitan area with a large number of contri buting salt water springs; a site after the river drains Lake Jesup, but before it enters Lake Monroe (Sanford: 02234440; drainage area = 6,734 km 2 ); and a final site right before the river enters Lake George (Astor: 02236125; drainage area = 8,624 km 2 ). T his watershed bears some similarity to the results seen from the Santa Fe and the Suwannee in that its total inorganic nitrogen flow weighted concentrations slightly increase from 0.05 0.09 mg/L over the river length analyzed (compared to 0.05 0.5 mg/L). H owever, the increase in drainage area (1.7 times) sampled by this study is more similar to that of the Withlacoochee (1.6 times), than to the Santa Fe (14.5 times) or Suwannee (3.9 times). Strangely enough, phosphorus concentrations exhibits a similar shap e to that seen by the other rivers, peaking at the central site before dropping off again (range = 0.07 0.09 mg/L). Calcium and magnesium increase by roughly two times between the first and last site. However, unlike previous results, CV C /CV Q all solutes at the St. previous results, log(C) log(Q) slopes for nitrogen tend to increase, instead of decrease, with increasing drainage area, but phosphorus shows a similar increasing slope trend. Land Cover The influence of agriculture on stream solute concentrations, revealed marked convergence of TIN, TP, and TOC to steady values with increasing agricultural cover (ca. 0.3 mg P/L, 0.6 mg N/L, and 23 mg C/ L; Figure 3 6 ). In contrast, base cations like magnesium and

PAGE 36

36 calcium (not shown) increase in concentration with increasing agricultural cover. Contrary to predictions, the CV C /CV Q for TIN increases with increas ing agricultural cover (Figure 3 7 ), though a less clear negative relationship was observed TP (p>0.5). Similar solute behavior was evident in respose to variation in percent wetland and percent urban cover. Wetland cover exerted clear control on flow weighted concentrations of nutrient s, carbon, and calcium (Figure 3 8 ), with increases in wetland cover inversely correlated with TIN, TP, and Ca export, but positively correlated with TOC and TKN. Wetland cover had no observable effect on CV C /CV Q for any of the solutes considered, but did affect the log (C) log(Q) slope for geogenic solutes. The log(C) log(Q) slope for both calcium and magnesium decreased nearly linearly with increasing wetland cover, going from ~ 0.05 at 10% cover to ~ 0.55 at 85% cover for both solutes. Urban cover had no clear infl uence on solute CV C /CV Q (Figure 3 9 ). The CV C /CV Q for TOC and calcium appears to declines with increasing urban cover, though the relationship is very weak (R 2 < 0.1). The CV C /CV Q for silica has a similarly weak, but positive relationship with urban cover and phosphorus has not observable trend with increasing urban cover. However, urban cover influences the log(C) log(Q) slopes for many solutes by forcing values close to 0 (Figure 3 10 ), particularly for geogenic solutes (sodium, chloride, magnesium, and c alcium). Urban cover did not impact solute flow weighted concentrations, except in the case of organic solutes (TKN and TOC), where increases were associated with an exponential decline from ~110 mg/L at 5% cover to ~10 mg/L at 80% cover for TOC. The flow weighted concentration for magnesium showed a similar exponential decline with urban cover, but exhibited more variability at lower values. Flow weighted Ca concentrations also converged as urban cover increased, though the relationship is weaker than for other solutes.

PAGE 37

37 Figure 3 6. Mean daily flow weighted concentration (on log scale) as a function of % agricultural cover for select solutes. TIN, TP, and TOC converge on stable concentrations with increasing agricultural cover, though the pat tern for TIN is distinct from TOC and TP. Magnesium generally increases in concentration with increasing cover.

PAGE 38

38 Figure 3 7. CV C /CV Q as a function of % agricultural cover for nutrients TIN & TP. There is an appearance of a general trend of increasing C V C /CV Q with increasing agricultural cover for TIN (p=0.02). CV C /CV Q for TP appears to decrease with increasing agricultural cover, but attempts at a best fit line yield p values of greater than 0.5.

