FLORIDA THE STATE UNIVERSITY COLLEGE OF ARTS AND SCIENCES
BIOGEOCHEMICAL CYCLING OF CARBON, PHOSPHORUS, AND TRACE METALS
By JENNIFER CLAIRE STERN
A Dissertation submitted to the Department of Geological Sciences in partial fulfillment of the Requirements for the degree of Doctor of Philosophy
Degree Awarded: Summer Semester, 2005
The members of the Committee approve the dissertation of Jennifer Claire Stern defended on 4/28/04
_________________________ Yang Wang Professor Co-directing Dissertation
_________________________ Vincent J.M. Salters Professor Co-directing Dissertation
_________________________ Jeffrey Chanton Outside Committee Member
_________________________ A. Leroy Odom Committee Member
Approved:
___________________________________ A. Leroy Odom, Chair, Department of Geological Sciences
The Office of Graduate Studies has verified and approved the above named committee members.
ii
ACKNOWLEDGEMENTS Many different people contributed their advice and support to this research. My situation was unique in that I had three distinct research projects supervised by three different advisors, and I would like to thank my major professors Vincent Salters and Yang Wang (who introduced me to isotope geochemistry), and committee member Jeff Chanton for giving me the opportunity to work on projects in their respective fields. I would also like to thank my other committee members Roy Odom and David Furbish for their academic and personal support. Every member of my committee has been a significant source of support in my time at Florida State. The Isotope Geochemistry Division at the National High Magnetic Field Laboratory has been a wonderful place to work, and I would like to thank Ted Zateslo and Afi Sachi-Kocher for their technical support. Special thanks to Michael Bizimis and Jeroen Sonke, who provided their insights and technical support on a daily basis. I would also like to thank all of the geochemistry graduate students for the long chats, the little favors, and camaraderie along the way. I would like to thank my parents for providing me the educational opportunities to get this far, and my wonderful husband Rob Lowe for the infinite amount of emotional (and sometimes financial!) support through our 6+ years in Tallahassee. This would not have been possible without him.
iii
TABLE OF CONTENTS List of Tables ................................................................................................................. vi List of Figures ...............................................................................................................vii Abstract.......................................................................................................................... ix INTRODUCTION......................................................................................................... 1 1. DEVELOPING A METHOD TO USE THE OXYGEN ISOTOPIC CONTENT OF PHOSPHATE TO TRACE THE SOURCE OF PHOSPHATE IN THE EVERGLADES 1.1. Introduction .................................................................................................. 3 1.2. Study Site...................................................................................................... 7 1.3. Methods........................................................................................................ 9 1.3.1. Isotopic analysis of fertilizer samples.............................................. 9 1.3.2. Filter Preparation ............................................................................ 9 1.3.3. Extraction of DP from filter for isotopic analysis .......................... 10 1.4. Results and Discussion................................................................................ 12 1.4.1. 18O of phosphate in fertilizers ..................................................... 12 1.4.2. 18O of dissolved phosphate.......................................................... 12 1.5. Conclusion .................................................................................................. 16 2. DISTRIBUTION AND TURNOVER OF DOC IN NATURAL AND CONSTRUCTED WETLANDS IN THE FLORIDA EVERGLADES 2.1. Introduction ................................................................................................ 17 2.1.2. Study sites .................................................................................... 18 2.2. Methods...................................................................................................... 20 2.2.1. Analytical methods ....................................................................... 20 2.2.2. Computational methods for estimating the DOC turnover rate in constructed wetlands ............................................................................. 21 2.3. Results and Discussion................................................................................ 24 2.3.1. Concentration and isotopic composition of organic carbon along a water quality gradient ................................................................ 24 2.3.2. Concentration and Stable Carbon Isotope Composition of Organic Carbon in two constructed wetlands .......................................... 30 2.4. Conclusion .................................................................................................. 35 3. SPECIATION OF TETRAVALENT METALS THORIUM, HAFNIUM, AND ZIRCONIUM BY CAPILLARY ELECTROPHORESIS INDUCTIVELY COUPLED PLASMA MASS SPECTROMETRY (CE-ICP-MS) AND EQUILIBRIUM DIALYSIS LIGAND EXCHANGE (EDLE) 3.1. Introduction ................................................................................................ 36 3.1.2. Single site binding approach ......................................................... 37 3.1.3. CE-ICP-MS .................................................................................. 37 3.1.4. Actinide chemistry........................................................................ 39 3.2. Methods and Materials ................................................................................ 40 3.2.1. CE-ICP-MS .................................................................................. 40 3.2.2. EDLE ........................................................................................... 45 3.3. Results and Discussion................................................................................ 49
iv
3.3.1. REEs ............................................................................................ 49 3.3.2. Tetravalent metals ........................................................................ 49 3.3.3. Comparison of experimentally determined log Kc with literature values..................................................................................................... 57 3.4. Conclusion .................................................................................................. 59 4. METHANE OXIDATION IN LANDFILLS 4.1. Introduction ................................................................................................ 61 4.2. Methods...................................................................................................... 63 4.2.1. Site description ............................................................................. 63 4.2.2. Flux.............................................................................................. 65 4.2.3. Stable carbon isotopes .................................................................. 65 4.3. Results and Discussion................................................................................ 67 4.3.1. Methane flux ................................................................................ 67 4.3.2. 13C.............................................................................................. 70 4.3.3. Oxidation...................................................................................... 71 4.4. Conclusion .................................................................................................. 76 4.5. Appendix .................................................................................................... 77 REFERENCES.............................................................................................................. 80 BIOGRAPHICAL SKETCH ......................................................................................... 92
v
LIST OF TABLES 1.1. Oxygen isotopic composition of phosphate in fertilizers ......................................... 12 1.2. Results from laboratory experiments using DI water spiked with KH2PO4 ....................... 13 1.3. Phosphate concentrations (in ppb) at sample sites ................................................... 13 1.4. Oxygen isotopic values of Everglades waters.......................................................... 15 2.1. Concentration and 13C of DOC in test cell inflow and outflow .............................. 31 2.2. 13C of vegetation in test cells................................................................................. 31 2.3. Estimated fraction of residual and new DOC in test cell outflow............................. 32 2.4. Estimated DOC turnover time and net DOC production of test cells ....................... 34 3.1. Selected characteristics of HA ................................................................................ 43 3.2. Summary of Binding constants used ....................................................................... 43 3.3. Hydrolysis constants for tetravalent metals ............................................................. 43 3.4. CE-ICP-MS run parameters .................................................................................... 45 3.5. Stock solution and sample solution compositions.................................................... 45 3.6. Analytical data for complexation studies of Th4+ with humic acid in 0.1 M NaNO3. ......................................................................................................................... 52 3.7. Analytical data for complexation studies of Hf4+ with humic acid in 0.1 M NaNO3. ......................................................................................................................... 52 3.8. Analytical data for complexation studies of Zr4+ with humic acid in 0.1 M NaNO3. ......................................................................................................................... 53 4.1. Results of ANOVA for flux data............................................................................. 69 4.2. ANOVA results ...................................................................................................... 73 4.3. Methane flux (g CH4/m2/day) average and standard deviations for Control and Biocover on each date ................................................................................................... 77 4.4. Average 13C values (per mil) used in Figures 4.6 and 4.7 ...................................... 77 4.5. Average oxidation without 100% uptake values used in Figure 4.8 ......................... 77 4.6. Average oxidation with 100% uptake values used in Figures 4.8 and 4.9 ................ 78 4.7. Average soil moisture values, grams water per grams soil....................................... 78 4.8. Average soil temperature values, C........................................................................ 78 4.9. Chamber 2B1 ......................................................................................................... 79 4.10. Chamber 2B3........................................................................................................ 79 4.11. Chamber 2D4 ....................................................................................................... 79
vi
LIST OF FIGURES Figure 1.1. Study site....................................................................................................... 8 Figure 1.2. Phosphorus Extraction Procedure ................................................................ 11 Figure 1.3. The isotopic composition of phosphate oxygen............................................ 14 Figure 2.1. Study Site .................................................................................................... 19 Figure 2.2. Study site detail of STA-1W........................................................................ 20 Figure 2.3. Schematic diagram showing the major components of the DOC cycle in lined, hydrologically controlled test cells ................................................................................ 22 Figure 2.4. DOC and TDP concentrations from northern to southern Everglades ........... 25 Figure 2.5. Variation in DOC content of STA-1W Cell 4 outflow over 4 months in 2001 ......................................................................................................................... 25 Figure 2.6a-b. 13C of (a)DOC and (b)POC from northern and southern Everglades ...... 26 Figure 2.7. 14C values of DOC, POC, and DIC in waters collected in August 2001 Periphyton samples collected in May 1998 and June 2002............................................. 26 Figure 2.8. Comparison of DOC and POC radiocarbon data from this study and Retentate and Permeate data from Wang et al. (2002). ................................................................. 27 Figure 2.9. Comparison of 14C values of DOC in various freshwater aquatic systems.. 29 Figure 2.10. Schematic diagram of SFWMDs test cell system within STA-1W ............ 30 Figure 2.11. The relative amount of new DOC in the test cell outflow ........................... 32 Figure 2.12a-b. (a) Turnover time for DOC in cell 9 and cell 15 compared to the North Pacific. (b) Comparison of net DOC production rate in the test cells, mg C/L/day ......... 33 Figure 3.1. Electropherogram showing separation of TmEDTA- and TmHA2+ species... 41 Figure 3.2. Schematic representation of experimental setup for EDLE........................... 46 Figure 3.3. Conditional binding constants, log Kc,LnHA as a function of REE and HS...... 49 Figure 3.4. Conditional binding constants, log Kc,MEHA as a function of pH and metal ... 50 Fig 3.5a-c. Conditional binding constant compilations for the tetravalent metals Th, Hf, and Zr............................................................................................................................ 50 Figure 3.6a-d. Electropherograms of Hf-EHA complexes .............................................. 54 Figure 3.7a-d. Electropherograms of Hf-PHA complexes .............................................. 55 Figure 3.8. Comparison of log K variation with [HA] from our study and Reiller et al... 56 Figure 3.9. Comparison of log K variation with [Th]/[COOH] from our study and Reiller et al. .............................................................................................................................. 57 Figure 3.10. All conditional binding constants generated in this study ........................... 57 Figure 3.11. Conditional binding constants for ThHS complexes taken from this study and published literature, corrected to 0.1 IS ................................................................... 59 Figure 4.1. Location of squares selected for control (no compost) and biocover (compost) sites............................................................................................................................... 63 Figure 4.2. Cross-section sketch of S1 biocover locations.............................................. 64 Figure 4.3. Experimental design for the biocover study ................................................. 65 Figure 4.4a-b. (a) Methane emission rates or flux from control and biocover areas (b) Control and biocover fluxes averages without large fluxes included ......................... 67 Figure 4.5. Methane emission rates at each chamber from March 24, 2004 to December 10, 2004. ....................................................................................................................... 68
vii
Figure 4.6a-b. (a) Methane emission rates averaged over 12 chambers in the control area vs. single chamber data from 2B1 (b) Methane emission rates averaged over 12 chambers in the biocover area vs. single chamber data from 2D1. ................................................. 68 Figure 4.7a-b. (a) Comparison of control flux including data from 2B1 and without data from 2B1. (b) Comparison of biocover flux including data from 2D1 and without data from 2D1....................................................................................................................... 69 Figure 4.8. Mean 13C values (per mil) of methane in the control and biocover compared to anoxic methane ......................................................................................................... 70 Figure 4.9a-b. Methane emission rate and 13C values (per mil) of CH4 emitted from the (a) control and (b) biocover ........................................................................................... 71 Figure 4.10a-c. Percent oxidation of CH4 in the control and the biocover, (a) calculated without using values of 100% for incidences of negative flux, (b) calculated using values of 100% for incidences of negative flux. (c) The number of samples per date for the control and the biocover that yielded negative fluxes. .................................................... 72 Figure 4.11a-f. Percent methane oxidation and flux (a) in the control, averaged across 12 chambers (b) in the biocover, averaged across 12 chambers (c) in chamber 2B1 from the control (d) in 2D4 from the biocover. Oxidation and soil moisture regression for (e) chamber 8B2 in the control (f) chamber 2D2 in the biocover. ........................................ 75
viii
ABSTRACT Stable and radiocarbon isotopes were used as tracers in the Florida Everglades to yield information on the sources and sinks of dissolved organic carbon in natural and constructed wetlands and provide a way to monitor ecosystem restoration efforts. Stable carbon isotopes were used to determine the source of DOC, POC, and DIC, and in a basic mass balance model to calculate turnover times of DOC in small constructed wetland cells. Radiocarbon was used to distinguish old DOC derived from historic peats from new DOC derived from modern primary production. Our study suggests that 14C measurements can be a useful indicator of the progress of ecosystem restoration in the Everglades. The oxygen isotope of phosphate (P) can also serve as an isotopic tracer in wetlands. Initial method development to use the oxygen isotope of phosphate extracted from natural waters is presented here. Preliminary data indicates that microbial recycling is a major means by which P stays in the water column despite reducing anthropogenic contributions. Stable carbon isotopes were also used to quantify the percent of methane oxidized within Tallahassee landfill soils. Treatment of carbon isotope and oxidation data collected over almost 9 months of monitoring methane emissions from landfill surfaces with and without a biocover is examined. These measurements, combined with measurements of methane flux, can help monitor the efficiency of various treatments in reducing methane emissions to the atmosphere by enhancing oxidation of methane by methanogenic bacteria. The presence or absence of DOC in the water column can determine whether trace metals will be present as a nutrient or as a toxicant. A novel method coupling capillary electrophoresis with ICP-MS was used to separate metal species and calculate binding constants of rare earth elements and Th, Hf, and Zr with humic substances at a range of pH and ionic strength of 0.1. Equilibrium dialysis exchange was performed to validate the CE-ICPMS method. Conditional stability constants of tetravalent metal-HA complexes are several orders of magnitude higher than lanthanide-HA complexes. Because thorium is often used as a proxy for the tetravalent actinides, Th-HA binding constants are useful in the study of sequestration of actinides in nuclear repository settings.
ix
INTRODUCTION Research on behavior of organic carbon in natural ecosystems is essential to our understanding of how nature will respond to anthropogenic effects on the carbon cycle. Each chapter in this dissertation addresses a topic relating to the source, transport, or fate of carbon. The importance of the carbon cycle goes beyond the climatic changes caused by greenhouse warming. Carbon is a substrate for microorganisms, supporting microbial communities that work to turnover other nutrients, such as nitrogen and phosphorus. Carbon compounds such as humic substances can strongly complex trace elements, affecting their bioavailability and how they are transported. Degradation of carbon recycles carbon back into the atmosphere in the form of CO2 or CH4 gas, which have different infrared activities and residence times in the atmosphere. Thus, no study of carbon cycling would be complete without taking into account the interactions of carbon with other elements and microorganisms. A variety of geochemical methods were used in this study. Stable and radiocarbon isotopes were used as tracers in the Florida Everglades to yield information on the sources and sinks of dissolved organic carbon in this area and provide a way to monitor ecosystem restoration efforts here. Initial research in the Everglades quantified the amount of carbon coming from two isotopically different sources using stable carbon isotopes and suggested that radiocarbon could fingerprint older carbon sources and provide a useful indicator of water quality (Wang et al. 2002). As part of my dissertation research, I analyzed stable carbon isotopes and radiocarbon from DOC, POC, sediments, and plants to characterize the source and turnover of DOC in the Florida Everglades. This study used radiocarbon as a tool to distinguish between old DOC from historic peats and new DOC from modern primary production. Stable carbon isotopes were used in this study to determine the relative contributions of C4 sugarcane and C3 wetland plants to bulk DOC, and in a basic mass balance model to calculate turnover times of DOC in small constructed wetland cells. The carbon cycle is intimately linked to other nutrient cycles, such as phosphorus. The oxygen isotope of phosphate has the potential to be a power tracer of microbial activity in certain ecosystems. (Mclaughlin 2000; Blake 2001; Paytan 2002; Stern 2002). Initial method development to use the oxygen isotope of phosphate extracted from natural waters is presented here. Phosphorus is bound very strongly to oxygen atoms in phosphate and is resistant to equilibrium isotopic exchange of oxygen with environmental water at low temperatures. However, the breaking of the P-O bond occurs via enzyme catalyzed reactions, thus isotopic fractionation of oxygen in phosphate is likely to occur during microbial uptake and remineralization of this nutrient. Hypothetically, inorganic phosphate from an isotopically distinct source, such as fertilizer should be distinguishable from naturally cycled phosphate, thought to be at some level of equilibrium with environmental water. Research supporting active microbial recycling of phosphorus in the Florida Everglades is presented here and can be linked to the labile nature of DOC.
1
Preliminary data indicate that microbial recycling is a major means by which P stays in the water column despite reducing anthroprogenic contributions. The importance of DOC in the complexation and transport of metals in aquatic systems is widely accepted and well studied. The DOC cycle is intimately linked to the cycling and bioavailability of trace metals. The presence or absence of organic carbon in the water column can determine whether trace metals will be present in limited quantities as a nutrient, or in surplus quantities as a toxicant. It is important to know how the presence of humic substances, which represent the refractory products of DOC degradation, affect metal speciation. Metal speciation is widely studied using many different methods and models to generate metal-humic binding constants. This is an obstacle when comparing binding constants from the literature. For this reason I compared two methods to measure partitioning of metals between humic acids and a competing ligand (EDTa). Capillary electrophoresis coupled with ICP-MS (CE-ICP-MS) has recently been identified as a rapid and accurate means to separate metal species and calculate binding constants metal-humic complexes (Olesik et al. 1998; Sonke 2003; Sonke and Salters 2004). To validate this method and because of the lack of data available in the literature for comparison, I ran equilibrium dialysis exchange experiments using membranes to separate metal species. The metals examined in this study were thorium, halfnium, and zirconium. Relatively little is known about how strongly these tetravalent metals bind to humic substances due to the experimental difficulties of working with them. Thorium is often used as a proxy for the tetravalent actinides, so information regarding Th-HA interactions is useful in the study of sequestration of actinides in nuclear repository settings. The reduction of greenhouse gases is one of the ultimate goals of research on the carbon cycle. Methane is a more potent greenhouse gas than carbon dioxide, despite the fact that carbon dioxide is present at greater concentrations. Because the balance of methane sources and sinks is almost equal, mitigation of anthropogenic sources could easily stabilize methane at the present atmospheric levels. Even a small reduction of methane emissions into the atmosphere would be rapidly noticeable due to the short residence time of CH4 in the atmosphere (about 9 years). Landfills would be a good candidate for mitigation by management practices. They are estimated to account for approximately 37% of annual anthropogenic CH4 emissions in the US and 10-19% of global anthropogenic emissions (Czepiel et al. 2003) It has been shown that a biologically active layer of compost or topsoil can significantly reduce methane emissions to the atmosphere (Bogner et al. 1997; Chanton and Liptay 2000; Barlaz et al. 2004). Stable carbon isotopes can be used to quantify the percent of methane oxidized within the landfill soil. These measurements, combined with measurements of methane flux, can help monitor the efficiency of various treatments in reducing methane emissions to the atmosphere by enhancing oxidation of methane by methanogenic bacteria. In my final chapter, treatment of carbon isotope and oxidation data collected over almost 9 months of monitoring methane emissions from landfill surfaces with and without a biocover is examined. Ultimately, we hope that this technology will be proven useful in reducing methane emissions to the atmosphere using waste products readily available at landfills (recycled glass and composted waste) and be adopted by landfills on a larger scale.
