Pnfl7erslry of HAWAII UBRAR'l'

PNfl7ERSlry OF HAWAII UBRAR'l'

CLIMATE CHANGE AND ANTHROPOGENIC EFFECTS ON SHALLOW-WATER

CARBONATE BIOGEOCHEMISTRY

A THESIS SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERSITY OF HAWAI'I IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

MASTER OF SCIENCE

IN

OCEANOGRAPHY

DECEMBER 2003

By Andreas J. Andersson

Thesis Connnittee: Fred T. Mackenzie, Chairperson Edward Laws Yuan-HuiLi TABLE OF CONTENTS

Acknowledgements vi Abstract viii List ofTables x List ofFigures xi

CHAPTER I: Shallow-water carbonate biogeochemistry 1 1.1 Introduction 1 1.2 Background ofresearch 5 1.2.1 Atmospheric C02 and global warming 5 1.2.2 CO2 in seawater and CaC03 saturation state in the ocean 9 1.2.3 CaC03 in natural environments 15 1.2.4 Production ofCaC03 17 1.2.5 The effect ofglobal warming and increasing atmospheric CO2 on marine calcareous organisms and communities 21 1.2.6 Opposing observations and alternative predictions: The Magnesian Salvation Theory .30 1.3 Briefoverview ofthesis content .35

CHAPTER 2: Shallow-water Ocean Carbonate Model (SOCM) .37 2.1 Introduction .37 2.2 SOCM description .38 2.3 Shallow-water carbonate masses, production and accumulation .44 2.3.1 Production and accumulation .44 2.3.2 Shallow-water carbonate budget .49 2.3.3 Shallow-water carbonate sediment mass .51 2.3.4 Carbonate sediment composition .52 2.4 Biogenic calcification 53 2.4.1 Biogenic calcification and DlC 53 2.4.2 Biogenic calcification and carbonate saturation state .54 2.4.3 Biogenic calcification and temperature .56 2.4.4 Combined effect ofDIC, carbonate saturation state and temperature on biogenic calcification in SOCM .57 2.5 Carbonate dissolution and precipitation reaction kinetics 58 2.6 Terrestrial Ocean aTmosphere Ecosystem Model (TOTEM) 68

CHAPTER 3: Solution ofshallow-water carbonates: An insignificant buffer against rising atmospheric C02 (article published in Geology, June 1, 2003) 72 3.1 Abstract 72 3.2 Introduction 73 3.3 Methods 75 3.4 Results and discussion 79

III 3.5 Conclusion 84 3.6 Acknowledgement 85 3.7 References cited 85

CHAPTER 4: Sensitivity analysis and validation ofSOCM 90 4.1 Introduction 90 4.2 Validation ofSOCM 91 4.2.1 Results ofstandard run 91 4.2.2 Comparison ofmodel results to observations from the natural environment 92 4.2.2.1 Carbonate saturation state 92 a) Bermuda Atlantic Time Series (BATS) 94 b) Hawaiian Ocean Time series (HOT) 95 4.2.2.2 Biogenic calcification 97 4.2.2.3 Carbonate mineral dissolution 100 4.2.2.4 Summary 101 4.3 Sensitivity analysis 102 4.3.1 Background 102 4.3.2 Sensitivity procedure 103 4.3.3 Sensitivity analysis results 105 4.3 .3.1 Organic matter deposition and remineralization 107 4.3.3.2 Carbonate reaction kinetics 118 a) Reaction order 118 b) Reaction rate constant... 122 c) Reaction inhibition factor 122 4.3.3.3 Initial model parameters 124 a) Initial carbonate sediment mass 124 b) Initial pore water dissolved inorganic carbon composition and carbonate saturation state 124 c) Average magnesian calcite composition 126 d) Magnesian calcite solubility 130 4.3.3.4 CO2 emission scenarios 134 4.3.3.5 Shallow-water ocean - open ocean exchange 137 4.3.4 Biological implications 139 4.3.4.1 Species and community specific response to decreasing carbonate saturation state 140 4.3.4.2 CO2 emission scenarios - biogenic calcification 141 4.3.4.3 Temperature scenarios - biogenic calcification 144 4.3.4.4 Combined effect ofcarbonate saturation state and temperature on biogenic calcification 146 4.4 Summary and concluding remark 148

CHAPTER 5: Summary and conclusions 150

IV APPENDIX A 153 A.l ForcingsofSOCM 153 A.2 DIC calculations 153 A.3 Model equations 156 A3.1 Mass balance equations 159 A3.2 Flux equations 160 A3.3 Flux equations: special cases 161 A.3.3.1 Atmosphere - surface water CO2 exchange 161 A.3.3.2 Marine photosynthesis 163 A.3.3.3 Marine biogenic calcification 163 A.3.3.4 Pore water-sediment system carbonate precipitation and dissolution 164 A4 Derivation ofinitial reservoir masses in SOCM 165 A5 Derivation ofinitial carbon fluxes in SOCM 168 References 178

v ACKNOWLEDGEMENTS

The completion ofthis thesis could never have been done without the support of my committee: Dr. Fred Mackenzie, Dr. Yuan-Hui (Telu) Li and Dr. Edward Laws. As a student I could never have asked for a better committee. The support, encouragement, and willingness ofthe committee members to be always available, no matter when, where or what, to answer questions or just discuss the existential questions oflife, and to share their knowledge and intellect were beyond what any student could wish for. I wish all students the same support that I received from my committee. Since the very first day when I started in the oceanography program, I have met with Fred almost on a weekly basis. Sometimes he even had to deal with me on many more occasions. In addition to my appreciation ofthe magnificent scientific guidance that Fred has given me, I'm extremely appreciative ofthe support and advice he has provided me in times when science has been out offocus for various reasons. Similarly, Telu and Ed have always made themselves available to me and I am tremendously grateful for their willingness to always put the student in focus and for offering their guidance and help. Furthermore I would like to thank the Mackenzie group: Mike Guidry, Dan Hoover, Stephanie Ringuet, May

Ver, as well as Jane Schoonmaker, for their unconditional support and encouragement;

Ricky Grigg for keeping me updated on the ongoing discussion about the Magnesian

Salvation Theory; Jim Cowen and Eric Hochberg for giving me valuable ship and field time; Bob Buddemeier, Jean-Pierre Gattuso and Dave Barnes for providing me with reprints and insightful information about coral reefs and calcification; Rudolf

Kloosterziel for mathematical assistance; Kathy Kozuma and her student helpers Kellie

VI Gushiken and Kellie Shek whom always have a smile to spare; and Varis Grundmanis and Christopher Winn for introducing me to the field ofoceanography. Finally, I would like to acknowledge my mother Ann-Birgitt Andersson, who has always been of tremendous support although I have always gone my own way and moved as far away from Sweden as I possibly could. Since I was born she has always taught me that "the more you know, the more you realize you do not know" and although I have always acknowledged these words ofwisdom, they have never been more clear to me as they are now after completing my master's thesis. Financial support ofmy thesis research was provided by National Science Foundation grants ATMOO-80878 and EAR02-235090 and partial assistance was also provided by Sigma Xi.

VB ABSTRACT

As a consequence ofanthropogenic activities, future projections suggest that the saturation state ofsurface ocean waters with respect to carbonate minerals will decline during the twenty-first century owing to increasing atmospheric C02. As a result calcareous organisms could have difficulty calcifying, leading to production ofweaker skeletons and their greater vulnerability to erosion, and ultimately leading to dissolution ofcalcareous sediments. At the same time, sea surface temperature could be significantly higher and the amount oforganic matter deposited within the coastal zone could also increase owing to human activities. Increased deposition oforganic matter and subsequent remineralization within the sediments ofthe coastal region could have implications with respect to the carbonate geochemistry ofthe pore water-sediment system, affecting rates ofcarbonate dissolution and precipitation. Increasing dissolution ofmetastable carbonate minerals, such as high magnesian calcite has been suggested as a mechanism to restore changes in saturation state and pH owing to increasing atmospheric

C02, acting as a buffer, and could counteract any negative effects on calcareous organisms and communities. In order to investigate the effects ofclimate change and anthropogenic activities on the carbonate biogeochemistry ofthe shallow water ocean enviromnent, a global physical-biogeochemical box model referred to as SOCM

(Shallow-water Ocean Carbonate Model) was developed. Numerical simulations demonstrated that biogenic calcification could decrease by 7-44% throughout the 21 5t century owing to a decrease in carbonate saturation state. Dissolution ofmetastable carbonate minerals could increase owing to increased deposition and remineralization of

Vlll organic matter, but will not result in the production ofsufficient alkalinity to buffer the carbon chemistry ofthe surface ocean water. However, a buffer effect was observed within the pore water system. Sensitivity analysis indicated that the extent ofdissolution was mainly controlled by remineralization oforganic matter rather than reaction kinetics.

In the current standard simulation, the metastable equilibrium ofthe pore water changed from 21 mol% magnesian calcite to 14 mol% magnesian calcite. Future changes in pore water carbonate saturation state could affect the average composition and rates of precipitation ofcarbonate cements in contemporary shallow-water sediments.

IX LIST OF TABLES

1.1 Carbonate mineralogy ofmajor carbonate organism groups 18

2.1 Calcium carbonate production and accumulation in the shallow- water ocean environment in the late Holocene ocean 46

2.2 Relative rate ofcalcification as a function ofaragonite saturation state ...... 55

2.3 Relative rate ofcalcification as a function oftemperature change 57

2.4 Kinetic rate constants and reaction order from selected carbonate precipitation and dissolution experiments 61

2.5 Constants adopted for inorganic CaC03 precipitation and dissolution rates 63

2.6 Specific surface area and reactive surface area for various carbonate grains 64

3.1 Constants adopted for inorganic CaC03 precipitation and dissolution rates 78

4.1 Ranking ofparameters according to their influence on surface water aragonite saturation state and net carbonate dissolution based on the maximum absolute ratio ofvariation (MAROV) 106

4.2 Parameter influence (sensitivity) on surface water buffer effect 106

A.l Carbon reservoirs in the shallow-water ocean environment and adjacent reservoirs in TOTEM 158

A.2 Estimates ofinitial reservoir masses within the shallow-water ocean environment at the onset ofthe model simulation in the year 1700 165

A.3 Estimates ofinitial carbon fluxes within the shallow-water ocean environment at the onset ofthe model simulation in the year 1700 168

x LIST OF FlGURES

1.1 Historical atmospheric CO2 concentration on different time scales 6

1.2 Range ofatmospheric CO2 and temperature in the 21 st century projected by IPCC Special Report on Emission Scenarios 8

1.3 Rate ofcalcification as a function oftemperature for Pocillopora damicornis and Porolithon gardineri , 22

1.4 Rate ofcalcification ofPorites colonies as a function ofannual mean sea surface temperature from different Indo-Pacific coral reefs 23

1.5 Rate ofcalcification as a function ofaragonite saturation state for individual calcareous species 25

1.6 Community net calcification as a function ofaragonite saturation state ...... 27

2.1 Schematic ofShallow-water Ocean Carbonate Model (SOCM) .39

2.2 Inhibition factor ofcalcium carbonate reactivity owing to humic acid and orthophosphate 66

2.3 Schematic ofTerrestrial Ocean aTmosphere Ecosystem Model 69

3.1 Pre-industrial carbon estimates ofshallow-water ocean environment 76

3.2 Changes in biogenic CaCO] production and carbonate saturation state in surface waters and pore waters between 1700 and 2100 80

3.3 Sensitivity analysis with respect to river transported organic carbon and carbonate dissolution rates 82

4.1 Changes in calcite saturation state between 1993 and 1996 at Bermuda Atlantic Time Series (BATS) 93

4.2 Seasonal changes and annual average calcite saturation state in the upper surface layer at BATS 94

4.3 Changes in calcite saturation state between 1989 and 2000 at Hawaii Ocean Time series (HOT) 96

Xl 4.4 Annual changes in average calcite saturation state in the upper 50 m around the main Hawaiian Islands between 1973 and 2000 97

4.5 Sensitivity analysis: sediment organic matter deposition via rivers and organic matter flux from the water column 109

4.6 Changes in total alkalinity as a function ofchanges in total dissolved inorganic carbon upon dissolution ofCaC03 ..••..••••.••••.•••••••••••••••.111

4.7 Sensitivity analysis: sediment organic matter flux via rivers and the effects on CaC03 production, dissolution and accumulation 112

4.8 Sensitivity analysis: sediment organic matter flux from the water column and the effects on CaC03 production, dissolution and accumulation 113

4.9 Organic carbon balance and CO2 gas exchange between the atmosphere and the surface water within the shallow-water ocean environment 117

4.10 Sensitivity analysis: carbonate reaction kinetics 120

4.11 Metastable equilibrium between pore water and the most soluble solid carbonate sediment phase 121

4.12 Sensitivity analysis: initial dissolved inorganic carbon speciation 125

4.13 Pore water saturation state with respect to magnesian calcite of various compositions and the effect on surface water carbon chemistry 127

4.14 Sensitivity analysis: magnesian calcite solubility 132

4.15 Range ofsurface water and pore water saturation state with respect to IS mol% magnesian calcite as an effect ofIPCC SRES C02 emission scenarios 135

4.16 Mauna Loa atmospheric CO2 record and model calculated C02 concentration 136

4.17 Sensitivity analysis: shallow water-open ocean water exchange 138

4.18 Changes in relative rate ofcalcification for various calcareous organisms and communities 141

xu 4.19 Range ofvariability in relative rate ofcalcification as a function ofIPCC SRES predicted CO2 emission scenarios 143

4.20 Range ofvariability in relative rate ofcalcification as a function ofIPCC SRES predicted temperature scenarios 145

4.21 Range ofvariability in biogenic carbonate production adopting a range ofdifferent relationships between calcification, carbonate saturation state and temperature 147

A.l Forcing parameters ofSOCM 154

A.2 Difference in partial pressure ofC02 between surface water and atmosphere 155

XIIl CHAPTER 1: SHALLOW-WATER OCEAN CARBONATE

BIOGEOCHEMISTRY

1.1 INTRODUCTION

Climate system and global carbon models have shown that by the middle ofthe

21 st century, the CO2 concentration in the atmosphere will be more than double the pre­ industrial concentration ofabout 280 ppmv (Houghton et aI., 1996,2001). At that time, the temperature in the atmosphere and surface ocean could be several degrees warmer than at present (Houghton et aI., 1996,2001). Increased atmospheric CO2 will lead to increased invasion ofCO2 into the surface ocean and will subsequently cause a decrease in pH and saturation state with respect to CaC03. Higher sea surface temperatures and decreased CaC03 saturation state could impair the ability ofcalcareous organisms such as corals, coralline algae, coccolithophorids and other taxa to produce skeletons, shells and tests out ofCaCOJ (Gattuso et aI., 1998, 1999a; Kleypas et aI., 1999,2001; Riebesell et aI., 2000; Langdon et ai, 2000; Mackenzie et aI., 2000; Leclercq et aI., 2000, 2002;

Andersson et aI., 2003). The difficulty posed by lower saturation states to the production ofCaC03 could result in calcareous organisms and structures being weaker and more vulnerable to environmental stress and erosion. Subsequently, calcareous organisms and communities could face significant impediments and alterations as a result offuture environmental conditions (Gattuso et aI., 1998, 1999a; Kleypas et aI., 1999; 2000;

Langdon et aI., Mackenzie et aI., 2000; Leclercq et aI., 2000, 2002; Andersson et aI.,

2003). Ultimately, decreased CaC03 saturation state could cause dissolution and alteration ofthe composition ofcarbonate minerals in the water column and within the

I pore water-sediment system. As an alternative hypothesis, it has been suggested that such dissolution, preferentially ofhigh magnesian calcites that are unstable relative to aragonite and calcite, and therefore more soluble, could restore the pH and carbonate saturation state ofthe surface water (Halley and Yates, 2000; Barnes and Cuff, 2000).

Dissolution ofhigh magnesian calcite and buffering ofseawater could protect calcareous organisms such as corals and coralline algae against changes owing to increased atmospheric C02 (Halley and Yates, 2000; Barnes and Cuff, 2000). Sometimes this hypothesis has been referred to as the Magnesian Salvation Theory (Buddemeier, personal communication).

The objective ofthe present thesis was to investigate the effects ofincreasing atmospheric CO2 and temperature on the carbonate chemistry ofthe shallow-water ocean environment including coastal zones, shelves, banks and reefs, and the calcareous organisms living within this region. The intention was to elucidate how changes in the inorganic carbon chemistry and the saturation state with respect to carbonate minerals in the water column and within the pore water-sediment system could alter biotic and abiotic production and preservation ofcarbonate minerals. The main objective was to clarify whether dissolution ofmetastable carbonate minerals such as high magnesian calcites could restore any changes in pH and saturation state owing to increased

atmospheric C02 and provide a temporal refuge for calcareous organisms in the shallow­ water ocean environment. In addition, the effect and magnitude ofany changes in

biogenic carbonate production and carbonate composition and dissolution/precipitation

chemistry within the pore water-sediment system were also investigated. To address

2 these problems, a biogeochemical box model representative ofthe shallow-water ocean environment, referred to as SOCM (Shallow-water Ocean Carbonate Model) was constructed. To facilitate inputs and outputs at the boundaries ofthis region, SOCMwas integrated into the global biogeochemical model TOTEM (Terrestrial Ocean aTmosphere

Ecosystem Model; Ver, 1998; Ver et aI., 1994, 1998, 1999a, 1999b; Mackenzie et aI.,

1998a, 1998b, 2000, 2002).

The hypotheses tested were:

I. Changes in CaC03 saturation state within the shallow-water ocean

environment owing to increased atmospheric CO2 are restored by

dissolution ofmetastable carbonate minerals.

II. Biogenic carbonate production among calcareous organisms such as corals

and coralline algae is not inhibited by increased atmospheric CO2 due to

dissolution ofmetastable carbonate minerals.

In addition to the hypotheses stated, multiple sensitivity analyses were carried out to investigate the robustness ofthe model results and to assess the importance and effect of various processes and variables such as initial carbonate mass, initial inorganic carbon chemistry and carbonate saturation state, carbonate dissolution and precipitation kinetics, river transport and remineralization oforganic matter.

The results ofthe simulation ranging from year 1700 to 2100 A.D. indicated that metastable carbonate minerals within the pore water-sediment system during the 21 5t century could dissolve owing to undersaturated conditions with respect to carbonate minerals. Dissolution was mainly driven by increased remineralization oforganic matter

3 originating from in situ production and river input rather than changes in the atmospheric

CO2 content. Increasing atmospheric CO2 caused a substantial decrease in surface water carbonate saturation state. However, the surface water remained supersaturated even with respect to the most soluble carbonate phase (15 mol% magnesian-calcite) used in the standard simulation by year 2100.

Dissolution ofmetastable carbonate minerals within the pore water-sediment system produced a buffer effect within this reservoir, slowing down any changes in the inorganic carbon chemistry imposed by remineralization oforganic matter and subsequent release ofC02. None ofthis buffer effect was observed in the overlying surface water. As a result, carbonate production by calcareous organisms was impaired by 7-44% during the extent ofthe standard simulation. Consequently, calcareous organisms and communities could in the near future be significantly weakened and altered relative to today. However, some uncertainty exists on how increasing temperature will affect biogenic calcification, and at present it is difficult to ascertain what the combined effect on calcareous organisms ofdecreasing carbonate saturation state and increasing temperature will be. In addition to the above findings, future decreases in pore-water saturation state could affect the composition and rates of precipitation ofcarbonate cements in recent shallow-water environments.

4 1.2 BACKGROUND OF RESEARCH

1.2.1 Atmospheric CO2 and global warming

th During the 20 century, the atmospheric CO2concentration has increased from

-315 ppmv in 1958 when measurements first started at Mauna Loa, Hawaii, to -371 ppmv in year 2001 (Fig 1.lA; Keeling and Whorf, 2002). Model predictions for the 21 st century suggest that the C02 concentration in the atmosphere by the end ofthe century will be close to -700 ppmv (Houghton et aI., 1996,2001). At this time, according to the results ofthe \PCC (Intergovernmental Panel ofClimate Change) Special Report on

Emission Scenarios (SRES; Houghton et aI., 2001), the CO2concentration is projected to range between 540 to 970 ppmv (Fig 1.2A), depending on the eVOlution ofthe "emission driving forces" such as population growth, economic growth, technological development and environmental protection (Houghton et al., 2001). Given present trends, economic forces and the attitude among policy makers, it is very unlikely that CO2concentrations outside these limits will be observed by year 2100. On a historical time perspective of hundreds ofthousands ofyears, the present atmospheric CO2concentration is higher than

any concentration observed during the last 420,000 years as inferred from analysis ofgas bubbles trapped in ice from the Vostok ice core (Fig l.1B; Petit et aI., 1999; Barnola et

aI., 2003). During this period oftime, the earth went through four glacial and five

interglacial stages. Atmospheric C02 concentration fluctuated between -180 ppmv

during glacial stages and -280 ppmv during interglacial stages (petit et aI., 1999; Barnola

et aI., 2003). Prior to the industrial revolution, the CO2concentration was -280 ppmv. On

an even longer time scale oftens ofmillions ofyears, atmospheric CO2concentration has

5 O..l- ..L. 6 108 5 108 4 108 3 108 2 108 1 108 a Time (years ago)

Figure 1.1. Historical atmospheric CO2 concentration (ppmv) on different time scales until present ranging from (A) decades (Mauna Loa record, Hawaii; Keeling and Whorf, 2002), (B) hundreds of thousands of years (Vostok ice core; Bamola et aI., 2003), and (C) hundreds ofmillions ofyears (Model results from the Geocarb ill model (Berner and Kothavala, 2001). Note the different scales in CO2 concentration. The C02 scales in (A) and (B) are the same to highlight the rapid increase of CO2 during the past decades and currently high atmospheric CO2 concentration relative to the last 420,000 years. Interannual fluctuation in (A) represents changes in atmospheric CO2 due to differences in seasonal effects between the northern and southern hemispheric winter and summer. In (B), high CO2 concentrations correspond to interglacial time stages and low concentrations to glacial stages.

6 at times been many times higher than at present (Fig l.IC; Berner, 1991, 1994; Berner and Kothavala, 200I; Crowley and Berner, 2001; and references therein). During those times the temperature ofthe atmosphere was also higher, and in fact climatic changes in the geological past have been attributed largely to changes in atmospheric CO2content

(Berner and Lasaga, 1999; Crowley and Berner, 2001). However, it is not clear whether

CO2is forcing the climate to change or vice versa. In fact, CO2 appears to lag changes in temperature by centuries in the Vostok ice core data (petit et aI., 1999; Caillon et al.,

2003). In addition, growing evidence suggests that celestial processes, such as changes in solar and cosmic ray activity at least to some extent may be responsible for climate variability on time scales ranging from days to millennia (Friis-Christensen and Lassen,

1991; Svensmark, 1998; Solanki, 2002; Shaviv and Veizer, 2003).

The climate relationship between earth surface temperature and atmospheric CO2 concentration comes from the fact that CO2 along with other gases such as H20, C~ and

N20 is a greenhouse gas. It is important to note that water is the most important greenhouse gas at present. The greenhouse mechanism allows short-wavelength radiation from the sun to penetrate the atmosphere but traps this energy when it is reradiated from the earth as long-wave radiation, causing a subsequent warming ofthe earth surface. The concern about global warming lies in the relationship between atmospheric CO2and temperature; the more C02, the higher the temperature. The greenhouse properties ofCO2

and the consequences ofadding it to the atmosphere were discussed by the Swedish scientist Svante Arrhenius in 1896, who stated that humans might be increasing the temperature ofearth by "evaporating our coal mines into the air" (Arrhenius, 1896, in

7 1000 A Scenarios • •• 900 -A1B .• > ·····A1FI •• -E --An •• a. 800 • S •• r:: -61 •- 0 •­ -.:; 700 -B2 • ltl ... -IS92A •-- r:: •- -0) 600 (.) .--­ r:: 0 (.) N 500 0 (.) 400

300 B 5 -(.) e....- O) ••.­ C) 4 .- r:: _ .. ltl .•• .c. (.) ••••• ...0) 3 . ::J .- ­ -...ltl - 0) 2 - a. .,.;; -- E 0) I- 1

0 2000 2020 2040 2060 2080 2100 Year

Figure 1.2. IPeC Special Report Emission Scenarios (SRES; Houghton et aI., 2001). Results ofa range ofdifferent model predictions for the 21 st century adopting different CO2 emission forcing scenarios (see Houghton et aI., 2001 for a detailed description ofthese scenarios). (A) Atmospheric e02 concentration (ppmv). (B) Temperature change (OC) relative to 1990. The gray shaded region represents the range ofthe results for several general circulation models.

Pilson, 1998). Current predictions from IPCC SRES scenarios suggest that by the end of the 21 st century, the earth surface mean temperature will range between 1.4 to 5.8 °c warmer than that ofyear 1990 (Fig 1.2B; Houghton et aI., 2001). In any discussion

regarding greenhouse gases and the greenhouse effect, it is important to remember that in

8 a situation with no greenhouse gases in the atmosphere, the temperature ofthe earth surface would be approximately -20°C, and the planet would be completely frozen

(Kasting et aI., 1988). In conclusion, to summarize the effects ofthe anthropogenically induced flux ofCO2 into the atmosphere, we can anticipate that the CO2 concentration will be anywhere between 540 to 970 ppmv and the global mean surface temperature 1.4-

5.8 °C above that ofthe year 1990 by the end ofthe 21 st century (Fig 1.2; Houghton et a!., 2001).

1.2.2 CO2 in seawater and CaC03 saturation state in the ocean

As the CO2 concentration in the atmosphere increases, the invasion ofCO2 into the ocean will also increase. The chemistry ofCO2 in the ocean is complex because CO2 gas reacts with water to form carbonic acid and other dissolved inorganic carbon species.

The chemical system involved is usually referred to as the dissolved inorganic carbon

(DIe) system or the carbonic acid system. The distribution ofchemical species in this system in seawater is a function oftemperature, salinity and pressure. The system can be described by the following equations:

k~ CO2(g) = CO2(aq) (1.1)

k' CO2(aq) + H20 :; H2CO~ (1.2)

k' o I _ + H 2 C03 = HC03 + H (1.3)

(1.4)

9 where the concentration ofC02(aq) is given by Henry's law with kH ' being the stoichiometric solubility coefficient ofCO2 in seawater, ko', k j ' and k2' are stoichiometric equilibrium constants, often referred to as the dissociation constants ofcarbonic acid.

These reactions can be added, resulting in the following overall equation:

CO2(g) + H20 + CO/- = 2HC03- (1.5)

The total dissolved inorganic carbon (DIC) is defined by the sum ofthe dissolved forms ofCO2, HC03-, and CO/-:

2 DIC = [C02] + [HC03-] + [C03 -] (1.6)

As can be inferred from equation (1.3) and (1.4), pH (= -log{W}) is strongly related to the equilibrium ofthe DIC system and the relative proportions ofthe carbonate species.

The carbonate system functions as a buffer for seawater pH and on time scales up to several thousands ofyears, the pH ofseawater is controlled mostly by this system

(Pilson, 1998). The ability ofcertain substances in seawater to react with hydrogen ions during titration with a strong acid until these substances essentially are completely protonated is referred to as the Total Alkalinity (TA):

(1.7)

+ minor components

The total alkalinity is strongly related to the charge balance ofseawater. The most important contribution to the TA comes from the carbonate species and is referred to as the Carbonate Alkalinity (CA):

CA = [HC03-] + 2[CO/-] (1.8)

10 The amount ofCO2 that can dissolve in water and thereby also the flux between the ocean and the atmosphere is determined by Henry's Law. At equilibrium, the partial pressure ofCO2 (PC02) in seawater is equal to the partial pressure ofthe gas in the atmosphere at sea level. Prior to the industrial revolution, the partial pressure ofCO2 in the surface ocean was on average higher than in the atmosphere, and the ocean served as a net source ofcarbon to the atmosphere (Smith and Mackenzie, 1987; Wollast and

Mackenzie, 1989; Wollast, 1998). At present, the human-induced increase ofCO2 in the atmosphere has caused a reversal ofthis flux so that the oceans now serve as a net sink of carbon dioxide (Smith and Mackenzie, 1987; Mackenzie et al., 2002). In general, at high latitudes and where the surface water is cold, there is a net flux ofC02 from the atmosphere into the ocean, and in low latitudes, centered around the equator and near­ equatorial upwelling regions, the opposite is usually observed (Pilson, 1998). As more and more CO2 is taken up by the ocean, its capacity to absorb additional CO2 will decrease or the Revelle factor (RF), i.e. the ratio ofrelative change in CO2 to the relative change in DIC will increase (Revelle and Suess, 1957; Broecker and Peng, 1982; Pilson,

1998; Zeebe and Wolf-Gladrow, 2002). Typical values ofRF in the ocean are between 8 and 15 (Zeebe and Wolf-Gladrow, 2002).

It can be inferred from equation (1.5) that as the atmospheric CO2 concentration increases and subsequently the invasion ofCO2 into the ocean, the COJ 2- concentration will decrease as these two species along with water will react to form bicarbonate. In other words "basic" COJ 2- will be titrated by "acidic" CO2, with no change in total

2 alkalinity. A decrease in C03 - concentration will have significant implications on

11 seawater saturation state with respect to carbonate minerals such as calcite (CaC03).