PAGE 39

39 Figure 3 8. Mean flow weighted concentr ation as a function of % wetland cover for select solutes. Nutrients tend to follow an exponential decline in concentration with increasing wetland cover, whereas TOC tends to increase according to a power law function. Magnesium appears to experience redu ced concentrations with more wetland cover, but best fit lines are not significant. P values clockwise from upper left: 0.0001, <0.0001, 0.2, and <0.0001.

PAGE 40

40 Figure 3 9. CV C /CV Q as a function of % urban cover for select solutes. There does not appear to be a generalizable behavior for silica or phosphorus, though for TOC and Ca there is a general decreasing trend in CV C /CV Q with increasing urban cover. No best fit lines were significant for p<0.05. P values clockwise from upper left: 0.5, 0.18, 0.8, and 0.3.

PAGE 41

41 Figure 3 10. Log(C) log(Q) slope as a function of % urban cover for select solutes. Increasing urban cover drives solutes towards chemostasis (slope = 0). Associations for TP and TOC were not statistically significant. However, we observe convergence on chemostasis with increasing urban cover.

PAGE 42

42 Geology The Hawthorn Group and limestone formations were found in 36% and 29% of the sites, respectively, with 9% of the sites containing both geologic formations. By far, the most commo n lithology in the study catchments was sand, which was in every catchment except for the Suwannee and Withlacoochee River catchments, where limestone and dolostone are ubiquitous. A comparison of mean flow weighted concentrations and CV C /CV Q with Hawthor n Group presence/absence revealed significant differences (Table 3 2 ). Presence of the Hawthorn group exerted a significant (p<0.05) negative effect on CV C /CV Q and increased flow weighted concentration. Limestone presence was not significantly correlated w ith either differences in mean concentration or variance in concentration for calcium. Table 3 2 Results from two sample t tests for differences in mean CV C /CV Q and flow weighted concentration (FWC) for phosphorus and calcium between sites with differin g geologies. a Indicates statistical significance at the p < 0.05 level.

PAGE 43

43 CHAPTER 4 DISCUSSION Results from this analysis suggest strong support for the streams as integrative filters conceptual framework first outlined in the introduction. At almost e very site, measured solute concentrations vary on the order of 1 3 magnitudes less than flow varies, indicating catchment scale buffering of point scale variance in soil water concentrations and source area heterogeneity. Furthermore, concentration dischar ge relationships exhibited coherent behavior across watersheds and were controlled by watershed scale structure and processes. Patterns in Concentration Discharge Relationships Geogenic solutes defined in this study as calcium, magnesium, silica, sodium and chloride clearly follow power law functional relationships with flow and tend to have slightly negative or near zero log(C) log(Q) slopes. This was an expected result as similar trends have been observed for streams and rivers all over the Unit ed States (Johnson et al. 1969; Godsey et al. 2009; Basu et al. 2010). Silica was by far the most chemostatic solute analyzed, suggesting a large source mass over the entire study region. On the other hand, consistent shallow negative slopes for major exch angeable base cations sodium, calcium, and magnesium indicate different mobilization and transport processes than for silica. All three of these solutes are significantly more soluble than silica and play a more important role in larger scale biological pr ocesses (though silica is important for diatom growth). Due to their higher solubility, one would expect a more transport limited response, resulting in a stronger dilution signal when compared to silica. This signal may be buffered however due to terrestr ial biological demand and base cation exchange in the soil profile such that its log(C) log(Q) slope is greater than would be expected for a pure dilution (slope = 1).