2
CHAPTER 1 DEVELOPING A METHOD TO USE THE OXYGEN ISOTOPIC CONTENT OF PHOSPHATE TO TRACE THE SOURCE OF PHOSPHATE IN THE EVERGLADES
1.1. Introduction Nutrient loading in freshwater and estuarine environments is the subject of much concern due to our dependence on the benefits offered by a healthy wetland. Wetlands serve to store and filter water, provide habitats for a diverse array of species, and support recreational activities such as fishing. Many coastal and freshwater wetlands have experienced eutrophication due to poor land use and excessive nutrients in agricultural or urban runoff upstream. Eutrophication is problematic as it causes an increase in growth of algae and exotic species, a decrease of oxygen in water leading to fish kills, and cyanobacteria blooms that make the water toxic to humans and animals (Sharpley 1999). While the source of these excess nutrients can sometimes be traced, the fate of these nutrients once they enter the wetland system is poorly understood. Phosphorus (P) is most often the limiting nutrient in freshwater environments. P is present in the sediment and water column as particulate organic phosphorus (POP), dissolved organic phosphorus (DOP), and dissolved inorganic phosphorus (DIP). DIP, most often in the form of orthophosphate, is the most bioavailable form of phosphate and may be quickly taken up by organisms. It may also be sequestered in soils by adsorption to Fe, Al, Ca, and Mg minerals. Organic P pools include plants, animals, soil organic matter, and microbial biomass. Microbes transform organic phosphorus from decaying organic matter into bioavailable inorganic phosphorus. These processes are largely governed by nutrient content (C, N, and P) of the soil as well as by the size of the microbial pool. In wetland soils, phosphorus accumulates as inorganic P (adsorbing to Fe, Al, Ca, and Mg minerals) and as P in organic detritus, remaining long after the cessation of Ploading upstream. The fate of this excess P is unknown, making it important to better define the biogeochemical processes governing the cycling of this element. Phosphorus loading is particularly a problem in the Florida Everglades, historically a P-limited wetland. The runoff of the Everglades Agricultural Area (EAA) is high in inorganic phosphorus (on average 20 times higher than P-levels in the pristine Everglades), directly influencing water quality throughout the Everglades. Plants and algae endemic to the Everglades are adapted to low phosphorus conditions, with organic
3
phosphorus accounting for the largest fraction of soil phosphorus in this area (Koch-Rose 1994). Typically, inorganic phosphorus exists only in very small quantities in natural fresh waters; natural cycling ensures that phosphorus is efficiently utilized during photosynthesis by plants and algae and subsequently regenerated by microbial action for later use. However, when phosphorus loading occurs due to the use of phosphate fertilizer, a surplus of bioavailable phosphorus (mostly inorganic) is created, resulting in changes in water quality and vegetation assemblages. The results of this phosphorus enrichment have been well documented in the Everglades. Perhaps the most noticeable ecological change has been the vegetation shift from P-limited sawgrasses to P-adapted cattail stands in the more polluted areas of the marsh (Davis 1994). In addition, increased net primary productivity and P storage by wetland vegetation (Craft et al. 1995), increased decomposition of detritus (Davis 1991), and increased organic soil accretion (Craft and Richardson 1993; Craft and Richardson 1998) have been documented. Increased decomposition rates indicate a faster rate of microbial recycling in the marsh. The decomposition of organic matter by microbes remineralizes organic phosphorus, making it bioavailable to plants and further increasing the production of organic matter. Although P enrichment in the Everglades due to fertilizer runoff has been extensively studied, less is known regarding the in situ microbially mediated reactions responsible for the uptake and sequestration of P and its subsequent remineralization. It has been suggested (Stern 2002) that the oxygen isotopic ratio of phosphate may be useful as a tracer to determine the source of phosphate in freshwater ecosystems. The rational is as follows. Fertilizer phosphate is marine in origin and is enriched in the heavy oxygen isotope (18O). Assuming oxygen isotopic systematics in freshwater ecosystems are controlled primarily by equilibrium isotopic exchange, phosphate that is processed by biological activities in freshwater ecosytems should be in equilibrium with the oxygen in freshwater, which is depleted in 18O with respect to seawater. Thus, phosphate in runoff waters flowing directly off agricultural lands should be enriched in 18 O if the source of the excess phosphate is fertilizer, whereas phosphate in more pristine areas should have less enriched 18O value. If isotopic exchange between phosphate oxygen and water oxygen are controlled by kinetic instead of equilibrium or massdependent isotopic exchange factors, phosphate in the pristine areas of the marsh may have a more depleted or a more enriched signature. Determination of the isotopic compositions of the potential sources and sinks of P would allow the rate of P production (source strength) and the rate of P removal (sink strength) to be calculated. These rate constants could be used in a coupled hydrological biogeochemical model to evaluate how changes in hydrology and P loading affect the P cycle in freshwater environment. The goal of this study was to determine whether the oxygen isotopic composition of phosphate could be utilized in highly eutrophic watersheds in order to understand the source and fate of phosphorus and to better assess the role of microbial recycling in the P cycle. By comparing the 18O of dissolved phosphate in Everglades water to the 18O of the water itself in pristine areas of the marsh, we should be able to determine whether microbial recycling of P results in total or partial re-equilibration of dissolved phosphate with environmental water. This information will help us better understand how the oxygen isotopic ratio of phosphate can be used as an environmental tracer, and give us insight into the time scales over which P is cycled.
4
Oxygen Isotopic Studies of Phosphate Initial studies of the oxygen isotopic composition of phosphate were fueled by the hypothesis that oxygen isotopic fractionations between phosphate and carbonate in marine shells could provide estimates of ocean paleotemperatures. Tudge (Craft and Richardson 1998) was the first to devise a method to measure the oxygen isotopic composition of orthophosphate. Because it was shown that the phosphate oxygen isotopic ratio does not provide paleotemperature data independent from that gleaned from carbonate oxygen isotopic studies, this tedious method slipped into relative obscurity. However, in the 1980s, a series of papers published by Kolodny et al. (Kolodny 1983) and Longinelli et al. (Choppin 1983) established the use of the oxygen isotope ratio of biogenic phosphate from the bones and teeth of fish as a proxy for paleoenvironmental change. Since then, many researchers have taken advantage of the resistivity of the P-O bond to low-temperature alteration and used the oxygen isotopic composition of phosphate as a tool in several different types of applications. In addition, analytical methods are constantly being refined due to the tedious nature of the analysis. At present, comparison of results from different laboratories is hindered by lack of standardization. The research of Tudge (1960) proved that isotopic exchange between aqueous inorganic solutions and phosphate ions is negligible at low temperatures over long periods of time. This has been shown repeatedly by other authors (Longinelli 1973; Baes and Mesmer 1976; Kolodny 1983; Luz 1984). In addition, Tudge demonstrated that rapid exchange between 18O of water and phosphate does occur during enzyme-catalyzed reactions. One such reaction is the formation of calcium phosphates such as bone and teeth. This is due to the hydrolysis of adenosine triphosphate (ATP) and other phosphates, which allows exchange of water oxygen with phosphate oxygen (Leonard 1969). Again, this has been shown repeatedly by other authors (Blake 1997; Lecuyer 1999; Paytan 2002). The first study analyzing the oxygen isotopic content of dissolved phosphate (DP) in the oceans was by Longinelli, in 1976. They found the 18O values of DP to be almost uniform in each ocean, 19.7 in the Atlantic and 20.6 in the Pacific. Longinelli speculated that DP may even be in near-isotopic equilibrium with average ocean water at a temperature of about 4 C. In addition, this study sampled organic matter (soft-tissue) of fish and mollusks from various different locations, finding consistent fractionations between organic P from fish and DP in ocean water (fish being enriched relative to DP by about 3). These findings indicated to the authors that the 18O of phosphate in soft tissue is not related to water temperature, and must be controlled by biological and not equilibrium processes. However, Kolodny et al. (1983) analyzed fish from different layers of thermally stratified Lake Baikal and found that fishbone apatite recorded the temperature of the ambient water it grew in. In addition, they experimented with feeding fish a phosphate food source with a distinct 18O, finding that this food source had no effect on the isotopic composition of fishbone apatite. They found that the precipitation of fishbone apatite was related to the water temperature in the holding ponds at approximately 15 C using the equation from Longinelli (1973): t = 111.4 4.3(p + w). Research by (Shemesh 1983) supports this temperature dependence. He studied the phosphate oxygen
5
isotopic values of very young marine phosphorites, finding that these values reflect the temperatures at which the phosphorites were precipitated, and not the dissolved phosphate in the water column. In the early 1980s, Longinelli assessed the feasibility of using oxygen isotopes in mammal bone phosphate in paleohydrological and paleoclimatological studies. Because mammals are warm blooded, their bone forms at a relatively constant temperature of 37 C (Luz, 1984). His research lead him to conclude that 18O of bone phosphate varied systematically with 18O of local meteoric water. Luz et al. (1984) found that 18O of mammal bone phosphate was not in equilibrium with drinking water, but with body water. Oxygen isotopic composition of body water is dependent on a number of factors, including the 18O of drinking water and the metabolic rate. More importantly, linear relationships exist between 18O of drinking water and 18O of body water and between 18O of drinking water and 18O of mammal bone phosphate. The latter relationship has found wide application in paleoclimate research (e.g., Fricke et al., 1995, 1996, 1996; Iacumin et al., 1996). More recent studies of 18O of phosphate take a closer look at the metabolic reactions that cause isotopic exchange of PO43- with water and whether this isotopic exchange results in equilibrium fractionation. The invention of new and easier analytical techniques that precipitate PO43- as Ag3PO4 (ONeil, 1994) have made these studies more feasible. The first study to examine the isotopic systematics of microbially mediated reactions of phosphate was by Blake et al., (1997). This study grew bacteria on substrates containing organophosphorus or inorganic phosphate as a P source. As the bacteria grew and metabolized the P source, they released PO43- which was precipitated as flourapatite, replacing carbonate in the substrate provided. They found that experiments using organophosphorus in the form of RNA as a growth medium resulted in some incorporation of oxygen from the water into the flourapatite, while experiments with the inorganic growth medium resulted in significant oxygen isotopic exchange with water in the presence of bacteria. The resulting values in both cases were shifted in the direction approaching estimated equilibrium values, suggesting that bacterially mediated isotopic exchange is governed by equilibrium rather than kinetic factors (Blake, 1997). In the case of the organic P, the mechanism by which this isotopic exchange occurs is enzyme catalysis. P in RNA is in the form of phosphate but is characterized by having two phosphodiester linkages. In other words two oxygen atoms are bound and two are free. To free phosphate from the organic superstructure, two hydrolytic cleavages must occur. During these cleavages, enzymes catalyze the breaking of the P-O bond to liberate inorganic P. This breaking of bonds results in two oxygens left behind. Two oxygens are retained, and two must be picked up from the surrounding water. Thus, Blake (1997) hypothesizes, based on the measured shifts in fractionations during these experiments, that 50% of the oxygens undergo isotopic exchange with the water. The systematics by which inorganic phosphate undergoes almost total isotopic exchange with water are less known and thought to perhaps involve repeated exchange of inorganic P between the cell cytoplasm and environmental fluids across the cytoplasmic membrane (Maloney 1992). As noted in Blake et al. (1997), very few measurements of 18O separated from organic matter have been reported. Those values of phosphate extracted from soft tissues
6
of fishes (Longinelli et al., 1976) and algae (Paytan 1990) that have been reported reflect a dependence on the 18O of environmental water. Unlike the fractionation between bone apatite and water, 18O of phosphate extracted from soft tissues is not temperature dependent between 0 and 35 C (Paytan, 1990). Lecuyer (Benitez-Nelson 1999) suggested that the oxygen isotopic ratio of PO43could be a useful tracer of source contributions to the phosphate pool, citing the fact that the 18O data of dissolved phosphate in seawater from Longinelli et al. (1976) shows that DIP is not in equilibrium with sea water based on equilibrium fractionation factors and thus must reflect the source of P or biological cycling. Markel et al. (Ayliffe 1994) used the 18O of phosphate in sediments in Lake Kinneret to estimate the contributions of igneous, sedimentary, and anthropogenic sources to lake sediments. Paytan et al. (1990) presented data from experiments involving oxygen isotopic exchange between 18Oenriched water (18O = 30), phosphate fertilizer (18O = 6), and soft tissue of algae, observing enrichment of algal biomass by 19 to 23 with respect to environmental water. This was attributed to intense nutrient recycling of P. An interesting observation made in this study regards the fractionation between the 18O of DIP in the water and the 18O of fishbone apatite (which was 1.4 higher than DIP), and between the 18O of fishbone apatite and the 18Op of soft-tissue (which was 2 higher than skeletal apatite). The former fractionation between DIP and apatite is possibly an indication that different fractionation factors are involved in the formation of different-P bearing species (Paytan, 2002). Alternatively, the difference in 18O of these different P pools could be due to differences in turnover times of P in both reservoirs. DIP would presumably have a faster turnover time than bone apatite, which could represent equilibrium at time averaged set of conditions, whereas 18O of DIP could represent equilibrium over shorter time scales. The latter difference between the 18O of bone apatite and soft-tissue could indicate that isotope fractionation is dependent on the specific biochemical/physiological processes involved in the synthesis of these compounds. The fact that both 18O of bone apatite and soft-tissue were higher than 18O of DIP indicates some degree of isotopic exchange between organic compounds and the water had to occur during the remineralization of these organic compounds and conversion to bioavailable SRP. 1.2. Study Site The Everglades region of Florida encompasses most of the southern Floridan peninsula and represents the largest freshwater wetland and subtropical ecosystem in the country. Prior to settlement, the hydrology of this region was controlled by seasonal cycles in rainfall causing sheetflow from Lake Okeechobee and flooding low lying south Florida. To encourage settlement and provide agricultural lands, drainage and reclamation projects were instituted at the turn of the last century, emplacing canals and dikes to control water flow in this area. The early 1950s ushered in an era marked by improvement of existing water control projects and the designation of the Everglades Agricultural Area (EAA) in the upper Everglades along with Water Conservation Areas bordering the EAA. Today these Water Conservation Areas are used not only to store water but as buffers between the EAA and the more pristine Everglades to the south. The Water Conservation Areas are the largest remnants of the original Everglades ecosystem,
7
although water quality here has been adversely affected by nutrient rich agricultural runoff from the EAA. Stormwater Treatment Areas have been established by the South Florida Water Management District (SWFMD) to treat water coming directly off the EAA and have proven useful in reducing P in effluent waters. Samples were taken from both polluted and non-polluted areas of the marsh (Fig. 1). Samples taken just upstream of Stormwater Treatment Area 1 West (STA-1W) represent the most polluted samples, as this water is coming directly off the EAA. Samples were also taken at the outflow of STA-1W. Samples were taken from the Hillsborough Canal between structures 39 and S10A. Because the canal receives untreated runoff from the EAA, these samples were rich in P. Unpolluted samples were taken from the center of WCA-2A at a dock used by Duke University and in the Everglades National Park, where P concentrations were below detection limits.
Figure 1.1. Study site
8
1.3. Methods 1.3.1. Isotopic analysis of fertilizer samples Fertilizer samples were prepared for oxygen isotopic analysis using established methodology (O'Neil 1994). This involved dissolving the fertilizer in dilute nitric acid and treating the solution with hydrofluoric acid to remove Ca2+ cations. Silver ammine solution was added and samples were heated at no more than 60 C for approximately four hours to precipitate trisilver phosphate (Ag3PO4). This solid was then combusted with a stoichiometric amount of graphite at 1200 C to produce CO2 for analysis on the stable isotope mass spectrometer. 1.3.2. Filter preparation Analytical methods to determine oxygen isotope ratios of the low levels of dissolved phosphate found in natural fresh waters are limited. Analysis to determine the oxygen isotopic composition of biogenic apatite (Kolodny 1983; O'Neil 1994) involves dissolving materials in which phosphate is already heavily concentrated. In order to get a sufficient amount of phosphate from freshwater for analysis, a scheme to concentrate and extract phosphorus must be implemented. Most of the research done to this end has been in marine systems. Such schemes have been implemented for collection of P for studying 32 P and 33P activities in seawater (Lal 1988; Lee 1992; Benitez-Nelson 1998). These studies involved the extraction of phosphate from seawater via Fe-impregnated filter. This extraction scheme has been shown to be equally efficient in extracting organic and inorganic forms of P (Benitez-Nelson 1998). Our procedure involved soaking polypropylene filters in hot sodium hydroxide (NaOH) and then hot iron (2+) chloride (FeCl2). The sodium hydroxide allowed the filters to retain the iron cations, effectively making the filter an ion exchange device, allowing the iron to scavenge phosphate anions from solution. PVC cartridges were filled with four of these filters. Water was pumped from the source through cartridge filters using marine pumps powered by a 12 volt battery. The first filter was a 25 micron filter to remove particulate matter. The second filter concentrated the phosphate on the polypropylene fabric that has been treated with iron chloride solution to scavenge phosphate. The pore size of this fabric was 45 microns. Ostensibly the 25 micron filter excluded the large particles and anything that gets through it will pass through the larger 45 micron filter. Although it is more customary to use a 0.45 micron filter to exclude particulate matter, we found it slowed the pumping considerably. These filters are extremely efficient and saturation has not been a problem due to the low concentration of phosphate in freshwater. Laboratory tests were conducted by pumping distilled water spiked with K2HPO4 (an inorganic phosphate standard) at concentrations several hundreds of ppb higher than those found in nature. These tests found the pump and filter system to be 99.6% efficient in extracting inorganic phosphate from distilled water. The amount of water that must be pumped through the filter depends on the phosphate concentration of the water. Usually pumping rates did not exceed 3-4 liters per minute, so several hours of pumping and multiple pumps were required to collect enough sample for isotopic analysis. In the future, a larger pump will be used to increase the flow rate in order to concentrate more phosphate on the filter in
9
less time. Once sample was collected, it was immediately frozen to halt any microbial reactions taking place. 1.3.3. Extraction of DP from filter for isotopic analysis Samples collected in the field were brought back to the lab for chemical and isotopic analyses. In the lab, the filters were removed from the cartridge and soaked in a solution of 8N hydrochloric acid and hydrogen peroxide for 12 hours. Immediately it became clear that the large amount of organic carbon found in the waters of the marsh inhibited the formation of the first precipitate, ammonium molybdate. This made it necessary to treat the eluted sample aggressively to remove as much organic carbon as possible. This required adding concentrated nitric acid to the solution to make aqua regia and subject it to treatment with ultraviolet radiation in a UV light box for six hours. Even after treatment with UV and strong acids, it was necessary to add additional hydrogen peroxide and stir the sample vigorously over low heat. The incomplete destruction of organic matter indicated that some refractory P was excluded from our analysis, affecting the final yield. The phosphate analyzed in this study was soluble reactive phosphate (SRP), which includes orthophosphate (DIP) as well as some labile organic phosphate hydrolyzed by acids in solution. Once the removal of organics was sufficient, the pH of the solution was adjusted to 9 by adding NH4OH and an excess of ammonium molybdate was added to form a precipitate of ammonium phosphomolybdate (Fig. 2). This was filtered on a vacuum filter and dissolved in concentrated NH4OH and collected. The solution was acidified and an excess amount of a reagent containing Mg(NO3) was added. Concentrated NH4OH was slowly added until a white precipitate was formed, with excess added to complete the formation of the precipitate magnesium ammonium phosphate. This precipitate was filtered using a vacuum filter and dissolved in dilute nitric acid. The solution was brought to a pH of 5 using NH4OH and reduced to a small volume (15 mL). It was then sent down a cation exchange column to remove Mg2+ from the solution. Mg2+ inhibits precipitation of Ag3PO4 because it competes with Ag for PO4. Because this procedure required the solution to be moved to so many different containers and distilled, a significant amount of P is lost. This made it necessary to take an aliquot to determine P concentration on the ICP-MS prior to adding Ag+ to the solution. If too much Ag+ is added, silver metal is formed along with Ag3PO4. When the sample is combusted with graphite, some oxygen may react with metal silver to form silver oxide (Ag2O) and possibly fractionate the oxygen isotopes. Once Mg2+ was removed from solution, concentrated NH4OH was added to make the solution basic. Ag+ in solution was added in a ratio of 0.6 ml silver ammine solution per milligram of PO4 (ONeil et al, 1994). The sample was heated to no more than 60 C to evolve ammonia gas and Ag3PO4 was precipitated as the pH drops. This took about 4.5 hours, or until the pH dropped to about 7-7.5. The sample was filtered and dried in an oven overnight, then measured to determine the P yield. Once solid Ag3PO4 was obtained, enough graphite was added to the sample to completely react to form CO2 gas (ONeil et al, 1994). Graphite was added to Ag3PO4 in quartz tubes, which are heated to drive off excess water vapor and sealed under vacuum. The samples were combusted at 1200 C to produce CO2, which was subsequently measured on a Finnegan Delta S stable isotope ratio mass spectrometer.