Calcium carbonate saturation state (0) can be defined as the product ofthe concentration ofcalcium ions and the concentration ofcarbonate ions divided by the stoichiometric solubility product (K'sp) at the in situ conditions oftemperature, salinity and pressure:

0= [Ca'+][CO;-] (1.9)

K;p(caco3 ) where K'sp(cacoy is determined from the equilibrium:

(1.10)

Since [Ci+] can be considered nearly conservative in seawater, the saturation state with respect to calcium carbonate will mainly be determined by the carbonate ion concentration. Similarly, saturation state with respect to magnesian calcite minerals can be defined as,

0= rCa,+ ](l-X) [Mg'+y[CO;-] (1.11) K;p(Mg-Calcite) where x refers to the magnesium content in mol% for the magnesian calcite phase under consideration and K'SP(Mg-caJCite) is determined from the equilibrium:

(1.12)

At present, all surface waters ofthe oceans are in general supersaturated with respect to calcite, whereas deep waters are undersaturated. Although the degree of undersaturation increases with depth, the rate ofcalcite dissolution remains relatively constant down to a certain depth referred to as the lysocline. At this depth the rate of calcite dissolution rapidly increases until the rate ofdissolution is equal to the rate at

12 which calcite is delivered to the seafloor. This depth is referred to as the calcite compensation depth (CCD). Below the CCD, the rate at which calcite particles dissolve is fast enough that no significant accumulation is found on the bottom. A similar trend, but at different depths, can be observed with respect to other carbonate minerals such as aragonite. Decreased saturation state and subsequent dissolution ofCaC03 with depth can be attributed to several factors. Because ofthe physical and chemical properties of calcium carbonate, it becomes more soluble as the temperature decreases. In a future of increased temperature and atmospheric C02, the temperature increase would counteract the effect ofincreased CO2 on carbonate saturation state, but the effect is minor. More important are the effects ofpressure and temperature on the dissociation constants of 2 boric and carbonic acid, which cause the pH and the C03 - concentration to decrease and therefore also the saturation state. In addition, and probably most important, respiration by organisms at depth produces CO2, which further decreases the pH and hence col­ concentration and carbonate saturation state. Because ofthe global circulation pattern of deep water, water in the deep Pacific Ocean is older and more corrosive with respect to carbonate minerals than the deep water ofthe Atlantic Ocean. Consequently, the carbonate compensation depth is shallower in the Pacific Ocean than in the Atlantic

Ocean (Morse and Mackenzie, 1990; Lisitzin, 1996).

Recently, it has been suggested that a large fraction ofcalcium carbonate (60­

80%) produced in the surface waters ofthe oceans is dissolved at a depth of500-I000 meters, a depth well above the chemical lysocline ofthe water column (Milliman et aI.,

1999; Iglesias-Rodriguez et al., 2002). The mechanism(s) responsible for this dissolution

13 is not completely understood. Hypotheses proposed include dissolution within microenvironments, such as sinking particles and organic flocs where bacterial respiration drives conditions toward undersaturation, or possibly, within the guts of zooplankton where acidic conditions also could facilitate undersaturation and subsequent dissolution (Milliman et a!., 1999). Alternatively or in addition to the above mechanisms, dissolution at shallow depths represents dissolution ofmetastable carbonate phases such as high magnesian calcite exported to the open ocean from banks and other shallow environments (Agegian et a!., 1988; Sabine and Mackenzie, 1995).

The global shallow-water ocean environment, which can be defined as the depth ofthe seafloor < 200 meters, i.e. the coastal zone, shelves, banks, reefs, etc is located well above the calcite chemical lysocline. At present, the surface water ofthese regions is significantly supersaturated with respect to calcite, aragonite and a range ofmagnesian calcite compositions; nevertheless with the possible exception ofa few locations such as the Great Bahama Bank and the Persian Gulf, no spontaneous precipitation ofcarbonate minerals is observed to occur within the water column in these regions. The lack of carbonate reactivity in seawater, i.e. the resistance ofcarbonate minerals to undergo dissolution or precipitation reactions, is most often accounted for by the presence and inhibition ofmagnesium ions (pytkowicz, 1965; Berner, 1975; Morse, 1983; Tribble and

Mackenzie, 1998), phosphate ions (Berner et a!., 1978; Morse, 1983; Mucci, 1986;

Burton and Walter, 1990; Dove and Hochella, 1993), and dissolved organic matter

(Chave and Suess, 1967; Berner, et a!., 1978; Morse, 1983; Lebron and Suarez, 1996,

1998). Although the shallow-water ocean environment is well above saturation with

14 respect to carbonate minerals, undersaturation can be found within certain microenvironments such as particles and organic floes, and also within the interstitial pore waters ofsediments (Morse and Mackenzie, 1990). In general the inorganic carbon chemistry ofthe pore water is significantly different from that ofthe overlying surface water. The concentration ofdissolved phosphate and dissolved organic matter within pore waters can be high due to decomposition oforganic matter and can significantly inhibit carbonate precipitation (Berner et al., 1978; Mucci, 1986; Burton and Walter, 1990;

Lebron and Suarez, 1996). Similarly, microbial processes in the pore water-sediment system can create conditions that either prevent or mitigate against, carbonate dissolution or precipitation (see subsequent discussion; Berner, 1971; Milliman, 1974; Ritger et aI.,

1987; Moulin et aI., 1985; Morse and Mackenzie, 1990).

1.2.3 CaCO, minerals in natural environments

In natural environments several forms ofcarbonate minerals are present. In modem, shallow-water carbonate sediments, aragonite and high magnesian calcite are the predominant carbonate phases whereas calcite or low magnesian calcite is dominant in pelagic sediments located above the calcite compensation depth. In ancient sedimentary rocks, the stable mineral assemblage oflow magnesian calcite and dolomite

(CaMg(C03)2) constitutes more than 90% ofnatural carbonate occurrences (Reeder,

1983). To some extent this is due to the fact that aragonite and high magnesian calcite are unstable relative to calcite and dolomite. Given the right conditions, aragonite and high magnesian calcite should spontaneously convert into the stable phases ofcalcite and

15 dolomite at earth surface conditions oftemperature and pressure (Land, 1967; Morse and

Mackenzie, 1990; Tribble et al., 1995).

Calcite or low magnesian calcite « 4 mol% Mg) has a rhombohedral mineral structure and is the stable form ofCaC03• Aragonite has an orthorhombic structure and is unstable at low temperature and pressure relative to calcite. Under earth surface temperature and pressure conditions, it is commonly a metastable phase, i.e. thermodynamically it is unstable, but it persists because ofkinetic factors. High magnesian calcite (> 4 mol% Mg) is also metastable under earth surface temperature and pressure conditions and is unstable relative to aragonite for phases with mol% Mg > 8 mol% (Bischoff et a!., 1993). Magnesian calcite is an isomorph ofcalcite. The instability ofhigh magnesian calcite relative to calcite and aragonite arises from the replacement of

2 2 Ca + ions with the significantly smaller Mg + ions, which subsequently causes variation and asynnnetry in the mineralogical structure and increases its solubility (Reeder, 1983;

Tribble et a!., 1995). However, the solubility ofmagnesian calcite is not only controlled by Mg concentration. Other physical and chemical properties exert a significant effect on stability such as the presence oftrace metals other than Mg, carbonate ion position or cation ordering, microstructural and surface defects, and adhered small particles

(Bischoff et al., 1993). High magnesian calcites are preferentially produced bybiological processes even though recrystallization into stable phases with a lower mol% Mg occurs abiotically during early diagenesis (Morse and Mackenzie, 1990; Bischoffet al., 1993;

Tribble et a!., 1995). In addition, experimental observations have indicated magnesian calcite overgrowths on calcite seeds submerged in seawater, although little evidence for

16 this exists in the natural environment (Wollast et aI., 1980; Tribble and Mackenzie,

1998).

The range ofmol% Mg present in naturally occurring magnesian calcites varies from -0 to 30 mol%, with an average around 12-15 mol% (Garrels and Mackenzie, 1981;

Morse and Mackenzie, 1990). The amount ofmagnesium present in the tests or skeletal parts ofcarbonate-secreting organisms is species specific, and dependent on such factors as skeletal mineralogy, temperature, salinity, and carbonate saturation state (Chave,

1954a; 1954b; Mackenzie et aI., 1983; Morse and Mackenzie, 1990; Rao, 1996).

1.2.4 Production ofCaC01

At present, the major producers ofcalcium carbonate in the oceans are calcareous organisms such as corals, bryozoans, coralline algae, coccoitihophorids, pteropods, and foraminifera, which produce skeletons, shells or tests out ofcalcium carbonate (Table

1.1; see chapter 2 for a detailed review ofthe carbonate budget ofthe shallow-water ocean environment). The reasons why some organisms calcify and what evolutionary advantage calcification and the production ofcalcareous skeletons, shells or tests might give to organisms are poorly known. One advantage afforded organisms with calcareous shells or tests comes from the obvious protection that hard parts provide against predators and grazers as well as structural stability (Milliman, 1974). Corals, which produce extensive skeletons and reefstructures out ofcalcium carbonate, could achieve advantages due to the large increase in surface area and availability ofspace that these structures provide (Falkowski, personal communication). Biochemical considerations suggest that calcification is an adaptation for autotrophs to utilize bicarbonate as a source

17 TABLE 1.1. CARBONATE MlNERALOGY OF MAJOR ORGANlSM GROUPS Organism Mineralogy Mol%MgC03 Annelid worms Calcite, aragonite 7.1 - 20.2 Asteroids Calcite 10.7 - 19 Bryozoa Calcite, aragonite 0.2 - 13 Calcareous algae Calcite, aragonite 8.3 - 35 Calcareous brachiopods Calcite 0.6 -10.7 Calcareous sponges Calcite 0.08 - 0.7 Cephalopods Aragonite Trace - 0.4 Coccolithophorids Calcite Crinoids Calcite 8.3 - 19 Crustaceans Calcite, calcium phosphate 1.2 -19 Echinoids Calcite 4.8 -19 Foraminifera Calcite <1.2 - 19 Gastropods Calcite, aragonite 0 - 2.4 Octacorals Calcite 0.4 - 19 Ophiuroids Calcite 0.2 - 13.1 Pelecypods Calcite, aragonite 0 - 3.6 Pteropods Aragonite Scleractinian corals Aragonite 5.9 - 16.7 Table modified after Morse and Mackenzie (1990)

ofcarbon to be used in the process ofphotosynthesis (Sikes et al., 1980; Sikes and

Wilbur, 1982; McConnaughey and Whelan, 1997; McConnaugheyet al., 2000) and/or a mechanism ofnutrient uptake powered by a proton pump (McConnaughey and Whelan,

1997; McConnaughey et al., 2000), which in both cases would be facilitated by the process ofcalcification. The latter suggestion might be supported by the fact that calcareous organisms and communities are commonly present in nutrient-poor, oligotrophic seawater conditions. For example, coral reefs thrive under such oligotrophic

18 conditions and coccolithophorids have been observed to bloom under nitrogen-depleted conditions following an initial bloom ofdiatoms (Fernandez et aI., 1996).

An intrinsic relationship between the process ofcalcification and photosynthesis seems to exist (McConnaughey and Whelan, 1997; Gattuso et aI., 1999a). Photosynthesis utilizes CO2, water and light to produce organic matter,

h, CO2 + H20 = CH20 + O2 (1.13)

As CO2 is withdrawn from the water column or within the calcifying cells ofan organism, the pH will increase and the equilibrium ofthe carbonic acid system will shift

2 in favor ofC03 - and a subsequent increase in carbonate saturation state. The higher the saturation state, the higher the rate ofcalcification and production ofcarbonate minerals.

The process ofcalcification will cause a release ofCO2 according to the following reaction:

(1.14)

The CO2 produced in the calcification process could then be utilized and reduced to organic matter by the process ofphotosynthesis. The net reaction is described by the following equation:

(1.15)

In scleractinian corals containing zooxanthellae, photosynthesis and calcification have been observed to be tightly coupled, as supported by the fact that the rate ofcalcification is several times higher during daylight hours than in darkness (Gattuso et aI., 1999a).

In some areas with favorable environmental conditions such as the Great Bahama

Bank and the Persian Gulf, CaC03 might spontaneously precipitate out ofthe water

19 colunm. The phenomenon is quite spectacular and can clearly be observed from an aerial view as a white cloud within the water column. The exact mechanisms ofthese precipitation events, which are referred to as whitings, are not known and have historically been the subject ofsubstantial controversy (Cloud, 1962; WeBs and Illing,

1964; De Groot, 1965; Bathurst, 1975; Morse et aI., 1984; Shinn et aI., 1989; Boss and

Neumann, 1993; Robbins et aI., 1997; Broecker et a!., 2000). The fundamental problem has been whether whitings represent directly precipitated calcium carbonate from the water colunm or simply resuspended sediments. Recent findings suggest that the major removal ofcarbonate from the water colunm ofthe Great Bahama Bank is dominated by the slow precipitation ofcalcium carbonate on suspended sediments, and that the major fraction removed is outside ofthe whitings (Morse et al. 2002; Morse et a!., 2003).

Another area where spontaneous precipitation ofcarbonate minerals occurs is within the interstitial waters ofsediments. Biological processes, for example microbial sulfate reduction or methane oxidation, can produce alkalinity and drive the conditions within the pore waters toward supersaturation with respect to CaC03 (Berner, 1971;

Milliman, 1974; Ritger et aI., 1987; Morse and Mackenzie, 1990). Subsequently, this increase in saturation state can facilitate inorganic carbonate precipitation and the formation ofcarbonate cements and aragonite muds. Although many complex reactions are involved, the process ofsulfate reduction can be represented as:

(1.16)

The reaction results in a decrease in sulfate concentration and pH, and an increase in total alkalinity, total dissolved inorganic carbon and hydrogen sulfide concentration. During

20 the initial stages ofsulfate reduction, the effect oflowered pH can be greater than that of increased alkalinity and the pore water saturation state with respect to carbonate minerals could actually decrease and initiate carbonate dissolution rather than precipitation (Ben­

Yaakov, 1973; Morse and Mackenzie, 1990). Near the base ofthe sulfate reducing zone, oxidation ofmethane coupled to sulfate reduction may produce alkalinity and initiate abiotic carbonate precipitation (Reeburgh, 1980; Devol and Ahmed, 1981; Ritger et aI.,

1987):

CH4 + sol- = HCO)- + HS- + H20 (1.17)

Although the resulting carbonate cements are referred to as inorganic or abiotic, it is important to note that this abiotic precipitation in modem sediments is most often driven by biological processes.

1.2.5 The effect of global warming and increasing atmospheric CO2 on

marine calcareous organisms and communities

The future predicted changes in atmospheric C02 and temperature and subsequent effects on seawater chemistry could have significant implications for marine organisms and ecosystems. Because calcareous organisms are sensitive to carbonate saturation state, they might be faced with significant stress in a future oflower saturation states, and have difficulty calcifying, leading to production ofweaker skeletons, and hence more vulnerability to erosion (Gattuso et aI., 1998, 1999a; Kleypas et aI., 1999,2001; Langdon et aI., 2000; Mackenzie et aI., 2000; Leclercq et aI., 2000, 2002).

It is not well known how calcareous organisms will respond to increasing temperatures. Experimental evidence indicates that the calcification rates for corals and

21 700 A Pocillopora damicornis ... --:c 600 • 1---1 "!- E

EM 500 0 U ctl U 400 0) c: • -Q) 300 (22.8°C) ctl • -~ c: :=0 200 r3 10: I--t • '0 100 (ij (22.6°C) U 0 •

-:c 1.6 c: -0 E E 1.4 E -Q) -~ 1.2 c: :=0 r3 10: 1.0 '0 (ij u 0.8

16 18 20 22 24 26 28 30 32 34 36 38 Temperature (0C)

Figure 1.3. Rate ofcalcification as a function oftemperature. (A) Hermatypic scleractinian coral, Pocillopora damicornis (Clausen and Roth, 1975). Temperatures shown in parenthesis represent the average temperature during the experimental incubation. Horizontal bars indicate the average temperature in the natural environment during time ofcollection ofcoral colonies. (B) Coralline algae, Porolithon gardineri (Agegian, 1985; Mackenzie and Agegian, 1989). The horizontal bar indicates the ambient temperature.

22 coralline algae exhibit a negative parabolic relationship to temperature (Fig 1.3; Clausen and Roth, 1975; Agegian, 1985; Mackenzie and Agegian, 1989). This relationship suggests that calcareous organisms most efficiently calcify within a narrow range of temperatures close to the ambient temperature ofthe environment in which the organism lives. Ifthe temperature either exceeds or goes below this temperature range, the rate of calcification will decrease. However, observational evidence suggests that a direct positive linear relationship exists between the rate ofcalcification and temperature among corals from different locations (Fig 1.4; Grigg, 1981, 1992; Lough and Barnes, 2000).

This observation suggests that a moderate rise in temperature within a restricted

..... ­'>. 2.0 N 'E 0) -~ • ~ 1.5 • • c: o ~ ;0:: .~ 1.0 U • Great Barrier Reef (Lough and Barnes. 2000) • Hawaiian Archipelago (Grigg, 1981, 1997) Phuket. Thailand (Scoffin et al.. 1992) 0.5 +-~---r--...... ----r--..,.-~--.---r-----.---r--~--+ 23 24 25 26 27 28 29 Annual mean temperature (OC) Figure 1.4. Rate ofcalcification (g m-2 yr-l) ofPorites colonies as a function of annual mean sea surface temperature eC) from different Indo-Pacific coral reefs (Lough and Barnes, 2000).

23 range possibly could have a positive effect on the rate ofcalcification in corals and other calcareous organisms. However, as inferred from experimental results, ifthe rate of change in temperature is rapid, the impact on the rate ofcalcification might be negative

(Clausen and Roth, 1975; Agegian, 1985; Mackenzie and Agegian, 1989).

The occurrence and distribution ofcoral reefs around the world are limited to areas with an average sea surface temperature> 18°C (Kleypas et aI., 2001). At present, the geographical distribution ofreefs does not appear to be limited by any maximum average sea surface temperature, but negative impacts such as coral bleaching have been strongly correlated with abnormally high sea surface temperatures (Cook et aI., 1990;

Gates, 1990; Lesser et aI., 1990; Glynn, 1993; Jones et aI., 1997; Lesser, 1997; Kleypas et aI., 2001). However, some evidence indicates that corals possibly can acclimate and adapt to bleaching caused by solar and temperature stresses (Buddemeier and Fautin, 1993;

Brown et aI., 2002). Exactly how corals and other calcareous organisms will respond to the predicted higher temperatures in the future is not known, but an important concern is how rapidly these temperature changes will take place.

Experimental investigations on coccolithophorids, corals, coralline algae and calcareous communities have shown that there exists a strong positive relationship between the rate ofcalcification and the saturation state ofseawater with respect to

CaCOJ. Some studies indicate that the relationship is linear (Borowitzka, 1981;

Mackenzie and Agegian, 1989; Gao et aI., 1993; Langdon et aI., 2000; Riebesell et aI.,

2000; Leqlercq et al.,2000, 2002) whereas other investigations indicate that the relationship is non-linear, and the rate ofcalcification eventually approaches an

24 , , I I I I 100 , 28 A B I '~ , I , ~ g 9.5 , ~ 24 - I · , I , I • ~ 20 ~ 90 , - I · , H u , I f f g 85 , 16 - ,I ~ , i 6.0 12 I . , t - I , f I g7.5 , ~ 8 ~ - , f - u'" , Emili,mla huxloyi :, GcphyroOOpf;:J ocoanica 7.0 . 4

, I I I , 550 , . C , D , • SOD - ~ - , I , • ~ 450 - I . , f f I , f I • • , ~ 400 , , f ~ 350- , · , , , 300 I , , - , 250 - I ~ 200- •, ~ •r::.;tyiDphora ,Ystil/Ilfa , A,.;mpora SIl . + 150 . I , I 22 I I f ,I 2.0 ~ ~ - - I sc , g 16 - I • - f I -• E 18 , ~ .to - , I. • c ¥ 1.4 I. • : § 1-2 - .. ~ ·0 , • 8 LO - , ~ PoroJithon gardineri u.~ .'I , • o 2 4 6 8

Arc1Qonite seturation $tate Figure 1.5. Rate ofcalcification as a function ofaragonite saturation state for individual calcareous species. (A-B) coccolithophorids (Riebesell et al., 2000; Zondervan et al., 2001), (C-D) scleractinian corals (calcification rates were nonnalized to protein content ofthe experimental subjects, nmol CaC03 (mg proteinfl h-1; Gattuso et al., 1998), (E) coralline algae (calcification rates were detennined based on uptake of45Ca and by measuring the activity: counts per minute (ct min-I) per tip per hour; Smith and Roth, 1979), (F) coralline algae (Agegian, 1985; Mackenzie and Agegian, 1989). Note that the scales and the units ofcalcification (y-axis) are different between species. Aragonite saturation states are estimated based on DIC data given in each experiment.

25 asymptote as the seawater becomes increasingly supersaturated with respect to calcium carbonate (Fig 1.5 and 1.6; Smith and Roth, 1979; Gattuso et al., 1998). In either case, as the carbonate saturation state decreases, the rate ofcalcification ofcalcareous organisms will also decrease. Based on biogeochemical models and experimental results on various calcareous organisms and communities exposed to elevated pCOz conditions similar to those predicted by the end ofthe 21 st century, the rate ofcalcification ofcalcareous organisms and communities is projected to decrease by 9-40% by year 2065 (Gattuso et al., 1998, 1999a; Kleypas et aI., 1999,2001; Langdon et al., 2000; Mackenzie et al.,

2000; Leclercq et aI., 2001, 2002). The maintenance and expansion ofa coral reefare dependent on the amount ofCaC03 produced by its calcifiers, minus the sum of chemical, physical and biological erosion rates. Considering a decrease in CaC03 production and a potential increase in chemical and physical erosion, which has been proposed as a possible effect ofclimate change, one can conclude that coral reefs and other calcareous organisms may be faced with significant difficulties to sustain themselves in a near future.

Because ofthe intrinsic relationship between calcification and photosynthesis, an important aspect to consider in making predictions for the future is how increased atmospheric COz will affect marine primary production and community production.

Because ofthe large concentration ofdissolved inorganic carbon present in seawater relative to the concentrations ofother plant macronutrients such as nitrate and phosphate, it was traditionally assumed that marine primary producers were not limited by carbon

26 400 0

0 III '7 300 >. <00

o

2 3 4 5 6 Aragonite saturation state

---B- Coral community - light (Leclercq et aI., 2002) ----.- Coral community - dark (Leclercq et aI., 2002) -B- Sand community - light (Leclercq et aI., 2002) -+- Soft bottom community -light (Boucher et aI., 1998) ____ Biosphere 2 - long term (Langdon et aI., 2000) ~ Biosphere 2 - short term (Langdon et aI., 2000) ....Ji;,}- Coral community -light (Leclercq et aI., 2000) -.- Coral community - dark (Leclercq et aI., 2000) -m- Coral community (Ohde and Woesik, 1999)

Figure 1.6. Community net calcification as a function ofaragonite saturation state. Experiments conducted in light and dark conditions. Negative values represent net dissolution ofcarbonate minerals. Solid lines represent the best linear fit to the data, except for the data ofOhde and Woesik (1999), which is represented by a power equation.

27 for photosynthesis (Riebesell et al., 1993; Raven, 1993, 1997). However, marine primary producers can either utilize dissolved molecular COz(aq) or HC03- as a source ofcarbon.

The largest fraction ofdissolved inorganic carbon in the ocean is in the fonn of bicarbonate (-2 mmol kg-I). Consequently, organisms utilizing bicarbonate are most likely not limited by this compound. However, organisms that utilize dissolved molecular

COz as a source ofcarbon for primary production are faced with greater difficulty, because the concentration ofdissolved molecular COz is relative low in seawater (-10

).lmol kg-I). Furthennore it should be noted that most photoautotrophs can utilize both fonns as a source ofcarbon (Raven, 1997). Experiments conducted on diatoms indicated that these siliceous phytoplankton were indeed limited by the supply ofCOz and that productivity increased with increasing pCOz (Riebesell et aI., 1993). Similarly, coccolithophorids exposed to elevated pCOz showed a small increase in organic production, although, in agreement with other experiments conducted on corals and coralline algae, the rate ofcalcification decreased with increasing COz (Riebesell et aI.,

2000).

An increase in photosynthesis and carbon fixation owing to increasing pCOz could counteract the negative effect ofpCOz increase on calcification (Gattuso et aI.,

1999a; Leclercq et al., 2002). Ifsuch a response were to occur, according to Leclercq et al. (2002), there are two different scenarios to be considered. Assuming that the response time ofphotosynthesis and calcification are similar, increased carbon fixation might partly balance the effect ofincreasing pCOz on calcification. The direct negative effect of increasing pCOz on calcification would in that case be partly masked by the increase in

28 photosynthesis. On the other hand, ifthe response ofphotosynthesis was longer than that ofcalcification, decreasing calcification could be considered an acute response to increasing pC02. In that case, calcareous organisms would adapt slowly and the negative response ofcalcification would diminish with time as photosynthesis increased (Leclercq et aI., 2002). Recent experiments conducted on biological communities in Biosphere 2

(Langdon et aI., 2002, submitted) and in experimental mesocosms (Leclercq et aI., 2002) indicated no significant increase in community production under elevated CO2 concentrations whereas the rate ofcommunity calcification consistently declined in each experiment. In conclusion, based on the present evidence, it is most likely that an increase in primary production and community production ofcoral reefs and other carbonate communities will not compensate for a decrease in calcification rate.

It has been proposed that some calcareous organisms might have an internal DIC regulator with which they themselves could regulate their internal carbon chemistry and thereby overcome the problem ofdecreased saturation state with respect to calcium carbonate ofthe surrounding water (Sikes and Fabry, 1982). The evidence for this suggestion is weak, and ifcalcareous organisms did have some form ofDIC pump, it would still be controlled to some extent by the carbon chemistry ofthe surrounding water, and the energetic investment and cost to calcify would increase along with decreasing saturation state.

In conclusion, based on current evidence the effects ofglobal warming and increasing atmospheric C02 on calcareous organisms and communities could be significant. It is important to note that our current knowledge is limited, but experimental

29 results indicate that calcareous organisms and communities in the future could have significant difficulty sustaining themselves and the structures ofthe communities as we know them today.

1.2.6 Opposing observations and alternative predictions: the Magnesian

Salvation Theorv

Considering current experimental results and biogeochemical models, the future outlook for corals and other calcareous organisms does not look very optimistic.

However, experimental investigations are conducted on relatively short time scales, and it could be argued that organisms could adapt to new conditions on a longer time scale. At present, some evidence exists for the ability ofcorals to acclimate to bleaching induced by environmental stresses such as solar radiation and temperature, which indeed suggests that adaptation possibly could occur under conditions ofglobal climate change

(Buddemeier and Fautin, 1993; Rowan et aI., 1997; Brown et aI., 2002). Based on changes in atmospheric CO2 concentration and subsequent changes in seawater carbonate saturation state since the onset ofthe industrial revolution, it has been predicted that the rate ofcalcification arnong calcareous organisms should already have declined (Kleypas et aI., 1999; Anderson et aI., 2003). However on the contrary, it has been found that the rate ofcalcification ofcorals on the Great Barrier Reefhas increased during this period of time (Lough and Barnes, 2000). Lough and Barnes (2000) attributed this increase in calcification to increasing temperature.

The question ofwhether or not corals and other calcareous organisms are able to adapt to different environmental conditions is highly dependent on how rapidly these new

30 conditions are imposed on the organisms. The burning offossil fuels and the subsequent increase in atmospheric CO2has been so rapid that at present there are no similar events in the geological record with which to make comparisons. The argument that corals and other organisms will adapt to future conditions might be true with respect to certain variables such as temperature, as suggested from the fact that the current global distribution ofcoral reefs does not seem to be restricted by an upper temperature limit

(Kleypas et al., 2001). However, it is difficult to ascertain how marine calcareous organisms will adapt to decreased carbonate saturation state. Even ifcalcareous organisms have some form ofDIC pump to control their internal carbon chemistry, at some point the difference between the internal carbonate saturation state and the surrounding water would be too great, and the energetic investment to calcifY would be too large for the organism to be beneficial.

An alternative hypothesis to the predicted negative impacts ofincreasing atmospheric C02 on coral reefs and calcareous organisms is that no changes or negative effects will be observed (Bames and Cuff, 2000). Instead, as the saturation state with respect to carbonate minerals decreases, the more soluble phases such as high magnesian calcite would initially dissolve and produce col- and alkalinity, which could restore any changes in carbonate saturation state and pH owing to increased invasion ofatmospheric

C02 (Bames and Cuff, 2000; Halley and Yates, 2000). In other words dissolution ofhigh magnesian calcites would act as a buffer to prevent changes in seawater chemistry imposed by increasing atmospheric CO2and thereby protect calcareous organisms and communities. Sometimes this overall process has been referred to as the Magnesian

31 Salvation Theory (MST; Buddemeier, personal communication). On a longer time scale ofmore than a few centuries, one also has to consider changes in the input ofalkalinity to the coastal zone via rivers. A significant source ofalkalinity transported via rivers originates from continental weathering ofcarbonate and silicate rocks (Berner et aI.,

1983; Bemer and Lasaga, 1989). Ifcontinental weathering were to increase owing to increased temperature and/or atmospheric CO2 content (Berner et aI., 1983; Berner and

Lasaga, 1989), the alkalinity delivered to the coastal zone would subsequently also increase and act as a buffer ofthe surface water ofthis region. However, changes in river transport ofalkalinity have not been observed or demonstrated during the past 300 years, although atmospheric CO2 has increased drastically during this time period (Mackenzie, personal communication).

In favor ofthe magnesian salvation theory, the distribution ofcarbonate minerals among various size classes in sediments, showing increasing mineral stability with decreasing grain size, has been attributed to selective dissolution ofunstable carbonate phases, i.e. high magnesian calcite and aragonite (Chave, 1962; Schmalz and Chave,

1963; Neumann, 1965). Schmalz and Chave (1963) suggested that dissolved carbonate activity in seawater is controlled by a metastable equilibrium between the seawater and the most soluble solid carbonate phase present in the sediments. Consequently, during early solutional diagenetic modifications on the seafloor owing to natural or anthropogenic processes, dissolution ofcarbonate minerals follows a sequence based on mineral stability, progressively leading to removal ofthe more soluble phases until the stable phases remain (Schmalz and Chave, 1963; Neumann, 1965; Wollast et al., 1980).

32 Based on experiments on the solubility and magnesium content ofcarbonate surface films precipitated on calcite seeds (Wollast et aI., 1980; Tribble and Mackenzie, 1998), one can conclude that high magnesian calcite minerals should dissolve in response to increasing atmospheric C02. In agreement with these previous observations, Mackenzie et al. (1980; unpublished data) observed ongoing dissolution ofcarbonate minerals, most likely high magnesian calcite, within the sediments ofMangrove Bay in Bermuda owing to respiration and remineralization oforganic matter.