PAGE 44

44 While geogenic solutes show consistent C Q trends, nutrients exhibit much more variabl e C Q relationships. The extent to which these cycles influence C Q relationships is illustrated in the comparison between total phosphorus (TP) and total inorganic nitrogen (TIN). Total inorganic nitrogen exhibits greater variability and more dilution wit h flow on average (b TIN = 0.13) than phosphorus, which tends to be nearly chemostatic on average (b TP =0.0003). This is a reasonable result, given that nitrogen is more strongly under biological control and includes processes (denitrification) that result in permanent removal from the system. Subsequently, nitrogen may experience much more temporal variability, both seasonally and daily, though this is not clear from the data. This raises the question of why TIN has a tendency to become more dilute with incre asing discharge when compared with TP. This may be attributed to the fact that high flows generate suspended particles to which particulate P is bound, subsequently increasing P concentrations and buffering the dilution response. There is sufficient eviden ce to support the hypothesis that TOC exhibits an enrichment effect with increasing discharge across all catchments within this study. Several lines of evidence exist to support this notion, but perhaps the most convincing is observation of a consistent po sitive slope of log(C) log(Q) plots for total organic carbon for nearly every site. The simplest mechanistic explanation of this phenomenon is that under high rainfall, the flow generating part of the watershed expands and begins to connect more isolated w etlands with downstream water fluxes. Larger flows are indicative of greater sampling of source areas and thus greater fluxes of carbon. This idea is reinforced by the fact that higher mean TOC concentrations were associated with lower variation in concent ration with respect to discharge. The onset of stationarity is observable for TP but not for TKN. It seems clear (Figures 5 & 6) that at flow weighted concentrations of ~1 mg/L, variation in phosphorus concentration

PAGE 45

45 with respect to variation in discharge i s dramatically less than for lower concentrations. The explanation for this can be attributed to increasing homogeneity of source areas which lead to less variation in transport to downstream waters, and also higher average concentrations because there is less phosphorus poor water to dilute streamflow. The lack of such a relationship for total inorganic nitrogen is curious, and may be due to the fact that there are few natural source zones for nitrogen (as opposed to sediments and clays for phosphorus) wit hin the landscape. Controls on Concentration Discharge Relationships Catchment Size Evidence from four longitudinal studies suggests an array of controls on C Q relationships. For instance, the Santa Fe and the Suwannee undergo calcium and magnesium conce ntration increases by ~15 and ~5 times, respectively along with increasing distance. Likewise, decreases in TOC concentrations, increases in nitrate concentrations, and peaks in phosphorus concentrations were nearly identical between rivers. The similarity of C Q behavior for all solutes between the Suwannee and the Santa Fe is indicative of watershed scale controls. I suggest that geology is the primary control on C Q relationships for these watersheds. As delivery of aquifer water to these rivers increase s downstream of the headwaters due to spring influence, mixing of source water solute concentrations tends to become weighted towards aquifer characteristics (i.e. high concentrations of Ca, Mg, TIN and low concentrations of TOC). The influence of aquifer source water is a direct result of underlying karst geology. Furthermore, peaks in phosphorus concentrations in mid reaches result from passage over the Cody Scarp, the geological boundary between limestone and sand lowlands to the south and the more claye y highlands to the north. The Hawthorn Group clays exposed at this escarpment are significantly enriched in TP, and have been shown elsewhere to be major sources of phosphorus to lakes and streams (SJRWMD report SJ2008 SP29).

PAGE 46

46 The impact of variable source water can be seen not only in concentrations of solutes but also in their relationship to discharge. Log(TIN) log(Q) slopes become much more negative with increasing distance downstream and TP and TOC slopes become much more positive. At higher flows the s prings that feed the Santa Fe and Suwannee undergo backwater effects, suppressing the impact of aquifer water on streamflow concentrations. Therefore, downstream C Q relationships are more likely to represent mixtures of source water at low flows, but are more likely to represent surficial drainage basin chemistry at higher flows. This means that these basins have strong enrichment effects for TP and TOC, but that these effects are mitigated by aquifer source water contribution. Total organic carbon may al so be controlled by its location within a watershed. My analysis of the Santa Fe, Suwannee, and Withlacoochee Rivers supports this contention. TOC flow weighted concentrations decrease with increasing drainage area, and variation in concentration with resp ect to variation in discharge increases dramatically at higher order sections of river when compared to headwaters. Therefore, headwaters tend to have a more consistent supply of carbon (likely from wetland leachate and allochthonous riparian inputs, Vanno te et al. 1980) than do higher order reaches which are more affected by variable source area contribution. In addition, log(C) log(Q) slopes show a consistent increasing trend along the river continuum, adding more support for this mechanistic framework, a nd implying that concentration effects of carbon become greater as one moves downstream and samples a larger potential leaching horizon. Another interesting result from this analysis is the observation of the onset of stationarity for total organic carbon at somewhere near 50 60 mg/L flow weighted concentration (Figures 5 and 6). This may be attributed to an increasing homogeneity of organic