10
Figure 1.2. Phosphorus Extraction Procedure
11
1.4. Results and Discussion 1.4.1. 18O of phosphate in fertilizers To test the hypothosis that the oxygen isotope 18O in phosphate can be used as a tracer in freshwaters, it was necessary to 1) determine the oxygen isotopic values of fertilizer phosphate, and 2) determine the oxygen isotopic values of DP in freshwater from both the polluted and unpolluted areas of the Everglades. Fertilizer phosphate is manufactured from marine apatite and should be enriched in the heavy oxygen isotope (18O). Our analysis of fertilizer phosphate show that fertilizer values are well within the 18Op range of marine phosphate and have a small standard deviation (Table 1.1). Table 1.1. Oxygen isotopic composition of phosphate in fertilizers. Fertilizer 18O () Esposito's Scotts EAA Kmart Sunniland Stay Green Peter's Professional 23.3 24.6 23.9 24.1 23.8 25.0 20.9
Average Commercial Fertilizer 23.7 1.3 The predicted 18O values of natural dissolved phosphate in the pristine areas of the Everglades are 18-21, based on equilibrium isotopic fractionation factors and the assumption that the 18O value of Everglades water is 3. The fertilizer 18O average of 23.7 1.3 indicates that fertilizer phosphate should be detectably different from microbially cycled phosphate (Figure 1.3). 1.4.2. 18O of dissolved phosphate The procedure developed to analyze the oxygen isotope composition of dissolved phosphate in freshwater (Fig. 2) was tested in the laboratory by spiking distilled water with potassium phosphate (KH2PO4). All samples were prepared and analyzed with rock standard NBS-120c of known oxygen isotopic composition (21.8 1.3, Blake). Results of this testing show that this method can quantitatively extract phosphate from water for isotopic analysis without fractionating the oxygen isotopes in phosphate (Table 2). The standard deviation of 0.6 indicates that the difference between fertilizer phosphate (18O = 23.7 1.3 ) and naturally cycled phosphate (predicted values of 18O = 18-21) can be resolved using this method.
12
Table 1.2. Results from laboratory experiments using DI water spiked with KH2PO4. Sample 18OP (corrected to standard NBS 120) PJ-12 15.7 PJ-13 15.6 NEWJS-3 16.1 PJ-14 16.2 PJ-19 15.5 PJ-21 16.9 PJ-29 16.8 PJ-30 16.7 PJ-31 15.8 PJ-32 16.1 average 16.1 0.6 Analyses of filtered water samples taken from the inlet and outlet of STA-1W, from Hillsborough Canal, from the interior of WCA-2A, and from the ENP (Figure 1.2) show that phosphate concentration ranged from ~74 ppb in the polluted area of the northern Everglades to below detections limits in the Everglades National Park (Table 1.3). Table 1.3. Phosphate concentrations (in ppb) at sample sites. Sample Location P concentration, ppb Sampling Date PW-1 STA-1W inflow 73.7 8/15/01 PW-2 STA-1W outflow 50.2 8/15/01 PW-5 ENP below detection 8/17/01 PW-7 Hillsborough canal 21.7 8/19/01 PW-9 WCA-2A interior 2.8 8/20/01 PW-11 WCA-2A interior 43.0 8/20/01 Using the procedure we developed (Fig. 1.2), we analyzed 3 replicates of PW-1, the sample taken from the inflow to STA-1W. This sample had the highest concentration of P due to its proximity to the EAA. This yielded a 18O value of 19.7 0.8. This value is less enriched than 18O values for fertilizer phosphate, falling within the predicted range of 18O values for natural DIP (Fig. 1.3). It would be useful to compare this data to oxygen isotopic values of phosphate in samples taken from the pristine Everglades. However, as shown in Table 3, the phosphate concentrations in these unpolluted areas (PW-5 and PW-9) are very low and we were unable to get sufficient phosphate to prepare for isotopic analysis.
13
The more depleted 18O values of dissolved phosphate extracted from the STA1W inflow canal (18O = 19.7), as compared to fertilizer phosphate (18O = 23.9), likely indicate a mixed signal, including both fertilizer phosphate (isotopically heavy), microbially remineralized phosphate (predicted to be isotopically light), and possibly some labile organic phosphate hydrolyzed by acids during sample processing. However, it is also possible that remineralization of phosphate occurs so rapidly due to increased microbial action that any fertilizer signature is erased. The latter is supported by studies by Paytan, et al. (2002), which indicate that while 18O of phosphate in aquatic systems reflect biological processes, the signature of the source DIP delivered into an ecosystem will be eliminated as P is cycled through the biomass. Thus, source contributions may be impossible to identify if the rate of isotopic exchange due to biological recycling is faster than the rate of input of a source with a unique 18O signature.
Everglades Natural Waters (predicted values) N. American Natural Waters (predicted values) EAA fertilizer 18 5 19.7 21
23.9 23.7
Fertilizer Range
STA1 Inflow
20
4
6
8
10
12
14 16
18
18
20 22 24
26
28
O
Figure 1.3. The isotopic composition of phosphate oxygen. Paytan, et al. (2002) presented results of experiments designed to measure the length of time it takes for phosphate in organic matter or DIP to achieve isotopic equilibrium with the surrounding water. Algae were grown in seawater tanks and fertilized with a food source having a distinct 18Op. Samples of DIP and algae were taken at regular intervals to analyze 18Op. With the addition of the fertilizer, DIP 18O values in the tanks dropped to that of the fertilizer, and algal 18Op decreased sharply due to uptake of fertilizer P. However, in all experiments the 18O of phosphate in organic matter and DIP rebounded within 30 hours back to previous values in equilibrium with the surrounding water. This appears to indicate that using 18O of phosphate as a tracer 14
will be difficult in any environment where rapid and intense recycling of nutrients is occurring, assuming equilibrium fractionation factors alone are responsible for isotopic exchange. These findings indicate that while the 18O of phosphate may be a good indicator of microbial activity and the extent of microbial recycling within a system, this parameter may not be a very good tracer in either nutrient limited or highly eutrophic systems such as the Florida Everglades, where we would expect 18Op to be in equilibrium with 18O of environmental water. The one situation where the source of DIP has been successfully traced is described in Blake (Blake 2001). Here, phosphate from a sewage contaminant plume was traced in a shallow groundwater aquifer. This was possible because of the high concentration of P in the contaminant plume coupled with low dissolved organic carbon concentrations in the aquifer. Blake states that the phosphate within the contaminant plume is not fully equilibrated with groundwater due to the lack of vigorous recycling found in nutrient limited environments. In addition, because freshwater in the Everglades is not very depleted in 18O with respect to seawater due to its proximity to the marine environment (Fig. 1.3), it would be useful to test this method elsewhere further inland where P loading is a problem. 18Ow values of Everglades waters collected in the summer were actually more enriched than we had initially assumed due to intensive evaporation (Table 1.4), which translate into even more enriched values for phosphate oxygen in pristine areas of the marsh (21-24). This suggests that the DP collected from the STA-1 inflow canal is not in isotopic equilibrium with environmental water and recycling had resulted in more depleted isotopic values than expected if equilibrium fractionation factors are solely responsible for oxygen isotopic exchange between water and phosphate. Thus, our limited data suggest that isotopic fractionation during microbial processing of P might be controlled by kinetic rather than thermodynamic equilibrium processes in the mash, contrary to what was observed in Paytan et al. (2002). However, more data are needed to resolve this issue. Table 1.4. Oxygen isotopic values of Everglades waters. Sample Location O18w () Sampling Date PW-1 STA-1W in -0.3 8/15/01 PW-2 STA-1W out 0.1 8/15/01 PW-5 ENP 0.1 8/17/01 PW-7 Hillsborough Canal -0.3 8/19/01 PW-9 WCA-2A 0.2 8/20/01 Our limited data indicate that despite the massive addition of fertilizer to the northern Everglades, at least part of the DIP has been processed by microbes due to the high concentration of DOC, which provides ample substrate to support the microbial pool. Unless there is some systematic variation in 18O of environmental water along a north-south transect in the Everglades, 18O of DIP should not vary as a function of differences in P concentration. Any deviation from this value could represent differences
15
in the sources of P and the rate at which P is recycled. DIP in the pristine Everglades should provide the end-member 18O values due to the nutrient limited nature of the system, which ensures rapid and efficient recycling of P. 1.5. Conclusion Using the method we developed to extract phosphate from freshwater for oxygen isotopic analysis, we successfully analyzed three replicate samples from the inflow to the Everglades restoration area. These analyses show that dissolved phosphate extracted from the STA1 inflow canal had 18O values of 19.7 0.8, clearly not as heavy as fertilizer phosphate (18O = 23.9). This suggests that biologically recycled DIP in the Everglades is more depleted in 18O than fertilizer PO43- and the DIP pool at this sampling site consisted of both biologically recycled P and fertilizer P. Alternatively, our limited data may indicate that remineralization of TDP occurs so rapidly due to increased microbial action that any fertilizer signature is erased. This study shows that the oxygen isotopic signature of phosphate is a potentially useful tool as an indicator of the degree of microbial cycling in freshwater ecosystems. In this particular wetland, the efficient recycling of phosphate by a large microbial pool makes using oxygen isotopes to determine the source of phosphate difficult. However, in order to conclusively determine whether or not the 18Op in the polluted areas of the marsh is isotopically enriched due to the presence of fertilizer phosphate, isotopic values of phosphate in pristine waters, the other end member, must be determined. To do this, a stronger pump must be used during sample collection so that more phosphate can be collected for analysis.
16
CHAPTER 2 DISTRIBUTION AND TURNOVER OF DOC IN NATURAL AND CONSTRUCTED WETLANDS IN THE FLORIDA EVERGLADES
2.1. Introduction Research on the carbon cycle is integral to our understanding of global climate change. Wetlands play a crucial role in carbon cycling, representing 15% of terrestrial organic matter (TOM) flux to the oceans, and are thus a large source of reduced carbon transported to the marine environment (Hedges et al. 1997). In addition, wetlands are often the primary source of humic substances to freshwater ecosystems (Ziegler and Fogel 2003). A large portion of dissolved organic matter (DOM) in wetlands is bioavailable and is thought to support elevated microbial activity (Bano et al. 1997; Zweifel 1999), with dissolved organic carbon (DOC) mineralization occurring predominantly by bacterial oxidation (Hansell et al. 1995). Cycling of DOC is linked to the cycling and bioavailability of phosphorus (P) and nitrogen (N) (Craft and Richardson 1998; Kalbitz and Geyer 2002) as well as bioavailability and transport of metals (Kinniburgh et al. 1999; Voelker and Kogut 2001; Tipping et al. 2002). Surface waters in the Florida Everglades are rich in DOC. Nutrient loading (particularly P) is a problem in the Florida Everglades, historically a nutrient-limited wetland. The runoff from the Everglades Agricultural Area (EAA) is high in inorganic P (on average 20 times higher than P-levels in the pristine Everglades), directly influencing water quality throughout the Everglades. Plants and algae endemic to the Everglades are adapted to low phosphorus conditions, with organic P accounting for the largest fraction of soil P in this area (Craft and Richardson 1998). Ecological effects of this nutrient enrichment have been well documented in the Everglades. Perhaps the most noticeable ecological change has been the vegetation shift from nutrient-limited sawgrasses to nutrient-adapted cattail stands in the more polluted areas of the marsh (Davis 1994). In addition, increases in net primary productivity (Craft and Richardson 1993), decomposition of detritus (Davis 1991), and organic soil accretion (Craft and Richardson 1993; Craft and Richardson 1998) have been documented. The decomposition of organic matter by microbes remineralizes DOC, making its nutrients bioavailable to plants and further increasing the production of organic matter. Determining the source and residence time of DOC in the Everglades will aid in
17
understanding the biogeochemical cycling of nutrients in impacted and non-impacted areas of the wetland and aid restoration efforts. Carbon isotopes can be used as tracers to determine the source of DOC in wetlands (Onstad et al. 2000; Kalbitz and Geyer 2002; Wang et al. 2002; Ziegler and Fogel 2003). Stable carbon isotopic signatures can distinguish between the two major metabolic pathways employed by plants during photosynthesis. The majority of terrestrial plants use the Calvin (C3) cycle to fix carbon, causing them to be depleted in the heavy 13C isotope, with 13C values ranging from 22 to 34 (Smith and Epstein 1971; Farquhar et al. 1989). In contrast, C4 plants fix carbon using the Hatch-Slack photosynthetic pathway and have 13C values ranging from 7 to 17 (Smith and Epstein 1971; Farquhar et al. 1989). Freshwater phytoplankton (algae) 13C values can be even more depleted than most C3 plants and range from 40 to 30 (Forsberg et al. 1993). Our study found average 13C values of 39.6 0.3 for microalgae and 28.4 0.8 for periphyton in the Northern Everglades. Bulk DOC and soil organic matter have 13C values that reflect their source material, and thus can be used to trace the origins of DOM in an ecosystem (Wang et al. 2002). Until the turn of the century, the Everglades was dominated by sawgrass (Cladium jamaicense) and other C3 wetland plants. This vegetation still dominates the non-impacted portion of the marsh, including Everglades National Park. However, agricultural runoff from the Everglades Agricultural Area (EAA) just south of Lake Okeechobee carries DOC from sugarcane (Saccharum officinarum), the primary crop grown in the EAA, as well as from historic wetland peats. Because sugarcane is a C4 plant, it has a distinct 13C signature and its relative contribution to the DOC pool can be assessed. Radiocarbon has been used to study the distribution and transport of marine DOC (Bauer et al. 1995; Mitra et al. 2000; Santschi et al. 2002) as well as degradation and cycling of freshwater DOC (Kalbitz and Geyer 2002; Wang et al. 2002). It is used in this study to distinguish between modern primary production and historic peat deposits, which have been accumulating over the past few thousand years (Stephens 1984). Historic peats contain significantly less 14C than modern primary production, which contains bomb radio-carbon derived from thermonuclear weapons testing in the 1950s and 1960s (Hedges et al. 1997; Wang et al. 1997). Thus 14C measurements are a means by which to quantify contribution to the DOC pool by old carbon sources versus modern carbon sources. In this study, we measured the stable and radiocarbon isotopic ratios of DOC, dissolved inorganic carbon (DIC) and particulate organic carbon (POC) in both natural and constructed wetlands in the Florida Everglades. Our objectives were (1) to examine the source and turnover rate of DOC in these wetland ecosystems; (2) to quantify the efficiency of carbon and P cycling in constructed vs. natural wetland; and (3) to determine if radiocarbon measurements could be effectively utilized to gauge the progress of ecosystem restoration in the Everglades. 2.1.2. Study Sites The Everglades region of Florida encompasses most of the southern Floridan peninsula and represents the largest freshwater wetland and subtropical ecosystem in the country. Prior to settlement, the hydrology of this region was controlled by seasonal
18
cycles in rainfall causing sheetflow from Lake Okeechobee and flooding low lying south Florida. To encourage settlement and provide agricultural lands, drainage and reclamation projects were instituted at the turn of the last century, emplacing canals and dikes to control water flow in this area. The early 1950s ushered in an era marked by improvement of existing water control projects and the designation of the Everglades Agricultural Area (EAA) in the upper Everglades along with Water Conservation Areas (WCAs) bordering the EAA (Figure 2.1).
Figure 2.1. Study Site
19
Figure 2.2. Study site detail of STA-1W. Today these Water Conservation Areas are used not only to store water but as buffers between the EAA and the more pristine Everglades National Park to the south. The Water Conservation Areas are the largest remnants of the original Everglades ecosystem, although water quality here has been adversely affected by nutrient rich agricultural runoff from the EAA. Stormwater Treatment Areas (i.e., constructed wetlands) have been established by the South Florida Water Management District (SWFMD) to treat water coming off the EAA and have proven useful in reducing P in effluent waters. Our study focused on the Water Conservation Area 2A (WCA-2A) and Stormwater Treatment Area 1W (STA-1W), previously the site of the Everglades Nutrient Removal (ENR) project. Samples for DOC and POC analysis were taken at the inlet (Post-BMP site) and the outlet (Post-STA site) of STA-1W (Figure 2.2). Several samples were also taken at the outlet of Cell 4, including DOC, POC, DIC, sediment, and plants. In addition, samples were collected from two test cells in the South Test Cell (Figure 2.2) and from the Hillsborough Canal at the north border of WCA-2A as shown in Figure 2.1. We also collected samples from unpolluted areas in the center of WCA-2A at a dock used by Duke University and in the Everglades National Park, where P concentrations were below detection limits.
2.2. Methods 2.2.1. Analytical methods Water samples for DOC were collected in 1 L Nalgene bottles by pumping water through a 1.0 micron filter and were immediately place on ice. Traditionally, DOC has
20
been operationally defined as organic compounds that can pass through a 0.2 m filter. Thus, the DOC defined here would include some larger organic molecules (in the size range of 0.2 1 m). DOC samples were prepared upon return to the lab following standard procedure (Peterson et al. 1994). 60 ml water samples were placed in centrifuge tubes, frozen, and subsequently freeze-dried to reduce to ~10-15 ml volume to concentrate the DOC. The water sample was then transferred to a precombusted (4 hours at 550 C) 9 mm O.D. Pyrex tube and was acidified with about 50 l of 6N HCl (until it no longer bubbles), and then completely freeze-dried. CuO, silver foil, and copper shots were added to the tube, which was sealed under vacuum and then combusted overnight at 580 C. The CO2 generated by this combustion was cryogenically extracted and its carbon isotopic ratios were measured on a Finnigan isotope ratio mass spectrometer (IRMS) at Florida State University. Organic carbon standards were prepared and measured alongside DOC samples. Radiocarbon content of the purified CO2 was analyzed at the National Ocean Sciences AMS Facility at the Woods Hole Oceanographic Institution. POC samples were collected on an in-line filter while collecting phosphate samples. Water was obtained using a small marine pump connected to a 12V battery and POC was collected on a 1.0 micron glass filter paper (Whatman). Upon returning to the lab, filter papers were dried in an oven at 60 C, then cut into small pieces and placed in a 6 mm O.D. Vycor tube and combusted with CuO, Cu and silver foil under vacuum at 875C for 2 hours to produce CO2. The CO2 was then purified cryogenically for isotopic measurements. Sediment samples were dried at 60 C and ground into a fine powder, which was subsequently treated with 10% HCl to remove carbonate. Plant samples were treated with 10% HCl, rinsed with distilled water, dried at 60 C, and then ground into a powder. Sediment samples were prepared as described and were analyzed at Isotope Services, Inc., Los Alamos, NM. Carbon isotope ratios of plant samples were measured on a new EA-IRMS (Finnigan MAT Delta Plus XP) at the Florida State University. Stable isotope data are reported relative to PDB standard as 13C = [(13C/12C)sample/(13C/12C)PDB 1] x 1000. Radiocarbon (14C) data are expressed in standard notation as 14C or radiocarbon date of years before present (years B.P.) (Stuiver and Polach 1977). The analytical precision (based on replicate analyses of standards processed with each batch of samples and on sample replicates) is 0.1 for 13 C analysis and better than 9 for 14C analysis. Total dissolved phosphate (TDP) was measured for each water sample on a Finnigan sector field ICP-MS. DOC concentration was determined from CO2 yield and volume of the water sample processed (30 to 60 ml). Average analytical precision for DOC concentration measurements was within 20%, with error most likely due to inaccurate volume measurements. 2.2.2. Computational methods for estimating the DOC turnover rate in constructed wetlands Turnover times and production of DOC were calculated in two hydrologically isolated wetland test cells using mass balance equations described below. The results are discussed in section 2.3.2.
21
CO2
DOC in
Bacteria
DOC out DOC
DIC Figure 2.3. Schematic diagram showing the major components of the DOC cycle in lined, hydrologically controlled test cells. The amount of DOC in the cell outflow is controlled by the balance between the amount of DOC flowing into and out of the cell and the decomposition and production of DOC within the cell (Figure 2.3). Assuming that the new DOC produced in the cell has the same 13C value as the vegetation growing in the cell, the dynamics of DOC cycling in the cell can be examined using a mass-balance model: [DOC]out = [DOC]in k . RTwater . [DOC]in + [DOC]new [DOC]13out = [DOC]13in k . RTwater . [DOC]13in + [DOC] 13new where (2.1) (2.2)
[DOC]out and [DOC]13out are the concentrations (mg/L) of total DOC and DOC13 in the outflow, respectively; [DOC]in and [DOC]13in are the concentrations (mg/L) of total DOC and DOC13 in the inflow, respectively; [DOC]new and [DOC] 13new are the net amounts (mg/L) of new DOC and DOC13 produced in the test cell (from the growth and decay of vegetation), respectively; k is the DOC turnover rate (day-1). RTwater is residence time of water in the cell (days).