Barnes and Cuff(2000) conducted experiments in which different carbonate phases such as laboratory grade calcite, coral skeletons (aragonite), magnesium carbonate and skeletons ofthe red alga Lithotamnion (high magnesian calcite) were mixed together and exposed to C02 concentrations and pH similar to those predicted by the end ofthe

21 st century. The calcite and aragonite mineral phases were unaffected by the low pH and carbonate saturation state whereas the magnesium carbonate and high magnesian calcite dissolved and produced alkalinity. Upon dissolution ofthese phases, the pH ofthe seawater approached or attained values higher than the initial pH ofthe experimental setup. Barnes and Cuff(2000) concluded that"...rising C02 will not greatly impact reef systems since any reduction in pH will be rapidly compensated by dissolution ofhigh magnesian calcite components (up to 50%) ofreefs rock." These authors suggested that dissolution ofhigh magnesian calcite will buffer the pH ofthe water in coral reefregions around 8.15 to 8.35 depending on the amount and composition ofhigh magnesian calcite available. Based on the observation that the pH and alkalinity actually increased in some cases relative to initial conditions, Barnes and Cuff(2000) proposed that solution of

33 metastable carbonates due to increasing atmospheric C02 is likely to enhance calcification among calcareous organisms and communities. In agreement, experiments conducted by Mackenzie et al. (1980; unpublished data) and Wollast et al. (1980) indicated a buffering effect ofseawater and a subsequent increase in pH and alkalinity upon dissolution ofmetastable carbonate minerals.

Field experiments conducted in incubation chambers in Molokai, Hawaii, investigating calcification, photosynthesis and respiration for different benthic communities exposed to elevated pC02 showed dissolution ofmagnesian calcite-rich, carbonate sediments, predominately taking place in substrates such as sand and coral rubble (Halley and Yates, 2000). Similarly, Leclercq et al. (2002) observed carbonate dissolution within sediments in experimental mesocosms exposed to elevated pC02, but concluded that the carbonate saturation state and inorganic carbon chemistry ofthe pore waters had little to do with the overlying surface water. Instead, undersaturated conditions and dissolution within the sediments were attributed to microbial respiration

(Leclercq et aI., 2002).

Whether or not dissolution ofmetastable carbonate minerals imposed by either rising atmospheric C02 or increased production and remineralization oforganic matter within the sediments could buffer future changes in the surface water carbon chemistry of the shallow-water ocean environment owing to increasing atmospheric C02 is not known and has not been quantitatively investigated. The primary objective ofthis thesis was to elucidate whether or not dissolution ofmetastable carbonate minerals such as high magnesian calcites could restore any changes in the dissolved inorganic carbon chemistry

34 and biogenic calcification within the shallow-water ocean environment imposed by increased atmospheric CO2 and lower carbonate saturation state.

1.3 BRIEF OVERVIEW OF THESIS CONTENT

To resolve the issues ofincreasing atmospheric CO2 and temperature on carbonate geochemistry and biogenic carbonate production within the shallow-water ocean environment, a biogeochemical box model representative ofthis region was developed (Shallow-water Ocean Carbonate Model; SOCM) and integrated into the global biogeochemical box model TOTEM (Terrestrial Ocean aTmosphere Ecosystem

Model) (Ver et aI., 1994; 1998; 1999a; 1999b; Mackenzie et aI., 1998a; 1998b; 2000;

2001; 2002). In Chapter 2 the methodology and the construction ofSOCM are described.

Assumptions, calculations and derivations ofreservoir masses and fluxes between the reservoirs are introduced. A synthesis ofcurrent estimates ofcarbonate mass, sources and production in different parts ofthe ocean is presented. The derivation ofequations pertinent to biogenic carbonate production and abiotic precipitation/dissolution reactions within the pore water-sediment system are reviewed and described. A briefdescription of TOTEM and its significant results are also presented.

In Chapter 3 a publication resulting from work discussed in this thesis is presented (Andersson, A. J., Mackenzie, F. T., and Ver, 1. M., 2003. Solution ofshallow­ water carbonates: An insignificant buffer against rising atmospheric CO2• Geology, vol.

31(6):513-516). The publication represents an overview ofthe work conducted and the major results and conclusions ofthe thesis, but does not include several important results

35 derived from comparison with actual data and sensitivity analyses, which are expanded on and discussed in depth in Chapter 4.

The objectives ofChapter 4 were to assess the reliability ofSOCM and to test the robustness ofthe major conclusions. The results ofthe standard simulation are reviewed and compared with observations from the natural environment. Multiple sensitivity analyses are conducted in order to investigate the robustness ofthe model results and the importance ofcertain properties and processes relevant to the inorganic carbon cycle in the shallow-water ocean environment. A wide range ofparameters such as initial inorganic carbon chemistry, initial carbonate satUTation state, carbonate dissolution/precipitation kinetics, magnesian calcite solubility, river transport and remineralization oforganic matter are investigated. In addition, the effects ofvarious CO2 emission and temperature scenarios on biogenic calcification are explored in detail.

In Chapter 5 a briefsnmmary and the main conclusions ofthe thesis are given.

36 CHAPTER 2: SHALLOW-WATER OCEAN CARBONATE MODEL (soeM)

2.1 INTRODUCTION

A box model representative ofthe shallow-water ocean environment referred to as SOCM (Shallow-water Ocean Carbonate Model) was constructed to evaluate the controversy whether marine calcareous organisms will be negatively impacted by increasing atmospheric C02 and temperature, or simply remain unaffected due to dissolution ofmetastable carbonate minerals. Major reservoirs and fluxes important to the cycling ofcarbon within this region were defined and derived from the literature. To facilitate input and output fluxes at the boundaries ofthe shallow-water ocean environment, SOCMwas integrated into the global biogeochemical model TOTEM

(Terrestrial Ocean aTmosphere Ecosystem Model) (Ver et aI., 1994, 1998, 1999a, 1999b;

Mackenzie et aI., 1998a, 1998b, 2000, 2001, 2002). Input fluxes ofcarbon from TOTEM to SOCM, as well as atmospheric CO2concentration and temperature were used as forcing functions ofthe latter (see Appendix A). To maintain compatibility with TOTEM and the successful results ofprevious simulations ofthis model, reservoir masses and fluxes already defined for the shallow-water ocean environment in TOTEM were not significantly altered. Simulations were initiated prior to the industrial revolution in the year 1700 and continued until year 2100. The model was assumed to be at an initial quasi-steady state in the year 1700. To evaluate the robustness ofSOCM, the major conclusions and the importance ofvarious processes within the shallow-water ocean

37 environment, multiple sensitivity analyses were conducted. These analyses are elaborated on in Chapter 4.

In the following sections ofthis chapter, the conceptual framework ofSOCM is presented; the current carbonate budget is introduced and reviewed; and derivations of biogenic carbonate production as a function ofdissolved inorganic carbon, temperature and carbonate saturation state are discussed along with abiotic carbonate precipitation and dissolution within the sediment pore water. Finally a briefdescription of TOTEM is presented. The mathematical logic and quantitative derivations ofmasses, fluxes and differential equations are presented in the Appendix A.

2,2 SOCM DESCRIPTION

The shallow-water ocean environment was defined for depths < 200 meters, including coastal zones, reefs, banks and shelves. In general, SOCM consisted oftwo major domains representing the surface water domain and the pore water sediment­ system domain (Fig 2.1). The surface water domain was subdivided into two major reservoirs referred to as surface water, i.e. the water column, and organic matter reservoirs. The total carbon content ofthe surface water reservoir was defined as the sum ofthe masses ofdissolved inorganic carbon (DIC; COz, HZC03, HC03-, COl-) aqueous species in this reservoir (Broecker and Peng, 1982; Ver, 1998; Ver et al., 1999). The organic matter reservoir was composed ofcarbon as particulate organic matter (fecal matter, floes, marine snow, etc), dissolved organic matter, and living biota (Williams,

1975; Lerman et aI., 1989; Murray, 1992; Ver, 1998; Ver et al., 1999). Flux ofcarbon

38 --- -"'-(221-)- - Atmosphere-surface water exchange -- , ... Rivers Surface water urface water-l OPen­ open ocean I ocean DIC 6000 • exchange. (32) :l:C02 =2000 ~M pHNBS = 8.37 • (477~)I'

organl,'-c-m-a-tt-e-r~-Y--B-iO-log-ic-al----'-----s-urf-a-c-e-w-a~terr_-p':"':(~:::~e71'):'"""w-at-e--'r Organic : remineralization uptake exchange matter ~ export (576) (600) r------I Reactive (18) " organic Organic matter Pore water matter 367 54 Upwelling 1:C02 = 3800 J.lM (26) (473.5) Biogenic pHNSS= 7.51 CaC03 production Organic matter C, A, M (24.5) remineraltzation Organic mailer (31) sedimentation ° Abiotic CaCOS CaCOS dlssolullon 0(32) precipitation R (5) l C. A. M (0.4) C, A. M (6.4)

Sediments (28.3 x 106 km2, 1 m depth. 50% porosity) Transport POC, PIC Organic matter , River-derived , Calcite I Aragonite 'Magnesian. to slopes , PIC (calcite) II calcite I-....;...'-'-':....;...;.~ 0(8) ° (0)1 R (15) 134.100 , 29.800 '5400 I 27,350 10.350 R, C, A, M (4) I Burial of land and marine (9) organic matter, river-derived °R (10) PIC, in situ produced CaCOS C, A, M (14.5)

Figure 2.1. Schematic ofthe Shallow-water Ocean Carbonate Model (SOCM) and estimated reservoir masses (1012 mol; shown in italics) and fluxes (1012 mol yr-1) at the onset ofthe simulation in the year 1700. The model was divided into two major domains, the surface water domain (white boxes) and the pore water­ sediment system domain (gray boxes). The model was integrated into the global biogeochemical model TOTEM (Terrestrial Ocean aTmosphere Ecosystem Model; Ver, 1998; Ver et aI., 1999; Mackenzie et aI., 2002; and references therein) and connected to adjacent reservoirs at the boundaries (dashed boxes).

39 from the surface water reservoir to the organic matter reservoir was through the process ofphotosynthesis (Smith and Hollibaugh, 1993; Wollast, 1998; Ver, 1998) owing to marine primary producers utilizing either dissolved molecular CO2 or HCO]- to produce organic matter (Raven, 1997). The reverse flux back to the surface water reservoir was due to respiration and remineralization oforganic matter (Lerman et aI., 1989), producing molecular CO2 that either dissolved or escaped to the atmosphere depending on the difference in partial pressure between the surface water and the atmosphere.

A substantial input ofcarbon was delivered to the shallow-water ocean environment via river transport. In the present model, river transport ofparticulate inorganic carbon (PIC), particulate organic carbon (POC), dissolved organic carbon

(DOC) and dissolved inorganic carbon (DIC) were considered (Meybeck, 1979; 1981;

1982; Smith and Hollibaugh, 1993). Particulate organic carbon was recognized as either reactive (flux to the surface water organic matter reservoir along with dissolved organic matter) or refractive (deposited within the sediment organic matter reservoir; Smith and

Hollibaugh, 1993). The riverine flux ofdissolved inorganic carbon was added to the surface water reservoir, whereas particulate inorganic carbon was deposited in the sediments. In addition to river input ofdissolved inorganic carbon to the surface water reservoir, fluxes ofcarbon to and from this reservoir were transferred through water exchange with the pore water-sediment system, water exchange with the surface layer of the open ocean, upwelling from the deep ocean, and gas exchange ofmolecular CO2 through the ocean-atmosphere boundary.

40 At the initial conditions, the shallow-water ocean environment was assumed to serve as a net source ofcarbon to the atmosphere owing to net heterotrophy and CO2 released from the process ofcalcification (Smith and Mackenzie, 1987; Smith and

Hollibaugh, 1993). In the process ofcalcification, carbon is removed from the surface water reservoir by calcareous organisms utilizing dissolved inorganic carbon (DIC) to produce shells, tests, and skeletons ofcalcium carbonate (Milliman, 1993; Wollast, 1994,

1998; Milliman and Droxler, 1996). The equations describing biogenic calcification were related to total dissolved inorganic carbon (Raven, 1997; Marubini and Thake, 1999), carbonate saturation state and temperature based on current experimental and observational data (see subsequent discussion; Gattuso et aI., 1999a; Agegian, 1985;

Mackenzie and Agegian, 1989, Lough and Barnes, 2000). In accordance with equation

(1.14), the production ofcalcium carbonate caused a release ofmolecular CO2 into the water. Depending on the buffering capacity ofthe seawater, the CO2 produced from the calcification process will either dissolve or evade to the atmosphere. Current estimates, taking seawater buffering capacity into account, suggest that 0.6 mol CO2 is presently released to the atmosphere for every mol CaC03 produced (Ware et aI., 1992;

Frankignoulle et aI., 1994). In the future this ratio will increase with increasing partial pressure ofCO2 and decreasing buffer capacity ofthe surface water owing to anthropogenic emissions and uptake ofthis gas by the ocean (Frankignoulle et aI., 1994).

In the present simulation, in order to maintain an initial quasi-steady state within the shallow-water ocean environment, it was assumed that for every mol ofCaCOJ produced, 0.8 mol ofCO2 was released to the atmosphere.

41 The pore water sediment-system domain was defined by six major reservoirs designated as organic matter, river-derived detrital particulate inorganic carbon (mainly calcite), in situ produced calcite, aragonite and 15 mol% magnesian calcite, and a pore water reservoir ofaverage composition. The initial composition ofthe pore water (TCOz

= 3800 J.lIllol L-I; pHNBs = 7.51) was based on a variety ofenvironments from the

Bahamas and elsewhere (Morse et aI., 1985; Morse and Mackenzie, 1990). It is important to note that the composition ofthe pore water reservoir represented an idealized average setting. In reality pore waters are quite heterogeneous, and regional compositional variability can deviate significantly from this average setting. The sedimentary reservoirs were all connected to the pore water reservoir. The carbon chemistry ofthe pore water was altered due to microbial remineralization oforganic matter, abiotic dissolution and precipitation ofcarbonate minerals, and mixing and replacement ofwater by the overlying surface water.

Inputs ofcarbon to the sediments originated from settling ofin situ produced organic matter (Lerman et aI., 1989; Smith and Hollibaugh, 1993) and calcium carbonate

(Milliman, 1993; Wollast 1994), and deposition ofriver transported detrital particulate organic matter and particulate inorganic carbon (Meybeck, 1981, 1982). Halfofthe detrital organic matter delivered to the coastal zone via rivers was considered refractive whereas the remaining halfwas considered labile (Smith and Hollibaugh, 1993). The detrital refractive organic matter was primarily removed and permanently buried within the sediments whereas the labile fraction and in situ produced organic matter were primarily remineralized through microbial processes (Smith and Hollibaugh, 1993).

42 Remineralization oforganic matter produces CO2 that dissolves in the pore water, changing the carbonate species equilibrium and causing a decrease in pH, whereas the total alkalinity remains nearly constant. In reality, a slight decrease in total alkalinity occurs because nutrients are released along with CO2 upon remineralization oforganic matter (see Zeebe and Wolf-Gladrow, 2002). The organic matter within the sediments was either permanently buried or remineralized within the shallow-water ocean environment. It was assumed that no organic matter from the sediments was exported to the continental slope. However, a substantial amount oforganic matter was exported to the open ocean from the surface water domain (Smith and Hollibaugh, 1993).

River derived detrital particulate inorganic carbon was assumed to originate entirely from continental erosion and to be primarily calcite (Meybeck, 1982), which was considered to be unreactive due to coatings by organic material and other inhibitory substances. Due to its refractive nature, detrital particulate inorganic carbon is most often omitted in carbonate budgets ofthe coastal zone since it remains unreactive and makes very little difference to the geochemistry ofthis region (Mackenzie, personal communication). In the present case, it was assumed that detrital particulate inorganic carbon mainly was deposited in regions where only scant carbonate minerals are found due to dilution by terrigenous constituents. Subsequently, due to extensive remineralization oforganic matter causing corrosive conditions with respect to carbonate minerals, it was assumed that one third ofthe refractive calcite was dissolved at the initial conditions ofthe model simulations. Similarly, due to the substantial dilution by

43 terrestrial components, it was assumed that an insignificant amount ofdetrital particulate inorganic carbon was exported to the continental slope and the open ocean.

The flux ofcarbon to the reservoirs ofcalcite, aragonite, and 15 mol% magnesian calcite within the pore water-sediment system domain was primarily produced by calcareous organisms living in the water column or in the benthos, utilizing dissolved inorganic carbon to produce calcium carbonate (Milliman, 1993; Wollast, 1994, 1998;

Milliman and Droxler, 1996; Iglesias-Rodriguez et aI., 2002). In natural environments, the magnesium content ofbiologically produced magnesian calcite ranges from -0-30 mol% with an average of 12-15 mol% (Garrels and Wollast, 1978; Garrels and

Mackenzie, 1981; Mackenzie et aI., 1983; Morse and Mackenzie, 1990). In the present model, the reservoir ofmagnesian calcite was assumed to be composed ofan average composition of 15 mol% magnesian calcite (Garrels and Mackenzie, 1981). In addition to biological production, minor proportions ofcarbonate minerals as cements were produced through abiotic precipitation within the pore water-sediment system. The majority ofcalcium carbonate minerals produced within the shallow water ocean environment accumulated within the sediments (Milliman, 1993; Milliman and Droxler,

1996; Wollast, 1994; 1998; Iglesias-Rodriguez et al., 2002). The remaining fraction dissolved or was exported to the sediments ofthe continental slope.

44 2.3 SHALLOW-WATER CARBONATE MASSES, PRODUCTION AND

ACCUMULATION

2.3.1 Production and accumulation

In recent years, the carbonate budget ofthe ocean has been discussed extensively

(Milliman, 1974, 1993; Berger, 1976; Morse and Mackenzie, 1990; Sabine and

Mackenzie, 1991; Opdyke and Walker, 1992; Wollast, 1994; Milliman and Droxler,

1996; Wollast, 1998; Iglesias-Rodriguez, 2002). The most comprehensive estimates of carbonate production and accumulation within the shallow-water ocean environment are those ofMilliman (Table 2.1; Milliman, 1993; Milliman and Droxler, 1996) and reviewed by Wollast (1994, 1998) and Iglesias-Rodriguez et al. (2002).

Milliman (1993) divided the shallow-water ocean environment into four major regions based on differences in carbonate production and accumulation. The regions were referred to as coral reefs, banks and embayments, carbonate-poor continental shelf, and carbonate-rich continental shelf. The majority ofcarbonate minerals produced in the shallow-water ocean environment is produced within coral reefregions (Milliman,

1993). Hermatypic coralgal reefs are considered to be the most productive carbonate habitat in the present day ocean (Milliman, 1993). Estimates ofglobal coral reef area range from 150 x 103 km2 to as much as 3930 x 103 km2 (Spalding and Grenfell, 1997).

According to Spalding and Grenfell (1997), the great variability is due, in part, to different definitions ofcoral reefs. The most widely quoted estimate, and the estimate used by Milliman (1993), is that ofSmith (1978), who estimated the global coral reef

45 TABLE 2.1. CALCIUM CARBONATE PRODUCTION AND ACCUMULATION IN THE SHALWW- WATER OCEAN ENVIRONMENT IN THE LATE HOLOCENE OCEAN (modified fromMiIIimm, 1993)

CaCO, CaCO, Area CaCO, flux Percent Habitat 2 2 Production Accwnulation (10' km ) (gm- yr-l) preserved (1012 mol yr") (1012 molyr-')

Coral reefcorrqilex 0.6 1500 9 7 80 Bank/bays 0.8 500 4 2 50 Non-carbonate shelves 15.3 25 4 I 25 Carbonate shelves 10 20-100 6 3 50 Halimeda biohenns 0.05 3000 1.5 1.5 100 Total 26.8 90.9 24.5 14.5 59.2

3 2 area to be approximately 600 x 10 km . A more recent estimate by Spalding and

3 2 Grenfell (1997) suggests a total area of255 x 10 km .

The majority ofcarbonate minerals within coral reefregions is produced by corals and calcareous algae, although locally benthic foraminifera can be important (Milliman,

2 1993). At present, 1500 g CaC03 m- yr'l are produced in global coral reefregions.

Assuming that approximately 20% is lost due to biological erosion, dissolution and

2 offshore transport, approximately 1200 g CaC03 m- yr'l accumulate within these regions

(Milliman, 1993). Assuming a coral reefarea of600 x 103 km2 (Smith, 1978), the global production and accumulation ofcarbonate minerals within coral reefenvironments

12 12 correspond to 9.0 x 10 mol C yr-l and 7.2 x 10 mol C yr-l, respectively.

In addition to coral reefenvironments, high rates ofcarbonate production and accumulation are observed in regions such as banks and embayments. Although carbonate production is not as high as in coral reefenvironments, the area ofthese regions is somewhat greater, covering approximately 800 x 103 km2 (Milliman, 1993).

The largest fractions ofcarbonate minerals produced within banks and embayments originate from benthic coralline algae, mollusks and benthic foraminifera (Agegian et al.,

46 1988; Agegian, 1989; Milliman, 1993). In certain regions, bryozoans and serpulids can be important, and in areas such as the s and the Persian Gulf, inorganic precipitation in the form ofwhitings is important (Milliman, 1993). At present, the average carbonate production within banks and embayments is approximately 500 g CaC03 m,2 yr,l

(Milliman, 1993). Assuming that halfofthis is dissolved or exported to the continental slope, 250 g CaC03 m,l yr,l accumulate within this region each year (Milliman, 1993).

12 Globally this corresponds to a production of4 x 10 mol C yr,l and an accumulation of2 x 1012 mol C yr,l.

Continental shelfregions can be separated into carbonate-poor and carbonate­

6 2 6 l rich regions, which cover an area ofapproximately 15.3 x 10 km and lOx 10 km , respectively (Milliman, 1993). There are few reliable data on carbonate production and accumulation within these regions. In general, carbonate-rich sediments are found in areas oflow terrigenous input (Land, 1967; Milliman, 1974) and in regions ofsignificant terrigenous input, none or very little carbonate is found within the sediments (Land,

1967). Significant carbonate production might occur in carbonate-poor shelfregions, although no evidence for this is found within the sediments (Smith, 1972). This indicates that most ofthe carbonate probably is dissolved or exported to the continental slope

(Smith, 1972; Milliman, 1993). A difficulty in estimating carbonate production and accumulation within carbonate-rich shelfregions arises due to the presence ofextensive amounts ofrelict carbonates deposited during times oflower sea level (Milliman, 1993;

Wollast, 1998). Milliman (1993) estimated carbonate production in carbonate-rich shelf regions to equal 60 g CaC03 m,2 yr,l, but due to dissolution and export, only halfofthis

47 12 accumulates. Globally, this corresponds to a production of6 x 10 mol yr.1 and an

12 accumulation of3 x 10 mol yr-l. Within carbonate-rich shelfregions, significantly higher production has been observed in habitats associated with the green alga Halimeda.

2 It has been estimated that these algae can produce as much as 3200 g CaC03 m- yr-I

(Milliman, 1993). Although the global surface area covered by this alga is unknown,

Milliman assumed an area of50 x 103knl and an average production and accumulation

2 of3000 g CaC03 m- yr-I. The area assumed by Milliman (1993) corresponds to 0.2% of the total shallow-water ocean environment, but the global production and accumulation

12 of 1.5 x 10 mol yr-I corresponds to approximately 6% ofthe total carbonate production and 10% ofthe accumulation within the shallow-water ocean environment. Thus, these algae are an important contributor to the global carbonate budget.

In carbonate-poor shelfregions, carbonate production is poorly known. Milliman

2 (1993) estimated a production of25 g CaC03m- yr-I and an accumulation of

2 12 approximately 8 g CaC03m- yr-I. This corresponds to a global production of3.8 x 10

12 mol C yr-I and an accumulation of 1.2 x 10 mol C yr-l.

In conclusion, based on Milliman's estimates, the global production ofcarbonate minerals within the shallow-water ocean environment is equivalent to approximately

12 12 24.5 x 10 mol C yr-I from which approximately 14.5 x 10 mol C yr-I accumulate. The

12 remaining 10 x 10 mol C yr-l is lost due to dissolution and export to the continental slope.

48 2.3.2 Shallow-water carbonate budget

In the standard model, the estimated carbonate production of24.5 x 101 2mol C yr-1 and net accumulation of 14.5 x 101 2mol C yr-l within the shallow-water ocean environment were adopted (Milliman, 1993; Wollast, 1994, 1998; Milliman and Droxler;

1996). In comparison, Morse and Mackenzie (1990) estimated carbonate accumulation within this region to be less than halfthat ofMilliman's estimate (6 x 1012 mol C yr-l).

Wollast (1998) attributed this difference to the fact that the data used by Morse and

Mackenzie (1990) were based on the geological record and were not representative ofthe present-day fluxes as estimated by Milliman. Currently, carbonate production and accumulation are unusually high due to the relatively high sea level ofthe present interglacial and the large surface area constituting the continental shelf(Milliman, 1993;

Wollast, 1994, 1998). Considering the global accumulation ofcarbonate minerals, which

Milliman estimates as equivalent to approximately 32 x 1012 mol C yr-1 and a total influx ofcalcium carbonate from rivers, hydrothermal activity and atmospheric transport as equivalent to approximately 15 x 1012 mol C yr-l, it is obvious that the global short-term carbonate cycle is not at a steady state. Although substantial uncertainty is associated with the global carbonate budget, and the existence ofpotential but poorly quantified input fluxes, such as carbonate input via ground water, these uncertainties do not make up for the deficit in the budget (Milliman, 1993; Wollast, 1994). Instead, because ofhigh sea level and extensive continental shelfarea, the carbonate budget ofthe oceans is currently most likely in a non-steady state.

49 In addition to in situ production ofcalcium carbonate within the shallow-water ocean environment, a significant flux ofparticulate inorganic carbon originating from continental erosion is transported via rivers to the ocean. The estimate ofdetrital calcium

12 carbonate input via rivers is approximately 15 x 10 mol yr-l (Meybeck, 1981, 1982).

Due to the fact that rivers deliver substantial amounts ofterrigenous components to the ocean, very little or no carbonate deposits are found in regions near major rivers because ofdilution by terrigenous sediments and dissolution ofcarbonate due to extensive remineralization oforganic matter (Land, 1967). The fate ofriver-derived particulate inorganic carbon is therefore poorly known. At the initial conditions ofthe standard model, it was assumed that two thirds ofthe input ofdetrital particulate inorganic carbon accumulated within the shallow-water ocean environment whereas the remaining proportion dissolved due to extensive decomposition oforganic matter. Thus in SOCM, the total accumulation ofcalcium carbonate within the shallow-water ocean environment, considering both in situ production and river input, is equivalent to 24.5 x

12 10 mol C yr-l, which is larger than previous estimates (Morse and Mackenzie, 1990;

Milliman, 1993; Wollast, 1994; Milliman and Drox1er, 1996, Wollast, 1998; Iglesias­

Rodrigues et aI., 2002). These previous estimates do not include the detrital particulate inorganic carbon input via rivers.

In addition to dissolution ofcarbonate minerals within the sediments ofthe shallow-water ocean environment, a certain fraction ofcarbonate is lost due to export to

12 the continental slope. Current estimates range from 4 x 10 mol yr-l (Milliman, 1993,

50 12 Wollast, 1994) to 8 X 10 mol yr-I (Morse and Mackenzie, 1990). In SOCM, the former

12 estimate of4 x 10 mol yr-I was adopted (Milliman, 1993, Wollast, 1994)

2.3.3 Shallow-water carbonate sediment mass

In the standard model, it was assumed that the top meter ofthe sediments in the shallow-water ocean environment was reactive and interacted with the pore water and the surface water on the time scale ofyears to centuries (Berner, 1980). Based on the rates ofcarbonate accumulation previously presented, the total carbonate mass in the top meter ofthe sediments can be derived by extrapolating these rates backward in time.

However, due to the currently unusual high rate ofcarbonate accumulation (Milliman,

1993; Wollast, 1994, 1998) and the fact that the sediments contain constituents other than carbonate minerals, this method might yield a significantly biased estimate. Instead, the data on sediment calcium carbonate content by weight percent presented by Milliman

(1974) were used to calculate the total carbonate mass in the top meter ofthe sediments ofthe shallow-water ocean environment. Milliman (1974) divided the area ofthe

12 2 shallow-water ocean environment into reefand bank regions (1.4 x 10 km ) and

12 2 continental shelfregions (26.9 x 10 km ). The carbonate content ofthese regions is 80 wt% and 15 wt%, respectively (Milliman, 1974). Assuming an average sediment porosity of50%, an average calcium carbonate mineral density of2.83 g cm-3 and adopting the previous weight percentages, the carbonate mass in the top meter ofthe shallow-water

12 ocean environment totals 72900 x 10 mol CaC03. Assuming that the accumulation ratio ofriver-derived particulate inorganic carbon to in situ produced calcium carbonate has remained approximately constant for the sediments under consideration, 41 % ofthis mass

51 represents river derived calcium carbonate, and the remaining fraction represents calcium carbonate produced in situ. With these rates ofcarbonate accumulation, it would take approximately 3000 years to accumulate this mass ofcarbonate minerals, which corresponds to an average sedimentation rate ofapproximately 30 cm kyr-) within the shallow-water ocean environment. In reality, sedimentation rates vary considerably between different regions, but in general this estimate agrees well with estimates in the literature. In coral reefregions, seismic profiles across lagoons and outer reeftracts, suggest a sedimentation rate during the Holocene of 10-150 cm kyr-) (Lidz et ai., 1991) and according to Kennet (1982), the sedimentation rate in continental margins and borderlands could well exceed 20 to 30 cm kyr-). It is important to note that substantial uncertainty is associated with the estimate oftotal carbonate mass within the shallow­ water ocean environment and it could be significantly in error. However in the present model, the uncertainties associated with this mass were evaluated by running sensitivity analyses adopting a range ofdifferent initial carbonate masses.

2.3.4 Carbonate sediment composition

The proportions ofcalcite, aragonite and magnesian calcite within the sediments ofthe shallow-water ocean environment were based on the composition of758 recent neritic carbonate sediments from Land (1967). Based on Land's data, the relative percentages ofcalcite:aragonite:magnesian calcite were on average 12.6% : 63.4% :

24.0%. At the onset ofthe model simulation, the accumulation ofcarbonate minerals produced within the shallow-water ocean environment was assumed to have the same percentage composition. River-derived particulate inorganic carbon was assumed to be

52 entirely comprised ofcalcite originating from continental erosion. As a comparison, in an assessment whether or not dissolution ofmetastable carbonate minerals could serve as a sink ofanthropogenic CO2, the magnesian calcite composition ofshelfsediments was estimated to equal 15% ofthe total carbonate content (Wollast and Mackenzie, 1981). In the same stndy, global magnesian calcite production was estimated to be 6.2 x 1012 mol C yr-l. In the standard model, adopting the carbonate production estimate ofMilliman

(1993) and the carbonate mineral ratio ofLand (1967), magnesian calcite production is equivalent to 5.9 x 10 12 mol C yr-l, which agrees well with the estimate ofWollast and

Mackenzie (1981).