PAGE 47

4 7 horizons in flow generating areas such that a river is sampling an essentially constant supply of carbon from the sur rounding landscape. At each of the four rivers, several major trends in phosphorus/nitrogen concentration discharge relationships arise: (1) total inorganic nitrogen flow weighted concentrations exponentially increase along a river length, (2) total phosp horus flow weighted concentrations gently rise to a peak at middle reaches before dropping back down, (3) log(TIN) log(Q) slopes decrease along a river length, (4) log(TP) log(Q) slopes increase along a river length, (5) CV C /CV Q for both nitrogen and phosp horus is dramatically higher at downstream sites than upstream sites. These patterns are strikingly similar across rivers, and may emerge from both geological and biogeochemical mechanisms. At the headwaters of a river, biological processes are generally l ight limited and thus streamflow concentrations of nutrients would be relatively stable and representative of average allocthonous inputs, though rain events may leach even more nutrients from the surrounding landscape (concentration effect). Further downs tream, however, where light is no longer limiting, biological processes become more significant and nutrient demand increases, but at the same time, denitrification rates decrease with increasing water depth (Basu et al. 2010b), and nutrient supply to the river may be increasing from increasing contributing area. Once biological demand for nutrients is met, all extra mass flux simply adds to flow weighted concentrations. The peak in TP concentrations not seen with TIN may be due to the fact that in lower re aches of the river, phosphorus is more limiting than nitrogen, which would tend to result in an increase in concentration of both nutrients up to the point where nitrogen demands are satisfied. At that point, phosphorus concentrations would decline, assumi ng trivial lateral contributions of phosphorus to streamflow. This conceptual framework seems to be a reasonable, but by no

PAGE 48

48 means only, explanation of observed trends: (1) nitrogen concentrations increase along a river due to increasing supply with decreas ing or constant demand, (2) phosphorus concentrations peak at a point where ecosystem light limitation ends and decline when phosphorus becomes limiting, (3) log(TIN) log(Q) slopes decrease along a river continuum because at the headwaters nitrogen supply is relatively constant, but downstream lateral inputs from variable source areas have a dilution effect on concentrations, (4) log(P) log(Q) slopes increase along a river continuum because supply at headwaters is relatively constant, but downstream large f low events tend to suspend sediments with sorbed P increasing concentrations, and (5&6) headwaters have little variability in nutrient inputs and may be considerably more flashy than higher order sections of river which also have much higher variability in nutrient concentrations. Land Cover Analysis of concentration discharge relationships raises the question of why distinctive signals that are consistent in functional form (i.e. power law), but dissimilar in slope (b, in C=aQ b ), arise across catchments. T he most obvious differences among catchments are (1) land cover, and (2) geology, both of which impact both flow and solute generation. The motivation behind the hypotheses generated for the impacts of land cover and geology on C Q relationships comes from the ideas of variable source area flow generation and mass legacies of solutes (see Basu et al. 2010 and Jawitz and Mitchell, 2011). These concepts were touched upon in the previous discussion but the details are elaborated here. Increasing agricultural c over was expected to increase flow weighted concentrations and decrease relative nutrient variation (CV C /CV Q ) for both TIN and TP. As continued application of fertilizer treatments saturates biological demand for nutrients, excess mass would be transported to downstream river systems, thereby reducing variability in concentrations sampled by streamflow. My results do not support the hypothesis that increasing agricultural cover leads to