Since 13C = (Rsample/RPDB 1 )x 1000, we have the following relations:
C13 [DOC]13out = [DOC]out 12 C + C13 out
22
= [DOC]out
13 C out + 1 R PDB 1000 = A [DOC] 13 out C out R PDB 1 + + 1 1000
(2.3)
where
+ 1 R PDB 1000 A= 13C out 1+ + 1 R PDB 1000
13C out
13 C in + 1 R PDB 1000 = B [DOC] [DOC]13in = [DOC]in 13 in C in R PDB 1 + + 1 1000
(2.4)
where
+ 1 R PDB 1000 B= 13C in 1+ + 1 R PDB 1000
13 C new 1000 + 1 R PDB = C [DOC] 13 new C new 1 + + 1 R PDB 1000
13C in
[DOC]13new = [DOC]new
(2.5)
where
+ 1 R PDB 1000 C= 13C new 1+ + 1 R PDB 1000
13C new
Substituting equations (2.3), (2.4) and (2.5) into equations (2.1) and (2.2), we get: [DOC]out = [DOC]in k . RTwater . [DOC]in + [DOC]new A . [DOC]out = B . [DOC]in k . RTwater . B . [DOC]in + C . [DOC]new Equations (2.6) and (2.7) can be solved simultaneously for k and [DOC]new. (2.6) (2.7)
23
(C - A) [DOC]out k= + 1 RTwater (B - C) [DOC]in
(2.8) (2.9)
[DOC]new = [DOC]out [DOC]in + k . RTwater . [DOC]in
2.3 Results and Discussion 2.3.1. Concentration and isotopic composition of organic carbon along a water quality gradient The concentration of DOC in samples collected in August 2001 generally decreased with distance from the EAA, with a maximum of 20.1 mg-C/L at the STA-1W inlet to a minimum of 5.9 mg-C/L in Everglades National Park (Figure 2.4). DOC concentrations in the impacted areas of the marsh were significantly higher than in Everglades National Park, indicating that the agricultural runoff from the EAA is a major source of DOC in this area. The average DOC concentration for samples collected in August 2001 was 14 mg-C/L (Figure 2.4), significantly lower than the average DOC concentration of 32 mg C/L for samples collected in December 1997 and May 1998 (Wang et al. 2002). This difference in DOC concentration is most likely a function of seasonality. Samples taken from the outflow of Cell 4 within STA-1W (Figure 2.5) indicate that there are significant seasonal variations in the flux of DOC. Over a 4 month period, DOC concentrations in this cell ranged from 14 mg C/L to 64 mg C/L (Figure 2.5). Recently, Gu et al. reported large temporal variation in DOC from Cell 4 inflow and outflow between 1994 and 1999 (Gu et al. in press). As shown in Figure 2.4, the DOC concentration at these sites generally mirrors the P concentration, which decreases along the water quality gradient. TDP concentrations were highest in proximity to the canals and input structures through which agricultural runoff flows, but decreased markedly with distance from those canals. This change in P concentration along the flow regime in the Everglades has been well documented (Vaithiyanathan and Richardson 1997) and has resulted in shifts in vegetation in the wetlands; the original sawgrass community in the wetlands adjacent to the canals and input structures has been outcompeted by cattails due to the large amount of bioavailable nutrients in the runoff waters in the canals (McCormick et al. 2001).
24
Enr inflow Enr outflow WCA-2A Canal Duke Site 1 ENP 0 20 40 mg-C/L, ug-P/L 60 80 TDP DOC
Figure 2.4. DOC and TDP concentrations from northern to southern Everglades.
70
DOC concentration (mg C/L)
60 50 40 30 20 10 0 9/20 10/4 10/18 11/1 11/15 11/29 12/13
Date
Figure 2.5. Variation in DOC content of STA-1W Cell 4 outflow over 4 months in 2001. Stable carbon isotope data from DOC (Figure 2.6a) reveal increasingly depleted values with distance from the EAA, reflecting the diminishing contribution of sugarcane (13C = -11.4 0.4) and the increasing contribution of wetland vegetation (13C = -27 1.6). Stable carbon isotope data from POC (Figure 2.6b) do not show a discernible trend, but reflect numbers consistent with the wetland vegetation, the peat, and the presence of algae on the filter. The latter results in very depleted 13C values for POC. The variation in 13C values in the two canal samples are due to the fact that samples were taken from a boat ramp with heavy traffic, which kicked up a large amount of sediment when the first sample was taken.
25
a) DOC
STA-1W inf low
b) POC
STA-1W inf low
Sample Site
Sample site
STA-1W outf low WCA-2A canal (1) WCA-2A canal (2) Duke site 1
STA-1W outf low cell 4 WCA-2A canal Duke Site 1
ENP
-15
-20 1 3 C ()
-25
-30
-15
-25 1 3 C ()
-35
Figure 2.6a) 13C of DOC and b) POC from northern to southern Everglades.
100 50 0 -50 25 130
>modern
>modern
C
14
-100 705 35 -150 -200 -250 -300 -350 ENR in ENR interior ENR out 1710 30 1860 30 1460 35
935 40
840 25
DOC POC DIC Periphyton
2010 35 1690 30
WCA-2A canal
Duke Site
ENP
Figure 2.7. 14C values of DOC, POC, and DIC in waters collected in August 2001. Periphyton samples collected in May 1998 and June 2002.
26
125 75 25 -25 DOC (2001) POC (2001) Retentate (May 1998) Permeate (May 1998) Retentate (Dec. 1997) Permeate (Dec. 1997)
C
-75
14
-125 -175 -225 -275 ENR in ENR out WCA-2A Duke Site canal ENP
Figure 2.8. Comparison of DOC and POC radiocarbon data from this study and Retentate and Permeate data from Wang et al. (2002).
14C values of DOC and POC in the Everglades ranged from +52 to -227 (Figure 2.7). Negative 14C values indicate a source of carbon much older than the 1950s, when thermonuclear testing released large amounts of bomb 14C into the atmosphere. Positive 14C values on the other hand indicate a significant amount of bomb 14C and therefore a modern carbon source. The very negative 14C values presented here correspond to radiocarbon dates as old as 2010 years B.P. and are found in areas near the EAA, indicating that the dominant source of DOC and POC in these samples is the historic peat deposits. This means that carbon previously sequestered in peat is being re-mobilized and introduced back into the carbon cycle. Release of this old carbon from peat deposits may result in re-mobilization of other nutrients and metals previously buried in peats. The most positive 14C values are found at the pristine site in Everglades National Park, indicating that the source of carbon here is modern, postbomb primary production coming from the uptake of modern atmospheric CO2 by plants. As shown in Figure 2.7, 14C values of DOC and POC are most negative in the inflow and outflow of the STA-1W and increase to the south approaching the more pristine interior of the WCA-2A and Everglades National Park (ENP). DOC 14C values are very close to modern at ENP (14C = -0.4). POC 14C values at both the Duke site in the interior of WCA-2A and ENP are modern, although DOC 14C values at these sites are not. This indicates that in situ decomposition of modern plant matter is not the only source of DOC at these sites; enough DOC is carried downstream from degradation of
27
historic peats in the EAA to prevent the DOC 14C values from being modern. Periphyton samples have the most positive 14C values, reflecting that their carbon source is modern, post-bomb carbon from the atmosphere and water. This data shows that the primary source of POC and DOC in the STA-1W and the WCA-2A canal is the degradation of historic peats in the EAA, whereas in the pristine areas, the decomposition of modern plants is the most significant POC source, with DOC representing a mix of modern and historic carbon sources. The positive 14C values of DIC at the Duke site and in the ENP indicate that dissolution of 14C-dead limestone is not a significant source of DIC in the area. Wang et al. (2002) suggested that the 14C content of DOC in wetlands could be a valuable indicator of the effectiveness of wetland restoration efforts. They presented a limited dataset that showed old radiocarbon ages for DOC from the STA-1W inlet and WCA-2A canal and modern radiocarbon ages for DOC in the interior of WCA-2A. This suggests that in impacted areas of the marsh, DOC is primarily coming from the degradation of historic peat deposits. They hypothesized that DOC in a pristine wetland should originate primarily from modern primary production within the wetland and therefore wetland restoration efforts should result in a rejuvenation of radiocarbon ages of DOC as a restored wetland moves toward its original or natural state. Our 14C dataset provides further evidence to support this hypothesis, as radiocarbon ages of DOC and POC become younger along the water quality gradient. Comparison of our data and data in Wang et al. (2002) shows a consistent pattern 14 in the C content of DOC along the water quality gradient in the northern Everglades (Figure 2.8). In Wang et al. 2002, permeate represents low molecular weight-DOC (LMW-DOC) with molecular weight of less than 1000 Dalton, and retentate is a mixture of high and low molecular weight-DOC. The age of the retentate does not decrease as drastically from the STA-1W to the WCA-2A canal as our bulk DOC results show. However, the May 1998 rententate 14C values go from being significantly negative in the WCA-2A canal to significantly positive at the Duke site. This drastic increase in 14C value is not present in our DOC data; in fact, 14C value of DOC at the canal site is more or less the same within the analytical precision as that found at the Duke site. Permeate, or LMW-DOC ages are consistently younger than DOC, POC, or rententate ages at a given site, which suggests that LMW-DOC has a shorter turnover time than these other compounds and is more microbially labile (Wang et al. 2002). However, our results show more significant downstream enrichment in 14C of POC as compared to DOC. That is, POC at sites further from agricultural runoff (i.e., the Duke site and the ENP site) has positive, or modern 14C values, whereas the DOC at these sites has negative, or old 14C values (Figure 2.8). This indicates that refractory DOC from historic peat deposits persists in the water column, contributing to the DOC pool along with DOC produced by the degradation of modern organic matter. POC, on the other hand, is more rapidly removed from the water column via settling. Thus, POC is not transported as far as DOC. Both datasets show decreasing age of organic carbon compounds with distance from agriculturally impacted areas, indicating that a significant source of organic carbon in impacted areas is the degradation of historic peat deposits in the EAA. Comparison of our radiocarbon data with data from other freshwater riverine ecosystems (Spiker and Rubin 1975; Hedges et al. 1986; Raymond and Bauer 2001)
28
suggests that a larger portion of Everglades DOC comes from pre-bomb sources than elsewhere on the eastern seaboard (Figure 2.9). Only the Hudson River has 14C values as depleted as the northern Everglades. This is most likely because the Hudson drains lands historically used for agriculture (Raymond and Bauer 2001). Even radiocarbon values for Everglades Natural Park were more depleted than the majority of the other aquatic ecosystems with available radiocarbon data, further suggesting the presence of refractory DOC from erosion of historic peats in the northern Everglades. The gradient in the concentration and 14C content of DOC along the flow regime in the Everglades indicates that old DOC (with negative 14C) from upstream must eventually be replaced by DOC from in situ modern primary production within the wetland, suggesting that this system acts as both a source and a sink for carbon. Using the approach taken by Raymond and Bauer (2001), and assuming the only significant DOC sources to the system are DOC coming from upstream and DOC produced by primary production within the system, an estimation of the percentage of DOC turnover can be made: % Turnover = (D14C U14C)/(-U14C + A14C) (2.10)
where D14C is the 14C measured at the downstream station (ENP), U14C is the 14C measured at the upstream station (ENR inflow), and A14C is the 14C content of autochthonous DOC produced by modern primary production. If we assume general value of ~90 for 14C of modern primary production within the wetland (Manning et al. 1990; Wang et al. 1997) and a value of -197 for 14C of DOC coming from upstream, equation 2.10 predicts ~60 - 70% turnover of DOC from the northern end to the southern end of the Everglades.
Northern Everglades Everglades National Park Parker River (MA) Amazon Hudson River (NY) York River (VA) West River (MD) Potomac River (DC)
-400
-200
14
0
200
400
C ()
Figure 2.9. Comparison of 14C values of DOC in various freshwater aquatic systems. Everglades data from this study. Parker, Amazon, Hudson, and York River data from Raymond and Bauer (2001). West and Potomac River data from Spiker and Rubin (1975). Additional Amazon River data from Hedges et al. (1986).
29
2.3.2. Concentration and stable carbon isotope composition of organic carbon in two constructed wetlands South Florida Water Management District (SFWMD) has set up a series of test cells - small constructed wetlands - within STA-1W to test nutrient removal technologies. Water flows from the main canal into a storage area in STA-1W and is pumped into the head cell from which it flows into each half acre cell (Figure 2.10). Each test cell is hydrologically isolated by a full liner to prevent seepage from the adjacent treatment wetlands, and to allow cellspecific control of water depth, hydraulic residence time and flow rate. Inflow water for the test cells is obtained from the surrounding treatment wetlands. Water is first pumped into an elevated storage cell (head cell) and then flows into a 30-inch feeder pipe and is delivered in parallel fashion to the test cells through 8inch lateral pipes, each fitted with one of several calibrated orifice caps. The rate of flow into each test cell is regulated by changing the orifice cap. Outflow from each test cell is controlled by an adjustable 90 v-notch weir. Raising or lowering the weir controls water depth within that cell (Chimney and Goforth 2001). Submerged aquatic vegetation (SAV)/Limerock nutrient removal technologies are being tested in cell 9, and cell 15 is a control. Water residence time was 7.5 days in cell 9 and 21 days in cell 15. Both test cells contain the plant species Typha spp., Sagittarria latifolia, and Hydrilla verticillata, but test cell 9 is predominantly Chara vulgaris, common stonewart, a member of the algae family associated with carbonate precipitation. SAV/Limerock works to reduce TDP because the vegetation takes up P while encouraging co-precipitation of P with CaCO3 due to elevation of pH related to photosynthetic reactions. At the outlet of the cell, a limerock berm removes particulate P and some DOP (Jorge 2002). These small well-controlled constructed wetlands also provide a field laboratory for the study of the cycling of DOC. Carbon and carbon isotopic analyses of biweekly DOC and POC samples collected from cells 9 and 15 for the period of 9/25/01 to 12/18/01 show that there are detectable differences in concentration and 13C of DOC between inflow and outflow as well as between cells, reflecting the turnover of residual DOC in the inflow and production of new DOC within the cell (Table 2.1).
Figure 2.10. Schematic diagram of SFWMDs test cell system within STA-1W. (Figure
courtesy of SFWMD)
30
Comparison of the average amounts of DOC in the inflow and outflow suggests that the test cells were generally a small net sink for DOC (Table 2.1). Assuming that the 13C of residual DOC from inflow had not been altered in the cell and the 13C of new DOC produced within a cell is the same as the weighted average 13C of vegetation in the cell (Table 2.2), we can estimate the relative amount of residual DOC in the outflow using a mass balance relationship:
out = residual * (1-f) + new * f
where out = 13C of DOC in outflow;
(2.11)
residual = 13C of residual DOC = 13C of DOC in inflow; f = fraction of new DOC produced in the cell; (1 f) = fraction of residual DOC new = 13C of new DOC produced within cell = 13C of vegetation in cell
Table 2.1. Concentration and 13C of DOC in test cell inflow and outflow. Date [DOC]in [DOC]out Net 13Cinflow 13Coutflow (mg/L) (mg/L) DOC () () change (mg/L) 9/25 25.1 28.8 -24.2 -24.8 Cell 9 10/26 26.7 26.1 -25.1 -24.5 11/8 44.2 27.6 -24.8 -24.7 12/18 30.0 20.2 -24.9 -24.9 Average 31.58.7 25.73.8 -24.80.4 -24.70.2 -5.8 9/25 25.1 27.6 -24.2 -25.0 Cell 15 10/26 26.7 23.7 -25.1 -26.6 11/8 44.2 20.6 -24.8 -25.3 12/18 30.0 25.4 -24.9 -25.2 Average 31.58.7 24.32.9 -24.80.4 -25.50.7 -7.2 Table 2.2. 13C of vegetation in test cells. Vegetation Chara Typha Sagitaria Latifolia Cell 9 () -18.8 -25.4 Cell 15 () -22.0 -26.7
31
Table 2.3. Estimated fraction of residual and new DOC in test cell outflow. Vegetation Average 13Cvegetation Average Cover f () 13Cinflow 13Coutflow Assumed () () 80% chara -20.1 -24.8 -24.7 .02 Cell 9 20% sagittaria 40% typha -22.6 -24.8 -24.7 .05 Cell 9 50% chara 10% sagittaria -25.8 -24.8 -25.5 .75 Cell 15 80% typha 20% chara -26.7 -24.8 -25.5 .37 Cell 15 100% typha
1-f .98 .95 .25 .63
Table 2.3 shows the estimated values of f and (1-f) for two different vegetation assemblages in each cell based on vegetation surveys taken by SFWMD in May of 1999 and in December of 1998 using the 13C values measured over the sampling period of 9/25/01 12/18/01. The estimated amount of new DOC produced in cell 9 accounted for about 1-2% of the total DOC in the outflow (Figure 2.11 and Table 2.3). That is, most of the DOC (>98%) in the outflow from cell 9 was residual DOC. The estimated amount of new DOC in cell 15 outflow were ~40 80% of the total DOC in the outflow (Figure 2.11 and Table 2.3), significantly higher than in cell 9. Thus the amount of residual DOC accounted for about 20 60% of the total DOC in cell 15 outflow.
Percentage of new DOC in outflow
80% 60% 40% 20% 0% Cell 9 Cell 15
Figure 2.11. The relative amount of new DOC in the test cell outflow. Bars indicate the range of variation calculated for two different vegetation assemblages in Table 2.2.
32
A.
50
B.
1 Net DOC production (mg C/L/day) 0.8 0.6 0.4 0.2 0
Cell 9 Cell 15 North Pacific
Turnover time (days)
40 30 20 10 0
Cell 9
Cell 15
Figure 2.12. (a) Turnover time for DOC in Cell 9 and Cell 15 compared to the North Pacific. (b) Comparison of net DOC production rate in the test cells, mg C/L/day. Using the equations (2.8) and (2.9) and the average concentrations and 13C of DOC measured over the sampling period (Table 2.1), we estimated the average DOC turnover time (=1/k) and production rate in cells 9 and 15 (Table 2.4, Figure 2.12). The estimated DOC turnover times were about 37-39 days in Cell 9 and about 2639 days in Cell 15 (Table 2.4, Figure 2.12a). These fairly rapid turnover times indicate that these freshwater wetlands contain a labile pool of DOC. However, these estimated DOC turnover times are longer than the water residence time in each cell, suggesting that most of the DOC will persist in the water column as the water flows out of the cell into another part of the ecosystem. The DOC turnover in these freshwater wetlands seems to be slower than that in marine environments where the estimated DOC turnover times were from 2-6 days in the North Pacific (Cherrier et al. 1996) to less than a day to over 200 days in the mid-Atlantic Bight (Hopkinson et al. 2002). The shorter turnover times of DOC in marine environments are likely due to limited nutrient supplies that necessitate faster DOC turnover. Results from these mesocosm experiments can be extrapolated over the larger wetland to indicate a high rate of carbon turnover in the Everglades which is also supported by the radiocarbon data presented in the previous section. Estimated net production of new DOC within the cells range from ~ 0.04 to ~0.89 mg/L/day (Figure 2.12b, Table 2.4). The net production of new DOC is higher in cell 15, the control, indicating that the SAV/limerock technology used in cell 9 reduces the DOC production in the cell.