2.4 BIOGENIC CALCIFICATION

2.4.1 Biogenic calcification and DIC

In the standard model, the carbon flux associated with biogenic calcification in the surface water was related to total dissolved inorganic carbon (DIC) concentration, carbonate saturation state and sea surface temperature. Biogenic carbonate production was related to DIC content by a first order dependence based on the observation that marine primary production is directly related to this parameter (Riebesell et aI., 1993;

Raven et aI., 1993; Raven et aI., 1997; see discussion in Chapter I). A doubling ofthe skeletal growth rate ofthe coral Porites porites submerged in aquaria with tropical seawater was observed upon addition of2 rnrnol bicarbonate (Marubini and Thake,

1999). However, the relationship between skeletal growth rate and DIC concentration is weak, and the response ofbiogenic calcification to intermediate increases in DIC is

53 poorly known. In the standard model simulations, the overall increase in DIC due to increased atmospheric CO2 is relatively small and has only a minor effect on calcification rate.

2.4.2 Biogenic calcification and carbonate saturation state

CO2 partial pressure and carbonate saturation state experiments have been conducted on various calcareous organisms (Fig 1.4) such as coccolithophorids

(Riebesell et aI., 2000; Zondervan et aI., 2001), foraminifera (Bijma et aI., 1999), . coralline algae (Smith and Roth, 1979; Borowitzka, 1981; Agegian, 1989; Mackenzie and

Agegian, 1989, Gao et aI., 1993) and scleractinian corals (Gattuso et aI., 1998; Marubini et aI., 2001). In addition, experiments on typical calcareous communities (Fig 1.5) have been conducted in incubation chambers and mesocosms (Halley and Yates, 2000;

Leclercq et aI., 2002), and on the artificial reefofBiosphere 2 (Langdon et aI., 2000,

Langdon et al., 2002). Although substantial variations have been observed between species and communities, the major results and conclusions ofall studies have been similar, and the rate ofcalcification has been observed to decrease as a function of decreasing carbonate saturation state (Fig 1.3; 1.4; Table 2.2). Experimental results indicate that biogenic calcification rates exhibit either a linear or a curvilinear relationship to carbonate saturation state (Table 2.2). According to Buddemeier (personal communication), current results tend to indicate that the relationship probably is more likely linear than curvilinear. In addition, there have been problems associated with the method ofmanipulating the carbonate saturation state in some ofthose experiments that

54 TABLE 2.2. RELATIVE RATE OF CALCIFICATION AS A FUNCTION OF ARAGONITE SATURATION STATE (from Gattuso et aI., 1999, and Leclercq et aI., 2002).

'Relative calcification rate Reference

Scleractinian cerals Porites compressa' -41.1 + 29.1fl 'Marubini and Atkinson

Porites porites 51 + 100 'Marubini and Thake (1999)

Stylophora pistillata 228(1-e.010"')_128 'Gattuso et al. (1998)

Coralline alga

Amphiroa fotiacea -1+210 'Borowitzka (1981) Bossiela orbigniana' 77.2(1-e·010 '4)_16.8 'Smith and Roth (1979)

Corallina pi/utifera' -37.1 + 44.50 'Gao et al. (1993)

Porolithon gardineri 29 + 150 'Agegian (1985); Mackenzie and Agegian (1989) Reef communities Biosphere 2 ocean -67 + 340 'Langdon et al. (2000) Coral community 31 + 140 'Leclercq et al. (2000)

Coral community 52 + 100 Leclercq et al. (2002) Okinawa reef fiat 4(0-1 ),.36 'Odhe and Woesik (1999)

Sand cemmunity -1533 + 3330 Leclercq et al. (2002) Sand community -287 + 790 'Boucher et al. (1998)

, Calcification rate expressed in percent of the maximum rate measured. In remaining studies, calcification is expressed as a percentage of the rate found at 0a",. =4.9 (preindustrial period; Leclercq et aI., 2002) , in Gattuso et al. (1999) , in Leclercq et al. (2002) resulted in a spurious curvilinear relationship e.g. Gattuso et al. (1998) (Buddemeier, personal communication).

In the standard model so as not to exclude any possibilities, and to explore the sensitivity ofthe calcification rate dependence on carbonate saturation state, two different scenarios were employed. In the first scenario, a linear relationship between calcification

55 rate and carbonate saturation state based on multiple data for various scleractinian corals and coralline algae was adopted (Gattuso et aI., 1999a):

Rn = 21.30 + 12. (2.1)

In the second scenario, the curvilinear relationship between calcification rate and carbonate saturation state ofStylophora pistillata was used (Gattuso et aI., 1998; Leclercq et aI., 2002):

Rn = 228(l-e·flj069) - 128 (2.2) o is the seawater saturation state with respect to aragonite (I ~ Oarg ~ 6) and R is the relative rate ofcalcification expressed as a percentage ofthe approximate rate found at initial conditions.

2.4.3 Biogenic calcification and temperature

Relatively few experiments have been conducted to date to evaluate the effect of temperature and calcium carbonate saturation state on biogenic calcification rates.

However, experimental results indicate that a negative parabolic relationship exists between biogenic calcification and temperature (Fig 1.3; Table 2.3; Clausen and Roth,

1975; Agegian, 1985; Mackenzie and Agegian, 1989) whereas growth banding in coral colonies collected along the Great Barrier Reefand the Hawaiian Island chain show a positive linear relationship (Fig 1.4; Table 2.3; Grigg, 1981; 1997; Lough and Barnes,

2000). In SOCM, the standard scenario simulation was run using the experimental results ofPorolithon gardineri (Agegian, 1985; Mackenzie and Agegian, 1989), which were normalized to the maximum rate ofcalcification:

(2.3)

56 TABLE 2.3. RELATIVE RATE OF CALCIFICATION AS A FUNCTION OF TEMPERATURE CHANGE

Relative calcification rate Reference/Comment Scleractinian corals

PociJlopora damicornis 100 - 1.196.T' Smith and Roth (1975) PociJIopora damicornis 100 -1.256.T' Smith and Roth (1975)

Pocillopora damicornis 100 -1.656.T2 Smith and Roth (1975) Calcification rate at multiple locations (Lough and Porites sp. 100 + 286.T Barnes, 2000; Grigg, 1981, 1997; Scoffin etal., 1992)

Coralline aiga

Porolilhon gardineri 100 -1.326.T' Agegian (1985); Mackenzie and Agegian (1989)

In a second scenario simulation, a positive linear relationship was adopted based on the observed rate ofcalcification ofmultiple coral colonies from the Great Barrier Reef,

Hawaii and Thailand (Grigg, 1981, 1992; Scoffin et aI., 1992; Lough and Barnes, 2000):

(2.4) where RT is the relative rate ofcalcification expressed as a percentage ofthe rate at the initial temperature, and !':1T is the change in temperature in °C relative to year 1700.

2.4.4 Combined effect ofDIC, carbonate saturation state and temperature on

biogenic calcification in SOCM

Combining the three independent variables ofDIC, Q and T, the overall equation describing the flux ofbiogenic carbonate production (F) is described by:

R o.t R Tt F= kC orc x x' (2.5) RO,t=1700 RT,t=1700' where F is flux ofcarbon in mol yr'l, COle is the total mass ofDIC in the surface water in mol C, and Rn and RTare the relative rates ofcalcification as a function ofcarbonate

57 saturation state and temperature, respectively, expressed as a percentage ofthe rate at the initial conditions. The rate constant k (yr-)) was based on the initial estimate ofthe flux ofcarbon owing to biogenic calcification (Milliman, 1993; Wollast, 1994) and the total dissolved inorganic carbon concentration (see Appendix A):

k = F'=1700 (2.6) C DlC,t=1700

2.5 CARBONATE DISSOLUTION AND PRECIPITATION REACTION

KINETICS

The inorganic carbon chemistry ofthe pore water-sediment system exhibits significantly different properties than those ofthe overlying surface water. Carbonate dissolution and precipitation reactions can interchangeably take place in various regions within the sediments depending on different microbial processes. In SOCM, the pore water-sediment system reservoir was assumed to have an average composition. Net dissolution and precipitation were based on an average carbonate saturation state. To account for dissolution and precipitation ofcarbonate minerals, the flux ofcarbon between the sediments and the pore water was related to kinetic rate equations describing these processes.

The rate ofdissolution and precipitation ofcarbonate minerals in seawater is controlled by several factors. One ofthe primary factors is the degree ofdisequilibrium ofthe solution (super or undersaturation), although several other variables that will be discussed later also have a significant influence on carbonate mineral reaction rate. The

58 saturation state (Q) with respect to carbonate minerals can be determined from equations

(1.6) and (1.7). IfQ < I, the solution is undersaturated and net dissolution is favored over precipitation, and ifQ > 1, the solution is supersaturated and net precipitation is favored.

In general, the rate ofdissolution or precipitation increases with increasing disequilibrium. The rate ofreaction is not directly proportional to the extent of disequilibrium due to many complicating factors and mechanisms involved in the reaction between the solid and solution (Morse, 1983). These are not explored here because they do not significantly affect the ultimate conclusions reached. In addition, due to the fact that seawater is a complex electrolyte with many different constituents and ion species present, carbonate reactivity, i.e. the rate ofreaction between solid and solution, can be significantly inhibited by components that complex other ions and change their activities or adsorb on to mineral surfaces (Morse, 1983; Morse and Mackenzie, 1990).

The components adsorbed to the surface may block mineral growth or dissolution and may even control the mineralogy ofthe carbonate phase that precipitates (Morse, 1983).

Magnesium is by far the component that has received most attention as an inhibitor due to its abundance in seawater and significant effect on carbonate reactivity (Pytkowicz,

1965; Berner, 1975; Morse, 1983; Tribble and Mackenzie, 1995; Zhang and Dawe,

2000). Other components known to inhibit carbonate precipitation and dissolution are sulfate (Morse, 1983), phosphate (Berner et al., 1978; Morse, 1983; Mucci, 1986; Burton and Walter, 1990; Dove and Hochella, 1993), organic material (Chave and Suess, 1967;

Berner, et a!., 1978; Morse, 1983; Lebron and Suarez, 1996, 1998), and heavy metals

(Morse, 1983; Morse and Mackenzie, 1990).

59 The most common approach used to describe the reaction kinetics ofcarbonate dissolution and precipitation is empirical (Morse, 1983). In general, the following equations have been successful in fitting and describing data relatively close to equilibrium for dissolution (Rl) and precipitation (Rp), respectively:

Rl = k(l - Q)", (2.8)

Rp = k(Q - 1)". (2.9)

R is the rate ofreaction, k is the rate constant, Q is the saturation state with respect to the carbonate mineral under consideration, and n is the order ofthe reaction. When the equations are log transformed they become linear and the reaction order and the rate constant can be determined by fitting data to these linear equations and evaluating the slope and the intercept, respectively (Table 2.4; Morse, 1983; Morse and Mackenzie,

1990). Because the majority ofcarbonate minerals in nature are produced by calcareous organisms, precipitation reaction kinetics (Morse, 1983; Burton and Walter, 1987; Zhong and Mucci, 1989, 1992; Zhang and Dawe, 1997; Zhang and Dawe, 1999) have not been studied as extensively as carbonate dissolution kinetics (Plummer et al., 1978; Keir,

1979; Morse, 1983; Walter and Morse, 1985; Morse and Arvidson, 2001). Similarly, calcite has received far more attention than aragonite and magnesian calcite.

Based on the carbonate budget adopted in SOCM (Milliman, 1993; Wollast, 1994;

Milliman and Droxler, 1996), at the onset ofthe simulation, the pore water-sediment

12 system was characterized by net dissolution ofcarbonate minerals (6x 10 mol yr-1).

Note that this estimate is the net effect ofabiotic carbonate precipitation and dissolution taking place within the pore water-sediment system. Carbonate dissolution is controlled

60 TABLE 2.4. KINETIC RATE CONSTANTS AND REACTION ORDER FROM SELECTED CARBONATE PRECIPITATION AND DISSOLUTION EXPERIMENTS

Precipitation rate kinetics from \erious studies Rate Reaction Rate of precipitation R Reference/Comments 2 1 constant k order n [R=k (n-lf] (j.Lmol m- h- ) (IlmoJ m-2 h-1) (oeal :; 1.84; narag = 1.33)

Calcite 3.9 1.9 2.8 Burton and Walter, 1987 0.5 2.8 0.3 Zhong and Mucci, 1989 1.6 2.2 1.1 Zhong and Mucci. 1993 Aragonite 40.6 1.7 6.2 Burton and Walter, 1987 12.3 2.4 0.9 Zhong and Mucci, 1989

Dissolution rate kinetics from various studies Rate Reaction Rate ofdissolution R Reference/Comments constant k order n [R=k(1-n)"J (% day-') (%day-1) (neal = narag = 0.9)

Calcite 501 3.0 0.50 Pacific whole sediments (Morse, 1978) 1260 4.5 0.04 Atlantic whole sediments (Morse, 1978) 2990 4.5 0.09 Ontong Ja\e Plateau whole sediments (Kier, 1980) 1309 4.5 0.04 Synthetic calcite (Kier, 1980) Aragonite 41.1 4.2 0.003 Ontong Java Plateau whole sediments (Kier. 1980) 318 4.2 0.02 Pteropods (Kier, 1980) 13260 4.2 0.84 Synthetic aragonite (Kier, 1980)

Dissolution rate kinetics ofshallow marine carbonate phases (Walter and MOISe, 1985} Rate Reaction Rate ofdissolution R Reference/Comments constant k order n [R=k (1-n)n] (~mol g_1 h-1 ) 1 (~mol g_1 h- ) (neal =narag = OMg.cal = 0.9)

Calcite 3388.4 2.96 3715.4 Synthetic calcite 186.2 2.89 239.9 Iceland spar 457.1 2.74 831.8 Balanus Aragonites 537.0 2.45 1905.5 Fungia 467.7 2.50 1479.1 Acropora 1258.9 2.43 4677.4 Halimeda 1000.0 2.54 2884.0 Strombus 912.0 2.50 2884.0 Hippopus Mg-ealcite 208.9 3.30 104.7 CJypeaster 12 moJ% Mg-calcite 660.7 3.51 204.2 PeneropJis 15 mol% Mg-ealcite 537.0 3.20 338.8 Goniolithon 18 moJ% Mg-ealcite

61 by carbonate saturation state, but is strongly affected and mediated by such processes as mechanical erosion (waves, currents) and biological erosion (borers, predators, grazers, browsers; Milliman, 1974; Morse and Mackenzie, 1990; Gao, 1996). At the initial conditions ofthe model, the pore water was on average supersaturated with respect to calcite, aragonite and 15 mol% magnesian calcite (Oe = 1.84; OA = 1.33; OM = 1.19). If the geochemistry ofthe pore water-sediment system were homogenous and the carbonate saturation state were the only factor determining whether carbonate minerals precipitated or dissolved, based on equations (2.8) and (2.9), net precipitation would be observed at the initial conditions. In reality, the pore water-sediment system is highly heterogeneous, and pore water carbonate saturation is commonly very low in the upper decimeter ofthe sediments, rapidly increasing with depth in the sediments (Berner, 1978; Morse and

Mackenzie, 1990). In addition, mechanical and biological erosion playa major role in breaking down carbonate minerals into smaller grain sizes, increasing the surface area to voluroe ratio and facilitating dissolution. Additional complications in predicting carbonate dissolution and precipitation arise due to inhibition by dissolved phosphate, dissolved organic matter, and other materials (Berner, 1978; Morse, 1983; Morse and

Mackenzie, 1990).

In the standard run ofSOCM, the initial net carbonate dissolution within the pore water-sediment system was based on a first order dependence on the calciuro carbonate mass. The proportionality constant was based on the initial net flux and the initial mass of the carbonate mineral under consideration (see model equations in Appendix A). In order to relate abiotic precipitation and dissolution to the carbonate saturation state, an

62 TABLE 2.5. CONSTANTS ADOPTED FOR INORGANIC CaCOs PRECIPITATION AND DISSOLUTION RATES (R) Rate constant (k) Reaction order (n)

Precipitation R = k (£1_1)" (Zhong and Mucci, 1989) Calcite 10-0·29 2.80 15 mol% Mg-calcite' 10-0·29 2.80 Aragonite 101.09 2.36

Dissolution R = k (1-£1)" (Walter and Morse, 1985) Calcite 102.82 2.86 262 15 mol% Mg-calcite 10 3.34 Aragonite 102.89 2.48

•Assuming same rate as calcite additional term based on equations (2.8) and (2.9) was added (see subsequent discussion).

For supersaturated conditions, carbonate precipitation was calculated based on the kinetic rate equation (2.8) and the experimental results ofZhong and Mucci (1989; Table 2.5).

The resulting rate ofprecipitation (mol m-2 h-I) was converted into mol yr-I. In order to make this conversion, the total mass ofcarbonate minerals had to be converted into specific surface area and reactive surface area (Table 2.6; Walter and Morse, 1984,

1985). The specific surface area ofmost biogenic grains ranges from 0.1-0.5 m2 g-I with extreme values ranging from 0.03-24.0 m2 g-I whereas reactive surface area ranges from

0.3-66% ofthe specific surface area (Walter and Morse, 1984, 1985). In SOCM, an intermediate specific surface area of0.3 m2 g-I and a reactive surface area of25%

(calculated average for biogenic carbonate grains) were adopted (Walter and Morse,

1985). For undersaturated conditions, carbonate dissolution was calculated based on equation (2.9) and the experimental results ofWalter and Morse (1985; Table 2.5).

63 TABLE 2.6. SPECIFIC SURFACE AREA AND REACTIVE SURFACE AREA FOR VARIOUS CARBONATE GRAINS (Walter and Morse, 1984, 1985). Specific surface area Reactil.e surface area (m2 g-1) (% of total area) Synthetic calcite 0.45 100 Iceland spar 0.03 81 Ba/anus 0.19 32 Fungia 0.22 32 Acropora 0.48 12 Halimeda 0.19 8 Strombus 0.2 66 Hippopus 2.04 63 Clypeaster 0.14 20 Tripneustes 0.15 33 Peneroplis 2.5 3.5 Gonolithon 24 0.3 Neogoniolithon 16.5 0.5

Because the data ofWalter and Morse (1985) were presented in mol per unit mass per unit time, no corrections for specific surface area and reactive surface area were necessary.

In order to simulate more realistically natural conditions, inhibition ofcarbonate precipitation and dissolution by dissolved organic matter and dissolved phosphate was taken into account. The precipitation rate equation was corrected for both dissolved organic matter and dissolved phosphate because the experimental results ofZhong and

1 Mucci (1989) were derived in low-phosphate «0.1Ilmol L- ) artificial seawater. It was assumed that the artificial seawater contained no or very little dissolved organic matter.

The dissolution rate equation was only corrected for phosphate inhibition since the experiments by Walter and Morse (1985) were carried out in low-phosphate

1 «O.lllmol C ) seawater, i.e. it was assumed that this seawater contained dissolved

64 organic matter. The uncertainty ofthe adopted kinetic rate equations and the factors inhibiting these reaction rates were investigated by sensitivity analyses.

Inhibition by dissolved organic matter was based on Berner et al. (1978).

Dissolved organic matter consists ofnumerous organic molecules such as amino acids, proteins, aromatic acids, fatty acids, humic acids, fulvic acids, and others. Each substance has a different effect on the carbonate precipitation process (Berner et aI., 1978). In the standard scenario ofSOCM, a best fit ofthe inhibition on aragonite precipitation by humic acid from marine mud was adopted (Fig 2.2). It was assumed that the pore water contained 10 mg C L-1 ofDaM as humic acid, inhibiting carbonate precipitation by a factor of0.008 (Inhibition factor, I = R!Ro; R is the rate ofprecipitation in seawater containing DaM and RD is the rate ofprecipitation in seawater with no DaM). In reality, dissolved organic matter concentrations in nature are quite variable. Dissolved organic matter concentrations in pore waters ofBahamian sediments ranged from less than I to

20 mg kg-1 with an average of7 mg kg-1 (Morse et aI., 1985), whereas the integrated average DaM concentration in the top 60 cm ofthe sediments ofLong Island Sound was

80 mg CL-I (Berner et aI., 1978).

Phosphate inhibition was derived based on a best fit ofdata from Berner et aI.

(1978; Fig 2.2). The concentration ofdissolved phosphate in pore waters is variable and ranges from concentrations below detection level to concentrations as high as 3000 /.lmol

L-1 (Berner et aI., 1978; Morse et aI., 1985; Orem et aI., 1997). In SOCM, a conservative estimate of 10 /.lmol L-1 was used, inhibiting carbonate precipitation by a factor 0.018 (I

= R!Ro; see previous discussion). As a comparison the dissolved phosphate concentration

65 1.0+----'-----+----...... , __--J'L-.__"'_'--+

y =0.39977 • x"(-1.6754) R=0.99577 \ I 0.8 I I I iO ~ 0.6 I I I I 0.4 I I I I I

0.2 ~

O.O+--'=-=-=-::-- -.=-=-=-=-=-,----- ~------r--+ o 5 10 15 20 25 30

Hum" acid concentration (mg C L') 1.0+----'----'----'----'----'---+

A I y = 1.4444' x {-1.9079) R=0.9864 I 0.8 ~ I I I •

0.4 0.2 -~------r-=-- -~ I -:...:....=...r=------:...::...;-=-=------O.O..------r - - - I I o 5 10 15 20 25 30

OrthOphosphate concentratiOn (umol L')

Figure 2.2. Inhibition factor ofcalcium carbonate reactivity i.e. carbonate dissolution or precipitation by dissolved organic matter and dissolved phosphate as a function ofthe concentration ofthe inhibitor. Inhibition ofaragonite precipitation by humic acid from marine muds (A) and dissolved orthophosphate (B) (Berner et aI., 1978). Ro is the rate ofprecipitation in the absence ofthe inhibitor, and R is the rate when the inhibitor is present. Fitted equations are representative for the solid black lines. Dashed lines are linear extrapolations to axis limits and are forced through an inhibition factor of 1 for x = o.

66 ofupwelling regions is approximately 1 I-lmol L-1 (Berner et aI., 1978). The higher phosphate concentration observed within the pore water-sediment system is due to extensive remineralization oforganic matter within the sediments.

With corrections for specific surface area (Ass), reactive surface area (ARS), dissolved organic matter inhibition (10 ), and dissolved phosphate inhibition (Ip), the dissolution and precipitation kinetic rate equations (2.8) and (2.9) were converted to the following equations:

RJ = [k(1 - Q)"] x Mx x Ip , (2.1 0)

Rp = [ken - 1)"J x Mx x Ass x ARS x lp x 10 . (2.11) where Mx refers to the total mass ofthe carbonate mineral under consideration, and RJ and Rp represent the total carbonate dissolution and precipitation within the shallow water-ocean environment. Using these equations, the net flux between each carbonate sediment reservoir and the pore water was described by the following equations (see

Appendix A):

F = kCx- PDx.1"'0 + PDx.t, (2.12)

k = [FFoICFoJ, (2.13)

(2.14)

Rxd = 0 for Q > 1; Rxp = 0 for Q < 1 where F is the net flux ofcarbon, k is the rate constant defined by the initial carbonate mass and the initial flux, Cx is the mass ofcarbon in the carbonate phase under

consideration (X = calcite, aragonite or 15 mol% magnesian calcite), and PDx is the difference between dissolution rate (Rxd) and precipitation rate (Rxp), which are described

67 by equations (2.10) and (2.11), respectively. At t=O (i.e. year 1700), the first term in the equation represents the first order net flux ofcarbon between the carbonate and pore water reservoirs (kex), including both carbonate dissolution and precipitation. The next two terms (PDx• HJ + PDx, t), which represent the net change in flux ofcarbon due to dissolution or precipitation as a function ofcarbonate saturation state, are at this time equal to each other and cancel out. As carbonate saturation state changes, this flux changes and is an additional flux added to any flux changes owing to changes in carbonate mass.

2.6 TERRESTRIAL OCEAN ATMOSPHERE ECOSYSTEM MODEL (TOTEM)

To facilitate the inputs and outputs at the boundaries ofthe shallow-water ocean environment, SOCMwas incorporated into the global biogeochemical model TOTEM

(Fig. 2.3; Terrestrial Ocean aTmosphere Ecosystem Model; Ver, 1998; Ver et aI., 1994,

1998, 1999a, 1999b; Mackenzie et aI., 1998a, 1998b, 2000, 2002). TOTEM is an Earth system model for the domains ofland, coastal and open ocean (including the marine sediments ofthese domains), and atmosphere. It is comprised ofthirteen reservoirs: the atmosphere; six terrestrial reservoirs (living biota, humus, inorganic soil, continental soilwater, shallow groundwater, and lakes); three coastal-zone reservoirs (organic matter, surface water, and sediments); and three open ocean reservoirs (organic matter, surface water, and deep water). TOTEMbegins numerical calculations at a quasi-steady state condition in the year 1700. The model is based on parameterization ofphysical, biogeochemical, and ecological processes affecting the global environmental system.

68 _a.-aus.c ~""cln;'i? :,::.;.! F(,S!1 "_~It:--I$::IYS. ce,.: U ... ~. ':.I"':'~ •• W•••••••••••••••••••••••••••••••••••••••••••• • ••• <':Ja~tal ~edl""ent :>en,tr '.:a~::r: Ni.

D-nouron '3, on "$;11. :;, OteOS':lCt" • CC2' -..H ... t-c ... b-IO'-:rtnl:: ~ N")(. NO~ :..ux. NO ... SO.

51-:' ::'91 :al ... ':'~ke: zi N:,)J' ;;'-0..; ~4 N z . o'" : Land l..se u : C~,a.";_~: •N... j. ,.. ..."'. •...... : ~C'c . .. ~ :,,~~~!~:~~~ .• ;, : Land ""ie Cna;;ge: .J : Ole, N:J3' ;;'04, ~(-'4

:rcrsIX/L;,-rjl,•• .. Si-ett~:': .­ or;an C ~ "'organ c :artl:.. lates

A;pQ.l : ... ~al. B.. nal ~e(tl zer • :.:aCOj .... NOJ. ~::::45 la-=: an: 7v1ar ne . :a gao-Ic '~~ a~e'

Figure 2.3. Conceptual schematic ofthe global biogeochemical model TOTEM (Terrestrial Ocean aTmosphere Ecosystem Model) (Ver, 1998; Ver et aI., 1999; Mackenzie et aI., 2002; and references therein). The highlighted region in the middle ofthe plot corresponds to the shallow-water ocean environment. For details on how this region was modified in SOCM see Figure 2.1.

69 Most significant for the modeling approach is the coupling ofthe cycles ofcarbon, nitrogen, sulfur and phosphorus for every biologically mediated transfer process, such as photosynthesis, autorespiration, decay and burial. In TOTEM the linkages ofthe individual cycles are achieved through the average ratios ofC:N:P:S in marine plankton and benthic marine plants (Redfield ratios), terrestrial plants, soil organic matter, and organic matter in sediments. It is assumed that these ratios are generic and apply over many different environments. In reality the ratios are variable and can be significantly different from the idealized ratios. Additionally, it is assumed that the elemental ratios remain constant and do not change with time on the decadal to century time scale ofthe past and future several centuries. Transfer processes between the reservoirs in the model are defined with linear and non-linear equations describing reaction mechanisms and physical transport processes.

Standard anthropogenic forcings in TOTEM ofthe quasi-steady state condition from 1700 forward include emissions ofCO2 to the atmosphere due to fossil fuel buming and land use, application ofN and P fertilizers to the landscape, releases ofN and S gases to the atmosphere due to fossil fuel burning, and N and P loading ofrivers and coastal areas from sewage. The historical record ofglobal mean surface temperature and future forecasts are also a forcing in the model. Significant results include reconstruction ofthe history ofatmospheric CO2 growth and the fate ofanthropogenic CO2 during the past 300 years (Mackenzie et ai., 1998; Ver et ai., 1999); quantification ofthe human-induced N,

P, and CO2 fertilization ofterrestrial ecosystems (Mackenzie et aI., 1998); first global assessment ofthe quantitative role ofthe coastal zone as a source or sink ofatmospheric

70 CO2 and projections for the future (Mackenzie et aI., 1998; Ver et aI., 1999); evaluation ofhow changes in thermohaline circulation ofthe ocean may affect future C02 concentrations (Mackenzie et aI., 2000; Ver et aI., 1999); and assessment offuture atmospheric CO2 concentrations using various CO2 emission scenarios (Ver et aI., 1999;

Mackenzie et aI., 2002).

71 CHAPTER 3: SOLUTION OF SHALLOW-WATER CARBONATE: AN

INSIGNIFICANT BUFFER AGAINST RISING ATMOSPHERIC CO2

This chapter was published as an article in the scientific journal Geology, June I,

2003 (Andersson, A.J., Mackenzie, F.T., and Ver, L.M., 2003. Solution ofshallow-water carbonates: An insignificant buffer against rising atmospheric CO2• Geology, 31 :513­

516). Some ofthe values ofmasses or fluxes shown in the model schematic (Fig 3.1) have been updated since the time ofsubmission and could therefore be different ofthose shown in the schematic in Chapter 2. However, only minor changes have been made and the results ofthe model have not been affected.

3.1 ABSTRACT

Model predictions suggest that the saturation state ofsurface ocean waters with respect to carbonate minerals will decline during the twenty-first century owing to increased invasion ofatmospheric CO2. As a result calcareous organisms may have difficulty calcifying, leading to production ofweaker skeletons and greater vulnerability to erosion. Alternatively, it has been suggested that there will be no significant impact on coral reefecosystems because any changes in saturation state and pH will be restored by dissolution ofmetastable carbonate minerals. To resolve this controversy, we employ a physical-biogeochemical box model representative ofthe shallow-water ocean environment. Numerical simulations demonstrate that the carbonate saturation state of surface waters could significantly decrease and hamper the biogenic production of

72 CaC03during the twenty-first century. Similarly, the average saturation state ofmarine pore waters could significantly decline, inducing dissolution ofmetastable carbonate phases within the pore-water-sediment system. Such dissolution could buffer the carbon chemistry ofthe pore waters, but overlying surface waters ofreefs and other shallow­ water carbonate environments will not accumulate sufficient alkalinity to buffer pH or carbonate saturation state changes owing to invasion ofatmospheric CO2•

Keywords: calcium carbonate, coral reefs, calcification, carbonate sediments, C02.