PAGE 49

49 higher flow weighted concentrations of nitrogen and phosphorus. Perhaps more interestingly, with increasing agricultural cover streamflow concentrations of phosphorus, nitrogen, and carbon converge towards some relatively constant values (~0.3, ~0.6, and ~23 mg/L, respectively). What is the significance of these values and wh y do concentrations converge instead of simply increasing? The extrapolated molar C:N:P ratio of streamflow is 200:4:1 which is reminiscent of terrestrial plant ratios, implying a strong biotic control on solute fluxes from agricultural watersheds. If we c an assume that elemental ratios of crops in Florida are near constant, then increasing crop cover would tend to converge ratios sampled by streamflow towards ratios present in standing crops. It seems unlikely that the elemental ratio presented here is ver y far off of the extrapolated C:N:P ratio presented here, which is reminiscent of the Redfield ratio (106:16:1). Increasing agricultural area resulted in decreases in CV C /CV Q values for both total inorganic nitrogen and total phosphorus, supporting the sec ond aspect of the land cover hypothesis. Decreases in CV C /CV Q were not observed for any solute with increasing wetland cover, though results indicate that greater wetland cover tends to increase organic carbon exports and decrease total inorganic nitrogen, phosphorus, and base cation exports. Wetlands are known to act as sinks for nitrogen and phosphorus in the landscape via particulate settling for phosphorus bound sediments and denitrification (Mitsch and Gosselink, 2007), and these results support that s ervice for the catchments studied here. At approximately 40% wetland cover, flow weighted concentrations of nitrogen and phosphorus approach minimum values and a similar behavior is seen for calcium and magnesium, though it is much less clear. The near lin ear increase in TOC concentrations with wetland cover is most likely a function of increasing organic matter availability within wetlands.

PAGE 50

50 Urban cover was linked to reduced variation in concentration with respect to discharge for TOC and Ca. The most likel y reason is that the impervious surfaces associated with urban landscapes tend to induce much more variability in flow to downstream waters, thereby increasing CV Q All other things being equal, an increase in CV Q will reduce CV C /CV Q values. Based on data derived in this study, there is little reason to believe that urban cover has a major effect on streamflow concentrations of most solutes, except for total organic carbon, which exhibits concentration declines with increased urban cover. This is not to say that a greater presence of the urban environment within a watershed has no effect on streamflow hydrochemistry, in fact just the opposite is true! Nearly every solute, except for inorganic nitrogen and silica, converges towards stationarity with increasin g urban covers, a phenomenon that is particularly clear above 50% (Figure 11). Thus, it may be possible to make an a priori assumption about chemical loadings from heavily urbanized catchments given only an average flow weighted concentration and daily to monthly flow data. Conclusions An analysis of long term hydrochemical measurements for 58 sites along rivers and streams in Florida indicates a nearly universal power law relationship between concentration and discharge. Generally, concentrations of all me asured solutes vary 1 3 orders of magnitude less than flow varies. Geogenic solutes experience nearly chemostatic or slightly dilute relationships with flow. Organic solutes experience consistent concentration based relationships with flow, and nutrients e xhibit a wide range of variability in their concentration discharge relationships. Results from case studies of river reaches generally coincide with the river continuum concept for in stream processing of nutrients and carbon. Land cover was a significant driver of observed concentration discharge relationships: (1) increases in agricultural cover were associated with convergent C, N, and P concentrations as well as reduced variation in C & P concentrations, (2)

PAGE 51

51 increases in wetland cover were associated w ith reduced export of P and TIN, (3) increases in urban cover decreased CV C /CV Q for nearly all solutes. Presence of the Hawthorn group in a catchment, a phosphorus rich clay layer, significantly increased phosphorus flow weighted concentrations and reduced variability of phosphorus concentrations sampled by streamflow. Presence of limestone in a catchment did not significantly affect either flow weighted concentrations for calcium or variability in calcium concentrations. I conclude that most catchments beh ave near chemostatically for many solutes and that this chemostasis is a result of chemical mass legacies present in flow generating areas.