33
Table 2.4. Estimated DOC turnover time and net DOC production in test cells. Vegetation 13Cvegetation Average Average Average Average Residence Assemblage [DOC] Time () 13Cinflo 13Coutflow [DOC] Inflow Outflow of water () w (days) () Cell 9 Cell 9 Cell 15 Cell 15 80% chara 20% sagittaria 40% typha 50% chara 10% sagittaria 80% typha 20% chara 100% typha -20.1 -22.6 -25.8 -26.7 -24.8 -24.8 -24.8 -24.8 -24.7 -24.7 -25.5 -25.5 31.5 31.5 31.5 31.5 25.7 25.7 24.3 24.3 7.5 7.5 21 21
Turnover time(days ) 37 34 26 39
Net DOC produced (mg/L/day ) 0.073 0.157 0.873 0.463
34
2.4. Conclusion 13C of DOC measured at sites along a water quality gradient in the Everglades reflects the decreasing influence of sugarcane agriculture with distance from the EAA. DOC concentration was greater in the agriculturally impacted areas than in the pristine areas of the marsh. This decrease in DOC concentration with distance from the EAA mirrored the decrease in phosphorus concentrations along the same gradient. In the areas that do not have a surplus of nutrients, carbon and phosphorus are efficiently cycled in situ. Our study shows that in pristine marsh areas in the Everglades, the 14C ages of DOC, POC and DIC are modern or near modern. In contrast, both DOC and POC in impacted areas have old 14C ages. These data indicate: (1) In pristine marsh areas both DOC and POC were produced primarily by in situ decay of modern organic matter. (2) In agriculturally impacted areas of the wetlands, the decomposition of historic peat deposits has contributed significantly to the DOC and POC pools. (3) Dissolution of 14C-dead limestone did not contribute significantly to the DIC pool. (4) DOC is transported farther away from its source than POC, as shown by negative (old) 14C values of DOC coexisting with positive (modern) 14C values of POC at two unpolluted sites in the Everglades. Our data show that 14C measurements can be a useful indicator of the progress of ecosystem restoration in the Everglades. If a wetland restoration project is successful, the 14C content of DOC should approach that of the present day atmosphere. Carbon turnover rates estimated using stable carbon isotopic measurements of DOC in two small constructed wetland test cells indicate active turnover of DOC within the cells. Our measurements of concentration and carbon isotopic composition of DOC also suggest that these test cells are overall net sinks of DOC. Continued monitoring of radiocarbon and stable carbon isotopes would be a useful tool to examine the progress of ecosystem restoration in the Everglades.
35
CHAPTER 3 SPECIATION OF TETRAVALENT METALS THORIUM, HAFNIUM, AND ZIRCONIUM BY CAPILLARY ELECTROPHORESIS INDUCTIVELY COUPLED PLANMA MASS SPECTROMETRY (CE-ICP-MS) AND EQUILIBRIUM DIALYSIS LIGAND EXCHANGE (EDLE).
3.1. Introduction The importance of dissolved organic carbon (DOC) in the complexation and transport of metals in aquatic systems is widely accepted and well studied. Marine DOC content varies from 0.5 to 1.2 mg/L (Choppin 1999). Freshwaters are more variable, with DOC values starting at 0.1 mg/L and going as high as 50 mg/L in organic rich waters; northern Everglades waters average 25-30 mg-C/L. Humic substances (HS) are estimated to make up a large portion of DOC and have been shown to be strong ligands that are very effective in complexing metals (Sonke and Salters 2004). Humic substances may be present as humic or fulvic acids in dissolved or colloidal form, and have been proven to influence sorption of trace metals to mineral surfaces in natural waters (Lenhart 1999; Artinger 2002; Reiller 2002). Humic and fulvic acids are polyelectrolytes consisting of 40-50% carbon and having a heterogeneous distribution of acid functional groups. Humic acids (HA) are operationally defined as humic substances soluble above a pH of 3, whereas fulvic acids (FA) are soluble over the whole pH range. These substances differ in their collection of functional groups as well, with FA having higher carboxyl contents than HA (Ritchie and Perdue 2003). Metal-humate interactions have been studied in relation to radionuclide mobility in risk assessment of uranium mining or nuclear waste repositories (Glaus 2000). However, much of the data in the literature has been collected under conditions of acidic pH, which do not represent natural groundwaters (neutral to alkaline pH). In addition, few studies have been carried out at trace metal concentrations. Most of these studies have involved the sorption of radionuclides onto organic colloids or humic-coated minerals in the presence or absence of HS (Lenhart 1999; Artinger 2002) and have shown sorption to be significantly enhanced in the presence of HA at low pH (Lenhart 1999). Metal binding of trivalent cations has been extensively studied, but less data exists for the actinides and tetravalent metals. Studies of trivalent metals have shown that even at low HS concentrations found in groundwater, metal-humate complexation takes precedence over hydrolysis and carbonate complexation (Glaus 1995). Concentration of tri- and tetravalent trace metals in groundwater has been found to be directly related to concentration of HS. It is thought that slow kinetics is at work here, with species in
36
slow equilibrium with one another. This makes metal-HS interactions important to account for in assessing transport of radionuclides in groundwater from nuclear repository sites. In addition, more data in the neutral to alkaline pH range for tri and tetravalent metals is needed to assess predictive speciation models. 2.1.2. Single site binding approach Stability constants describing proton and metal binding by humic substances has been described by a large range of models based on the assumption of 1 to 8 different binding sites (Sonke and Salters 2004). The most simple model, the single site or discrete ligand model simplifies the complexities represented by the existence of different types of binding sites on the HS molecule by assuming metals bind to one discrete site, of the carboxylic (COOH) type. The justification for this is that carboxylic groups are 80100% ionized in natural waters as opposed to phenolic groups, which remain protonated, and less likely to bind metals (Choppin 1999). The binding constants calculated using this type of model are referred to as conditional binding constants, as they are only valid at the experimental conditions under which the experiment analysis was performed (pH, temperature, IS). This model also lacks any consideration of electrostatic interactions between metal ions and acid functional groups on the HS molecule not directly bound to the metal. Polyelectrolyte models attempt to include these electrostatic, coulombic effects between HS and all ions, including H+ ions, which compete for metal ions for binding sites. These models take into account the charge neutralization effect that metal ions and protons impart to HS when they bind to HS, decreasing the overall affinity of HS to bind ions. One of the most widely used models is the NICA-Donnan model, which uses two binding sites to represent carboxyl acid functional groups and phenol acid functional groups. The NICA (non-ideal competitive adsorption) equation, which describes the heterogeneity of HS, is coupled with the Donnan model, which takes into account electrostatic affinity of HS based on ionic strength (Christl et al. 2001). Although the NICA-Donnan model had been shown to be very accurate and robust (Milne, 2003), conditional stability constants are still used to demonstrate the effects of pH and IS on the metal-humate system. I will be presenting conditional stability constants calculated using a one site model to explore how the tetravalent ions bind at various pH and comparing this to data in the literature and the REE data generated using the same methodology by Jeroen Sonke and myself. Conditional binding constants continue to be an important contribution to the literature because they are used to validate the polyelectrolyte models. 3.1.3. CE-ICP-MS Capillary electrophoresis coupled with invectively coupled plasma mass spectrometry (CE-ICP-MS) has shown to be an extremely useful tool in quantitative studies of trace metal-humate interactions. Low detection levels of ICP-MS allow experiments to be carried out under natural conditions of very low metal concentrations. This relatively new technique is still in development but has been used to quantitatively measure metal speciation between lanthanide REEs and HS (Sonke and Salters 2004). Previous studies focused on method development (Olesik 1995; Olesik 1998). The ligand competition approach used in Sonke and Salters, 2004 allows accurate
37
measurements to be taken despite the fact that CE is a disequilibrium speciation technique, where the original equilibrium of the system is disturbed to separate the species (M, L, and ML) in time and/or space in order to measure them. By using two strong ligands, HA or FA and EDTA, stable complexes are formed that are resistant to dissociation and can be measured accurately (Sonke and Salters 2004). Some of the initial problems associated with using capillary electrophoresis coupled with inductively coupled plasma mass spectrometry include instability in siphoning due to backpressure produced by the nebulizer causing laminar flow within the capillary (Caruso and Montes-Bayon 2003). Laminar flow produces band broadening, or increase in peak width, and thus a decrease in resolution (Sonke et al. 2003). There are several ways to avoid siphoning. Several researchers have had success with reducing the capillarys diameter. More commonly, a concentric sheath flow around the capillary delivered by syringe pump is tuned to counter backpressure created by nebulizer aspiration. Theoretically, it is possible to tune an experiment to 0 siphoning (Sonke et al. 2003). However, laminar flow can be used to decrease analysis times or aid in the detection of positive and negative ions (Olesik 1998; Sonke et al. 2003), albeit at the expense of resolution. Sonke et al., 2003 found that while laminar flow velocity was less influential in causing band broadening than analyte diffusion coefficient and capillary diameter, it was still a significant source of increase in peak width. Essentially, there is a trade-off between separation resolution, sensitivity, and speed of analysis, and each experiment will ultimately require optimization of the most critical parameter at the expense of the others. The benefits of CE-ICP-MS over traditional speciation methods include the speed of analysis, small sample volumes, the low detection limits of ICP-MS (ppb range), and the ability to monitor multiple elements and isotopes at once via ICP-MS. Other electrochemical methods such as ion-selective electrodes (ISE) and Anodic Stripping Voltommetry may only be applied to specific elements (Reiller 2005), while chromatographic based methods such as high performance liquid chromatography (HPLC) have on the order of 10 times lower resolution than CE. In addition, liquid chromatographic methods such as HPLC introduce organic solvents that are pyrolized by the argon plasma in the ICPMS and can clog the interface (Caruso and Montes-Bayon 2003). Ion exchange chromatography is also frequently used to study metal speciation. However, separation times can be long and depend highly on pH, limiting the pH range over which this method can be used. To validate CE-ICP-MS we performed equilibrium dialysis ligand exchange (EDLE) experiments at the same range of pH, IS, and metal concentrations as CE-ICPMS. EDLE is a speciation technique whereby the system is allowed to reach equilibrium without disturbance; effects of species dissociation caused by disequilibrium induced during CE-ICP-MS will be apparent by comparison. EDLE is a membrane technique commonly used to measure metal-DOM interactions, and was initially developed to measure radionuclide-DOM interactions (Glaus 1995; Hintelmann et al. 1997; Lead et al. 1998; Glaus 2000; Peters et al. 2001). It is based on the size differences between the two competing ligands, in this case soil HA (generally >2000 Da) and EDTA (<500 Da). The larger ligand is confined to one side of the membrame while the smaller ligand will be present on both sides.
38
3.1.4. Actinide chemistry It is well established that adsorption onto biogenic material is one of the primary mechanisms controlling trace metal concentration in seawater. Because of the release of over 2 * 1020 Bq of radioactivity into the atmosphere and oceans due to nuclear weapons testing since 1945, it is necessary to study the mobilization dynamics of actinides. It is estimated that approximately three quarters of this radioactivity has been sequestered in ocean bottom sediments (Choppin 2003). Because the actinides are redox sensitive, the potential for remobilization of radioactive elements, most significantly plutonium, is high. Pu in natural systems exists in oxidation states III, IV, V, and VI. The dominant oxidation state of Pu in oxic waters is Pu(V)O2+, while in reducing waters, Pu(III) is the dominant soluble state (Choppin et al. 1986; Choppin 1991). Several trivalent lanthanides can be used as Pu(III) analogs. Thorium is often used as an analog to study the complexation behavior of tetravalent Pu, which is dominated by hydrolysis in natural waters at neutral and alkaline pH. Although Th(IV) exhibits weaker hydrolysis than Pu(IV), complexation constants of Th(IV) can be multiplied by the ratio of the ionic radii of Th(IV)/(Pu(IV) to closer approximate those of Pu(IV). Dissolved Pu(IV) decreases with increased DOC content in oxic waters, indicating that Pu is either complexed or reduced to the tetravalent state and hydrolyzed, at which point it may be sorbed to colloidal or particulate matter (Choppin 2003). Pu(IV) and Th(IV) both migrate primarily as colloids. Many studies have examined thorium adsorption to particulate and colloidal matter using parameters such as thorium complexing capacity, or ThCC (Hirose, 1995). ThCC is defined as the moles of added Th4+ complexed per liter of sample in 0.1 mol/l HCl solution. This is measured by adding free Th4+ to POM filtered from seawater and measuring the absorbed and free Th4+. Studies of ThCC and POM in seawater suggest the presence of a strong organic ligand in oceanic particulate matter(Hirose and Tanoue 1999; Hirose and Tanoue 2001; Quigley 2001). Laboratory studies of complexation between thorium and cultured heterotrophic bacteria (Hirose and Tanoue 1999; Hirose and Tanoue 2001) in acidic media (pH=1) indicate that Th forms a 1:1 complex with bacteria yielding conditional stability constants of log k = 6.63 to 7.07. It is thought that this strong ligand found in bacteria is the same strong ligand controlling complexation in marine POM and DOM. Complexation is thought to occur either due to the introduction of intracellular or extracellular chelators in response to the presence of metals, or via complexation sites on or at the surface of the cell membrane. The presence of a strong ligand in marine microorganisms is the result of adaptation to low concentrations of bioavailable metals. Assuming this ligand is primarily composed of carboxylic functional groups, the logarithmic conditional stability constant for Th complexation with this strong ligand in seawater (pH 8.2, IS 0.7) is estimated to be 20.7, very close to that of Th-DTPA (21.1) (Hirose and Tanoue 2001). Another study found that addition of EDTA to Th-COM complex was unable to displace Th complexed with COM in artificial seawater (pH 8, IS 0.7), suggesting log Kc for the Th-COM complex is on par with that of the Th-EDTA complex (23.2)(Quigley 2001). One would thus expect high log Kc values for Th complexation with humic substances in freshwaters as well. Several studies have examined Th and REE binding with different size fractions of humic substances to gain information regarding actinide dynamics at nuclear waste repository sites. Results of these vary. Casartelli and Miekeley (2003) used the highly
39
weathered environment Morro de Ferro in Brazil as an analogue for a nuclear waste repository. They found that Th and REE were highly labile and had been assimilated by plants, via the incorporation of humic complexes formed in soil water due to the absence of competing ligands in this highly weathered environment. Size exclusion chromatography (SEC) was coupled to ICP-MS to determine fractions of Th and REE in different HS size fractions. The largest proportion of Th and REEs were found in the high molecular mass (>10,000) fraction of the HS, supporting the widely held view that thorium is primarily associated with material. colloidal In another study, Wu, et al. (2004) found most of the Th to be associated with the 500-1000 MW fraction of HS from streams in south-central Ontario, Canada. This study found metal distribution among molecular weight fractions of HS closely related to binding strength; metals with high binding affinity were associated with high MW fractions, while those low binding affinity were associated with low MW fractions. For example, Th was associated with lower MW fractions than Fe, and higher MW fractions than U. In a study by Zhang, et al. (1997), gel permeation fractionation was used to examine different actinide (Th, U, Np, Pu, Am) interactions with humics from the Needles Eye in Scotland, a natural analog site for radioactive waste repositories, Thorium was the most strongly associated with the high MW (~49,000 Da) fraction of HS. Am(III) and Pu(IV) were also associated with high MW fractions, but were more evenly distributed among the whole size range. Np(V), Np(VI) and U were associated with smaller MW fractions. If metal binding strength is correlated with size distribution, then we would expect Th to have some of the highest conditional binding constants among the actinides, possibly followed by Am(III) and Pu(IV). While studies such as those described above do not yield actual log K values, they complement studies measuring log Kc values such as ours in that they give some indication of how thorium-humic interactions relate to the other actinides. 3.2. Methods and Materials Humic acid standards were obtained from the International Humic Substances Society and were dissolved in water with the addition of a small amount of 1N NaOH to make stock solutions of 1000 mg/L. Reagents of the highest purity available were obtained through Johnson-Matthey. These included Na2H2EDTA, MES, TRIS and NaNO3. ICP standard solutions (100.05 ppm in 2% HNO3) from High Purity Standards, Inc. were used for metal (Th, Hf, Zr, REE) spikes. Milli-Q water was used in the making of all reagents. Spectra/Por 6 regenerated cellulose membranes with a MWCO of 1000 were used in EDLE experiments. Prior to using the membranes, they were cut into 10 cm lengths and soaked overnight in Milli-Q water, as per the manufacturers instructions. A chelating rinse was used to remove possible metal contaminants in some experiments. Membranes were sealed by tying knots on one side of the tubing and using nylon membrane closures on the other side. 3.2.1. CE-ICP-MS Capillary electrophoresis can separate metal species of different charges and sizes by application of a voltage gradient during the transport of metal species over the length
40
of the capillary. EDTA and HA are strong ligands that form complexes stable over the time scale of the experiments and additionally prevent metals from interacting with the silanol groups that line the capillary wall. Metal complexation with the capillary wall can also be avoided by running experiments at low pH and/or high ionic strength so that there are enough cations in solution to react with the silanol groups. Species are separated based on their difference in electrophoretic mobility. The source (inlet of capillary) was negative potential, while the capillary outlet was kept at ground potential. The net flow for all species, even the positively charged, in the capillary was from the inlet to the outlet due to siphoning and all species were eluted. The neutral (MEDTA0) or negative (MEDTA-) complex was eluted first, followed by the positive MHA+ complex.
Figure 3.1. Electropherogram showing separation of TmEDTA- and TmHA2+ species. Figure 3.1 shows an electropherogram produced by a CE-ICP-MS experiment. Quantitative determination of binding constants is done by integration of the peak areas of both metal species and determining the percentage of metal bound to each ligand. This assumes that the peak areas represent the actual partitioning of the metal into the two species, and that these peak areas are not affected by disequilibrium effects resulting from the CE separation. Metal dissociation due to non-equilibrium effects has been studied and modeled (Sonke and Salters 2004). Results of these studies indicate that disequilibrium will not significantly effect the concentrations of the two species during the time scale of a ligand competition experiment in which two strong ligands such as HA and EDTA are used. Using a one-site binding model, stability constants for the tetravalent ions of Zr4+, Hf4+, and Th4+ can be calculated the same way as stability constants for trivalent REEs (Sonke, 2003). M + HA = MHA
4+ 3+
[MHA3+ ] with KMHA = [M 4 + ][HA- ]
(3.1)
Where [MHA3+] is the molar concentration of metal bound to humic acid, and [M4+] and [HA-] are the molar concentrations of free metal and humic acid, respectively.