3.2 INTRODUCTION

Future projections from climate system and carbon models suggest that atmospheric CO2 concentrations by the end ofthe twenty-first century will be close to

700 ppmv (-370 ppmv at present) and global mean surface temperature could be 1.4-5.8

°c warmer than A.D. 2000 (Houghton et al., 2001). Increased atmospheric CO2 will subsequently lead to increased dissolved inorganic carbon in the mixed layer ofthe ocean due to the increased uptake ofgaseous CO2 in seawater according to CO2 + H20 + col­

= 2HC03-. Consequently col- concentration will decrease, thereby lowering the saturation state ofseawater with respect to carbonate minerals. Calcium carbonate saturation state (0) is determined by the product ofthe concentrations of[Ca2+J and

[COl-j divided by the stoichiometric solubility product (Ksp').

Experimental evidence indicates that the rate ofcalcification ofcoralline algae and corals exhibits a strong positive correlation to saturation state (Smith and Roth, 1979;

73 Mackenzie and Agegian, 1989; Gao et aI., 1993; Gattuso et aI., 1998; Langdon et aI.,

2000) and a negative parabolic dependence on temperature (Clausen and Roth, 1975;

Mackenzie and Agegian, 1989). Studies comparing corals from different locations show that a positive correlation between rate ofcalcification and temperature exists (Lough and

Barnes, 2000), suggesting that a moderate temperature rise could facilitate increased coral growth. On the basis ofexperimental results and geochemical modeling, it has been suggested that calcification by coral reefs will decrease significantly by the middle ofthe twenty-first century owing to increased atmospheric C02 and subsequent decreased carbonate saturation state (Gattuso et aI., 1999a; Kleypas et aI., 1999; Mackenzie et aI.,

2000; Langdon et aI., 2000; Leclercq et al., 2002). However, experimental field and laboratory results indicate that dissolution ofmetastable carbonate phases, such as high­ magnesian calcites, potentially could restore any changes in carbonate saturation state and pH owing to increased atmospheric C02 (Barnes and Cuff, 2000; Halley and Yates,

2000).

In an attempt to provide some resolution to this issue, we constructed and employed a box model representative ofthe global shallow-water ocean environment.

The objective was to investigate the hypothesis that dissolution ofmetastable carbonates can restore the alkalinity and pH ofthe surface water and provide a temporal refuge for calcifying organisms against anthropogenically induced changes in the dissolved inorganic carbon system ofseawater.

74 3.3 METHODS

The shallow-water ocean environment was defined to include coastal zones, reefs, banks, and shelves. The model consisted oftwo major domains representing the surface water and the pore-water-sediment system (Fig 3.1). The surface-water domain included a water-column reservoir and an organic-matter reservoir (particulate organic carbon, dissolved organic carbon, and living biota). The pore-water-sediment system was defined by an organic-matter reservoir, reservoirs ofbiologically produced calcite, aragonite, 15 mol% magnesian calcite, and river-transported CaC03 (mainly refractive calcite), and a pore-water reservoir ofaverage composition (C02,tOla] = 3800 ~mol/L; pHNBS = 7.51) based on a variety ofenvironments from the Bahamas and elsewhere (Morse et aI., 1985;

Morse and Mackenzie, 1990). The parameters for the pore-water-sediment system were calculated by assuming a reactive sediment depth of 1 m with 50% porosity (Berner,

1980) and a total area of28.3 x 106 km2 (Milliman, 1974). Ofthe total area, 95% was assumed to contain an average carbonate content of 15 wt% whereas the remaining 5% contained an average carbonate content of80 wt% (Milliman, 1974). The proportions of

biologically produced calcite, aragonite, and high-magnesian calcite were calculated on the basis oftheir proportion in present-day neritic sediments (12.6%:63.4%:24.0%; Land,

1967), Note that the sediment parameters represent an idealized average setting, and that

local or regional variability can be quite large. Initial net CaC03 fluxes were mainly

adopted from Wollast (1994). Fordetails on reservoirs and fluxes not described here, see

description ofTOTEM and references therein (see subsequent discussion).

75 --- -+(251-)- - Atmosphere-surface water exchange -- .. Rivers, Surface water urfaee water--{ OPen­ open ocean ,ocean I DIC 6000 • exchange. ~C02 (32) = 2000 I-lM (477-+) I pHNBS =8.37 I (421") Organic malter Biological Surface water-pore water Organic I remlnerallzalion uptake excha_ng~e matter ~ ,.------export I (576) (600) (18) I Reactive orgamc Organic matter Pore water I matter 367 54 Upwelling (26) C02 = 3800 IJM (473.S) Biogenic pHNBS=7.51 CaC03 production Organic matter C, A. M (24.S) remineralization Orgamc matter o (31) sedlmentalon Abiotic CaC03 CaC03 dissolution 0(32) precipitation R (S) L C, A. M (0.05) C,A,M (6)

Sediments (28.3 x 106 km2, 1 m depth. 50% porosity) Transport POCo PIC Organic matter River-derived Calcite Aragonite I Magnesian- to slopes PIC (calcite) calcite I-.:..::.....:=~~ 0(8) o (0)1 R (1S) 134.100 29.800 5400 27.350 10.350 R, C, A. M (4) I Burial of land and marine (9) organic matter, river-derived °R (10) PIC, in situ produced CaC03 C, A, M <14.5)

Figure 3.1. Preindustrial carbon estimates ofshallow-water ocean environment used in present simulation. Model consists oftwo major domains representing surface water (white boxes; surface-water and organic-matter) and pore-water­ sediment system (gray boxes; O-organic matter; R-river-derived particulate inorganic carbon; C-----ealcite; A-aragonite; M-magnesian calcite, and pore­ water). Reservoir masses (shown in italics) are in 1012 mol ofcarbon. Arrows denote carbon fluxes between reservoirs in 1012 mol·yr-l. Straight arrows denote physical processes, whereas wavy arrows denote biological or chemical processes. Shallow-water model was integrated into global biogeochemical model TOTEM (Terrestrial Ocean aTmosphere Ecosystem Model) and linked to appropriate reservoirs (open dashed boxes).

76 Equations describing biogenic carbonate production were related to DIC (dissolved inorganic carbon) concentration, temperature, and carbonate saturation state. Biogenic carbonate production was related by a first-order dependence on DIC content (CD/c) to account for the assumption that community production is dependent on this parameter

(Raven, 1993; Marubini and Thake, 1999). However, this assumption is weak and the response ofbiogenic calcification to intermediate changes in DIC is poorly known. In the present simulation, it has a minor effect. Relative biogenic carbonate production (R [in percent]) was related to temperature (1) according to the relationship ofPorolithon gardineri (Mackenzie and Agegian, 1989):

R = 100 - O.00013LlT- 1.3Llr. (3.1)

Sensitivity ofthe calcification rate dependence on saturation state (0) was explored by employing two different scenarios using the curvilinear relationship ofStylophora pistil/ata (Gattuso et aI., 1998; Leclercq et aI., 2002),

nJO 69 R = 228(1 - e- . ) - 128, (3.2) and a relationship derived by a linear best fit to multiple data for various coralline algae and corals (Gattuso et aI., 1999a),

R = 21.30 + 12. (3.3)

The overall equation describing the flux ofbiogenic carbonate production (F) was described by:

F= CD/C x (FinitiaVCDlC,initial) x Ro x RT • (3.4)

77 Inorganic carbonate precipitation (Zhong and Mucci, 1989) and dissolution

(Walter and Morse, 1985) within the pore-water-sediment system were defined by equations ofthe general form

R=k(Q-l)" (3.5) and

R =k(1- Q)", (3.6) respectively, where R = rate ofchange, k = rate constant, Q = saturation state, and n = empirical reaction order (Table 3.1). Precipitation equations were normalized to reservoir masses byusing best estimates ofspecific surface area and reactive surface area (Walter and Morse, 1984, 1985). Inhibition by dissolved phosphate (Berner et aI., 1978; Mucci,

1986) and dissolved organic matter (Berner et aI., 1978) was also taken into account.

Solubilities ofcalcite and aragonite were obtained from Plummer and Busenberg (1982), and the solubility of IS mol% magnesian calcite was taken from Bischoffet aL (1993).

TABLE 3.1. CONSTANlS ADOPTED FOR INORGANIC Caco3 PRECIPITATION AND DISSOLUTION RATES (R) Rate constant (k) Reaction order (n)

Precipitation R = k(a-1)n (Zhong and Mucci, 1989) Calcite 10-0·29 2.80 15 mol% Mg-<:alcite' 1a.o·2• 2.80 , Aragonite 10 .09 2.36

Dissolution R = k (1-a)" (Walter and Morse, 1985) Calcite 102.82 2.86 15 mol% Mg-<:alcite 102.62 3.34 Aragonite 102.89 2.48

'*Assuming same rate as calcite

78 To facilitate the inputs and outputs at the boundaries ofthe shallow-water ocean environment, the model was incorporated into the global biogeochemical model TOTEM

(Terrestrial Ocean aTmosphere Ecosystem Model; Ver et al., 1999; Mackenzie et al.,

2000; and references therein). TOTEM is an Earth system model for the domains ofland, coastal and open ocean, and atmosphere. It links the behavior ofthe carbon cycle to those ofthe elements ofnitrogen, phosphorus, and sulfur, and begins numerical calculations at a quasi-steady-state condition in the year 1700. Several anthropogenic emission parameters and the historical record ofglobal mean surface temperature and future forecasts are used as forcings. In the current investigation, multiple sensitivity analyses were conducted to investigate the importance ofcarbonate precipitation-dissolution kinetics, initial pore-water carbonate saturation state and carbon chemistry, magnesian calcite solubility, flux oforganic matter from rivers or in situ production and its subsequent remineralization.

3.4 RESULTS AND DISCUSSION

Our numerical simulations showed that the saturation state ofsurface marine waters with respect to carbonate minerals decreased between 1700 and 2100. The trend was observed in both surface and pore waters (Fig 3.2). Even though the saturation state ofthe surface water decreased by -20% by the year 2100, it remained well above saturation with respect to all carbonate phases under consideration. Decreased saturation state ofthe surface waters led to a reduction in biogenic carbonate production by 70/.­

44%, depending on whether a linear or a curvilinear saturation-state relationship was

79 7 A 8.4

6 • • 8.2 5 ...... 8.0 I d' 4 -111------11--.-....1----.:: Q. 7.8 3 -OpH "Calcite 7.6 2 ....Aragonile ~!~ .!1:!~I~~ !'!!l.-~~~c!t~ 1 _ _ 7.4 7 1.8 7.50 <.~.::~_...... - B 8.4 6 .~ eo...... ~"\. e-. 8.2 5 1.4 """'-\.. 7.45 -.- .~\ 8.0 ...... ,.."1'--...... c:X4 ""- -. I 1.0 ...... -'-"'.-tII 7.8 Q. 3 7.40 1950 2000 2050 2100 7.6

7.4

-Q- Linear sat. stale dependence -10 J:;}- Nonlinear sat. slale dependence 1700 1800 1900 2000 2100 Year

Figure 3.2. Saturation state (Ox) with respect to carbonate minerals on left y­ axis and pH on right y-axis observed in (A) surface water and (B) pore water between year 1700 and 2100. Results between 1950 and 2100 have been inserted to highlight apparent buffer effect taking place within pore water upon dissolution of 15 mol% magnesian calcite at end ofsimulation. (C) Changes in biogenic carbonate production (x 1012 mol'yr-l) during simulation.

80 used in the calculations (Fig 3.2). This result is in good agreement with previous estimates (140/0-40% by year 2065) derived from experimental work and geochemical modeling (Gattuso et aI., 1999a; Kleypas et aI., 1999; Langdon et aI., 2000). The smaller reduction observed by the adoption ofa curvilinear relationship between biogenic calcification and saturation state can be attributed to the fact that the carbonate saturation state remained above the threshold for significant change in calcification throughout most ofthe simulation. A small increase in biogenic carbonate production was observed from the beginning ofthe simulation until the middle ofthe twenty-first century, which was due to the increasing DIC content ofthe surface water (Fig 3.2). Relative to the results obtained by using a curvilinear biogenic calcification saturation-state dependence, adoption ofa linear relationship produced a significantly greater reduction in biogenic carbonate production by the year 2100; however, no significant difference was observed in surface water or pore-water carbon chemistry.

At the end ofthe simulation, the average pore water became undersaturated with respect to IS mol% magnesian calcite (Fig 3.2). Consequently, magnesian calcite dissolved and produced alkalinity, leading to a significant decrease in the rate ofchange in saturation state with respect to carbonate minerals within the pore-water-sediment system, i.e. it acted as a buffer. None ofthis buffering effect was observed in the overlying surface water which we attribute to its well-mixed character and significantly greater reservoir size compared to the pore water.

The problem ofwhether dissolution by metastable carbonate phases can buffer the overlying water column is dependent on the rates ofphysical mixing processes and the

81 1.20 :=""~~::;::=-----~kT 4000 3500 1.15 3000 1.10 2500 ~ "0 2000 E 1.05 N 1500 ~a,.... ()) 1000 ~ ~1.00 J 1 ~ "0 ~ 0.95 ~~~~~~~~~~~~::=='::~~ 0 ~ C *~ B 400 0 ~ 1.00 ....------.--...-- ...------e ~ ~ C ~ I :;::; 01 0.99 "5 200 E:0 ()

100 0.98

1700 1800 1900 2000 Year Figure 3.3. Sensitivity analyses with respect to (A) river-transported organic carbon (I x-triangles, 2x-squares, and lOx---diamonds) to shallow-water ocean environment and its subsequent remineralization and (B) calcium carbonate dissolution rates (Ix-triangles, lOx-squares, and 100x~amonds). In each scenario, solid symbols represent saturation state (0) with respect to 15 mol% magnesian calcite, and open symbols show cumulative amount ofcarbon (x 1012 mol) dissolved. In (B), initial average carbon chemistry ofpore water was adjusted so that saturation state with respect to 15 mol% magnesian calcite was close to 1.0, i.e., at saturation.

82 extent ofcarbonate dissolution, which is mainly controlled by the chemistry ofthe sediment pore-water (Leclercq et ai., 2002). Ifexchange with the open ocean is slow enough and dissolution ofmetastable carbonates is quantitatively sufficient and fast enough, it is possible that a buffer effect on a relatively short time scale could be observed in certain regions differing from the idealized average environmental setting of the present simulation. The extent ofcarbonate dissolution will mainly be governed by the remineralization oforganic matter and subsequent production ofCO2 rather than by the kinetics ofcarbonate mineral dissolution (Morse and Mackenzie, 1990). Thus, sensitivity analysis with respect to organic matter transported via rivers and its subsequent remineralization, assuming a first-order dependence, resulted in significant increases in the amount ofcalcium carbonate dissolved as the load oforganic matter increased (Fig 3.3). Changes in the dissolution rates ofcarbonate minerals by up to two orders ofmagnitude resulted only in minor changes in the amount ofcalcium carbonate dissolved (Fig 3.3). In either case, although as much as -40% ofthe global reservoir of magnesian calcite dissolved in the worst-case scenario, no significant buffer effect was observed in the overlying surface water.

It is important to note that although the average pore water did not become undersaturated with respect to 15 mol% magnesian calcite until the last decades ofthe simulation, magnesian calcite with>15 mol% MgCOJ was already subjected to dissolution before this event. Sensitivity analysis indicates that even ifthe initial pore­ water carbon chemistry was different and dissolution of15 mol% MgCOJ was already taking place at the onset ofthe simulation, or another more soluble composition

83 magnesian calcite was used in the simulation (Bischoff et ai., 1993), there still would not be any significant buffering effect seen in the overlying water column. However, decreased saturation state and dissolution ofmetastable carbonate minerals within the pore-water-sediment system could have significant implications for the rate of precipitation and composition ofcarbonate cements in shallow-water sediments. Given time and depending on the extent ofenvironmental change, metastable carbonate phases will most often dissolve and recrystallize into more stable phases (Morse and Mackenzie,

1990; Tribble et ai., 1995). In the future, it is possible that the average MgC03 content of magnesian calcite cements and the bulk magnesium content ofcontemporary sediments could gradually decrease as an effect ofchanges in the average pore-water saturation state.

3.5 CONCLUSIONS

Our results confinn the hypothesis that metastable carbonates could dissolve in the future owing to increased invasion ofatmospheric CO2 (Barnes and Cuff, 2000;

Halley and Yates, 2000), but the surface water ofthe global shallow-water marine environment will not accumulate sufficient alkalinity to buffer pH or carbonate saturation state. Thus, calcification by calcareous marine organisms and the development of carbonate reefs could be negatively affected as a consequence ofrising anthropogenic

C02 and lowering ofthe saturation state ofseawater with respect to carbonate minerals

(Gattuso et al., 1999a; Kleypas et al., 1999; Mackenzie et ai., 2000; Langdon et al., 2000;

Leclercq et ai., 2002). In addition, future decrease in pore-water saturation state could

84 affect the composition and rates ofprecipitation ofcarbonate cements in contemporary shallow-water marine sediments.

3.6 ACKNOWLEDGMENT

This research was supported by National Science Foundation grants ATMOO­

80878 and EAR02-235090. This is University ofHawaii School ofOcean and Earth

Science and Technology contribution 6113.

3.7 REFERENCES CITED

Barnes, DJ., and Cuff, C., 2000, Solution ofreefrock buffers seawater against rising

atmospheric CO2, in Hopley, D., Hopley, M., Tamelander, J., and Done, T.,

Proceedings ofthe Ninth International Coral Reef Symposium Abstracts, State

Ministry for the Environment, Indonesia, p. 248.

Berner, RA., 1980, Early diagenesis: A theoretical approach: Princeton, New Jersey,

Princeton University Press, 241 p.

Berner, RA., Westrich, J.T., Graber, R, Smith, J., and Martens, C.S., 1978, Inhibition of

aragonite precipitation from supersaturated seawater: A laboratory and field

study: American Journal ofScience, v. 278, p. 816-837.

Bischoff, W.D., Bertram, M.A., Mackenzie, F.T., and Bishop, F.C., 1993, Diagenetic

stabilization pathways ofmagnesian calcites: Carbonates and Evaporites, v. 8,

p.82-89.

85 Clausen, C.D., and Roth, A.A., 1975, Effect oftemperature and temperature adaptation

on calcification rate in the hermatypic coral Pocillopora damicornis: Marine

Biology, v. 33, p. 93-100.

Gao, K., Aruga, Y., Asada, K., and Kiyohara, M., 1993, Influence ofenhanced CO2 on

growth and photosynthesis ofthe red algae Graciliaria sp. and G. chilensis:

Journal ofApplied Phycology, v. 5, p. 563-571.

Gattuso, J.-P., Frankignoulle, M., Bourge, 1., Romaine, S., and Buddemeier, R.W., 1998,

Effect ofcalcium carbonate saturation of seawater on coral calcification: Global

Planetary Change, v. 18, p. 37--46.

Gattuso, J.-P., Allemand, D., and Frankignoulle, M., 1999a, Photosynthesis and

calcification at cellular, organismal and community levels in coral reefs: A review

on interactions and control by carbonate chemistry: American Zoology, v. 39,

p.160-183.

Halley, R.B., and Yates, K.K., 2000, Will reefsediments buffer corals from increased

global C02, in Hopley, D., Hopley, M., Tamelander, J., and Done, T.,

Proceedings ofthe Ninth International Coral ReefSymposium Abstracts, State

Ministry for the Environment, Indonesia, p. 248.

Houghton, J.T., Ding, Y., Griggs, DJ., Noguer, M., van der Linden, PJ., Dai, X.,

Maskell, K., and Johnson, C.A., oos., 2001, Climate change 2001: The scientific

basis. Contribution ofWorking Group I to the Third Assessment Report ofthe

Intergovernmental Panel on Climate Change: Cambridge, Cambridge University

Press, 881 p.

86 Kleypas, J.A., Buddemeier, R.W., Archer, D., Gattuso, J.-P., Langdon, C., and Opdike,

B.N., 1999, Geochemical consequences ofincreased atmospheric carbon dioxide

on coral reefs: Science, v. 284, p. 118-120.

Land, L.S., 1967, Diagenesis ofskeletal carbonates: Journal ofSedimentary Petrology,

v. 37, p. 914-930.

Langdon, C., Takahashi, T., Sweeney, C., Chipman, D., Goddard, J., Marubini, F.,

Aceves, R., Barnett, R., and Atkinson, M., 2000, Effect ofcalcium carbonate

saturation state on the calcification rate ofan experimental coral reef: Global

Biogeochemical Cycles, v. 14, p. 639-654.

Leclercq, N., Gattuso, J.-P., and Jaubert, J., 2002, Primary production, respiration, and

calcification ofa coral reefmesocosm under increased CO2 partial pressure:

Linmology and Oceanography, v. 47, p. 558-564.

Lough, J.M., and Barnes, D.J., 2000, Environmental control on the massive coral Porites:

Journal ofExperimental Marine Biology and Ecology, v. 245, p. 225-243.

Mackenzie, F.T., and Agegian, C.R., 1989, Biomineralization and tentative links to plate

tectonics, in Crick, R.E., ed., Origin, evolution, and modern aspects of

biomineralization in plants and animals: New York, Plenum Press, p. 11-27.

Mackenzie, F.T., Ver, L.M., and Lerman, A., 2000, Coastal zone biogeochemical

dynamics under global warming: International Geology Review, v. 42, p. 193­

206.

Marubini, F., and Thake, B., 1999, Bicarbonate addition promotes coral growth:

Linmology and Oceanography, v. 44, p. 716-720.

87 Milliman, J.D., 1974, Recent sedimentary carbonates, Part I: Marine carbonates: Berlin,

Springer-Verlag, 375 p.

Morse, J.W., and Mackenzie, F.T., 1990, Geochemistry ofsedimentary carbonates:

Amsterdam, Elsevier, 707 p.

Morse, J.W., Zullig, J.J., Bernstein, L.D., Millero, FJ., Milne, P., Mucci, A., and

Choppin, G.R., 1985, Chemistry ofcalcium carbonate-rich shallow water

sediments in the Bahamas: American Journal ofScience, v. 285, p. 147-185.

Mucci, A., 1986, Growth kinetics and composition ofmagnesian calcite overgrowths

precipitated from seawater: Quantitative influence oforthophosphate ions:

Geochimica et Cosmochimica Acta, v. 50, p. 2255-2265.

Plummer, L.N., and Busenberg, E., 1982, The solubilities ofcalcite, aragonite and

vaterite in CO2-H20 solutions between 0-90°C and an evaluation ofthe aqueous

model for the system CaC03-C02-H20: Geochimica et Cosmochimica Acta,

v.46,p.1011-I040.

Raven, J.A., 1993, Limits on growth rates: Nature, v. 361, p. 209-210.

Smith, A.D., and Roth, A.A., 1979, Effect ofcarbon dioxide concentration on

calcification in the red coralline alga Bossiella orbigniana: Marine Biology, v. 52,

p.217-225.

Tribble, J.S., Arvidson, R.S., Lane, M., III, and Mackenzie, F.T., 1995, Crystal

chemistry, and thermodynamic and kinetic properties ofcalcite, dolomite, apatite

and biogenic silica: Applications to petrologic problems: Sedimentary Geology,

v. 95, p. 11-37.

88 Ver, L.M., Mackenzie, F.T., and Lerman, A., 1999, Biogeochemical responses ofthe

carbon cycle to natural and human perturbations: Past, present, and future:

American Journal ofScience, v. 299, p. 762-801.

Walter, L.M., and Morse, J.W., 1984, Reactive surface area ofskeletal carbonate during

dissolution: Effect ofgrain size: Journal ofSedimentary Petrology, v. 54,

p. 1081-1090.

Walter, L.M., and Morse, J.W., 1985, The dissolution kinetics ofshallow marine

carbonates in seawater: A laboratory study: Geochimica et Cosmochimica Acta,

v.49,p.1503-1513.

Wollast, R., 1994, The relative importance ofbiomineralization and dissolution ofCaC03

in the global carbon cycle, in Doumenge, F., ed., Past and present

biomineralization processes: Considerations about the carbonate cycle: Monaco,

Bulletin de l'Institut oceanographique, numero special 13, p. 13-34.

Zhong, S., and Mucci, A., 1989, Calcite and aragonite precipitation from seawater

solutions ofvarious salinities: Precipitation rates and overgrowth compositions:

Chemical Geology, v. 78, p. 283-299.

89 CHAPTER 4: SENSITIVITY ANALYSIS AND VALIDATION OF soeM

4.1 INTRODUCTION

The objective ofthis chapter is to provide validation for the Shallow-water Ocean

Carbonate Model (SOCM), i.e. to assess the reliability ofthe model and to test the robustness ofthe main conclusion that dissolution ofmetastable carbonate minerals will not produce sufficient alkalinity to buffer surface ocean waters against increasing atmospheric C02. In the first section ofthe chapter, a summary ofthe results ofthe standard model ofSOCM is given and compared to observational data from the Hawaiian

Ocean Time series (HOT; http://hahana.soest.hawaii.edu/hot/hot jgofs.html) and the

Bermuda Atlantic Time Series (BATS; http://www.bbsr.edu/cintoo/bats/bats.html).

Additional observations are also presented and discussed relative to the results ofSOCM.

The remaining sections ofthe chapter are dedicated to sensitivity analyses to evaluate the robustness ofthe main conclusions and the relative importance ofthe model parameters and processes within the shallow-water ocean environment. The theory behind sensitivity analysis and the method utilized in the current analyses are briefly discussed.

The results ofthe sensitivity analysis are presented and discussed in relation to observations. Finally, the impact ofincreasing atmospheric CO2 and temperature on biogenic calcification is investigated by adopting a range ofrelationships between the rate ofcalcification and carbonate saturation state and temperature derived from the literature. The atmospheric CO2 concentration and temperature were altered by adopting a range ofCO2 emission and temperature scenarios presented by the IPCC SRES

(Houghton et aI., 2001).

90 4.2 VALIDATION OF SOCM (Shallow-water Ocean Carbonate Model)

4.2.1 Results of standard run

The results ofthe standard simulation indicated that the carbonate saturation state ofthe surface waters ofthe shallow-water ocean environment has decreased since the onset ofthe industrial revolution, and in particular during the last halfofthe 20th century until present owing to increased invasion ofatmospheric CO2 (Fig 3.2). Numerical calculations showed that the carbonate saturation state could be expected to continue to decrease throughout the 21 st century owing to continued increases in atmospheric CO2•

As a consequence, carbonate production by calcareous organisms could be expected to be negatively impacted by decreasing carbonate saturation state (Fig 3.2). As opposed to the hypothesis that any changes in surface water inorganic carbon chemistry will be restored by dissolution ofmetastable carbonate minerals (Barnes and Cuff, 2000; Halley and

Yates, 2000), the results ofthe standard simulation indicated that surface ocean waters will not accumulate sufficient alkalinity owing to dissolution ofcarbonate minerals to counteract the effect ofincreasing atmospheric CO2• However, the results did show that dissolution ofmetastable carbonate minerals could occur or might already be taking place within the pore water-sediment system (Fig 3.2). A carbonate buffer effect was observed within the pore water, but none or very little ofthis effect was observed in the overlying surface waters. Consequently, the negative effects ofdecreasing carbonate saturation state ofsurface ocean waters on calcareous organisms such as corals and coralline algae will not be prevented by dissolution ofhigh magnesian calcite minerals.

91 4.2.2 Comparison ofmodel results to observations from the natural

environment

The results ofthe standard simulation suggest that a decrease in carbonate saturation state ofthe world oceans should already be apparent in historical datasets of the marine carbon system. Consequently, depending on the relationship between biogenic calcification and carbonate saturation state, as well as the influence on calcification of temperature and other variables such as nutrient concentrations, light, etc, it is possible that a decrease in biogenic calcification also could be observed in historical records. In addition, the model results suggest that metastable carbonate minerals currently should be subject to dissolution within the pore water-sediment system.

4.2.2.1 Carbonate saturation state

The Mauna Loa CO2record (Fig 1.1; Keeling and Whorf, 2002) is one ofthe best time series illustrating how atmospheric CO2concentrations have increased during the past decades owing to human activities (Keeling et aI., 2002). Increasing atmospheric

C02 will subsequently lead to increased invasion ofthis gas into the surface ocean and cause an increase in the total dissolved inorganic carbon ofthis reservoir. Indeed, although the existing records are relatively short, observations from the Hawaiian Ocean

Time series (HOT) and the Bermuda Atlantic Time Series (BATS) indicate a gradual increase in mixed layer total dissolved inorganic carbon concentration with time (Bates et a!., 1996; Winn et a!. 1998; Bates, 2001). Consequently, a decline in carbonate saturation state would be anticipated for at least the shallower mixed layer waters at these locations.

92 0 7 A 50 6 gl00 .s::. a. 5 ~ 150

200 4

250 3 B • Average sat. slate - Linear regression 7.0 -MODELl ···MODEL2 6.5 . tft. #1• 6.0 -. I_ I ". - - ~ I.~ Ij I ! ~ • 5.5 . , 1n l: 0 ! ::::I C • Annual~~ i - Linear regression ~ 7.0 -MODELl

6.0 -~ 5.5 ---- I 1993 1994 1995 1996 Year

Figure 4.1. Changes in calcite saturation state between 1993 and 1996 at Bennuda Atlantic Time Series (BATS). (A) Vertical contour profile ofcalcite saturation state between the surface and 250 meters depth (color bar indicates calcite saturation value), (B) average calcite saturation state (Linear regression: red line, y= -0.0418 + 89.1833; R2 = 0.04;p = 0.19), and (C) annual average calcite saturation state (Linear regression: red line, y = -0.0200 + 45.6218; R2 = 0.07;p = 0.74) ofthe upper 50 meters calculated based on the carbonic acid system ofRoy et al. (1993). Error bars indicate one standard deviation. Blue lines indicate calcite saturation state predicted by the model simulation. Model 1 was calculated based on activity coefficients (Nagy, 1988; Morse and Mackenzie, 1990) and the carbonic acid system as defined by Plummer and Busenberg (1982). Model 2 was calculated based on the stoichiometric constants ofRoy et al. (1993).