PAGE 52

52 A P PENDIX TABLE OF SITE CHARACTERISTI CS

PAGE 53

53 Appendix. ( continued )

PAGE 54

54 Appendix. ( continued )

PAGE 55

55 Appendix. ( continued )

PAGE 56

56 Appendix. ( continued )

PAGE 57

57 LIST OF REFERENCES Anderson, S.P., W.E. Dietrich, AND G.H. Brimhall. 2002. Weathering profiles, mass balance analysis, and rates of solute loss : Linkages between weathering and erosion in a small, steep catchment. Geological Society of America Bulletin 114 (9): 1143 1158. Basu, N.B., Destouni, J.W. Jawitz, S.E. Thompson, N.V. Loukinova, A. Darracq, S. Zanardo, M. Yaeger, M. Sivapalan, A. Rinaldo, AND P.S.C. Rao. 2010. Nutrient loads exported from managed catchments reveal emergent biogeochemical stationarity. Geophysical Research Letters 37 L23404, doi: 10.1029/2010GL045168. Basu, N.B., S.E. Thompson, AND P.S.C. Rao. 2011a. Hydrologic and biogeo chemical functioning of intensively managed catchments: A synthesis of top down analyses. Water Resources Research 47 doi: 10.1029/2011WR010800. Basu, N.B., P.S.C. Rao, S.E. Thompson, N.V. Loukinova, S.D. Donner, S.Ye, AND M. Sivapalan. 2011b. Spatiotemp oral averaging of in stream solute removal dynamics. Water Resources Research 47 W00J06, doi: 10.1029/2010WR010196. Botter G., E. Bertuzzo, AND A. Rinaldo. 2010. Transport in the hydrologic response: Travel time distributions, soil moisture dynamics, and the old water paradox. Water Resources Research 46 W03514, doi: 10.1029/2009WR008371. Cohen M.J., S. Lamsal, L. Korhnak, AND L. Long. 2008. Spatial nutrient loading and sources of phosphorus in the Newnans Lake watershed. SJRWMD report SJ2008 SP29. Clo w, D.W. AND J.I. Drever. 1996. Weathering rates as a function of flow through an alpine soil. Chemical Geology 132 : 131 141. Darracq, A., G. Lindgren, AND G. Destouni. 2008. Long term development of phosphorus and nitrogen loads through the subsurface and surface water systems of drainage basins. Global Biogeochemical Cycles 22 (3), GB3022. De Montety, V., J.B. Martin, M.J. Cohen, C. Foster, AND M.J. Kurz. 2011. Influence of diel biogeochemical cycles on carbonate equilibrium in a karst river. Chemical Geo logy 283 (1): 31 43. Evans, C. AND T.D. Davies. 1998. Causes of concentration/discharge hysteresis and its potential as a tool for analysis of episode hydrochemistry. Water Resources Research 34 (1): 129 137. Feng, X., J.W. Kirchner, AND C. Neal. 2004. Spe ctral analysis of chemical time series from long term catchment monitoring studies: Hydrochemical insights and data requirements Water, Air, and Soil Pollution 4 : 221 235. Godsey, S.E., J.W. Kirchner, AND D.W. Chow. 2009. Concentration discharge relation ships reflect chemostatic characteristics of US catchments. Hydrological Processes 23 : 1844 1864.