41
MEDTA + HA- = MHA3+ + EDTA4with Kexch =
(3.2)
[MHA ][EDTA ] [MEDTA][HA- ]
3+
4+
[MHA3+] and [MEDTA] are measured. Free [HA-] is determined by equations 3.3-3.5 and free [EDTA4-] is determined by equation 3.6. Protonation of HA is described by: H+ + HA- = HHA, where KH1 =
[HHA] [H + ][HA - ]
(3.3)
The total HA is the sum of free HA, protonated HA, and metal bound HA, given by: HAtot = [HA-]+[HHA]+[MHA3+] [COOH ]tot [MHA 3+ ] [HA ] = 1 + K H 1[H + ]
-
(3.4) (3.5)
where [COOH]tot is the molar concentration of carboxylic binding sites on the humic acid, and KH1[H+] is the protonation constant for HA+, both given in table 3.1, and [H+] is the proton concentration. Free [EDTA4+] is determined by: [EDTA4-] = (EDTA tot [MEDTA]) , where =
1 1+
[H]
i =1
n
i
i
H
(3.6)
The parameter describes the stepwise binding of protons to EDTA. The values for log iH are found in table 3.2. From there we can substitute our calculated values into equation 3.2 and solve for log Kexch. This value is used solve the equation (using literature values for log KMEDTA, table 3.2): log KMHA = log Kexch log KMEDTA (3.5)
The justification for using this model is that carboxylic groups account for the majority of binding sites on HA and are less likely to be protonated at pH<7 than phenolic groups (Stamberg et al. 2003). This is an assumption commonly made for single site binding models (Ritchie and Perdue 2003; Stamberg et al. 2003; Sonke and Salters 2004). Additionally, a recent X-ray photoelectron spectroscopy study shows that Th is predominantly bound to carboxylic groups of humic acid (Schild and Marquardt 2000). HA and EDTA are strong ligands, therefore there is no free metal ion in solution. In addition, at the low concentrations of Th used in this experiment, free metal should be too scarce to form hydroxides (Zhang et al. 1997). Tetravalent metals are known to
42
hydrolyze in the absence of competing ligands (Moulin et al. 2001; Neck and Kim 2001; Bentouhami et al. 2004), and in interpreting our results we will explore this possibility. Table 3.1. Selected characteristics of HA. Mw Mn KDa KDa Humic Substance
Elliot Soil HAb 3.8 7.9 b Pahokee Peat HA 7.3 14.9 Summit Hill Soil HAc 3.65 7.76 d Leonardite HA NA 3.98 Suwannee River NOMe NA 2.3 a IHSS and (Ritchie and Perdue 2003) b Molecular weight data from (Phillips and Olesik 2003) c Molecular weight data from (Hur and Schlautman 2003) d Molecular weight data from (Karanfil et al. 1996) e Molecular weight data from (Meier et al. 1999)
Carboxyl % Ca content a (mol/kg C) 58.13 8.28 56.37 9.01 54.00 7.14 63.81 7.46 52.47 9.85
Protonation Constant a Log KA1 at 0.1 IS 4.36 4.22 4.47 4.59 3.94
Table 3.2. Summary of Binding constants used. Reaction Log at 0.1 IS H+ + HA- = HHA See table H+ + EDTA4- = HEDTA310.24 2H+ + EDTA4- = H2EDTA216.25 + 43H + EDTA = H3EDTA 19.05 + 44H + EDTA = H4EDTA 21.54 Th4+ + EDTA4- = ThEDTA 23.19 4+ 4Hf + EDTA = HfEDTA 30.15 Zr4+ + EDTA4- = ZrEDTA 29.25 Table 3.3. Hydrolysis constants for tetravalent metals. Species Log at 0.1 IS Reference 3+ Th(OH) -3.36 (Baes and Mesmer 1976) Th(OH)22+ -8.36 (Bentouhami et al. 2004) Th(OH)3+ -11.63 (Bentouhami et al. 2004) Th(OH)4 -18.24 (Bentouhami et al. 2004) Th2(OH)7+ -24.32 (Bentouhami et al. 2004) Hf(OH)3+ -0.41 (Baes and Mesmer 1976) Hf(OH)4 -16.69 (Baes and Mesmer 1976) Zr(OH)3+ 0.14 (Baes and Mesmer 1976) Zr(OH)4 -15.49 (Baes and Mesmer 1976)
Reference (Ritchie and Perdue 2003) (Morel and Hering 1993) (Morel and Hering 1993) (Morel and Hering 1993) (Morel and Hering 1993) (Martell and Smith 1998) (Martell and Smith 1998) (Martell and Smith 1998)
43
Experimental design The coupling CE apparatus to the ICP-MS was tested and used to study REEhumic interactions by Jeroen Sonke (Sonke 2003). The setup consists of the sample injection system, the capillary tubing, the sheath flow interface, and the nebulizer/spray chamber assembly. Sample and buffer are pumped to a 6-port injection valve (Valco Instruments, Co., Houston, TX) via peristaltic pump. The injection valve is controlled by a Microneb 200 control unit (CETAC, Omaha, NE). Sample or buffer flows out of the injection valve into a microcross (Upchurch Scientific, Oak Harbor, WA) into which one end of the capillary (48 m I.D., 124 m O.D., Polymicro Technologies, Phoenix, AZ) is inserted. The capillary is guided through a second microcross and into PFA tubing containing the sheath flow of 1 ppm In in 2% HNO3, which is pumped into the microcross via a syringe pump (Harvard Apparatus 22). This tubing containing sheath flow and capillary is inserted into a Glass Expansion micromist nebulizer (100 L/min). The end of the capillary is positioned approximately 0.5 cm from the tip of the nebulizer. The nebulizer is inserted into a specially machined 20 ml teflon spray chamber attached to the ICP-MS. A Finnigan MAT ELEMENT, high resolution sector ICP-MS was used as the detector. Experimental conditions are listed in Table 3.4. Sheath flow provided an internal standard (In) to monitor nebulizer and plasma stability and sensitivity (typically 300,000 to 350,000 cps per1 ppb In at 20 L/min). Na+ in sample and buffer solutions provided a standard to monitor flow inside the capillary. Laminar flow was controlled by the length of the capillary, and tuned daily to maximize separation efficiency by adjusting the back pressure valve and peristaltic pump flow (typically 90-250 L/min). Siphoning was measured by injecting samples for 10 seconds without applying voltage. This also gave a measure of peak area and thus concentration. Once flow had been optimized for maximum resolution of MHA and MEDTA species, the sample was injected and 10-15 kV potential difference was applied to the capillary. The source (inlet of capillary) was negative potential, while the capillary at the nebulizer was at ground potential. Sample preparation Buffers were prepared for pH 3.5, 4, 5, 6, 7 and 8. Buffer composition was 0.1 M NaNO3 and 0.01 MES for pH 3.5-7 or TRIS for pH 8. Humic acid, EDTA, and metals (Th, Hf, Zr, REE) were added from stock solutions to 60 ml poly bottles. Concentrations of stock solutions, buffers, and samples are listed in table x. Bottles were then filled with 40 ml of buffer, shaken, and allowed to equilibrate for at least 2 days. The pH of buffers and samples were measured by a Denver Instruments pH meter and combination electrode after calibration against NIST standards (pH 4, 7, and 10).
44
Table 3.4. CE-ICP-MS run parameters ICP-MS Nebulizer Gas Auxiliary Gas Cool Gas Rf Power Sensitivity CE Sheath Flow Buffer/sample flow rate Capillary length Capillary diameter High Voltage Background electrolyte Injection time
1.2-1.8 L/min 0.88-1.02 L/min 13 L/min 1250 W 300-350 kcps for 1 ppb In 20-25 L/min 90-200 L/min 45-60 cm 48 m I.D., 110 m O.D. 0-15 kV 0.1 NaNO3 10 sec
Table 3.5. Stock solution and sample solution compositions Stock solutions Concentration HA EDTA stock A EDTA stock B Metals NaNO3 Sample solutions HA (COOH-) EDTA for Th samples EDTA for Hf and Zr samples EDTA for REE samples M NaNO3 MES, TRIS 1000 mg/L 10-4 M 10-2 M 10 ppm 0.5 M 10 mg/L (48 M) 8.0 M 0.4 M 0.4-4.0 M 10 ppb 0.1 M 0.01 M
3.2.2. EDLE An equilibrium dialysis ligand exchange method was used to measure MHA binding constants of tetravalent metals for comparison to CE-ICP-MS. Equilibrium dialysis ligand exchange is commonly used to measure metal-DOM interactions. Initially it was designed to measure radionuclide-DOM interactions (Glaus 1995) and was later modified to include leakage of DOM across the membrane (Glaus 2000). It has recently been used to measure conditional binding constants for Hg-DOM interactions (Haitzer et al. 2002; Haitzer et al. 2003). We have adapted this method to include information
45
regarding the concentration and availability of binding sites of soil and peat humic acids used in this study.
MHA complexes MEDTA complexes
Figure 3.2. Schematic representation of experimental setup for EDLE. MEDTA complexes are able to penetrate and diffuse back and forth across the membrane, but MHA complexes are too large and remain inside the outer compartment. At dialysis equilibrium, MEDTAout = MEDTAin. In an equilibrium dialysis exchange experiment the distribution of the metal between the inner and outer compartment of the dialysis bag is measured. The outer compartment of the bag contains both MHA and MEDTA complexes, whereas the inner compartment only contains MEDTA complexes. It is initially assumed that MHA complexes are too large to diffuse across the 1000 Da membrane. The metal concentration of the inner compartment is the concentration of metal complexed by EDTA, while subtracting the inner compartment concentration from the outer compartment concentration yields the concentration of metal complexed by HA. EDTA was chosen as the reference ligand because - it competes with HA in the range of pH (37) explored in this experiment; - it does not form ternary complexes (HA-M-EDTA) (Glaus 1995; Haitzer et al. 2002), binding constants of M-EDTA complexes are readily available in the literature. Binding constants were calculated following Glaus. To reiterate the definition of the metal-humic binding constant in equation 3.1: [MHA3+ ] with KMHA = (3.6) M4+ + HA- = MHA3+ [M 4 + ][HA- ] Where [MHA3+] is the molar concentration of metal bound to humic acid, and [M4+] and [HA-] are the molar concentrations of free metal and humic acid, respectively. To calculate KMHA, the concentration of [M4+] and [HA-] on the inside and the outside of the membrane must be determined and described in a series of mass balance equations: [M4+]tot, in = [M 4+ ] + [MEDTA i ] + [MHA 3+ ]in
i =1 n
(3.7)
46
[M ]tot, out = [M ] + [MEDTA i ] + [MHA 3+ ]out
4+
4+ i =1
n
(3.8)
The total metal concentrations on the inside and the outside of the membrane is the sum of the free metal, the metal complexed by EDTA, and the metal complexed by HA in each compartment. We can write similar equations for the inner and outer compartments summarizing the total HA concentrations: [HA-]tot,in = [HA-]in + [MHA3+]in [HA-]tot,out = [HA-]out + [MHA3+]out (3.9) (3.10)
Where [HA+]in/out is the concentration of unbound HA and [MHA3+]in/out is the concentration of HA bound to metal on the inside and outside of the membrane. To calculate KMHA, Glaus introduces the parameter Q, which describes the partitioning of the metal between the inner and outer compartment. In our case, [M4+]tot,in represents the concentration of MEDTA, which is equally distributed between both sides of the membrane. [M4+]tot,out is the sum of both the MEDTA and MHA complexes. The concentration of [MHA3+], then, is [M4+]tot,out minus [M4+]tot,in, the MEDTA component:
Q=
[M 4+ ]tot ,out [M 4+ ]tot ,in [M 4+ ]tot ,in
[MHA 3+ ] = [MEDTA]
(3.11)
Next, equations 3.7 and 3.8 can be substituted into equation 3.11 which becomes:
Q=
[MHA 3+ ]out [MHA 3+ ]in [M ] + [MEDTA i ] + [MHA ]in
4+ 3+ i =1 n
(3.12)
Now the expressions for [MHA3+]in/out can be replaced using the relationship [MHA3+] = KMHA[M4+][HA+] from equation 3.6:
Q=
K MHA [M 4+ ][HA ]out K MHA [M 4+ ][HA - ]in [M 4+ ] + [MEDTA i ] + K MHA [M 4+ ][HA - ]in
i =1 n
=
K MHA ([HA ]out [HA ]in ) M + K MHA [HA + ]in
(3.13)
The parameter m is a measure of the complexation of M by the reference ligand, EDTA. Under the experimental conditions of low to neutral pH and [M]tot<<[EDTA]tot, m can be defined by (Glaus 2000; Haitzer et al. 2003):
47
M =
[M 4+ ] + [MEDTA i ] [M 4+ ]
i
n
i =1
n EDTA tot M = 1 + i m H i =1 1 + [H]i i j =1
n H i =1
(3.14)
where [H]i i = K HEDTA 2 [H + ] + K HEDTA3 [H + ]2 + K HEDTA4 [H + ]3 + K HEDTA5 [H + ] 4 where i M is the binding constant for the MEDTA complex and j H and KHEDTAi are the binding constants describing the stepwise binding of H+ to EDTA (table 3.2). By inserting this expression for M into equation 3.13 and solving for KMHA we obtain: KMHA =
[HA ]
-
tot ,out
Q m , HA - tot ,in (Q + 1)
[
]
(3.15)
where Q is determined experimentally, m is calculated from literature values and the experimentally defined EDTA concentration, and [HA-] concentrations are determined sprectrophotometrically. To make the equations of Glaus more specific to the particular HA- used in the experiment, the concentration of carboxyl groups [COOH-] specific to each HA (Table 3.1) is used instead of HA concentration in our modified equation. To express [HA-]out/in in terms of carboxyl group concentration, [HA]in = [COOH ]in , 1 + K H 1[H + ] (3.16)
[COOH ]out [HA]out= 1 + K H 1 [H]+ where K H 1 [H + ] is the protonation constant specific to each HA.
(3.17)
Sample preparation Dialysis experiments were performed in 50 ml Fisherbrand centrifuge tubes. Centrifuge tubes were filled with 40 ml buffer (0.1 M NaNO3, 0.01 M MES). To this buffer, 400 l stock HA solution, an optimized amount of EDTA (50 l EDTA stock solution B for Th, 200 l EDTA stock solution A for Hf and Zr), and 50 l of 10 ppm metal stock were added. Table x shows the buffer and sample concentrations at different pH, which were the same as used for CE-ICP-MS experiments. A dialysis bag containing 4 ml buffer was immersed into the solution in the tubes. Samples were placed in a shaker
48
bath for at least 48 hours in order to reach equilibrium. Each sample was run alongside a control containing only EDTA and a metal spike to ensure the dialysis equilibrium of MEDTA complexes had been reached.
Analytical The metal concentrations were determined for the inside and outside solutions on a Finnigan MAT ELEMENT, high resolution sector ICP-MS. DOC concentrations for the inside and outside solutions were determined using a Spec 20 spectrophotometer to measure the absorbance at 340 nm. The initial and final pH of outside solutions were measured by a Denver Instruments pH meter and combination electrode after calibration against NIST standards (pH 4, 7, and 10). 3.3. Results and Discussion 3.3.1. REEs Binding constants were calculated for four kinds of humic acid and for Suwannee River NOM (Figure 3.1). The variation in log Kc among these different HS was found to be relatively small, varying only by approximately two orders of magnitude. Suwannee River NOM was on the lower end of the scale.
log Kc of various LnHS complexes
20 19 18 17 16 15 14 13 12 11 10 9
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu LnEDTA EHA (soil) LHA (soil) PHA (peat) SHHA (soil) SRNOM (aquatic)
Figure 3.3. Conditional binding constants, log Kc,LnHA as a function of REE, and HS, at fixed pH (7) and fixed IS (0.1 M), as calculated from data from CE-ICP-MS.
3.3.2. Tetravalent metals Conditional stability constants were calculated for Th, Hf, and Zr and humic complexes (EHA) at pH 3.5, 4.7, 5.9 (Figure 3.2). Overall, the conditional stability constants of the tetravalent metal-HA complexes are several orders of magnitude higher than the lanthanide-HA complexes, with Hf having the greatest log Kc values. The
log Kc,LnHS
49
pH variation in log Kc,MEHA
25 23 21
Th Hf Zr REE
log Kc,MHA
19 17 15 13 11 9 3 4 5 6 7
pH
Figure 3.4. Conditional binding constants, log Kc,MEHA as a function of pH and metal, at fixed IS (0.1 M), as calculated from data from CE-ICP-MS.
a) Th-HA binding constants
21 20 19
log Kc,ThHA
18 17 16 15 14 13 3 4 5 6 Th-CE Th-EDLE
pH
7
c) Zr-HA binding constants
25 24 23
log Kc,ZrHA
b) Hf-HA binding constants
26 25 24 23 22 21 20 19 18 17 3 4 5 pH 6
21 20 19 18 17 3 4 5
pH
Zr-CE Zr-EDLE
log K c,HfHA
22
Hf-CE Hf-EDLE
6
7
7
Fig 3.5a-c. Conditional binding constant compilations for the tetravalent metals Th, Hf, and Zr, as a function of pH and speciation method, IS 0.1 M.
50
tetravalent metals have a larger charge and generally smaller atomic radii than the lanthanides. This makes for a higher charge density, and could explain the reason why thorium, halfnium, and zirconium bind more strongly to humic substances than the lanthanides. Figures 3.5a, b, and c show the compilation of log Kc,MHA data for the tetravalent metals Th, Hf, and Zr. Both CE-ICP-MS and EDLE consistantly produced very similar conditional binding constants, indicating that both methods are appropriate for measuring metal-humic interactions at pH<6. We were able to obtain conditional binding constants at pH 7 with EDLE, while metal complexes, especially Th, stuck to tygon tubing above pH 7 in CE-ICP-MS experiments. While our yields with EDLE were less than 100%, binding constants were calculated using the ratios of inside concentration to outside concentration, assuming equal loss of MHA and MEDTA complexes to the membrane or vessel walls.
Metal-tubing interactions Th is known to stick to tubing in experimental set-ups (Macka et al. 1999). We tested our experimental setup at a range of pHs to determine whether metal complexes were sticking to the tubing supplying sample to the capillary or to the capillary itself. We first cleaned the capillary with 2% nitric acid until Th4+ concentrations in the eluted solution were at background values. We then sent solution containing ThHA and ThEDTA complexes through the tygon tubing bypassing the capillary and eluting into a waste container. Finally, 2% nitric acid was sent through the same tygon tubing and through the clean capillary. Thorium was present in the eluted solution for ThHA complexes, but not for ThEDTA complexes. This is most likely because the walls of the tubing compete with HA for Th4+, and Th4+ can become dissociated from the HA complex and bind to the wall. The binding constant for ThEDTA is 12 orders of magnitude higher than that predicted for ThHA complexes, so Th4+ does not become dissociated from EDTA even in the presence of a competing negatively charged surface (the tubing walls). Therefore, we argue that when both ThEDTA and ThHA are in solution, all metal will be bound to either of the two ligands and not to the tubing walls. The reproducibility of our results indicate that this is a valid assumption (Tables 3.6, 3.7, 3.8). To test if Th complexes stuck to the capillary, solution containing Th complexes were sent through a clean capillary and eluted. Then 2% nitric acid was sent through the capillary from clean tygon tubing. No significant amount of thorium was eluted with the acid. Less than 1% of the Thorium complexes stuck to the capillary. The effects of M4+ hydrolysis are theoretically not significant over the pH range of the experiment in the presence of two strong ligands. However, when solutions containing free Hf and Zr metal and buffer at pH 4 were run through the capillary, a peak approximately half the size of either the MHA or MEDTA peak was observed, most likely representing a hydroxide complex because of the lack of competing ligands in solution. If the sample was run under voltage, a small peak appeared before the large peak, possibly indicating free metal. The large peak indicates that Hf and Zr are being complexed to some extent, that not all metal in solution is free metal.
51
Table 3.6. Analytical data for complexation studies of Th4+ with humic acid in 0.1 M NaNO3. Concentration of Th4+ was 50 nM.
Sample Th1 Th2 Th3 Th4 Th1-1 Th2-1 Th3-1 Th4-1 Th5-1 1C 2c Th1 Th1 Th2 Th4 10B Th1 Th2 4B 11B Method CE CE CE CE CE CE CE CE CE EDLE EDLE CE CE CE CE EDLE CE CE EDLE EDLE HA EHA EHA EHA EHA PHA PHA PHA PHA PHA EHA EHA EHA EHA EHA EHA EHA EHA EHA EHA EHA HA (mg/L) 9.91 9.83 9.94 9.95 9.48 9.44 9.44 18.38 36.64 9.57 9.60 10.02 9.96 9.97 9.50 9.5 9.9 9.9 9.7 9.7 [COOH] (mol/L) 4.77E-05 4.73E-05 4.78E-05 4.79E-05 4.81E-05 4.80E-05 4.79E-05 9.33E-05 1.86E-04 4.60E-05 4.62E-05 4.82E-05 4.80E-05 4.80E-05 4.57E-05 4.59E-05 4.78E-05 4.74E-05 4.68E-05 4.68E-05 [EDTA] (mol/L) 6.51E-06 1.29E-05 1.96E-05 2.61E-05 8.60E-06 1.71E-05 3.43E-05 3.34E-05 3.33E-05 7.98E-06 8.02E-06 1.06E-05 1.01E-05 1.96E-05 1.04E-04 8.03E-06 1.09E-05 1.07E-04 8.18E-06 8.10E-06 pH 3.58 3.58 3.58 3.57 3.50 3.50 3.50 3.50 3.50 3.34 3.84 4.17 4.72 4.72 4.69 4.72 5.94 5.94 5.9 6.9 log K 14.09 14.500.01 14.53 14.49 14.120.03 14.19 14.25 14.58 15.520.35 14.53 14.99 15.810.06 16.71 16.38 16.59 16.56 18.270.07 18.53 18.44 20.36
Table 3.7. Analytical data for complexation studies of Hf4+ with humic acid in 0.1 M NaNO3. Concentration of Hf4+ was 65 nM.