93 a) Bermuda Atlantic Time Series (BATS)

A slight, gradual decline in carbonate saturation state is observed in the upper 50 meters ofthe water column at BATS between 1993 and 1996 (Fig 4.1). The rate of change is in relatively good agreement with the predicted trend ofthe current standard model, but is offset by approximately 0.5 units. The offset can be explained by the uncertainty associated with the initial estimate ofsurface water carbonate chemistry of

• Seasurface sat. statel - Linear regression 7.0 lA - MODEL1 --- MODEL2 I 6.5t_

6.0 . .... 1 i ~ 5.5 . c: e0 :::I m B • Annual mean .s - Linear regression '0 7.0 - MODEL1 (ij --- MODEL2 (J

6.5 ' -~ 6.0 -'r-- f -i- i-I 5.5

1993 1994 1995 1996 1997 1998 Year Figure 4.2. (A) Seasonal changes in calcite saturation state (Linear regression: red line, y = -0.0146 + 35.0656; R2 = O.OI;p = 0.39), and (B) the annual average calcite saturation state (Linear regression: red line, y = -0.0079 + 2 21.6794; R = 0.03; P = 0.73) in the upper surface layer « 8m) at BATS between 1993 and 1998. Error bars indicate one standard deviation. Blue lines indicate calcite saturation state predicted by the model simulation. Model 1 was calculated based on activity coefficients (Nagy, 1988; Morse and Mackenzie, 1990) and the carbonic acid system as defined by Plummer and Busenberg (1982). Model 2 was calculated based on the stoichiometric constants ofRoy et al. (1993).

94 SOCM and the fact that this estimate represents a global average but there could be significant regional differences. The correlation between time and carbonate saturation state is not significant (Kendall's rank correlation; Laws, 1997), but the record encompasses only a limited period oftime. The carbonate saturation state is strongly influenced by seasonal variability, which is much greater than the annual variability; thus, a longer record is needed to make any robust statistical conclusions. Water column measurements for depths < 8 m from the same location between 1993 and 1998 show no obvious decreasing trend in saturation state (Fig 4.2).

b) Hawaiian Ocean Time series (HOT)

The annual mean ofthe average carbonate saturation state for the upper 50 meters ofthe water column at HOT between 1989 and 2000 shows a consistently decreasing trend (Fig 4.3). This is in good agreement with the predicted trend ofthe standard model simulation, but shows an offset similar to that ofBATS (see previous discussion). Using non-parametric statistical tests (Kendall's rank correlation; Laws, 1997), the correlation between time and carbonate saturation state was found to be significant. Qualitatively, a shoaling ofthe contour lines ofequal carbonate saturation can be observed extending from the surface to a depth of250 meters during this time period particularly since 1994

(Fig 4.3). Ifdata from HOT (ALOHA: N22°45.0' WI58°00.0') are compared with measurements from nearby regions taken in previous years during the GEOSECS cruise

(GEOSECS-206: N21 °09.7' WI53°50.7'; Takahashi et al., 1980) and the CO2 dynamics cruise (ENP-16: N26°01.7' WI50007.1'; ENP-19: N20001.1' W149°57.8'; Chen et al.,

1986), a consistent decline in surface water carbonate saturation state is observed from

95 0 7 A 50 6 g 100 J::.a. 5 Q) Cl 150 4 200

250 3 B • Average sat. state - Linear regression 7.0 -MODEL1 -- MODEL2 6.5

6.0

Q) CiS iii 5.5 c 0 --- ~ ~ CiS C • Annual mean CIl - Linear regression Q) 7.0 'u - MODEL1 co- -- MODEL2 () 6.5 - - - 6.0 ~~'~rt +:~:+- 5.5 +_1-r +r------

1990 1994 1998 2002 Year Figure 4.3. Changes in calcite saturation state between 1989 and 2000 at Hawaii Ocean Time series (HOT). (A) Vertical contour profile ofcalcite saturation state between the surface and 250 meters depth (color bar indicates calcite saturation value), (B) average calcite saturation state (Linear regression: red line, y = -0.0208 + 47.2314; R2 = 0.17;p < 0.01), and (C) annual average calcite saturation state (Linear regression: red line, y = -0.0252 + 56.1284; R2 = 0.82;p« 0.01) ofthe upper 50 meters calculated based on the carbonic acid system ofRoy et al. (1993). Error bars indicate one standard deviation. Blue lines indicate calcite saturation state predicted by the model simulation. Modell was calculated based on activity coefficients (Nagy, 1988; Morse and Mackenzie, 1990) and the carbonic acid system as defined by Plummer and Busenberg (1982). Model 2 was calculated based on the stoichiometric constants ofRoy et al. (1993).

96 • Annual mean - Linear regression 1 - - - Linear regression 2 7.0 - MODEL1 .s GEOSECS --- MODEL2 ctI 1i) c: - o 6.5 ENP ~ ~ co::J (/J HOT .s 6.0 '0co () ------5.5 ---

1970 1975 1980 1985 1990 1995 2000 2005 Year Figure 4.4. Annual changes in average calcite saturation state in the upper 50 meters around the main Hawaiian Islands between 1973 and 2000 based on data from the Hawaiian Ocean Time series (http://hahana.soest.hawaii.edu/hot/hot jgofs.html), the GEOSECS cruise (Takahashi et aI., 1980), the CO2 dynamics cruise (ENP; Chen et 2 aI., 1986). Linear regression 1: red line, y = -0.0366 + 78.7661; R = 0.95;p« 0.01. Because the data from GEOSECS and ENP are represented by a few data points, the linear regression (2) based on only HOT data is also shown (red dashed line; equation is shown in Figure 4.3). Error bars indicate one standard deviation. Blue lines are explained in Figure 4.3

the early 1970s to 2000 (Fig 4.4). The correlation between time and carbonate saturation

state is significant (Kendall's rank correlation; Laws, 1997), but the rate ofdecline is

significantly faster than predicted by the model simulation. However, the data of

GEOSECS and ENP are represented by a single or few data points during a single

sampling ofthe water column in summer 1973 and 1981, respectively, compared to the

average ofmonthly and multiple samples at HOT. The seasonal variability observed in

97 the HOT data suggests that the estimates ofGEOSECS and ENP could have an error of approximately ± 0.5 units, which either could improve the agreement between the model and observations or make it worse. Because the water samples ofGEOSECS and ENP were taken during sununer (early September and July, respectively), it is most likely that the carbonate saturation state at this time is higher than the aunual mean. Thus, the slope ofthe regression line describing the observed rate ofchange might be too steep and in reality might be more similar to the predictions ofthe model, although one must be cautious in not overly interpreting the trend line because locations differ and the marine carbon system was analyzed by different investigators.

4.2.2.2 Biogenic calcification

Based on the observed decline in carbonate saturation state at HOT and possibly at BATS during the last decade, current relationships between calcification and saturation state suggest that a decline in the production ofparticulate inorganic carbon could already be in effect in oceanic regions near these locations. However, continuous measurements ofsurface water production and flux ofparticulate inorganic carbon have not been taken at HOT or BATS. Thus, we do not know how these parameters have changed during this period oftime.

A recent compilation ofcoral reefstudies conducted in the Caribbean for several decades found that the average hard coral cover has declined by 80% in the last three decades (Gardner et aI., 2003). Although regional variability was significant, some degree ofconsistency and synchrony on a regional scale was detected. The authors concluded that local factors, both natural (e.g. disease, storms, temperature stress,

98 predation) and anthropogenic (e.g. over-fishing, sedimentation, eutrophication, habitat destruction) were responsible. No convincing evidence was found indicating that the decline could be attributed to global warming or increasing atmospheric CO2 (Gardner et ai., 2003). However, although local factors could be the cause, coral colonies would become more susceptible to stress ifthe carbonate saturation state declined during the same period oftime (Smith and Buddemeier, 1992; Buddemeier and Smith, 1999;

Kleypas et al., 1999). As opposed to a predicted decrease in biogenic calcification during the last century (Kleypas et ai., 1999; Andersson et ai., 2003), growth banding in Porites colonies along the Great Barrier Reefindicated an increase in calcification of approximately 4% between 1880 and 1989 (Lough and Barnes, 1997; 2000; Lough et ai.,

1999). Lough and Barnes (2000) suggested that this increase matched an increase of approximately 0.25°C in the average annual sea surface temperature during the same time period and concluded that calcification rates along the Great Barrier Reefhave significantly increased in response to global climate change.

There are currently no data available that confirm decreased calcification or deterioration ofcalcareous organisms in the natural environment owing to decreasing carbonate saturation state. However, the matter is complicated because many factors play an important role such as nutrients, light, temperature, etc. In addition, the carbonate saturation state is at present still relatively high, and negative effects might not be apparent until the saturation state has decreased further, and especially, ifthe relationship between calcification and carbonate saturation state is curvilinear (Smith and Roth, 1979;

Gattuso et ai., 1998). The results ofthe standard simulation, adopting a curvilinear

99 relationship, showed an increase in biogenic carbonate production until approximately year 2050 (Fig 3.2). At this time, the carbonate saturation state became too low and biogenic calcification started to decline.

Ifcurrent experimental relationships between calcification and carbonate saturation state are correct and no other parameters have stimulated biogenic calcification or will do so in the future, a decrease in this process must already have taken place or will take place as the carbonate saturation state decreases further. Because the process of calcification causes a release ofCO2 to the atmosphere, a decrease in biogenic calcification would consequently act as a negative feedback to increasing atmospheric

CO2• However, the magnitude and effect would be relatively small compared to annual anthropogenic CO2 emissions (Riebesell et aI., 2000; Zondervan et aI., 2001).

4.2.2.3 Carbonate mineral dissolution

In agreement with the results ofthe standard model, observations from natural carbonate sediments indicate that metastable carbonate minerals such as high magnesian calcite already undergo dissolution (Chave, 1962; Schmalz and Chave, 1963; Neumann,

1965; Mackenzie et aI., 1980; Wollast et aI., 1980; Moulin et aI., 1985; Tribble and

Mackenzie, 1995; Halley and Yates, 2000; Leclercq et aI., 2002). Analysis ofcarbonate mineral distribution among various grain sizes ofcarbonate sediments shows a decrease in the ratio ofhigh to low magnesian calcite from coarse to fine grain sizes (Chave, 1962;

Neumann, 1965). Thus, mineral reactivity appears to increase with decreasing grain sizes.

It is believed that this pattern reflects selective removal ofunstable, fine grained carbonate phases, which follows a sequence ofcarbonate mineral stability (Chave, 1962;

100 Neumann, 1965; Sclnnalz and Chave, 1963). Consequently, metastable carbonate phases in sediments can be progressively removed by solution until the most stable phase remains (Sclnnalz and Chave, 1963). Recent observation, experimental studies, and theoretical considerations confirm that dissolution ofmetastable carbonate minerals is ongoing within shallow-water sediments (Ben-Yaakov, 1973; Mackenzie et aI., 1980;

Moulin et aI., 1985; Halley and Yates, 2000; Leclercq et aI., 2002). For example, Moulin et aJ. (1985) observed carbonate dissolution driven bymicrobial oxidation oforganic matter within the sediments ofGulfofCalvi, Corsica; Halley and Yates (2000) observed dissolution, preferentially ofhigh magnesian calcite within the sediments ofa coral reef environment in Molokai, Hawaii; and Leclercq et al. (2002) observed dissolution within the sediments ofexperimental mesocosms comprised ofcarbonate sand, rock, coralline algae, and scleractinian corals.

4.2.2.4 Summary

Current ideas concerning the carbonate satnration ofthe ocean, biogenic calcification and carbonate dissolution agree to some extent with the results ofthe standard model simulation ofSOCM. The biggest unknown is biogenic calcification whereas observations ofcarbonate saturation state and carbonate dissolution agree relatively well with model predictions.

101 4.3 SENSITIVITY ANALYSIS

4.3.1 Background

The objectives ofmodeling global biogeochemical cycles and processes, among others, are to describe and predict how these cycles interact, function and behave in the natural environment. In order to accomplish these objectives, mathematical models are utilized to approximate the immense complexity ofnatural processes and systems. To confirm the adequacy ofa model, comparison ofmodel output to actual data is necessary.

The ability ofa biogeochemical model to describe adequately the complexity ofnatural systems depends on two things, the model structure and the values ofthe initial parameters. Making the assumption that the structure ofthe current standard model adequately describes the fundamental processes within the shallow-water ocean environment, the relative importance ofparameters used in the model can be evaluated through sensitivity analysis (Hamby, 1994). In general, sensitivity analyses are used to assess the relationship between variations in input parameters and variations in output parameters. More specifically, sensitivity analyses may be conducted in order to determine the following: which parameters require additional research; which parameters are insignificant and can be eliminated; which parameters have the strongest influence on the output; which parameters are most strongly correlated with the output; and what consequences result from altering a certain input parameter (Hamby, 1994). There are several different methods ofsensitivity analyses, e.g. differential sensitivity analysis, one­ at-a-time sensitivity analysis, factorial sensitivity analysis, and Monte Carlo sensitivity analysis. The simplest method, and the one utilized in the current work, is the one-at-a-

102 time sensitivity analysis (Gardner et aI., 1980; Crick et aI., 1987; Hamby, 1994; Dubus and Brown, 2002). Conceptually, this method is based on repetitious alterations ofone parameter while holding the other parameters fixed. The main objective ofthe current sensitivity analysis was to investigate the strength ofthe major conclusion ofthe present study; that is, whether dissolution ofmetastable carbonate minerals has the potential to buffer the surface waters ofthe shallow-water ocean environment against rising atmospheric CO2.

4.3.2 Sensitivity procedure

The one-at-a-time sensitivity method (Hamby, 1994) was utilized to assess the relative importance ofvarious parameters adopted in the standard model, and the effect of these parameters on the carbonate geochemistry ofthe surface water and the sediment pore water. One parameter was altered at a time while holding all others equal to the values ofthe standard scenario. The sensitivity ofeach parameter was assessed numerically by utilizing the maximum absolute ratio ofvariation index (MAROV; Dubus and Brown, 2002):

MAROV = Max (0 - °BC) X I BC (4.1) (I - I BC ) 0BC where 0 is the output value, OBe is the output value for the standard run or what is referred to as the base-case scenario, I is the input value, and IBe is the original input value for the standard run. The larger the resulting MAROV index for a parameter, the larger the potential influence ofthat parameter on model output. IfMAROV=I, changes in the input parameter by x% will at most result in the same variation in the output

103 parameter. It is important to regard the results ofthe MAROY analysis with caution due to the distinction between 'important' and 'sensitive' parameters. 'Important' parameters are those parameters whose uncertainty contribute substantially to the uncertainty in assessment results, and 'sensitive' parameters as those that have a significant influence on assessment results (Crick et aI., 1987; Hamby, 1994). A 'sensitive' parameter has a large influence, but is not necessarily important because it is well constrained and does not add substantial variability to the output whereas an 'important' parameter is always sensitive because its uncertainty will not appear in the output unless the model is sensitive to the input (Hamby, 1994).

In the current study, sensitivity analyses were conducted on the following input parameters to investigate the influence on surface water carbonate saturation state and sediment carbonate dissolution:

I) organic matter deposition in sediments from river input and surface water

primary production and subsequent remineralization

2) carbonate dissolution kinetics

a) reaction order

b) rate constant

c) inhibition factor

3) initial model parameters

a) initial sediment carbonate mass

b) initial pore water dissolved inorganic carbon composition and

carbonate saturation state

104 c) average magnesian calcite composition

d) magnesian calcite solubility (stoichiometric solubility product)

4) fossil fuel burning and CO2 emission

5) shallow-water ocean - open ocean exchange

Ranges ofparameters were based on ranges ofestimates in the literature, but also on hypothetical scenarios in order to investigate whether or not surface waters could be buffered by dissolution ofmetastable carbonate minerals. Detailed discussion ofeach parameter, including ranges ofalteration, results and consequences are discussed in the subsequent sections.

4.3.3 Sensitivity analysis results

In summation, the results ofthe sensitivity analysis confirmed the main conclusion ofthe standard simulation that dissolution ofmetastable carbonate minerals will not buffer the surface water against rising atmospheric CO2• The relative importance and influence ofeach parameter on surface water carbonate saturation state and carbonate dissolution according to the maximum absolute ratio ofvariation (MAROV; Dubus and

Brown, 2002) are shown in Table 4.1. The results should be considered with caution due to the difference between 'sensitive' and 'important' parameters (see previous discussion). Table 4.2 illustrates parameters that are considered 'sensitive' with respect to their influence on the surface water buffer capacity. With one exception, none ofthe changes in parameter values classified as 'sensitive' (Table 4.2) resulted in any significant buffering ofthe surface water, despite the fact that substantial changes were made in values. The one exception was in alterations made to the value ofthe

105 TABLE 4.1 RANKING OF PARAMETERS ACCORDING TO Tl-IEIR INFLUENCE ON SURFACE WATER ARAGONITE SATURAllON STATE AND NET CARBONTE DISSOLUllON BASED ON Tl-IE MAXIMUM ABSOLUTE RAllO OF VARlAllON (MAROV)

Surface water aragonite Net carbonate Parameter saturation state dissolution MAROV Rank MAROV Rank

Ion acti\tty product 1.487 55.282

Atmoo pheric CO2 0.567 2 0.043 10

Organic matter remineralization 0.391 3 14.296 2

Coastal ocean-Qpen ocean exchange 0.168 4 0.065 9

Initial DIG chemistry 0.061 5 2.431 3

Surface water organic matter flux 0.042 6 1.552 4

Riwr input flux of organic matter 0.015 7 0.571 6

Magnesian calcite comJX)Sition 0.014 8 0.760 5

Reaction order 0.004 9 0.225 7

Inhibition factor 0.002 10 0.002 11

Initial carbonate mass 0.001 11 0.111 8

Rate constant 0.000 12 0.000 12

TABLE 4.2 PARAMETER INFLUENCE (SENSITIVITY) ON SURFACE WATER BUFFER EFFECT

Process/Property Sensitiloe Not sensiti\e

Atmaspheric CO2 X

Coastal ocean-Qpen ocean exchange X

Initial carbonate mass X

Initial DIG chemistry X

Ion acti",ty product X

Magnesian calcite composition X

Organic matter remineralization X

Reaction kinetics

Reaction order X

Rate constant X

Inhibition factor X

River input flux of organic matter X

Surface water organic matter flux X

106 stoichiometric solubility product ofmagnesian calcite, the most 'sensitive' parameter

(Table 4.1), which caused instability and a subsequent collapse ofthe model. However, based on observations from the natural environment (Wollast and Garrels, 1980; Morse and Mackenzie, 1990) and recent experimental work (Bischoff et aI., 1987; Busenberg and Plummer, 1989; Bertram et aI., 1991; Bischoffet aI., 1993; Tribble et aI., 1995), the solubilities ofmagnesian calcites have been relatively well constrained compared to the degree ofalteration ofthis parameter in the sensitivity scenario. The changes in the value ofthe magnesian calcite stoichiometric solubility product do not add any significant variability to the model output and do not refute the main conclusion (see subsequent discussion).

4.3.3.1 Organic matter deposition and remineralization

Carbonate dissolution was observed to increase significantly as a result of increasing deposition oforganic matter delivered to the pore water-sediment system or due to increasing rates ofremineralization ofthe organic matter already present within this reservoir. Remineralization oforganic matter produces CO2, causing a decrease in pH, carbonate ion concentration and subsequently carbonate saturation state whereas the total alkalinity remains relatively unchanged (Morse and Mackenzie, 1990; Zeebe and

Wolf-Gladrow, 2002). In SOCM, the rate ofremineralization oforganic matter was assumed to be related to a first order dependence on the total amount oforganic matter present within the pore water-sediment system. Consequently as the flux oforganic matter to the sediments increased, remineralization and production ofC02 also increased.

107 In order to investigate the robustness ofthe standard model and the importance of organic matter deposition within the shallow-water ocean environment, the flux of organic carbon to the pore water-sediment system was altered by increasing the amount deposited via rivers by as much as one order ofmagnitude and in a second scenario by increasing the flux from the water column by a factor oftwo. Consequently, the production ofCO2 from remineralization oforganic matter within the pore water­ sediment system significantly increased, resulting in corrosive conditions with respect to carbonate minerals and inducing substantial dissolution in both scenarios (Fig 4.5). By the end ofeach simulation, the cumulative amount ofcalcium carbonate that had dissolved upon pore water undersaturation corresponded to approximately 66% and 30% ofthe initial amount ofmagnesian calcite present within the shallow-water ocean environment in the two scenarios, respectively. A small buffer effect was observed in the surface water (Fig 4.5), although it made little difference in preventing any negative effects on calcareous organisms from rising atmospheric CO2. The maximum absolute ratio ofvariation (MAROV; Dubus and Brown, 2002) indicates that remineralization of organic matter can have a strong influence on the surface water carbonate saturation state and carbonate dissolution relative to other parameters tested, but the buffer effect ofthe surface water is not significant (Table 4.1). In fact, disregarding the stoichiometric solubility product (see subsequent discussion), remineralization oforganic matter was the most influential driver ofcarbonate dissolution (see Morse and Mackenzie, 1990). The relative importance ofthe surface water organic matter flux was greater than the river

108 1750180018501900195020002050 175018001850190019502000 2050 Year Year Figure 4.5. Sensitivity analysis: sediment organic matter deposition via rivers (A; flux ofC in standard run xI-diamonds; x2-squares; xlO-triangles) and organic matter flux from the water column (B; flux ofC in standard run x1­ diamonds; x2-squares). Black symbols in upper plots (A-I, B-1) indicate pore water saturation state with respect to 15 mol% magnesian calcite and open symbols indicate cumulative amount ofcarbonate dissolved as carbon upon undersaturation. Black symbols in lower plots (A-2, B-2) indicate surface water aragonite saturation state, and open symbols indicate surface water pH.

109 flux because ofthe initial higher value ofthe flux from the surface water (32xlQ12 mol C yr-l and 8xlOl2 mol C yr-l, respectively).

In agreement with the current observation that dissolution ofcarbonate minerals is strongly driven by remineralization oforganic matter, Moulin et al. (1985) demonstrated dissolution ofcarbonate minerals within the pore water ofsediments in the GulfofCalvi,

Corsica, driven by microbial mediated oxidation oforganic matter. Ifcarbonate dissolution were initiated due to microbial processes under aerobic conditions, one would expect a similar increase in total alkalinity and TC02upon dissolution. On the other hand, ifdissolution were simply associated with carbonate undersaturation, the increase in total alkalinity would be twice that ofthe increase in TC02 (Morse and Mackenzie,

1990). Moulin et al. (1985) observed approximately the same increase in total alkalinity and TC02, attributing the observed carbonate dissolution to microbial oxidation of organic matter. As would be anticipated, a similar relationship was observed in the current standard model (Fig 4.6). According to Morse and Mackenzie (1990), the extent ofcarbonate dissolution is mainly driven by the remineralization oforganic matter rather than carbonate reaction kinetics.

In the two sensitivity scenarios, results ofnet ecosystem calcification (carbonate precipitation - carbonate dissolution) indicated that net dissolution was dominant at the end ofboth simulations (Fig 4.7; 4.8). Consequently the mass ofcalcium carbonate was diminishing in the shallow-water ocean environment whereas in the standard scenario, accumulation ofcalcium carbonate was significantly greater than calcium carbonate removal throughout the simulation. However, biogenic carbonate production was

110 1.0 --Curve fit of model results - Curve fit of data from Moulin et al. (1985)

--~ a ~ 0.5 slope = 0.96 -«­

slope = 1.04

0.0 0.0 0.5 1.0 ~LC02 (meq L'l) Figure 4.6. Changes in total alkalinity (meq L-1) as a function ofchanges in total dissolved inorganic carbon (meq L-I) upon dissolution ofcalcium carbonate in Gulfof Calvi sediments (Moulin et aI., 1985) and as calculated in the standard model simulation. The one to one relationship indicates that dissolution is driven by microbial remineralization oforganic matter (see text for explanation). The results of Moulin et aI. (1985) do not pass through the origin. The authors concluded that a small amount ofCO2( aq) originating from bacterial respiration was added to the pore water prior to dissolution ofany carbonate minerals. impaired more in the standard run than in the sensitivity scenarios since a small buffering effect ofthe surface water was observed in the sensitivity runs. Thus, in discussions regarding future effects on coral reefs, carbonate reefs and calcareous organisms, it is important to distinguish between the effects on biogenic calcification, reef accretion and carbonate accumulation.

It is important to note that a doubling ofthe flux oforganic matter from the surface water to the sediments or an increase in the river deposition oforganic matter by one order ofmagnitude throughout the entire simulation are substantial increases relative to the standard scenario. The extreme fluxes ofthe sensitivity scenarios are greater than

111 50 A • CaC0 production 3 • CaC0 dissolution 40 3

() 30 "'5 E N ~ ...... Q...... -0 20 ...... -. linear sat stale- - - 0.: dependence 10

0 B 20

10

() Calcium carbonate accumulation "'5 0 E Calclum carbonate N ~ removal .. 0 ...... - -10 ...... -20 ...... -30 +-r"T""T""1...,.-,-"T""T""1...,.-,-.,.....,,....,.-r'T""T""",....,.-r'T""T"""..,..,..'T""T"""..,..,....,...,....T""T"""I'...... 1700 1750 1800 1850 1900 1950 2000 2050 2100 Year

Figure 4.7. Sensitivity analysis: sediment organic matter flux from river input (flux ofC in standard run x I-open circles; x I O-no symbols). (A) Blue lines indicate annual calcium carbonate dissolution in each scenario. Red lines represent annual calcium carbonate production: solid lines, adopting a curvilinear relationship between biogenic calcification and carbonate saturation state, and dashed lines, using a linear relationship. In the standard run, carbonate production was greater than dissolution throughout the simulation. The difference observed in biogenic calcification between the standard run and the sensitivity scenario is due to the greater buffer effect and subsequently less inhibition ofthe process ofcalcification observed in the latter. (B) Annual net calcium carbonate accumulation (precipitation-dissolution) within the shallow-water ocean environment for the scenarios described in (A). Negative values indicate net removal.

112 A curvilinear sat state dependence 25i-o-oO--...-o-~~~~:::;__~.. =o:====<~c... ---0 ...... 20 linear saL slale dependence

(5 • CaCO3 production E 15 • CaC0 dissolution 3 .....o 10

5

0 B 20 - --- 15 u 10 (5 -...... E 5 .. .. N ...... Calcium carbonate accumulation .. 0 ..... 0 -- - . ------... -- --. ------.. - - - - - '"-~ -- - Calcium carbonate removal ...... -5 ...... -10 '"..

-1 5 -+-r"""""'I""T"'T""T""T'"T""T""'1"""T""T'""""""'r-T""T""T"T'"T""T""'1"""T""T""T""T"""r-T""T""T""T'",...,..,-r-r"",,-,+ 1700 1750 1800 1850 1900 1950 2000 2050 2100 Year

Figure 4.8. Sensitivity analysis: sediment organic matter flux from the water column (flux ofC in standard run xl-open circles; x2-no symbols). (A) Blue lines indicate annual calcium carbonate dissolution in each scenario. Red lines represent annual calcium carbonate production: solid lines, adopting a curvilinear relationship between biogenic calcification and carbonate saturation state, and dashed lines, using a linear relationship. (B) Annual net calcium carbonate accumulation (precipitation-dissolution) within the shallow-water ocean environment for the scenarios described in (A). Negative values indicate net removal.

113 the maximum range ofestimates in the literature. As an example, the initial amount of particulate organic carbon deposited to the sediments from rivers in the standard run was

l2 equivalent to 8xlO mol C yr-l (Meybeck, 1982; Smith and Hollibaugh, 1993) compared to 80 mol C yr-l in the most extreme sensitivity scenario. The resulting total organic carbon load transported via rivers at the onset ofthe simulation then became 106xl012

12 mol C yr-l compared to 34xlO mol C yr-l in the standard scenario. Estimates in the literature offluvial loading of organic matter to the oceans range from IOXl012 to

l2 l2 83xlO mol C yr-l with an average around 39xlO mol C yr-l (Smith and Hollibaugh,

1993). Despite the substantial dissolution ofcalcium carbonate minerals induced by the substantial increases in organic matter deposition and remineralization, and the fact that no significant buffer effect was observed in the surface water in the current analysis, it can be concluded that the model results are robust and dissolution ofmetastable carbonate minerals will not buffer surface waters against rising atmospheric CO2• In addition, the results indicate an important coupling between the organic and inorganic carbon cycles and show how changes in the fluvial transport oforganic matter to the coastal ocean or changes in new production can have a significant impact on the carbonate geochemistry within the pore water~sediment system (Chave, 1967; Smith,

1972; Morse and Mackenzie, 1990; Milliman, 1993).

The amount oforganic matter deposited in the sediments could be altered due to changes in primary production within the water column and the benthos, or changes in the amount oforganic matter delivered to the coastal zone via fluvial transport and runoff. Since the onset ofthe industrial revolution there has been a significant increase in

114 the transport oforganic matter from land to the coastal zone (Likens et al., 1981; Berner,

1982; 1989; Smith and Hollibaugh, 1993; Ver et aI., 1999; Rabouille et al., 200I). In addition, the global transport ofN and P in rivers has increased substantially due to anthropogenic activities, such as the use ofagricultural fertilizers, phosphorus detergents, sewage discharge, and human-induced increases in erosion rates (Vitousek, 1994; Jordan and Weller, 1996; Schlesinger, 1997; Rabouille et aI., 2001; Mackenzie, 2002). Increased loading ofnutrients to the coastal zone can lead to eutrophic conditions that stimulate increased production oforganic carbon within this region and subsequently increase the flux oforganic matter to the sediment system. As the gross primary production increases within the coastal zone, observations indicate that net metabolism decreases i.e. the increase in organic matter being oxidized is greater than the increase in organic matter being produced (Smith and Hollibaugh, 1993). In other words the net heterotrophy ofthe system increases. Predictions for the future suggest that the net heterotrophy ofthe coastal region will continue to increase (Ver et al., 1999; Mackenzie, 2002).