PAGE 58

58 Godsey, S.E., W. Aas, T.A. Clair, H.A. de Wit, I.J. Fernandez, J.S. Kahl, I.A. Malcom, C. Neal, M. Neal, S.J. Nelson, S.A. Norton, M.C. Palucis, B.L. Skjelva le, C. Soulsby, D. Tetzlaff, AND J.W. Kirchner. 2010. Generality of fractal 1/f scaling in catchment tracer time series, and its implications for catchment travel time distributions. Hydrological Processes 24 : 1660 1671. Guan, K., S.E. Thompson, C.J. Harm an, N.B. Basu, P.S.C. Rao, M. Sivapalan, A.I. Packman, AND P.K. Kalita. 2011. Spatiotemporal scaling of hydrological and agrochemical export dynamics in a tile drained Midwestern watershed. Water Resources Research 47 W00J02, doi: 10.1029/2010WR009997. H effernan, J. B., AND M.J. Cohen. 2010. Direct and indirect coupling of primary production and diel nitrate dynamics in a subtropical spring fed river. Limnology and Oceanography 55 (2): 677 688. Hornberger, G.M., T.M. Scanlon, AND J.P. Raffensperger. 2001. Modeling transport of dissolved silica in a forested headwater catchment: the effect of hydrological and chemical time scales on hysteresis in the concentration discharge relationship. Hydrological Proceses 15 : 2029 2038. Jawitz, J.W. AND J. Mitchell. 20 11. Temporal inequality in catchment discharge and solute export. Water Resources Research 47 W00J14, doi: 10.1029/2010WR010197. Johnson, N.M., AND G.E. Likens. 1969. A working model for the variation in stream water chemistry at the Hubbard Brook Experi mental Forest, New Hampshire. Water Resources Research 5 (6): 1353 1363. Kirchner, J.W., X. Feng, AND C. Neal. 2000. Fractal stream chemistry and its implications for contaminant transport in catchments. Letters to Nature 403 : 524 527. Kirchner J.W., X. F eng, C. Neal, AND A.J. Robson. 2004. The fine structure of water quality dynamics: the (high frequency) wave of the future. Hydrological Processes 18 : 1353 1359. Likens, G.E. 2004. Some perspectives on long term biogeochemical research from the Hubbard Br ook ecosystem study. Ecology 85 (9): 2355 2362. McDonnell, J.J., K. McGuire, P. Aggarwal, K.J. Beven, D. Biondi, G. Destouni, S. Dunn, A. James, J. Kirchner, P. Kraft, S. Lyon, P. Maloszewski, B. Newman, L. Pfister, A. Rinaldo, A. Rodhe, T. Sayama, J. Seib ert, K. Solomon, C. Soulsby, M. Stewart, D. Tetzlaff, C. Tobin, P. Troch, M. Weiler, A. Western, A. Worman, AND S. Wrede. 2010. How old is streamwater? Open questions in catchment transit time conceptualization, modeling and analysis. Hydrological Processe s 24 : 1745 1754. McGuire, K.J., and McDonnell. 2010. Hydrological connectivity of hillslopes and streams: Characteristic time scales and nonlinearities. Water Resources Research 46 W10543, doi:10.1029/2010WR009341.

PAGE 59

59 Murphy, J.C., G.M. Hornberger, AND R.G Liddle. 2012. Concentration discharge relationships in the coal mined region of the New River basin and Indian Fork sub basin, Tennessee, USA. Hydrological Processes, doi: 10.1002/hyp.9603. Walling, D.E., AND B.W. Webb. 1986. Solutes in river systems. I n Solute Processes Trudgill ST (ed). John Wiley AND Sons: Chichester, 251 327. Zhang, Y., AND K. Schilling. 2005. Temporal variations and scaling of streamflow and baseflow and their nitrate nitrogen concentrations and loads. Advances in Water Resources 28 : 701 710.

PAGE 60

60 BIOGRAPHICAL SKETCH J a ke h as always been a scientist. He enjoy s asking the difficult questions and even more so, e njoys answering them. The satisfaction of coming to his own conclusions about the world is short lived: the more you know, the more h is youth, he found him self fascinated by the na tural world and spent many of his waking hours playing in streams and getting lost in the woods near h is home. Later on he realized that those early experiences were his most treasure d and h e knew he needed to go farther down the rabbit hole: h e had to know more a bout the beautiful settings of his childhood. After highschool, he decided that environmental en gineering would not only sate his desire for an understanding of the natural world, b ut that it would do so in a rigorous and mathematical way. Early on in his first year of college, he met a fellow wetl and enthusiast who encouraged him to become her mentee in an undergraduate research program with wor k in the Everglades. How could he say no? Little did he know how big of an impact that experie nce would have on the rest of his life. Because of that ye ar as an undergraduate mentee, he ha s met the most intelligent and stimulating people, performed research at the cutting edge of knowledge, attended an international conference, and enjoy ed (almost) every minute of it. He is grateful for all of his experiences here at the University of Florida, and i s indebted to many who have come before hi m go!