Sample Hf1 Hf2 Hf1 Hf2 Hf3 Hf4 Hf5 7B 20B Hf1 Hf1 Hf2 Hf3 Hf4 13B 22B Hf1 Hf2 9B 14B Method CE CE CE CE CE CE CE EDLE EDLE CE CE CE CE CE EDLE EDLE CE CE EDLE EDLE HA EHA EHA PHA PHA PHA PHA PHA EHA PHA EHA PHA PHA PHA PHA EHA PHA EHA EHA EHA EHA HA (mg/L) 10.04 9.88 9.29 9.41 9.26 18.47 36.29 9.55 9.99 9.85 9.46 9.38 18.65 17.91 9.77 9.91 9.93 15.23 9.77 9.67 [COOH] (mol/L) 4.83E-05 4.75E-05 4.72E-05 4.78E-05 4.70E-05 9.38E-05 1.84E-04 4.6E-05 5.07E-05 4.74E-05 4.8E-05 4.76E-05 9.47E-05 9.1E-05 4.7E-05 5.03E-05 4.78E-05 7.33E-05 4.7E-05 4.7E-05 [EDTA] (mol/L) 2.25E-07 4.43E-07 4.37E-07 8.72E-07 1.74E-06 2.15E-06 1.71E-06 4.24E-07 2.03E-07 4.42E-07 4.4E-07 8.82E-07 1.73E-06 1.68E-06 4.12E-07 4.06E-07 4.41E-07 4.39E-07 4.12E-07 4.08E-07 pH 3.48 3.52 3.50 3.50 3.50 3.50 3.50 3.19 3.32 4.67 4.63 4.66 4.67 4.70 4.67 4.56 5.94 5.94 5.92 6.86 log K 19.01 19.330.02 19.40 19.510.06 19.99 19.820.004 19.900.02 17.91 19.32 21.260.01 21.66 21.78 21.61 22.06 20.82 20.81 23.480.02 23.400.02 23.19 25.11
52
Table 3.8. Analytical data for complexation studies of Zr4+ with humic acid in 0.1 M NaNO3. Concentration of Zr4+ was 125 nM.
Sample Zr-a Zr1 Zr2 Zr3 Zr4 Zr5 15B 21B Zr-b 23B Zr-c 17B 24B 18B 25B Method CE CE CE CE CE CE EDLE EDLE CE EDLE CE EDLE EDLE EDLE EDLE HA EHA PHA PHA PHA PHA PHA EHA PHA EHA PHA EHA EHA PHA EHA PHA HA (mg/L) 9.94 9.46 9.31 9.23 18.53 36.26 9.45 9.94 9.93 9.89 15.23 9.58 9.90 9.62 9.93 [COOH] (mol/L) 4.78E-05 4.80E-05 4.73E-05 4.76E-05 9.41E-05 1.84E-04 4.55E-05 5.05E-05 4.78E-05 5.02E-05 7.33E-05 4.61E-05 5.03E-05 4.63E-05 5.04E-05 [EDLE] (mol/L) 4.46E-07 4.40E-07 8.76E-07 1.73E-06 1.74E-06 1.68E-06 3.99E-07 4.08E-07 4.46E-07 4.06E-07 4.39E-07 4.04E-07 4.06E-07 4.06E-07 4.07E-07 pH 3.51 3.5 3.5 3.5 3.5 3.5 3.48 3.36 4.69 4.37 5.92 5.98 5.96 6.85 6.85 log K 18.430.01 18.510.02 18.45 18.83 18.770.03 19.02 18.12 18.38 20.37 19.86 22.53 22.15 22.63 23.92 24.45
Electropherograms of MHA complexes An unexpected result of this study was the detection of an additional humic hump in Hf and Zr ligand competition experiments. This was not the case in Th or REE experiments. Hf-EHA complexes, when run on CE, dissociated into two peaks, while Hf-PHA complexes did not (Figure 3.6a, 3.7a). When samples with Hf, EHA, and EDTA were analyzed, species detected (Figure 3.6). Because the two peaks are present in samples containing only the Hf-EHA complex and no EDTA (Figure 3.6a), this cannot be a mixed ligand complex (i.e. M-HA-EDTA). Although Hf- and Zr-PHA complexes did not appear to separate into two humic species in the absence of a competing ligand, evidence of three species was present in these experiments as well, especially with high concentrations of PHA. Figure 3.7a shows an electropherogram of HfPHA as one species. Figure 3.7b and 3.7d show that under high PHA concentrations (~40 mg/L), three species are present. Even at lower PHA concentrations of ~10 mg/L, three species are suggested (Figure 3.5c). CE coupled with UV detection has been used in the fingerprinting of humic and fulvic acids. In general, electropherograms of HA show 1 or more humps corresponding to the structure of the HA (Garrison et al. 1995; Pompe et al. 1996; Pokorna et al. 2000; Schmitt-Kopplin and Junkers 2003; Ubner et al. 2004). Each HA will have a unique electropherogram at a given pH. There are many reasons for this, including differences in charge/size ratio and resulting differences in electrophoretic mobility among HA, as well as structural differences. Age or degree of humification of the humic substance can also be a factor in electropherograms of humic substances (Garrison et al. 1995). In these experiments, separation is achieved by using a buffer known to chemically complex HA. A result of this complexation is that component parts of HA can be isolated, and care must be taken to determine which components represent structural features of HA and which are artifacts of chemical interactions between HA and other analytes present in solution (Schmitt-Kopplin et al. 1998). It appears that
53
something similar occurs when Hf and Zr complex with EHA and PHA. Because of the polydisperse nature of the HA molecule, it is difficult to say which structural components are represented in electropherograms of ThHA. It is possible that these humps represent metal binding to different functional groups on the HA molecule, such as carboxylic or phenolic groups.. It is interesting to note that variations in the shape of the HA peak was only observed for HfHA and ZrHA complexes, and that for Th and the REEs, a Gaussian-like distribution around the average electrophoretic mobility (AEM) of the HA was observed.
Figure 3.6. Electropherograms of Hf-EHA complexes. a) Hf-EHA at pH 3.5. b) HfEDTA and Hf-EHA at pH 3.5. c) HfEDTA and Hf-EHA at pH 4.6. d) HfEDTA and Hf-EHA at pH 5.9.
54
Figure 3.7. Electropherograms of Hf-PHA complexes. a) Hf-PHA at pH 3.5. b) HfEDTA and Hf-PHA at pH 3.5, 36 mg/L PHA. c) HfEDTA and Hf-PHA at pH 4.6, 9.3 mg/L PHA d) HfEDTA and Hf-PHA at pH 4.6, 36 mg/L PHA.
The effects of [M]/[HA] on Log Kc,MHA CE experiments to determine the effect of HA (COOH- group) concentration on log Kc,MHA were also performed. During thorium-HA complexation experiments at pH 3.5, increasing HA from approximately 10 mg/L to 40 mg/L produced a large positive shift in log Kc,ThHA. This trend was not seen in experiments at the same pH with Hf or Zr, and further experiments will be performed using the EDLE technique to explore this apparent dependence of log Kc,THHA on [Th4+]/[HA] ratios. The final data presented in figures 3.4, 3.8, and 3.9 are all for samples containing 10 mg/L HA, with varying amounts of EDTA. The concentration of EDTA used in CE-ICP-MS experiments was optimized in order to maximize separation resolution, and ranged from 8.6 M to 34 M for thorium and from 0.4 m to 1.7 M for hafnium and zirconium. By increasing the ligand concentration in these experiments, we are effectively decreasing the free metal concentration. It has been shown that in general, increasing free metal concentrations or M/HA ratios leads to a decrease in log K c,MHA (Seibert et al. 2001). Specifically, log K c,MHA decreases with increasing log[M] by slope-1 (Hummel et al. 2000). Experiments performed under these conditions of high metal loading will underestimate log K c,MHA. As free metal concentrations decrease, log K values will increase until they reach a threshold value at some low metal loading. Here, log K will have no dependence on M/HA ratios (Glaus 2000). At these trace metal levels, protonated binding sites dominate, and thus, a distinct increase in log K will be seen with increases in pH, as more sites are available on the humic molecule at higher pH due to the
55
decrease in competition of H+ with metals. Under low metal loading conditions, a slope of 1 is expected for the line describing the variation of log K c,MHA with pH (Hummel et al. 2000). A slope <1 would indicate that metal-humic complexation is not entirely determined by proton exchange at strong sites on the HA molecule but that the reactions occurring here to bind metal to HA follow a continuous distribution model (Hummel et al. 2000). Alternatively, a slope <1 on a plot of log K c,MHA as a function of pH under circumstances of metal loading would indicate that the data is showing more of a dependence on free metal concentration (log[M]) than on pH. Because all of our results have slopes greater than 1 in plots of log K c,MHA as a function of pH, we suspect that our experiments have been conducted at the appropriate trace metal concentrations. However, only a systematic measurement of log K c,MHA at different M/HA ratios ranging over several orders of magnitude can prove this. Ligand competition experiments by (Reiller et al. 2003) where binding of Th to silica is measured in the presence and absence of competing HA at very low Th concentrations (1.15 10-12 M) show variations in log Kc,ThHA with concentration of HA. However, the trend displayed in their data is the opposite of what we find. In our data, an increase in HA and concomitant decrease in free [Th4+] results in increasing values of log Kc,ThHA, whereas Reiller finds log Kc,ThHA values decrease with increases in [HA-] (Figure 3.8). Log Kc,ThHA as a function of [Th4+]/[HA] is plotted in Figure 3.9. Apparently, the relationship between [Th4+]/[HA], and log Kc,ThHA needs to be explored further.
24 22 Log Kc,ThHA 20 18 16 14 12 0.E+00 pH 8.1, Reiller et al. 2003 pH 6.8, Reiller et al. 2003 pH 3.5, this study
1.E-04 [COOH]
2.E-04
3.E-04
Figure 3.8. Comparison of log K variation with [HA] from our study and Reiller et al. (Reiller et al. 2003).
56
24 22 Log Kc,ThHA 20 18 16 14 1.E-10 1.E-08 1.E-06 1.E-04 1.E-02 12 1.E+00
pH 8.1, Reiller et al. 2000
pH 6.9, Reiller et al. 2000
pH 3.5, this study
Log [Th]/[COOH]
Figure 3.9. Comparison of log K variation with [Th]/[COOH] from our study and Reiller et al. (2003).
3.3.3. Comparison of experimentally determined log Kc with literature values.
25 23
y = 1.7489x + 12.963 R2 = 0.9734 y = 1.6746x + 12.598 R20.9912 = y = 1.7236x + 8.3773 R2 = 0.9917
log Kc,MHA
21 19 17 15 13 3 4 5
Th Hf Zr
pH
6
7
Figure 3.10. All conditional binding constants generated in this study, IS 0.1M. Figure 3.10 shows that conditional binding constants for Th, Hf, and Zr can be predicted by linear equations: log Kc,ThHA(0.1IS) = 1.7236 pH + 8.3773
(3.11) (3.12) (3.13)
log KcHfHA (0.1IS) = 1.7489 pH + 12.963 log Kc,ZrHA (0.1IS) = 1.6746 pH + 12.598
57
The slopes of these lines are all very similar, with log KcHfHA having the steepest slope. The compatibility of these elements with humic substances mirrors their compatibilities with silicates. Zirconium and halfnium have almost the same ionic radius (0.72 and 0.71 ) and thus should behave similarly. Zirconiums slightly larger atomic radius makes it slightly less compatible with binding sites in silicates and here with humic acid binding sites. Thorium, which has the same charge as zirconium and halfnium, has a larger ionic radius (0.972 ) than either and is significantly less compatible with silicate binding sites due to its larger size. The fact that log Kc,ThHA is 4 orders of magnitude lower than the log Kc,HfHA and log Kc,ZrHA might be caused by a similar size limitation in humic binding sites. Our data is on the high side on literature values, but not outside of the range. Our study represents the most comprehensive examination of log Kc,ThHA over the widest and most continuous range of pH. Nash and Choppin (1980) used a solvent extraction technique to measure conditional binding constants log 1 and log 2 for the 1:1 and 1:2 thorium-humic complexes. Schuberts method, which measures the distribution of metal between a solution and a solid phase (usually ion exchange resin) in the presence and absence of a dissolved ligand (HA), was used to calculate these binding constants. However, total metal concentrations are not given in this paper, making comparison to our data difficult. If [Th4+] was too high, the binding constants given by Nash & Choppin (Nash and Choppin 1980) would be underestimated, as described previously. Hummel et al. (2000) suggests that these values may be used as a rough estimate for Th-HA interactions, indicating that Th-HA complexation must be 4 to 6 orders of magnitude stronger than Eu-HA complexation . Our EuHA experiments (using EHA) yield values of log K = 12.4 at pH 7, while our ThHA experiments (with EHA) yield values of 20.4 at pH 6.85, 8 orders of magnitude larger than EuHA binding constants. Reiller et al. (2003) extrapolated data from Nash and Choppin (1980) to pH 7-8 for comparison with their own study. At pH 6.75, data from Nash and Choppin (1980) give an extrapolated log K of 15.4, while experiments conducted by Reiller et al. (2003) produced at log K of 16.9 1.2 at the same pH. This data falls several orders of magnitude below our data. Reiller et al. (2003) studied the ternary system HA-Th-Silica to investigate the influence of HA on thorium retention onto the surface of silica colloids. This is essentially a ligand competition experiment where binding of Th(IV) to silica is measured in the presence and absence of competing HA. Metal concentrations were very low, 1.15 10-12, so the binding constants are probably not overestimated, and strong pH dependence is displayed. Because silica complexes thorium weakly (log Kc,ThS = ~2) compared to a strong ligand like EDTA (log Kc,ThEDTA = 23.2, IS 0.1), hydrolysis must be taken into account in the calculation of log KcThHA in the HA-Th-Silica ternary system. The values from Reiller et al. (2003) in Figure 3.10 represent a range of hydrolysis constants. If we extrapolated the data from the present study to pH 8, our log Kc,ThHA (22.04) would be very close to their values. Linearization of data from Reiller et al. (2003) produces a line (m = 4.63) with a much steeper slope than ours.
58
This study (EHA, PHA)
22 20
log Kc,ThHS
Moulin, 1986 (Aldrich HA)
18 16 14 12 10 3 4 5
pH
Hirose and Tanoue, 2001 (marine DTPA type ligand) log B1, Nash and Choppin 1980, (lake sediment HA) log B2, Nash and Choppin 1980, (lake sediment HA) extrapolated from Nash and Choppin by Reiller, 2003
6
7
8
Reiller et al., 2003 (Aldrich HA)
Figure 3.11. Conditional binding constants for ThHS complexes taken from this study and published literature, corrected to 0.1 IS. Hirose and Tanoue (2001) extrapolated laboratory measurements of thorium binding of a strong marine ligand at pH 1 to seawater values (pH 8, IS 0.7 M), assuming that this ligand was primarily composed of carboxylic functional groups. When corrected for IS using the Davies equation, this yields log KcThHS = 22.28 plotted in Figure 3.5, very close to our pH 8 extrapolated log KcThHA of 22.04. This suggests that terrestrial HA has the ability to complex thorium on par with that of ligands produced by marine bacteria. This would indicate that humic complexation of thorium is an important phenomenon in the transport and bioavailability of this metal. Th binding constants presented here and in the relevant literature (Nash and Choppin 1980; Reiller et al. 2003) are several orders of magnitude higher than published data for other actinides. Conditional binding constants for Am(III)-humic complexes have been reported as 6.230.11 at pH 6, IS 0.1 using ultrafiltration and absorption spectroscopy (Czerwinski 1995) and 6.60.2 at pH 5 over a range of IS using electrophoretic ion focusing (Franz 1997). An average log Kc for UO22+-HA interactions has been given as 6.160.13 over a range of pH and IS (Czerwinski 1994). The elevated log Kc values found in this study for ThHA complexes indicate that this species is likely to form in the absence of competing ligands at pH values in the range of natural waters (6-7). This study supports the assertion that thorium-humate complexes are significant up to pH 6.5 for 0.1 mg/L HS, and up to pH 7.5 for 10mg/L (Moulin et al. 1992).
3.4. Conclusion
Our investigation of thorium, halfnium, and zirconium-humate interactions indicate that tetravalent metals can strongly complex humic substances in the absence of
59
competing ligands at pH values relevant to natural systems (pH 3.5-7). Hydrolytic products were not detected by CE-ICP-MS. CE-ICP-MS and EDLE yielded nearly identical binding constants that fall reasonably close to published values for ThHA complexes, indicating that both methods are appropriate for examining metal speciation at low pH (<5). All metal-humic complexes measured exhibited dependence of log KMHA on pH. Metal species, especially thorium-HA, were prone to sticking to the tygon and capillary tubing, making measuring speciation data above pH 5 somewhat difficult. This study represents the first side-by-side comparison of CE-ICP-MS to an established speciation method. Additionally, it will contribute useful conditional binding constant data to aid in calibration of polyelectrolyte models. Finally, this study has contributed to the body information on thorium-humic interactions, which can be related to other actinide-humic interactions. The strong affinity of tetravalent metals for the humic component of DOC, along with the persistence of DOC in the water column discussed in Chapter 2, suggests that tetravalent actinide-humic complexes, in addition to more commonly studied actinide-colloid complexes, could be more important than previously thought in assessing transport of radionuclides in groundwater from nuclear repository sites.
60
CHAPTER 4 METHANE OXIDATION IN LANDFILLS
4.1. Introduction
Methane is known to be a more effective greenhouse gas than carbon dioxide, despite its lower concentration in the atmosphere. Atmospheric methane has more than doubled in concentration over the last 150 years to its present level of 1.75 ppmv (Schlesinger 1997; Dlugokencky et al. 2003). However, over a period of 100 years, the greenhouse warming potential, or GWP, of methane is 21 times that of an equal mass of CO2 (IPCC 1994). In the late 1980s methane emissions to the atmosphere were increasing the alarming rate of 1% per year although they have decreased recently possibly due to the repair of natural gas pipelines in the former Soviet Union (Schlesinger 1997). Thus, determining the sources and sinks of methane is an important effort. Landfills are estimated to account for 36% of anthropogenic methane emissions and 8% of total emissions to the atmosphere. They represent a large methane source with a potential for mitigation through management practices. Because the difference between sources and sinks of methane is less than 6% of the total methane production, even a small reduction in anthropogenic methane emissions would be significant (Dlugokencky et al. 1994; Dlugokencky et al. 1998; Etheridge et al. 1998; Dlugokencky et al. 2003). In addition, methanes relatively short residence time in the atmosphere (7-10 years) would mean the effects of a landfill management scheme to reduce emissions could be observed in a relatively short period of time. Methane is produced in landfills via anaerobic microbes during the degradation of organic matter. Two pathways are used by methanogenic bacteria to produce methane. Acetate fermentation represents possibly the most primitive metabolic pathway in which acetate is split into carbon dioxide and methane. CO2 reduction occurs via reduction of CO2 by H2, yielding methane and water. Both pathways are found among bacteria inhabiting wetlands and coastal sediments (Schlesinger 1997). Methane produced by acetate fermentation is depleted in 13C with 13C = -30 to 60, while methane produced by CO2 reduction is enriched in 13C with 13C = -60 to 110. Methane produced in landfills ranges from 13C = -62 to 53, suggesting acetate fermentation is a significant component of methanogenesis in landfills. Methane emissions from landfills are in part controlled by the rate of oxidation as methane travels through the outer most aerobic soil cap on top of the landfill. Oxidation
61
is controlled by several factors, including soil temperature, moisture, and texture, as well as pH and nutrient content (Kightley et al. 1995; Boeckx et al. 1996; Chanton and Liptay 2000; Borjesson et al. 2001). Methane oxidation in landfills can be enhanced by the emplacement of a biologically active compost cover (Humer and Lechner 1999; Hilger and Humer 2003; Barlaz et al. 2004). Oxidation of methane is achieved by methanotrophs, aerobic bacteria that consume CH4 and oxidize it to CO2. In doing so, they discriminate against the heavy carbon isotope in methane (13CH4). Residual methane, as it travels through landfill soil cap, will become increasingly enriched in methane due to the activity of methanotrophic bacteria. This isotopic fractionation can be taken advantage of in a mass balance relationship in which we can calculate the percentage of methane oxidized in the soil cap by measuring the shift in the carbon isotopic composition of the residual methane relative to the unaltered methane produced in the anoxic zone of the landfill. Previous studies have shown that the percentage of oxidation is seasonal, being greater during warmer months (Chanton and Liptay 2000; Borjesson et al. 2001), and greater for mulch-type soils than for clay. In addition, there appears to be an optimum soil moisture for methane oxidation, 10-20% at temperatures from 25 C to 30 C (Whalen et al. 1990; Boeckx et al. 1996). One incubation study of composted municipal solid waste used as landfill cover showed high percentage oxidation at soil moisture content of 45% (Hilger and Humer 2003). Soil composition is also an important parameter, as soil texture and grain size affect the ability of oxygen to penetrate the soil. Coarser grained soils and porous compost have been found to be superior to finer grained soils and clays. The current study examines the flux and percent of methane oxidized in landfills with and without a biocover, a biologically active layer of compost placed on top of landfill sediments. The purpose of this biocover is to optimize the environment for methanotrophic bacteria so that the maximum amount of methane can be oxidized as it travels through the soil cap. The biocover must be permeable enough for oxygen to penetrate, but also be able to hold an appropriate, optimum amount of moisture. The deeper oxygen can penetrate, the larger the zone of methane oxidation can be, and at greater depth, oxidation can proceed under more stable moisture and temperature conditions (Hilger and Humer 2003). Humner and Lechner (1999) showed that a 1 m layer of sewage sludge composted with woodchips or well composted solid waste can eliminate CH4 emissions from landfill soils. Positive results were also seen from a biocover consisting of 1 m yard waste compost underlain by 0.15 m tire chips and 0.15 m clay emplaced at the Outer Loop landfill in Louisville, KY (Barlaz et al. 2004). At the Tallahassee landfill, the site of our study, it has been shown that just six inches of mulch (yard waste and woodchips) is significantly more effective in oxidizing methane than a 1 m thick layer cover (Chanton and Liptay 2000). We hypothesize that a thin (~50 cm) layer of biologically active compost will likewise be more effective in oxidizing methane than untreated landfill soils. Here we present the results of flux and percent oxidation measurements for the biocover and the control over a nine month period beginning in March 2004 and ending December 2004.