An interesting aspect related to the trophic status ofthe shallow-water ocean environment and the increasing concentrations ofatmospheric CO2 is whether the former can be considered a net sink or a source ofC02 to the atmosphere. Prior to human influences, the global ocean and in particular the coastal ocean probably were net sources ofCO2 to the atmosphere because ofremineralization oforganic matter and production and accumulation ofcalcium carbonate in the sediments (Smith and Mackenzie, 1987;

Wollast, 1994; Wollast and Mackenzie, 1989; Smith and Hollibaugh, 1993).

Anthropogenic activities, such as burning offossil fuels and the subsequent increase in

115 atmospheric CO2, have caused a reversal ofthis flux. On average the open ocean is now recognized as a major sink ofCO2 (Smith and Mackenzie, 1987; Mackenzie et al., 2002).

Whether or not the global coastal region at present serves as a net source or a sink of atmospheric CO2 is debatable and the net flux is different between different systems such as estuaries, coral reefs and continental shelves (Frankignoulle et aI., 1996; 1998; Gattuso et aI., 1996; 1997; 1999b; Boehme et al., 1998; Frankignoulle and Borges, 2001).

Although the concentration ofCO2 in the atmosphere has increased, the net heterotrophy with respect to the organic carbon balance within the coastal zone has also increased. In addition, biogenic carbonate production could have decreased owing to decreasing carbonate saturation state and subsequently the flux ofCO2 to the atmosphere originating from this process (Riebesell et al., 2000; Zondervan et al., 2001). The results ofthe standard simulation ofSOCMindicated that the shallow-water ocean environment currently may serve as a net source ofCO2 to the atmosphere and remain such until the year 2100. However, it is important to note that this net flux is highly dependent on the magnitude and direction ofthe initial estimate, which is not well constrained. In fact, depending on the initial estimate ofCO2 evasion owing to the balance between organic production and respiration, and the magnitude ofbiogenic production and accumulation ofcalcium carbonate, the net CO2 flux could be significantly different and the shallow­ water ocean could currently serve as a net sink (Fig 4.9). However, no matter ifthe shallow-water ocean currently serves as a net source or a net sink ofCO2 to the atmosphere, as the C02 concentration ofthe atmosphere increases, the flux ofCO2 into the ocean will increase.

116 10 A

autotrophIC 0-+------+ heterotrophic ~, ------.. ..,"', ..,-----.... --- ..... -10 ------, 1,\.' " ~ . " ","~. 'r---- -.,.. ... -20 , -',,- >- , o , E -30 N... o T'"" 10 B , '-" , x , , :;] , ;;:: ,~-- net sink , oN O-i------,~----+- () net source ,, ,.. ' -10 ...... ,." .------

-20 L- ----...... ",.,ttItI" , ," '---.-... , , ,".. .-.,~, -30 ------,,-. 1700 1750 1800 1850 1900 1950 2000 2050 2100 Year

Figure 4.9. (A) Organic carbon balance and (B) CO2 gas exchange between the atmosphere and the surface water within the shallow-water ocean environment. Solid lines indicate the range ofCO2 flux in the standard run adopting either a linear relationship or a curvilinear relationship between the rate ofcalcification and the carbonate saturation state ofthe surface water. Dotted lines in (A) represent ±50% ofthe initial estimate ofthe trophic state ofthe shallow-water ocean environment (Smith and Hollibaugh, 1993). In (B) dotted lines represent the uncertainty in the net C02 flux associated with the initial estimate ofthe flux due to calcium carbonate precipitation (8.4x1012 mol C to 18.5xlQ12 mol C) and the organic carbon balance.

117 4.3.3.2 Carbonate reaction kinetics

Sensitivity analysis with respect to carbonate reaction kinetics indicated that alterations ofthe kinetic parameters adopted in the present study such as reaction inhibition factor, rate constant and reaction order did not change the major conclusions.

The maximum absolute ratio ofvariation (MAROV; Dubus and Brown, 2002) indicated that the influence ofkinetic parameters on surface water carbonate saturation state and carbonate dissolution were minimal (Table 4.1; 4.2). This can be explained on the basis ofthe previously stated conclusion that the extent ofcarbonate dissolution is mainly governed by remineralization oforganic matter rather than reaction kinetics (Morse and

Mackenzie, 1990). In SOCM, carbonate dissolution was a function ofsaturation state

(degree ofundersaturation). In addition, the magnitude ofdissolution was dependent on the adopted estimates ofthe constant parameters k (rate constant), n (reaction order) and i

(inhibition factor). All these parameters have significant uncertainty associated with them

(see Chapter 2 for a detailed discussion ofthese parameters and the carbonate dissolution and precipitation equations). Although alterations ofthe kinetic parameters did not produce sufficient alkalinity to buffer the surface water against increasing atmospheric

CO2, the results had important implications with respect to the carbonate geochemistry of the pore water-sediment system.

a) Reaction order

Sensitivity analyses with respect to carbonate dissolution reaction kinetics indicated that reaction order was the most 'sensitive' variable and had the greatest influence on the extent ofcarbonate dissolution (Table 4.1). The reaction order of

118 dissolution kinetics ofcarbonate minerals is not well constrained and ranges for calcite are between I and 5 (Hales and Emerson, 1997; Keir, 1980; Jansen et a!., 2002). Even a small change in this parameter will lead to a large change in the amount ofcalcium carbonate dissolved upon undersaturation.

In the current analysis, the reaction order was varied between 1.5 and 5 (Fig 4.10).

A reaction order of5 resulted in a very low reaction rate, and no buffer effect was observed in the pore water upon undersaturation. A reaction order of 1.5 caused rapid dissolution that instantaneously buffered the pore water reservoir against further changes owing to remineralization oforganic matter. The rapid dissolution maintained the pore water close to saturation and in metastable equilibrium with the dissolving phase. In a hypothetical scenario with initially low carbonate mineral masses, the equilibrium persisted until the entire reservoir ofthe carbonate phase had dissolved and was followed by a decrease in pore water carbonate ion concentration until the next carbonate phase started to dissolve and a new metastable equilibrium state was established (Fig 4.11). In reality dissolution ofvarious phase compositions is probably a continuous process rather than the stepwise changes illustrated in Figure 4.11. In agreement with these results,

Schmalz and Chave (1963) suggested that the dissolved carbonate activity ofseawater is controlled by a metastable equilibrium between seawater and the most soluble solid carbonate phase present in the sediments.

119 1.10-l-_--l.__...1...._---JL....-_-L._-+ 40 A 35 1.05 30 25 20 15 0.95 10 5 ~ 0 0 .$ E ~ N 35 ~o c: ~ o ~ ~ 1.05 30 al .a 25 ~ '" '" .~'" 1.00 ------_.-.,. 20 '5'" c: ~ 15 Ol £ :E 0.95 10 <> '"ell "$. 5 > o E 0 :;"'" '"E '"~ C 35 ()" 1.05 30 25 20 15 0.95 10 5 0.90~~~~~~~~~~:t 0 2050 2060 2070 2080 2090 2100 Year Figure 4.10. Sensitivity analysis: reaction kinetics. Black symbols indicate pore water saturation state with respect to 15 mol% magnesian calcite, and open symbols indicate cumulative amount ofcarbonate dissolved as carbon upon undersaturation. (A) Reaction order (n=1.5-squares; n=3.34-diamonds; n=5-triangles), (B) Reaction rate constant (kx I-diamonds; kx lO-squares; kx100-triangles; kx1000-circles), (C) Reaction inhibition factor (no inhibition -squares; 0.018-diamonds; 0.00014­ triangles).

120 Dissolution of 15 mol% Mg-calcite •! Q) to -iii ~ <: N.. 0 M :;:; 0 15 mo/% Mg-calcite at equilibrium ~ Q. t :::l Ol J: to <: 0. -til 'iii Q) to to ~ -<: 0 Aragonite at equilibrium -e0 -=-<: ------'Jj.------to ~ () / Dissolution of aragonite

Time_

Figure 4.11. Pore water carbonate saturation state as a function oftime. The figure illustrates a simulation ofhow the carbonate ion concentration is controlled by a metastable equilibrium between the pore water and the most soluble solid carbonate phase in the sediments. Initially, a decrease in carbonate saturation state is observed until dissolution of 15 mol% magnesian calcite is initiated. At this stage, the pore water is buffered and maintained close to equilibrium with this phase. The equilibrium persists until the phase has completely dissolved. The carbonate saturation state decreases until a new metastable equilibrium state is established with the next phase (in this case aragonite). In the current scenario, the consumption ofcarbonate ions was driven by remineralization oforganic matter. 2 2 The calculation assumes Ca + and Mg + concentrations remain constant in the pore water.

121 b) Reaction rate constant

Results ofnumerical simulations varying the rate constant ofthe carbonate dissolution equation showed that this parameter had a rather minor effect on carbonate dissolution. Although the rate constant was increased by as much as three orders of magnitude, the cumulative amount ofcalcium carbonate dissolved only doubled (Fig

4.10). As expected a small increase in the buffer effect ofthe pore water was observed with a higher reaction rate constant, but the effect was insignificant relative to the substantial increase in the reaction rate required to produce this buffer effect. The maximum absolute ratio ofvariation (MARGV; Dubus and Brown, 2002) indicated that the rate constant had the lowest relative influence on carbonate dissolution compared to the other parameters tested. In support ofthis numerical result, based on a numerical experiment simulating dissolution ofcarbonate particles sinking through the water column, Jansen et al. (2002) concluded that variations in the rate constant had very little effect on carbonate dissolution rate.

c) Reaction inhibition factor

Inhibition ofcarbonate reactivity, i.e. dissolution and precipitation reactions, by various dissolved chemical species is subject to substantial uncertainty and debate. In general it is known that carbonate dissolution and precipitation are inhibited by a number ofcompounds such as magnesium ions, dissolved organic matter, dissolved phosphate, and certain trace metals (Pytkowicz, 1965; Chave and Suess, 1967; Berner, 1975; Berner

et al., 1978; Morse, 1983; Mucci, 1986; Burton and Walter, 1990; Morse and Mackenzie,

1990; Dove and Rochella, 1993; Tribble and Mackenzie, 1995; Lebron and Suarez, 1996,

122 1998; Zhang and Dawe, 2000). Mechanistically and quantitatively very little is known about how these compounds inhibit the progression ofcarbonate dissolution or precipitation. The concentration ofthese compounds in the natural environment is variable, and even the composition ofdissolved organic matter is very heterogeneous, being composed ofmany different organic molecules that all exert a different influence on carbonate reactivity (Berner et aI., 1978). In the current analyses, the dissolution

4 inhibition factor was varied from no inhibition to an inhibition of lAx10- • This value was derived from quantitative estimates ofinhibition by dissolved phosphate and dissolved organic matter from studies ofsediments in Long Island Sound (See Chapter 2;

Berner et aI., 1978). An inhibition ofthis magnitude resulted in insignificant amounts of calcium carbonate dissolving, and virtually no observable buffer effect within the pore water upon undersaturation (Fig 4.10). In the standard scenario and in the no-inhibition scenario, a significant buffer effect ofthe pore water was observed. Considering the orders ofmagnitude difference between the high inhibition and no-inhibition scenarios and the moderate response relative to this difference, it can be concluded that the inhibition factor has a relatively small influence on carbonate dissolution compared to other parameters such as reaction order. Similarly, the maximum absolute ratio of variation (MAROV; Dubus and Brown, 2002) indicated that the inhibition factor had very little influence on carbonate dissolution, although it had a greater influence than the rate constant. (Table 4.1).

123 4.3.3.3 Initial model parameters

a) Initial sediment carbonate mass

The initial carbonate sediment mass present within the shallow-water ocean environment was calculated based on estimates and assumptions that are only moderately well constrained. In the current analysis, the initial carbonate mass was altered from being one order ofmagnitude smaller than the initial carbonate mass ofthe standard model to being one order ofmagnitude larger. The maximum absolute ratio ofvariation

(MAROV; Dubus and Brown, 2002) indicated that the initial carbonate mass had a minor relative influence on the surface water carbonate saturation state and carbonate dissolution. Alterations to the initial carbonate mass did not have any significant impact on the major conclusions ofthe present study or any other aspects ofthe model.

b) Initial pore water dissolved inorganic carbon composition and carbonate

saturation state

The initial dissolved inorganic carbon chemistry ofthe pore water was based on an average composition derived from data from the Bahamas and elsewhere (see Chapter

2; Morse et aI., 1985; Morse and Mackenzie, 1990). It is important to note that this pore water composition represents an average ideal setting and could be significantly different in other locations since substantial heterogeneity is commonly observed in pore water chemistry both temporally and spatially. In the current sensitivity analysis, the initial dissolved inorganic carbon composition ofthe pore water was adjusted so that the saturation state with respect to 15 mol% magnesian calcite was 1.20 (standard scenario),

1.10 or 1.00 (at saturation), respectively. Because undersaturated conditions were reached

124 $ ~ 1.15 <:: .2 "§ 200 ] .a 1.10i--- ...... ~ ~ $ 150 ·0 1 m1.05 <:: 't ~ 100 ~ ::2'" lF''*'~::..:.I~~~.:~= ~ i 1.00r...... 50 - />O~ ~ '" ::> () ~ 0.95 ~~~~~~~~~8q~~~ 0 1700 1750 1800 1850 1900 1950 2000 2050 2100 Year Figure 4.12. Sensitivity analysis: initial dissolved inorganic carbon speciation/carbonate saturation state (Te02 = 2.0 mmol kg-\ initial saturation state: Oarag=1.20-diamonds; O....g= 1.1 O-squares; O....g=1.0-triangles). Black symbols indicate pore water saturation state with respect to 15 mo1% magnesian calcite, and open symbols indicate cmnulative amount carbon dissolved upon undersaturation. at an earlier stage in these sensitivity scenarios, the initial lower carbonate saturation state led to more calcimn carbonate being dissolved (Fig 4.12). The pore water was observed to be buffered upon undersaturation in each scenario, but no significant effect was observed in the surface water. Maximum absolute ratio ofvariation (MAROV; Dubus and Brown, 2002) indicated that the relative influence ofpore water initial carbonate saturation state on surface water carbonate saturation and carbonate dissolution was high relative to other parameters tested (Table 4.1). Although classified as a 'sensitive' parameter, alterations to the initial dissolved inorganic carbon composition or the carbonate saturation state did not affect the major conclusions obtained from the standard model run.

125 c) Average magnesian calcite composition

Naturally occurring magnesian calcites have a magnesium content ranging from

-0 to 30 mol% with an average ofabout 12-15 mol% (Garrels and Wollast, 1978; Morse and Mackenzie, 1990). In SOCM, the magnesian calcite reservoir was assumed to have an average content of 15 mol% magnesium in the standard scenario. In the standard run, the average pore water became undersaturated with respect to 15 mol% magnesian calcite in the second halfofthe 21 st century, initiating dissolution ofthis carbonate phase. Prior to this, the average pore water was undersaturated with respect to carbonate phases with a magnesium content greater than 15 mol% and consequently already subject to dissolution. Figure 4.13 illustrates how the pore water saturation state varied with respect to a range ofmagnesian calcite minerals ofdifferent magnesium composition, ranging from 10 mo1% to 20 mol%, throughout the simulation and the resulting cumulative amount ofcalcium carbonate dissolved upon undersaturation by the year 2100. The solubility product ofeach magnesian calcite phase was adopted from the intermediate solubility curve ofBischoffet al. (1993). Throughout the simulation the pore water saturation state was supersaturated with respect to magnesian calcite phases of 10 mo1%

Mg and 12 mol% Mg, but became undersaturated with respect to magnesian calcite phases with 15 mol% and higher magnesium contents. At the onset ofthe simulation, the average pore water was close to saturation with respect to 20 mol% magnesian calcite and became undersaturated with respect to this phase around year 1950, initiating dissolution. Although the cumulative amount ofcalcium carbonate dissolved by the year

2100 ranged widely between scenarios, the effect on surface water carbon chemistry was

126 200 A ~ "5 E N -0 150 ~ ~ "C Q) > "5 II) II) 100 '5 c: 0 .0 ~ '"0 50 Q) ;f!. > ~ ~~~.~ ~ 1.0 .. ..-- ...------"3 E X ::> 0 () B

O03 0.003 i . § i;002 0.002 ~ ::> :x: til -+-0 ~ ~001 ---pH 0.001 o

f 0 -I-e--tl--tll-ll---ll--l~t-t'-""'''''~~ o

1700 1750 1800 1850 1900 1950 2000 2050 2100 Year

Figure 4.13. (A) Pore water saturation state (black symbols) with respect to magnesian calcite with a magnesium content of10 mol% (dianlOnds), 12 mol% (squares), 15 mol% (triangles j.), 18 mol% (circles), and 20 mol% (triangles T), and the cumulative amount carbonate as carbon dissolved upon undersaturation with respect to the phase under consideration (open symbols). (B) Observed difference in surface water aragonite saturation and pH between model scenarios using an average magnesian calcite composition of 10 mol% and 20mol%, respectively.

127 insignificant (Fig 4.13). The results ofthe maximum absolute ratio ofvariation

(MAROY; Dubus and Brown, 2002) indicated that the average magnesian calcite composition had a relatively strong influence on the extent ofcarbonate dissolution, although the surface water buffer effect was not significant. In addition, the average sediment magnesian calcite composition is constrained relatively well (12-15 mol%,

Garrels and Wollast, 1978; Morse and Mackenzie, 1990) and does not add any substantial variability to the main conclusions ofthe standard model.

The results ofthe current analysis indicated that magnesian calcite minerals with greater mol% MgC03 than the average sediment composition of 15 mol% were subject to undersaturated conditions and underwent dissolution prior to the onset ofdissolution with respect to this average phase composition. In other words, carbonate minerals with greater solubility than the carbonate phase in metastable equilibrium with the surrounding seawater were subject to dissolution. Ongoing dissolution ofmetastable carbonate minerals has been observed at various locations such as within the sediments ofa coral reefenvironment in Molokai, Hawaii (Halley and Yates, 2000) and within the sediments ofGulfofCalvi, Corsica (Moulin et al., 1985). In addition, selective dissolution of unstable carbonate minerals can be inferred from the observation that the average ratio of low to high magnesian calcite minerals, i.e. average carbonate mineral stability, increases from coarse to fine grain sizes in a wide range ofnatural environments (Chave, 1962;

Neumann, 1965). Based on the results ofthe current analysis, the carbonate composition ofthe pore water-sediment system could be expected to change as the carbonate saturation state in this reservoir decreases. Carbonate phases that are unstable will

128 dissolve and maintain a metastable equilibrium state witb the surrounding seawater until tbey have completely dissolved and a new equilibrium with respect to another phase is established (Fig 4.11; Chave, 1962; Neumaun, 1965; Schmalz and Chave, 1963; Wollast et al., 1980; Tribble and Mackenzie, 1998).

Results ofthe standard simulation suggest tbat the pore water carbonate saturation state could be expected to decrease in the future, but mainly not from increasing atmospheric CO2. Sensitivity analysis witb respect to various CO2 emission scenarios shows that the pore water essentially was decoupled from changes in the surface water owing to increasing atmospheric CO2 (see subsequent discussion). Changes in pore water carbonate saturation state were mainly controlled by changes in tbe depositional rate of organic matter and its subsequent remineralization. This rate has increased since the onset ofthe industrial revolution and probably will continue to increase into the near future

(Likens et aI., 1981; Berner, 1989; Smith and Hollibaugh, 1993; Mackenzie, 2002).

The amount ofmagnesium present in abiotic and biogenically produced magnesian calcite is controlled by several factors. Although vital effects have to be considered, it seems that the major controls are exerted by temperature (Chave, 1954;

Mackenzie et al, 1983; Morse and Mackenzie, 1990), magnesium to calcium ratio

(Stanley et al., 2002), and carbonate saturation state (Agegian, 1985; Mackenzie and

Agegian, 1989; Morse and Mackenzie, 1990). In a future scenario ofhigher sea surface temperature and lower carbonate saturation state, how will these organisms and tbe production ofhigh magnesian calcite minerals be affected? Will organisms secreting hard parts ofhigh magnesian calcite be inhibited at an earlier stage than organisms secreting

129 hard parts ofa low magnesian calcite composition, and/or will the calcareous hard parts ofthe high magnesian calcite producers change composition?

d) Magnesian calcite solubility (stoichiometric solubilityproduct)

A potential problem in predicting the influence ofhigh magnesian calcite minerals on the carbonate geochemistry ofthe shallow-water ocean environment arises from the heterogeneous character ofthese minerals. Consequently, substantial variability is observed between different solubility experiments (Plummer and Mackenzie, 1974;

Mackenzie et al., 1983; Walter and Morse, 1984; Bischoffet aI., 1987; Busenberg and

Plummer, 1989; Bischoffet al., 1993; Tribble et aI., 1995). Mackenzie et aI. (1983) attributed the observed variability between experimentally derived stoichiometric solubility products to differences in experimental procedures, materials, solutions and even the use ofdifferent models for the aqueous carbonic acid system. Even a small change in the stoichiometric solubility product will produce substantially different results in calculations ofthe equilibrium state with respect to a specific magnesian calcite phase.

Consequently, predictions and calculations ofmagnesian calcite saturation state in the current model could be significantly different due to small deviations from the adopted intennediate solubility estimates ofBischoffet aI. (1993). In the current sensitivity analysis, a range ofsolubility estimates for 15 mol% magnesian calcite minerals was adopted:

• biogenic material with high surface free energy subjected to minimal

preparation, i.e. material being compositional impure with positional

130 disorder - (high solubility; Plummer and Mackenzie, 1974; Bischoffet aI.,

1993)

• biogenic materials from which reactive materials were removed, the

samples were ground and annealed, but material still contained

compositional impurities and positional disorder - (intermediate

solubility; adopted in standard scenario; Walter and Morse, 1984; Bischoff

et aI., 1987; Busenberg and Plummer, 1989; Bischoffet aI., 1993).

• synthetic phases prepared at high temperature and pressure and relatively

free ofstructural defects, essentially no trace constituents and low

positional disorder - (low solubility; Bischoffet aI., 1987; Busenberg and

Plummer, 1989; Bischoffet aI., 1993).

It was not possible to complete the analysis ofthe high solubility scenario. In this scenario at the onset ofthe simulation, the pore water saturation state with respect to 15 mol% magnesian calcite was so low that the entire magnesian calcite reservoir dissolved in less than a month, leading to instability and a collapse ofthe model. It is not realistic to initiate the model with a pore water composition substantially undersaturated with the average solid magnesian calcite composition ofthe sediments. Presumably, the magnesian calcite composition ofthe sediments will to some extent reflect a metastable equilibrium state between the most soluble solid phase and the carbonate ion activity of the surrounding seawater (Schmalz and Chave, 1963; Neumann, 1965; Wollast et aI.,

1980; Tribble and Mackenzie, 1995). The results indicate that a 15 mol% magnesian calcite could not have been present within the pore water-sediment system at the initial

131 conditions, given the stoichiometric solubility product ofPlummer and Mackenzie

(1974).

The solubility sensitivity analysis was repeated with an average magnesian calcite composition of9 mol%, which at the onset ofthe simulation was in equilibrium with the pore water in the high solubility scenario (Fig 4.14). The results indicated that the pore water saturation state with respect to 9 mol% magnesian calcite remained above saturation throughout the simulation for both the intermediate and the low solubility scenario. In the high solubility scenario, undersaturated conditions were reached almost immediately, initiating dissolution and buffering ofthe pore water. No significant effect

~ o 300 E Til 2.0 N tl ~o c: ~ 250 ~ ~ 1.8 :l ~ .sl 1.6 ----t..-...... I-·-IIH...... ,__...... k J;! 1 ~, 1.4 OJ ~ '#. 1.2 '0 E

0> 1.01:=:=:=:::::::~~~:g:::i=:t=~~tt+O

1700 1750 1800 1850 1900 1950 2000 2050 2100 Year

Figure 4.14. Sensitivity analysis: magnesian calcite solubility. Pore water saturation state with respect to 9 mol% magnesian calcite (black symbols; high solubility-diamonds; intermediate solubility-squares; low solubility-triangles; see text for a detailed description ofthe solubility estimates), and cumulative amount ofcarbonate as carbon dissolved upon undersaturation (open symbols).

132 was observed in the overlying surface water. It is important to note that the analysis was conducted with an average magnesian calcite composition of9 mol% and not 15 mol%.

In the high solubility scenario, magnesian calcite phases with a magnesium content >9 mol% would already be subject to significant dissolution or non-existent because they had already completely dissolved. Thus, uncertainty in the stoichiometric solubility product could have significant implications for model results. The calculated value ofthe maximum absolute ratio ofvariation (MARGV; Dubus and Brown, 2002) confirmed this conclusion, indicating that the solubility product had the strongest relative influence on surface water carbonate saturation state and carbonate dissolution ofall the parameters tested. However, observations from the natural environment do not conform to the reactivity ofthe high solubility scenario (Garrels and Wollast, 1978; Mackenzie et al.,

1983; Morse and Mackenzie, 1990). The results ofthe high solubility scenario suggested that no or very little magnesian calcite with a magnesium content >9rnol% would be present within the sediments. In reality, the magnesium content ofnatural magnesian calcite can be as high as 30 mol% (Chave, 1954; Garrels and Wollast, 1978; Morse and

Mackenzie, 1990). In addition, recent work has significantly improved and constrained the solubility ofmagnesian calcite minerals (Bischoff et aI., 1987; Busenberg and

Plummer, 1989; Bertram et aI., 1991; Bischoffet aI., 1993; Tribble et aI., 1995). Thus, the high solubility scenario can be ruled out as unrealistic because the applied solubility is not within the observed constraints. The stoichiometric solubility product is stilI considered the most 'sensitive' parameter, but it is not considered 'important' since it is relatively well constrained and does not add any significant variability to the model

133 output (Hamby, 1994). Consequently, the results ofthe stoichiometric solubility product analysis do not refute the main conclusion ofthe standard model.

4.3.3.4 C02 emission scenarios

In the current analysis, the CO2 emission rates adopted in TOTEM for the 21 st century were altered according to the IPCC SRES low (B I) and high (AIFI) emission scenarios (Houghton et aI., 2001). The standard scenario adopted the "business as usual"

IS92A emission scenario (Houghton et aI., 1996; 200I). Because C02 emission and not atmospheric C02 concentration was used as a forcing in TOTEM, the numerically calculated atmospheric CO2 concentration ofthe model between the year 2000 and 2100 differed somewhat from the results ofthe IPCC calculated scenarios. The atmospheric

C02 concentration ofthe IPCC scenarios ranged from 540 ppm to 970 ppm whereas the numerical calculations of TOTEM combined with SOCMranged from 460 ppm to 950 ppm by the year 2100 (Fig 4.15). The reasons for this slight disagreement probably are due to several factors inherent in the different models (see Mackenzie et aI., 2001). The atmospheric C02 concentration calculated by TOTEM combined with SOCMuntil year

2002 agrees well with the observational record ofCO2 obtained at the Mauna Loa observatory (Fig 4.16).

As a consequence ofthe wide range ofatmospheric CO2 concentrations resulting from the various IPCC SRES emission scenarios (Houghton et aI., 200I), the surface water carbonate saturation state in SOCM varied substantially and had a significantly different impact on biogenic calcium carbonate production in the various scenarios (see subsequent discussion on biological implications). Although alterations to the

134 1000 6 .....Q) .....<0 en 800 c:: 5 0 -E :0= a. ...<0 a...... :::::J 600 -N <0en 4 0 () Q) ..... 0 °0 0;:: <0 Q) 0 3 400 ~ I a. 0) en ~ 0 ~ E 0 ..... 0 2 « E 200 .....LO

I----,----r--r---,-----.--.------.--~O 1700 1750 1800 1850 1900 1950 2000 2050 2100 Year

Figure 4.15. Effect ofIPCC SRES CO2 emission scenarios (AIFI-squares; IS92A­ diamonds; B I-triangles) on model calculated atmospheric CO2 (gray symbols), surface water (black symbols) and pore water (open symbols) saturation state with respect to 15 mol% magnesian calcite. The results indicate that changes in the surface water owing to variations in atmospheric CO2 have essentially no effect on the pore water. On a global scale, the surface water and the pore water can essentially be viewed as two decoupled systems.

atmospheric CO2 concentration trend had important implications for surface water carbon chemistry and the calcareous organisms living within the water column, the pore water of the sediment system remained unaffected in all scenarios (Fig 4.15). The maximum absolute ratio ofvariation (MAROV; Dubus and Brown, 2002) showed that atmospheric

CO2 had the strongest relative influence (disregarding the magnesian calcite solubility product) on surface water carbonate saturation state compared to the other parameters tested (Table 4.1). However, it should be emphasized that the effect was negative, i.e. increasing CO2 caused a decrease in carbonate saturation state. Furthermore, the analysis

135 indicated that atmospheric CO2 had very little influence on carbonate dissolution (Table

4.1). The results ofthe current sensitivity analysis indicated that the pore water-sediment system and the overlying surface water essentially can be considered as two decoupled systems. The surface water has very little influence on processes within the pore water- sediment system and vice versa. In mesocosm experiments with calcareous communities exposed to elevated pC02, Leclercq et aI., (2002) observed that dissolution ofcalcareous sand was not affected by surface water carbonate chemistry, but rather by the biogeochemistry ofthe pore water-sediment system.

> 380 -E Q. Q. o-N 360 o .u.:::: Q) .r:::. 340 Q. f/) o E 320 --Mauna Loa record < Model prediction

1960 1970 1980 1990 2000 Year

Figure 4.16. Annual mean atmospheric CO2 concentration (ppmv) at the Mauna Loa observatory, Hawaii (Keeling and Whorf, 2002), and the model calculated global average concentration.