62
4.2. Methods 4.2.1. Site description This study was conducted at Tallahassee Municipal Solid Waste landfill in Leon County, Florida. Previous studies to characterize methane emissions at this landfill were conducted by Chanton & Liptay (2000), who found some seasonality in methane emission and oxidation. The S1 grid was set up over waste that had been covered for 8 years by 15 to 100 cm of sandy clay and 15 cm of sandy loam, with a thin (45 cm) soil layer on top which was thickly vegetated during the time of this study. It is located on the side slope of the landfill. The S1 grid measured 60.8 m (200 ft) per side and divided into 64 squares measuring 7.6 by 7.6 m (25 by 25 ft). Following preliminary spatial assessment of methane emissions at the site (Abichou et al. 2004), a portion of S1-grid was chosen based on the location of methane hotspots to emplace the biocover (Figure 4.1). The biocover consists of a 10-cm gas dispersion layer of gravel-sized recycled glass covered with 46 to 51 cm of compost extending 3.8 m beyond the edges of the experimental squares and filling the spaces between the squares (Figure 4.2). Compost was provided by the landfill, and consisted of chipped yard waste that had been windrowed and turned for three years. The experiment was designed to consist of three untreated control squares (2B, 4B, and 8B) and three biocover squares (2D, 4D, and 6D) each with four collars and one probe nest (Figure 4.3).
Figure 4.1. Location of squares selected for control (no compost) and biocover (compost) sites.
63
For a year prior to this study and the emplacement of the biocover (Feb 2003-Feb. 2004), methane emission rates were measured for S1-grid to establish baseline methane flux from a typical older closed landfill with vegetated cover. Studies of methane emissions from S1 grid prior to the emplacement of the compost indicate an uneven pattern of flux across the surface of the grid. Prior to emplacement of the biocover, the mean flux was 24.6 63.3 g/m2/day, and the mean oxidation was 35.2 23.1% (Abichou et al. 2004).
CROSS SECTION OF S1 BIOCOVER STUDY
Compost Squares 2D, 4D, and 6D 120 100 80 Vertical (cm) 60 40 20 0 -20 -40 0 500 1000 1500 2000 2500 3000 3500 North to South (cm) Solid Waste 47 to 100 cm of sandy clay cover 10 cm of recycled glass (diameter 0.32 to 0.95 cm) 49 to 78 cm of sandy clay cover No compost Squares 2B, 4B, and 8B
46 to 51 cm of compost
Figure 4.2. Cross-section sketch of S1 biocover locations. Note the 11X vertical exaggeration.
64
Experimental Design for the Biocover Study 3 Control Squares 3 Compost-Treated Squares
4 flux locations 2ft X 2ft One probe nest
25 ft 25 ft
Figure 4.3. Experimental design for the biocover study. Each control square contains 4 flux collars and one probe nest.
4.2.2. Flux Methane emissions were measured using a static chamber technique. Collars (Chamber bases) were installed at each flux site at the initiation of the project. Chambers were placed on top of the collars and samples were collected from the headspace using 60 mL disposable syringes fitted with plastic stopcocks. Samples were collected over a period of 20 minutes, starting at time = 0 and every 5 minutes thereafter. An additional sample was taken at time = 0 and time = 20 minutes for isotopic analysis. Methane concentrations for each sample were measured on a gas chromatograph with a flame ionization detector. Concentrations were plotted vs. elapsed time to determine methane flux. Stable isotope ratios were only measured when flux was positive, indicating methane was being emitted from the landfill surface. 4.2.3. Stable Carbon Isotopes Stable isotopes for initial and final samples from each chamber were collected using 60 mL disposable syringes fitted with plastic stopcocks and subsequently transferred to evacuated glass vials. Samples were only analyzed when flux was positive in order to determine the carbon isotopic composition of residual methane. This was done using the equation:
R =
( F C F ) ( I C I )
CF CI
(4.1)
65
where R is the 13C value of the residual methane emitted from the landfill, F and I are the final and initial 13C values of methane measured at t = 0 and t = 20, and CF and CI are the final and initial methane concentrations measure at t = 0 and t = 20. The value computed for R and measured for anoxic methane by probes, A, can be used to calculate the percentage of methane oxidized provided we know the carbon isotopic fractionation factor for bacterial oxidation. This parameter, , is a measure of the bacterias preference for the light isotope over the heavy isotope, given by: ox = kL/kH (4.2)
where kL and kH refer to the rate constants of the light (12CH4) and heavy (13CH4) isotopes. This parameter is determined by closed system incubation experiments in which landfill soils of different compositions are placed in flasks and stoppered, then placed into baths of different temperatures. The isotopic composition and concentration of the residual methane in the flasks is measured over time and is calculated using the equation from Coleman et al., 1981: 13Ct = 1000(1/-1) * ln(M/M0) + 13Ct=0 (4.3)
The values for of clay and mulch used in the present study were determined in this way by Chanton & Liptay (2000). They are given by the equations: mulch = -0.000438 * T + 1.0411 clay = -0.000433 * T + 1.0421 (4.4) (4.5)
where T is temperature. The fraction of methane (fo) oxidized in upward transit through the landfill cover soil is then given by:
fo =
( R A )
1000 ( ox trans )
(4.6)
where R is calculated using equation 4.5 and A is the carbon isotopic content of anoxic methane from landfill probe data, and ox and trans are the isotope fractionation factors appropriate for the soil type and associated with transport of methane, respectively. The parameter trans is assumed to be 1 because transport of methane up through the soil cap is assumed to be dominated by advection, a process that does not cause isotopic fractionation (Bergamaschi et al. 1998; Liptay et al. 1998). This assumption is supported by observations of a negative relationship between methane emission and atmospheric pressure (Czepiel et al. 1996; Czepiel et al. 2003), suggesting a pressure gradient drove gas under high pressures in the landfill upward through the soil cap. Recent laboratory experiments have shown that this approach underestimated methane oxidation by a factor of 2 to 4 by not taking into account diffusive flux of methane (De Visscher et al. 2004). However, the effects of diffusion were less apparent in the field. At high fluxes, when advection is more important in gas transport than diffusion, trans will be closer to 1, while
66
at low flux, when diffusion dominates, trans will be greater than one, and isotopic fractionation will occur. Because Tallahassee MSW landfill has no gas collection system, the assumption of advective flux dominating is a good one, and supported by the observation of gas bubbling through cracks in the landfill surface after rain storms (De Visscher et al. 2004). Stable carbon isotopes were measured by direct injection into a Hewlett Packard Gas Chromatograph coupled via combustion interface to a Finnigan Mat Delta S-Gas Chromatograph Combustion Isotope Ratio Mass Spectrometer (GCC-IRMS) following Merrit et al. 1995. Samples with small concentrations (<1000 ppm) were cryogenically focused using a device coupled to the front end of the GC. Replicates were analyzed for most samples, yielding a standard deviation of approximately 0.15.
4.3. Results and Discussion 4.3.1. Methane Flux Flux from both treatments averaged 20 g CH4 m-2d-1, with measured flux from the control ranging from -0.111 to 740 g CH4 m-2d-1 and measured flux from the biocover ranging from -0.152 to 1159 g CH4 m-2d-1 (Figure 4.4a). These values are on the order of those (0 to 9000 g CH4 m-2d-1) reported by Chanton & Liptay (2000) for the same landfill over mulch and clay soil. The negative flux values have been interpreted by some authors to indicate that the landfill is actually acting as a methane sink for atmospheric methane (Bogner et al. 1997). This will be discussed below.
120 100 Flux (g/m 2/day) 80 60 40 20 0 March 24 May 18 June 10 July 27 Oct. 8 Dec. 10 control Flux (g/m2/day) biocover
16 14 12 10 8 6 4 2 0
M ar ch Ap 24 r il 2 M 8 ay 1 Ju 8 ne Ju 3 ne J u 10 ne Ju 29 ly 2 Se 7 pt . O 3 ct. N ov 8 . De 19 c. 10
control w /o 2B1 biocover w /o 2D1
-2
Figure 4.4. (a) Methane emission rates or flux (g CH4 m2/day) from control and biocover areas. Values for control and biocover at each date are the represent the average of 12 chambers. Averages and standard deviations are listed in Table 4.1 in the Appendix). (b) Control and biocover fluxes averages without large fluxes included. Studies of methane emissions from S1 grid prior to the emplacement of the compost indicate an uneven pattern of flux across the surface of the grid, with an average flux of 24.6 g CH4 m-2d-1 and fluxes ranging from -6.07 to 330 g CH4 m-2d-1 (Abichou et al. 2004). A plot of flux vs. chamber location from March 2004 to December 2004
67
(Figure 4.5) shows that 2B1 in the control area and 2D1 in the biocover area consistently had the highest fluxes, while the rest of the chamber locations had significantly lower fluxes. A plot of flux from the control and biocover without these large fluxes included is shown in 4.4b. If flux from chamber 2B1 is plotted with the flux averaged across the entire control site (Figure 4.6a), it is evident that flux at this one chamber is orders of magnitude higher than the average. The same can be done for 2D1 in the biocover (Figure 4.6b). Flux was plotted in Figure 4.7 for the control and biocover including and excluding data from 2B1 and 2D1. It is evident here that the fluxes at these two chambers strongly influence the calculation of mean flux for the control and the biocover.
2500 2000 Flux (g/m /day) 1500 1000 500 0 8B4 8B3 8B2 8B1 4B4 4B3 4B2 4B1 2B4 2B3 2B2 2B1 6D4 6D3 6D2 6D1 4D4 4D3 4D2 4D1 2D4 2D3 2D2 2D1
24-Mar 19-Apr 28-Apr 18-May 3-Jun 10-Jun 29-Jun 27-Jul 3-Sept 8-Oct 19-Nov 10-Dec
2
Chamber
2
Figure 4.5. Methane emission rates (g CH4 m /day) at each chamber from March 24, 2004 to December 10, 2004.
10000 1000 log flux (g/m2/day) 100 10
10000
control 2B1
log flux (g/m2/day)
A
1000 100 10 1 March 24 June 3 July 27 Oct. 8
biocover 2D1
Nov. 19
June 10
June 29
June 10
June 29
Nov. 19
March 24
April 28
Dec. 10
May 18
0.1
0
Figure 4.6. (a) Methane emission rates (g CH4 m2/day) averaged over 12 chambers in the control area vs. single chamber data from 2B1. Note the logarithmic scale. (b) Methane emission rates (g CH4 m2/day) averaged over 12 chambers in the biocover area vs. single chamber data from 2D1.
68
April 28
Dec. 10
May 18
July 27
June 3
Sept. 3
Sept. 3
1 Oct. 8
A
80 70 flux (g/m 2/day) 60 50 40 30 20 10 0 Nov. 19 March 24 June 10 June 29 Dec. 10 May 18 July 27 June 3 Oct. 8 Sept. 3 April 28 control f lux control w /o 2B1
B
120 100 flux (g/m 2/day) 80 60 40 20 0 May 18 June 10 March 24 June 29 July 27 June 3 Oct. 8 Nov. 19 Dec. 10 April 28 Sept. 3 biocover flux biocover w /o 2D1
Figure 4.7. (a) Comparison of control flux including data from 2B1 and without data from 2B1. (b) Comparison of biocover flux including data from 2D1 and without data from 2D1. Table 4.1. Results of ANOVA for flux data.
All flux data from March to December 10 not different p = 0.985 n mean std dev std error control 144 18.6 74.7 6.22 biocover 143 105.6 105.6 8.8 All flux data from and including June 29 through December not different p = 0.154 n mean std dev std error control 73 30.6 102 12.03 biocover 72 11.45 43.9 5.17 All flux data with 2D1 and 2D1 removed from and including June 29 through December Control significantly greater than biocover, p = 0.004 n mean std dev std error control 66 8.94 21.6 2.69 biocover 65 1.05 3.34 0.41
Table 4.1 lists the results of statistical analyses of variance (ANOVAs) performed on landfill flux data over the time period of the experiment including and excluding data from the high fluxes 2B1 in the control and 2D1 in the biocover. With the high fluxes included, there is no significant difference between flux from the control and biocover over the March to December period or from the period from July 29 to December. However, when data from 2B1 and 2D1 are excluded, the mean fluxes drop significantly to 8.94 g CH4 m-2d-1 for the control and 1.05 g CH4 m-2d-1 for the biocover, and there is a significant difference in mean flux from the control and the biocover (p=0.004).
69
The spatial variation in methane emissions from S1 grid has been examined by (Abichou et al. 2004). This spatial variation can come from differences in methane generation by the trash layer of the landfill as well as the heterogeneity of the overlying material. Surface cracking of both clay and compost soil was observed, as were methane hotspots where gas bubbles were observed in standing water after rain.
4.3.2. 13C Stable isotope data indicates that the biocover had significantly more enriched 13C values than the control from July through December, indicating that more oxidation was occurring in the biocover relative to the control several months after its initial emplacement (Figure 4.8). Oxidation in both layers is occurring except perhaps in the control on December 10, when 13C of residual methane is very close to that of anoxic methane (-55.38), taken from the base of the trash. Chanton and Liptay (2000) found that 13C of anoxic methane did not vary seasonally because the temperatures at which it is produced in the landfill are constant. Our results show that the 13C of landfill methane is more depleted during the summer months than during the fall, with the exception of December. This is the opposite of what Chanton and Liptay found. However, we find the same antipathetic relationship between flux and 13C, indicating emission rates are controlled by bacterial oxidation of methane (Figure 4.9).
-60 -55 -50
13 C
Control Biocover anoxic methane
-45 -40 -35 -30
ar ch Ap 24 ril M 28 ay 1 Ju 8 ne Ju 3 ne J u 10 ne J u 29 ly Se 27 pt . O 3 ct No . 8 v De . 1 9 c. 10
Figure 4.8. Mean 13C values (per mil) of methane in the control and biocover compared to anoxic methane.
M
70
a) Control
120 flux (g/m 2/day) 100 80 60 40 20 0
ar c Ap h 2 r 4 M il 28 ay Ju 18 J u ne n 3 Ju e 1 ne 0 Ju 29 l Se y 27 pt O .3 c No t. 8 v. De 1 c. 9 10
b) Biocover
-30 flux d13C
120 flux (g/m 2/day) 100 80 60 40 20 0
-35 -40 13C -45 -50 -55 -60
flux d13C
-30 -35 -40 -45 -50 -55 -60 13C
Figure 4.9. (a) Methane emission rate and 13C values (per mil) of CH4 emitted from the control. (b) Methane emission rate and 13C values (per mil) of CH4 emitted from the biocover.
4.3.3. Oxidation To determine the percent oxidation occurring at control and biocover sites, temperature data was used in equations 4.4 and 4.5 to calculate ox. In this study, clay was found to range from 1.022 to 1.035, and compost from 1.018 to 1.032, calculated over a soil temperature range from 17 to 52 C. These values were used in equation 4.5 along with the 13C of residual and anoxic methane to calculate percent oxidation. During data analysis, the question of how to deal with negative flux measurements arose. Negative flux values have been assumed to indicate uptake of methane from the atmosphere by the landfill (Bogner et al. 1997), and thus oxidation values for these fluxes are assumed to be 100%. If a negative flux indicates a downward flux of atmospheric gas into the landfill, all of the methane coming from below must be oxidized before it reaches the landfill surface. Graphs were plotted including (Figure 4.10a) and excluding (Figure 4.10b) 100% oxidation values. Whether or not negative fluxes are assumed to represent 100% oxidation of landfill methane primarily affects the amplitude of the trend seen in either graph. Both graphs show that before June 29, the control appears to oxidize methane better than the biocover, and that after June 29, the biocover is more effective in oxidizing methane than the control. The numbers of negative fluxes measured for the control and the biocover are shown in Figure 4.10c. This plot supports the interpretation of negative fluxes as the consumption of methane by the cover soil indicating 100% oxidation, and shows that the biocover was frequently a sink for atmospheric methane. From April to June, the control shows more occurrences of negative fluxes, while in Figures 4.10a and b, the control appears to be oxidizing more methane than the biocover at this time. After June 29, when the biocover becomes more effective at oxidizing methane than the control, there are more occurrences of negative fluxes in the biocover, which makes sense if those negative fluxes represent 100% oxidation. Thus, it was decided to assume negative fluxes
71
M ar c A h2 pr 4 M il 28 ay Ju 18 Ju ne n 3 Ju e 1 n 0 Ju e 29 l Se y 27 pt O .3 c N t. 8 ov D .1 ec 9 .1 0
M
A
100% % oxidation 80% 60% 40% 20% 0%
ch Ap 24 ril 2 M 8 ay 1 Ju 8 ne Ju 3 ne Ju 1 0 ne 2 Ju 9 ly 2 Se 7 pt .3 O ct N .8 ov . D 19 ec .1 0
control biocover
M ar
B 100% % oxidation 80% 60% 40% 20% 0%
Ap 24 ril 2 M 8 ay 1 Ju 8 ne Ju 3 ne Ju 1 0 ne 2 Ju 9 ly 2 Se 7 pt .3 O ct N .8 ov . D 19 ec .1 0 ch
control biocover
M
ar
C
12 10 8 6 4 2 0
ar ch Ap 24 r il M 28 ay Ju 18 n Ju e 3 ne Ju 1 0 ne Ju 29 ly Se 27 pt . O 3 c N t. 8 ov . D 19 ec .1 0
n
control biocover
Figure 4.10. (a) Percent oxidation of CH4 in the control and the biocover, calculated without using values of 100% for incidences of negative flux (thought to indicate uptake of methane from the atmosphere by the landfill soil cap). (b) Percent oxidation of CH4 in the control and the biocover, calculated using values of 100% for incidences of negative flux. (c) The number of samples per date for the control and the biocover that yielded negative fluxes.
M
72
represent 100% oxidation values, and data in the subsequent graphs in this paper reflect this. Over the period of this study, the difference between methane oxidation values of the control and the biocover was statistically significant. A two way ANOVA was performed on oxidation data including and excluding 100% values assumed for uptake of methane. Over the entire timespan represented by the data, with 100% values excluded, the biocover had a mean oxidation of 33.3%, while the control had a mean oxidation of 21.0% with a significant difference of p=0.004. Using the results after and including June 29, when our results show the biocover became more effective than the control, the mean oxidation of the biocover was 36.5%, and the mean oxidation of the control was 17.5%, with a significant difference of p = 0.001 When 100% values are included, the mean oxidation is 56.5% for the biocover and 45.5% for the control over the entire period from March to December, with p < 0.001. From June 29 to December, the mean oxidation for the biocover was 65.8%, and the mean oxidation for the control was 33.6%, with p = 0.015. Before June 29, the control (59.8%) was actually more efficient at oxidizing methane than the biocover (45.3%), with p=0.049, which is also supported by the 13C data (Figure 4.8). In all cases, the biocover is twice as successful in oxidizing methane than the control...