136 4.3.3.5 Shallow-water ocean - open ocean exchange

The exchange rate between the shallow-water ocean and the open ocean was based on the residence time ofwater in the former and the oceanic hydrologic cycle assuming a steady state balance between input via rivers and upwelling and output via net flow to the open ocean (Mackenzie, 1987). Evaporation and precipitation were assumed to be approximately equal and relatively small compared to upwelling and surface water exchange, and were not considered in SOCM. In a previous model ofthe hydrologic cycle, Mackenzie (1987) suggested that the average residence time ofwaters in the near· shore region is approximately 11 years. However, based on recent estimates of continental margin and coastal upwelling rates (Wyrtki, 1963; Chavez and Barber, 1987;

Pelegri and Csanady, 1991; Walsh, 1991; Tomczak and Godfrey, 1994; Brink et aI.,

1995; Chavez and Toggweiler, 1995), the average residence time ofwater within the shallow-water ocean environment is approximately 2-3 years. In the current sensitivity scenario, the residence time ofwater within the shallow-water ocean environment was allowed to vary from 2 years to as much as 24 years. The shorter the residence time, the smaller the effect ofcarbonate dissolution on surface water chemistry because any changes in chemical composition ofthis reservoir will be rapidly diluted by the much larger open ocean reservoir. Consequently, in the current analysis the buffer effect ofthe surface water carbon chemistry owing to dissolution ofcarbonate minerals increased as the residence time became longer and vice versa as it became shorter (Fig 4.17). The maximum absolute ratio ofvariation (MAROV; Dubus and Brown, 2002) indicated that the exchange rate between the shallow-water ocean and the open ocean could have a

137 relatively large influence on surface water carbonate saturation state (Table 4.l).

However, in the current analysis the residence times required to allow for a significant buffer effect were too long (> 24 years) to be realistic and applicable on a global scale.

On a regional scale and in areas oflimited circulation and high carbonate dissolution, it is possible that surface seawater to some extent could be buffered against changes owing to increasing atmospheric CO2•

4.5 8.3 2 ~ § 4.0 ~ 8.2 ::> :I: ~ 3.5 Co 2 8.0 '8 g'3.0 «... -'-t=2yrs -+-, = 12 yrs 7.9 2.5 ,. 24 yrs !---r---r--..,....--.,..-...... ---.,--,...--+7.8 1700 1750 1800 1850 1900 1950 2000 2050 2100 Year

Figure 4.17. Sensitivity analysis: the effect ofshallow water-open ocean water exchange on surface water aragonite saturation state (closed symbols) and pH (open symbols). The exchange rate was altered based on the residence time ('t) ofwater within the shal1ow-water ocean environment. The more rapid the exchange rate with the open ocean, the smaller the buffer effect ofthe shallow water owing to dissolution ofcarbonate minerals.

138 4.3.4 Biological implications

It is difficult to ascertain how calcareous organisms and communities will be affected by future environmental changes since the response appears to be substantially different on different time scales (Buddemeier and Smith, 1999). However, based on the current existing relationships between calcification and carbonate saturation state (Smith and Roth, 1979; Borowitzka, 1981; Mackenzie and Agegian, 1989; Gao et a!., 1993;

Gattuso et a!., 1998; Langdon et a!., 2000; Riebesell et a!., 2000; Leqlercq et al., 2000;

2002), it appears likely that calcareous organisms are directly threatened by increasing atmospheric CO2and subsequent lower carbonate saturation state. The effect of temperature is not clear and according to currently existing relationships might either enhance (Grigg, 1981; 1997; Lough and Barnes, 2000) or impede (Clausen and Roth,

1975; Agegian, 1985; Mackenzie and Agegian, 1989) the biogenic calcification process.

The objective ofthe current analysis was to investigate the range ofresponses of calcareous organisms to different CO2 emission and temperature scenarios (Houghton et a!., 2001), adopting currently existing relationships between the rate ofcalcification and carbonate saturation state and temperature. In order to investigate the response of calcareous organisms and communities, the following analyses were performed:

1) Species and community specific response to decreasing carbonate

saturation state.

2) The effect ofIPCC SRES AIFI and BI CO2emission scenarios (extreme

C02 emission i.e. maximum and minimum, respectively) on atmospheric

C02, seawater carbonate saturation state and marine biogenic calcification.

139 3) The effect ofIPCC SRES predicted minimum and maximum temperatures

on biogenic calcification.

4) The combined effect ofdecreasing carbonate saturation state and

increasing temperature on biogenic calcification adopting various

combinations ofequations relating these parameters to calcification.

4.3.4.1 Species and community specific response to decreasing carbonate

saturation state

Current experimental results indicate that the relationship between calcification and carbonate saturation state is either linear or curvilinear (Smith and Roth, 1979;

Borowitzka, 1981; Mackenzie and Agegian, 1989; Gao et aI., 1993; Gattuso et aI., 1998;

Langdon et aI., 2000; Riebesell et al., 2000; Leqlercq et al.,2000, 2002). The response to decreasing carbonate saturation state is significantly different between different calcifying organisms, species and communities. In the current analysis, the relative rate ofcalcification was calculated for a variety ofcalcareous species and communities based on the equations presented in Table 2.2 and the aragonite saturation state predicted by the standard model simulation ofSOCM. Based on these equations, changes in the rate of calcification showed substantial variability (Fig 4.18). The rate ofnet calcification decreased in each scenario, and ranged from being almost unaffected to being completely inhibited. The wide range ofvariability illustrates the difficulties at present in making predictions on how calcareous organisms and communities will respond to lower carbonate saturation states in the future. It is important to note that the equations in Table

2.2 represent the best fits to a limited number ofdata points that mayor may not be

140 I--_.....I..-_---'-__.L...-_-.L..._-----I__...... _...... &.._--+ 5.0

~ 80 c:: o ~ ~ --()-- Aragonite sat state ~ 60 --Porites porites' Stylophora pistil/ata2 ro() --+- 3 _o --Amphiroa foliacea 2 40 --Porolithon gardinen4 ~ Bisosphere 2 oceans Q)> --Coral communitl ~ 20 --Coral community' 3.0 Q) --Okinawa reef flal a:: Sand communit? --+- Linear best fit of multiple studies'o o+---.------.--~-"""T"""-~--r----r---+ 2.5 1700 1750 1800 1850 1900 1950 2000 2050 2100 Year Figure 4.18. Changes in relative rate ofcalcification (expressed as a percentage ofthe approximate rate at pre-industrial conditions: Qaragonite ~ 4.9) for various calcareous organisms and communities based on experimentally derived relationships (Leclercq et aI., 2002) as a function ofsaturation state between 1700 and 2100. Note the wide range ofresponses ofthe organisms from being nearly unaffected to being incapable ofcalcifying. Diamonds indicate the relationships adopted in two different model scenarios. 1) Marubini and Thake (1999), 2) Gattuso et al. (1998), 3) Borowitzka (1981), 4) A~egian (1985); Mackenzie and Agegian (1989), 5) Langdon et al. (2000), ) Leclercq et al. (2000), 7) Leclercq et al. (2002), 8) Odhe and Woesik (1999), 9) Boucher et al. (1998), 10) Gattuso et al. (1999).

representative or sufficient for extrapolation to the full extent ofthe carbonate saturation state predicted by the current model.

4.3.4.2 C02 emission scenarios - biogenic calcification

Model predictions suggest that the atmosphere most likely will reach a CO2 concentration of700 ppmv by the end ofthe 21 5t century. Depending on the development ofthe CO2 driving forces such as population growth, economic development, etc, the

141 C02 concentration could very well be substantially lower or higher than this estimate.

The !PCC SRES suggests an atmospheric CO2 concentration between 540 to 970 ppmv by the year 2100 (Houghton et a!., 2001). Consequently, the effects on seawater carbonate saturation state and marine biogenic calcification could be significantly different depending on the resulting atmospheric CO2concentration ofthe !PCC scenario adopted. In the current analysis, the !PCC SRES emission scenarios AIFI and BI were implemented as CO2 emission forcings in TOTEM. The model calculated atmospheric

C02 concentration varied between 460 to 950 ppmv (see previous discussion) and caused a subsequent decrease in surface water carbonate saturation state in SOCM (Fig 4.19). By the year 2100, seawater aragonite saturation state was 2.2 and 3.3 for the CO2forcing scenarios AIFI and Bl, respectively. Based on the calculated aragonite saturation state in each scenario, the relative rate ofcalcification was calculated for each scenario adopting the curvilinear relationship ofStylophora pistillata (Gattuso et a!., 1998) and a linear relationship derived from multiple studies (Fig 4.20; Gattuso et a!., 1999a). For each C02 emission scenario, the rate ofcalcification was minimally affected in the curvilinear scenario and ranged from 93% to 98% relative to the rate in the year 1700. In the linear scenario, the rate ofcalcification was significantly affected and decreased to rates ranging from 56% to 75% relative to the rate in the year 1700 (Fig 4.19).

The small effect on the rate ofcalcification in the curvilinear scenario could be explained by the fact that the aragonite saturation state remained above the critical threshold ofsignificant change for both emission scenarios. Ifthe aragonite saturation state were to drop below this threshold, the rate ofcalcification would very rapidly

142 decline. In conclusion, depending on the actual relationship between biogenic calcification and carbonate saturation state, the results indicate that the rate of calcification could remain almost unaffected to being suppressed by as much as 44% owing to changes in carbonate saturation state between year 1700 and 2100.

4.5 100 ~ -0 4.0 a:: Q) c: 90 as -0 1;)- :;:; 3.5 c: r3 :;0 &0:: '0 80 ~ (ij ;:, Q) R=21.3Q+12 as > L- :;:; 0I0 69 as R = 228(1-e· · )-128 « Q) 60 2.0 a:: Aragonite saturation state

50 1.5 1950 2000 2050 2100 Year

Figure 4.19. Range ofvariability in relative rate ofcalcification relative to year 1700 as a function ofIPCC SRES predicted CO2 emission scenarios (A2, B1) (Houghton et aI., 2001) and subsequent changes in aragonite saturation state as predicted by the standard model. The gray area shows the maximum and minimum range ofaragonite saturation state predicted based on SRES emission scenarios A2 and B1. The red area represents the results adopting a linear relationship between calcification and aragonite saturation state (R=21.3Q + 12; Gattuso et aI., 1999) nonnalized to the initial saturation state in the year 1700. The blue area represents the result adopting a curvilinear nJO 69 relationship (R=228(1_e· . )- 128; Gattuso et aI., 1998; Leclercq et aI., 2002) nonnalized to the initial saturation state in the year 1700.

143 4.3.4.3 Temperature sceuarios - biogenic calcification

During the past century, the global mean temperature increased by approximately

1°C. In the 21 51 century, the global average temperature has been predicted to increase 1.4 to 5.8°C relative to 1990 (Houghton et aI., 2001). It is not known how calcareous organisms will respond to increasing temperatures. On the one hand, experimental investigations indicate a relationship between calcification and temperature that would imply a decrease in calcification with increasing temperature (Clausen and Roth, 1975;

Agegian, 1985; Mackenzie and Agegian, 1989). On the other hand, observational data suggest a positive linear relationship between the rate ofcalcification ofcorals from different locations and the annual average temperature ofthese locations (Grigg, 1981;

1997; Lough and Barnes, 2000). In the current analysis, changes in the rate of calcification due to temperature were investigated based on the equations presented in

Table 2.3. The IPCC SRES predicted minimum and maximum temperature changes during the 21 51 century were used to predict the extent ofpotential changes in the relative rate ofcalcification (Fig 4.20). A linear positive relationship between the rate of calcification and temperature caused an increase of65% to 190% relative to the rate in the year 1700 for the minimum and maximum temperature scenario, respectively. As expected, a negative parabolic relationship between the rate ofcalcification and temperature caused a decrease ranging from 7% to 75% relative to the rate in the year

1700 for each temperature scenario, respectively. In conclusion, depending on the response ofcalcareous organisms to increasing temperature, the rate ofcalcification could be significantly enhanced or significantly inhibited owing to the predicted

144 temperature changes ofthe 21 st century. However, it is important to note that temperature changes in tropical and subtropical latitudes will be moderate and the average temperature will probably stay relatively constant as opposed to high latitudes. An increase in sea surface temperature could potentially also expand the habitable region for coral reefs.

300 +-__...... __...... __...... __--L....__---'-_____+_ 7

R =100 + 2MT - 6 ?fl 250 R =100 - a.H2 £i Temperature () -c: 5 -0 ~o Q) 200 -C) B c: ~ 4 m .s=. '0 0 150 B 3 ~ '0 ::::J m... oS -Q) ~ 1001-----_~,.._. 2 ~ E ~ ~ ~ 1 a:; 50 0:: 0 O-+---~---,...---__r----r---__r--____I 1950 2000 2050 2100 Year

Figure 4.20. Range ofvariability in relative rate ofcalcification relative to year 1700 as a function ofIPee SRES predicted temperature scenarios (Houghton et aI., 2001). The gray area shows the minimum and maximum temperatures predicted for the 21 st century. The red area represents the results using a positive linear relationship between calcification and temperature (R = 100 + 28L\T; Lough and Barnes, 2000; Grigg, 1981, 1997; Scoffin et aI., 1992). The blue area represents a negative parabolic relationship between calcification and temperature 2 (R = 100 - aL\T ; a = 1.19 to 1.65) (Smith and Roth, 1975; Agegian, 1985; Mackenzie and Agegian, 1989).

145 4.3.4.4 Combined effect of carbonate saturation state and temperature on

biogenic calcification

Based on current experimental and observational data, the relationship between calcification, temperature and carbonate saturation state is not well constrained.

Consequently, the response ofcalcareous organisms and communities to future predicted lower carbonate saturation states and higher temperatures could be significantly different depending on how the organisms actually respond to these changes. In order to investigate the possible outcomes, the various relationships and equations adopted in the various model scenarios were combined in all possible ways. The standard CO2 emission and temperature forcings ofthe model were used (IS92A; Houghton et aI., 1996; 2001).

Total biogenic carbonate production was related to DIC, temperature and carbonate saturation state (see Chapter 2 for derivation and detailed description ofthe equation).

The relationship between biogenic calcification and DIC was offirst order and was not altered in any ofthe analyses. The total DIC increased by approximately 8% throughout the simulation as an effect ofincreased invasion ofatmospheric C02 into the ocean and caused biogenic carbonate production to increase by the same percentage. The equation describing biogenic carbonate production was altered by combining the adopted equations between calcification rate and carbonate saturation state (curvilinear and linear), and calcification rate and temperature (negative parabolic and positive linear) in all possible combinations (Fig 4.21). The results indicate that the combined effect of increasing DIC, temperature and decreasing carbonate saturation state could cause global

146 carbonate production to increase by almost 100% or decrease by approximately 40% depending on the relationships adopted.

70 - curvilinear/parabolic ....., -~ - curvilinear/linear >. 60 () -linear/linear 0 - linear/parabolic E 50 .....N maximum range 0 (IPCC SRES A1FI) T"" --c 40 0 U :::J "C 30 0 ~ c...... Q) 20 ro c 0 .c ~ ro 10 () 0 1700 1800 1900 2000 2100 Year

Figure 4.21. Biogenic carbonate production from 1700 to 2100 adopting different relationships between calcification, carbonate saturation state and temperature. Rate ofcalcification was related to carbonate saturation state based on two different equations: R=228(l-e-Q10,69) - 128 (curvilinear; Gattuso et aI., 1998; Leclercq et aI., 2002); R=21.3Q + 12 (linear; Gattuso et aI., 1999). Calcification was also related to temperature based on two equations: R = 100 + 28~T (linear; Lough and Barnes, 2000); P = 100 - 1.32~T2 (parabolic; (Agegian, 1985; Mackenzie and Agegian, 1989). The results ofthe different scenarios represent the following relationships with respect to carbonate saturation state and temperature, curvilinear/parabolic, curvilinear/linear, linear/parabolic, linear/linear. Biogenic carbonate production was also related to the total DIC content ofsurface waters by a first order relationship, but the DIC concentration ofthe waters varies little in the current model.

147 4.4 SUMMARY AND CONCLUDING REMARKS

The standard model appears to represent adequately the shallow-water ocean environment since relatively good agreement is observed between the standard simulation and observations from the natural environment. Results from sensitivity analyses confirmed the major conclusion ofthe standard simulation that the surface water ofthe shallow-water ocean environment will not accumulate sufficient alkalinity from dissolution ofmetastable carbonate minerals to buffer any changes in surface waters owing to increasing atmospheric CO2• Although substantial dissolution ofcarbonate minerals was initiated in several ofthe sensitivity scenarios conducted and a significant buffer effect was observed within the pore water, no substantial buffer effect was ever observed in the overlying surface water. Small changes could be observed in some extreme scenarios, but the magnitudes ofthese changes were not sufficient to exert any significant restoration ofsurface water carbonate chemistry to protect calcareous organisms against rising atmospheric CO2• The absence ofa buffer effect within the surface water could be explained by the well mixed character and the significantly larger size ofthis reservoir relative to the pore water. Any significant changes taking place within the pore water were rapidly diluted and dissipated in the much larger surface water reservoir. Consequently, based on existing relationships between biogenic calcification rates and carbonate saturation state, the ability ofcalcareous organisms to calcify could be significantly impeded as a consequence ofincreasing atmospheric CO2 and lower carbonate saturation state. The biological response to decreasing carbonate saturation state and increasing temperature could be significantly different for different calcareous

148 species. In general, all species will be negatively affected by lower carbonate saturation state, but it is not known how calcareous organisms will respond to increasing temperature. Current evidence suggests that increasing temperature could either stimulate or impede future biogenic calcification rates.

149 CHAPTER 5: SUMMARY AND CONCLUSIONS

The results ofthe current work showed that increasing atmospheric CO2and temperature will have significant implications for the carbonate chemistry ofthe shallow-water ocean environment and the calcareous organisms living within this region during the 21 st century. The standard model simulation ofSOCM indicated that the surface water carbonate saturation state has decreased owing to increased invasion of atmospheric C02 since the onset ofthe industrial revolution until present. Time-series observations from the Hawaiian Ocean Time series (HOT) between 1989 and 2000 accord well with the predicted trend ofthe standard model and were statistically significant according to non-parametric tests. The results ofthe model indicated that surface water carbonate saturation state will continue to decrease as an effect of anthropogenically sustained increases in atmospheric C02 content. Consequently, calcareous organisms will have difficulty calcifying. In addition, calcareous organisms and communities could be significantly weakened and subject to increased mechanical and biological erosion ifnot dissolution ofmetastable carbonate minerals could restore the imposed changes in carbonate saturation state and pH owing to the predicted increase in atmospheric CO2(Halley and Yates, 2000; Barnes and Cuff, 2000). The results of

SOCM indicated that dissolution ofmetastable carbonate minerals is already taking place within the pore water-sediment system (Mackenzie et aI., 1980; Molin et aI., 1985;

Halley and Yates, 2000; Leclercq et aI., 2002), consequently acting as a 'buffer' and maintaining the pore water in a metastable equilibrium with the most soluble solid carbonate phase present in the sediments (Schmalz and Chave, 1963; Wollast et aI., 1980;

150 Tribble and Mackenzie, 1995). Sensitivity analyses indicated that dissolution of carbonate minerals was mainly driven by remineralization oforganic matter rather than carbonate reaction kinetics (Moulin et aI., 1985; Morse and Mackenzie, 1990), and was solely controlled by the biogeochemistry ofthe pore water (Leclercq et aI., 2002). In fact, the two systems could essentially be regarded as decoupled. Although substantial dissolution ofcarbonate minerals was initiated in multiple sensitivity scenarios, producing a significant buffer effect within the pore water, none or very little ofthis effect was ever observed in the overlying surface water. Consequently, the surface water ofthe shallow-water ocean environment will not accumulate sufficient alkalinity owing to dissolution ofmetastable carbonate minerals to counteract the effect ofincreasing atmospheric C02.

The absence ofa buffer effect within the surface water could be attributed to its well-mixed character and significantly greater reservoir size compared to the pore water.

Any significant changes in the pore water carbonate chemistry will be rapidly diluted in the much larger surface water reservoir. Consequently, calcareous organisms and communities will not be protected from any changes in the surface water carbonate chemistry owing to increasing atmospheric CO2• In the standard run ofSOCM, biogenic production ofcalcium carbonate was impaired by 7-44% by the year 2100 depending on whether a curvilinear (Smith and Roth, 1979; Gattuso et aI., 1998) or a linear

(Borowitzka, 1981; Mackenzie and Agegian, 1989; Gao et aI., 1993; Langdon et aI.,

2000; Riebesell et aI., 2000; Leclercq et al., 2000; 2002) relationship between carbonate saturation and calcification was adopted. Although decreasing carbonate saturation state

151 will have a negative effect on biogenic calcification in the future, the effect of temperature is not clear. Sensitivity analyses adopting current existing relationships between calcification and temperature indicated that calcification could either be enhanced or inhibited as an effect ofincreasing temperature.

Changes in pore water carbonate chemistry were strongly driven by organic matter deposition and its subsequent remineralization in the sediments by microbes

(Mackenzie et al., 1980; Moulin et al., 1985; Morse and Mackenzie, 1990). The results of

SOCM suggest that the dissolved carbonate ion activity ofthe pore water is controlled by a metastable equilibrium with the most soluble solid carbonate phase present within the sediments. Consequently, during early solutional diagenetic modifications on the seafloor owing to natural or anthropogenic processes, dissolution ofcarbonate minerals follows a sequence based on mineral stability, progressively leading to removal ofthe more soluble phases until the most stable phases remain. In the standard run ofSOCM, the composition ofmagnesian calcite at metastable equilibrium with the pore water changed from 21 mol% Mg to 14 mol% Mg. Future decrease in average pore water saturation state is likely to alter this metastable equilibrium and could affect the average composition and rates of precipitation ofcarbonate cements in contemporary shallow-water sediments.

152 APPENDIX A

A.I FORCINGS OF THE SHALLOW-WATER OCEAN CARBONATE MODEL

(SOCM)

To facilitate input and output fluxes at the boundaries ofthe Shallow-water

Ocean Carbonate Model (SOCM), the model was integrated into the global biogeochemical model TOTEM (Ver, 1998; Ver et aI., 1999). The model calculated atmospheric C02 concentration, total carbon delivered to the shallow-water ocean environment via rivers and upwelling, and the temperature scenario ofTOTEM were used as forcing functions in SOCM (Fig A. I). Output fluxes from SOCMvia surface water­ exchange with the open ocean, export ofsediments to the continental slope, and atmospheric exchange were fed back into the appropriate reservoirs ofTOTEM.

A.2 DIC CALCULATIONS

DIC calculations ofthe surface water were calculated based on TCOz and pCOz

(Morse and Mackenzie, 1990). The calculations were based on the assumption that the speciation ofdissolved inorganic carbon in the surface water was in equilibrium with the partial pressure ofCO2 in the atmosphere. Equilibrium was assumed to be established instantaneously at each time increment (dt). At the initial conditions ofSOCM, the surface water was assumed to be a net source ofCO2to the atmosphere, which implies that the surface water and atmosphere are not in equilibrium with respect to this gas.

Consequently, it could be argued that the assumption ofequilibrium is invalid. However, based on the net flux ofcarbon from the surface water to the atmosphere, the

153 I------'-----'---'---.l...------...L-...... ----oj- 3.0 A -700> -Temperature change E 2.5 Q. --- _. Atmospheric CO ,, .9: 600 2 , 2.0 N ,, o ,, 1.5 6 ~ 500 ,, o ';:: , -~ Q) , 1.0 r. , ~ 400 ,, o , 0.5 E , , ...., o c:t:- 300 -----~.-_.--_.---_ ... B 6480 o E478 N ~o :=..476 en c: =474 ~ 1------:§"472

C 60 -DIC -u --Reactive organic matter 0 E 50 - Unreactive organic matter N --PIC ~ 0..... 40 -:::l 30 -Q. .5 ... 20 Q) > ~ 10

O-l--,.----r------,r-~----yo-"""'T'"---r--_+_ 1700 1800 1900 2000 2100 Year

Figure A.I. Forcing parameters ofSOCM (calculated by TOTEM; Ver, 1998; Ver et aI., 1999). (A) Atmospheric CO2 concentration and changes in temperature relative to year 1700, (B) input flux ofcarbon via upwelling from the deep ocean, and (C) river input.

154 0.14

0.12 > -E 0.10 0. 0. 0.08 -N 0 () 0. 0.06 <::I 0.04

0.02 1700 1800 1900 2000 2100 Year Figure A.2. Difference in partial pressure ofCO2 (ppmv) between surface water and atmosphere based on the net flux ofcarbon from the surface water to the atmosphere assuming that equilibrium was reached instantaneously.

corresponding difference in partial pressure ofCO2 can be calculated (Fig A.2). The results indicate that the magnitude ofthe disequilibrium is insignificant for the purpose of the current calculations and the assumption ofequilibrium can be safely employed.

In the pore water, DIC calculations were based on the TC02 and pC02 ofthis system (Morse and Mackenzie, 1990). The initial dissolved inorganic carbon speciation ofthe pore water was based on the average composition ofpore waters from a variety of locations in the Bahamas and elsewhere (TC02 = 3800 mmol L-1; pH = 7.51; Morse et a!., 1985; Morse and Mackenzie, 1990). Changes in dissolved CO2(aq) and pC02 were assumed to equal changes in the release ofCO2 owing to variations in remineralization of organic matter.

155 DIC calculations in SOCM (both surface water and pore water) were based on activity coefficients in seawater with a salinity of35 psu (Nagy, 1988; Morse and

Mackenzie, 1990) and the carbonic acid system as dermed by Plummer and Busenberg

(1982). Although calcium and magnesium concentrations have been observed to deviate from their predicted conservative proportions in certain coastal regions and within pore waters, these proportions were assumed to remain constant in SOCM. Calcium carbonate solubility (Ksp) for calcite and aragonite were based on the estimates ofPlummer and

Busenberg (1982). Magnesian calcite solubility was based on Bishop et a!., 1993. In order to make a comparison to data at the Hawaiian Ocean Time series (HOT) and the

Bermuda Atlantic Time Series (BATS), the surface water DIC system was recalculated based on concentrations and the stoichiometric carbonic acid system defined by Roy et al.

(1993). Calculations were conducted using the program C02SYS (Lewis and Wallace,

1998).

A.3 MODEL EQUATIONS

The mathematical logic behind the construction ofthe box model ofSOCMwas adopted from TOTEM (Ver, 1998; Ver et a!., 1999). The basic assumptions ofbox modeling such as conservation ofmass and well mixed reservoirs were adopted. At the onset ofthe simulation, the system was assumed to be in a quasi-steady state. In general, the model consisted ofa set ofsimultaneous differential equations that were solved in incremental time steps. Euler's method ofintegration was used to project the equations between each time step. In the present case, the model was run from year 1700 until year

156 2100 using a time step (dt) of0.0125 years. TOTEM was constructed, compiled and run using the model simulation software STELLA®. The current model was also constructed and integrated into TOTEM using STELLA® research version 6.0.1 (High Performance

Systems, Inc).

The general form ofthe differential equations is (Lerman et aI., 1975; Ver, 1998):

dM; =" F .. _ " F. (A.l) dt ~J/ ~IJ J J where Mis the mass ofa particular constituent in reservoir i, Fij is the flux ofthe constituent from reservoir i to reservoir}, and Fj; is the flux ofthe constituent from reservoir} to reservoir i. At the initial steady state, the equation is:

d;;! = LFp - LFij =o=> LFj; = LF,j (A.2) t",oJ j j j

Processes facilitating fluxes between reservoirs in TOTEM and within SOCM were related to essential parameters such as temperature, atmospheric CO2, and land use emissions. The most commonly adopted flux between reservoirs is described by equations offirst order kinetics:

Fij = kijM; (A.3) where Fij is the flux ofthe constituent from reservoir i to reservoir}, which is linearly dependent on the mass ofthe constituent in reservoir M;, and kij is the rate constant. In the present model, kij was defined based on the initial reservoir mass and the initial flux adopted from the literature, i.e:

k.=M; (AA) y P- IJ t=O

157 In the following, the mass balance equations and the flux equations for the shallow-water ocean environment are presented. Reservoirs are annotated C, where Ci refers to the total mass ofcarbon in reservoir i (see table A.l). Fluxes are annotated CFij, where CFij refers to the flux ofcarbon between reservoirs i and).

TABLE A.!. ANNOTATIONS OF RESERVOIRS IN THE SHALLOW-WATER OCEAN ENVIRONMENT AND ADJACENT RESERVOIRS IN TOTEM. CORRESPONDING NOTATIONS USED IN TOTEM ARE ALSO SHOWN.

Reservoir SOCM TOTEM Surface water I 4 Surface water organic matter 2 5 Pore water 3 Sediment organic matter 4 6 Sediment river derived PIC 5 6 Sediment calcite 6 6 Sediment aragonite 7 6 Sediment magnesian calcite 8 6 Atmosphere A 10 Sediment permanent burial B B Open ocean o 7/9 Rivers R R

158 A.3.t Mass balance equations

Surface water dC1(t)/dt = CFR1(t) + CFA1(t) + CF01 (t) + CFou1(t) + CFJ1 (t) + CF11(t) - CF1A(t) - CFIO(t)

- CFu(t) - CFdt) (A5)

Surface water organic matter

(A6)

Pore water dC 3(t)/dt = CFI3Ct) + CF43(t) + CFS3(t) + CF63(t) + CF73(t) + CFS3 (t) - CF31 (t) - CF3S(t)-

CF36(t) - CF37(t) - CFJs(t) (A?)

Sediment organic matter dC4(t)/dt = CFR4(t) + CF24(t) - CF4J(t) - CF4B(t) - CF4o(t) (A8)

Sediment river derived PIC dCs(t)/dt = CFR5(t) - CFsJ(t) - CFSB(t) - CFso(t) (A.9)

Sediment calcite dC6(t)/dt = CFI6(t) + CF36(t) - CF63 (t) - CF6B(t) - CF6o(t) (A.10)

Sediment aragonite dC7(t)/dt = CF17(t) + CF37(t) - CF7J(t) - CF7B(t) - CF7o(t) (A.11)

Sediment magnesian calcite dCs(t)/dt = CF1S(t) + CF3S(t) - CFS3 (t) - CFSB(t) - CFso(t) (A.l2)

159 A.3.2 Flux equations

CFdt) = kFlZ. pO xIN! x!PI (A13)

CF13(t) = kC 13 x C1(t) (A14)

CF16(t) = [kC16 x C1(t)] x R(nAcagonite) x R(T) (A15)

CFl7(t) = [kCl7 x C1(t)] x R(nC.leile) x R(T) (A16)

CF1s(t) = [kC18 x C1(t)] x R(nMg-

CFllA(t) = dC/dt - [CFR1(t) + CFou1(t) + CF31(t) + CFZ1(t) - CFlO(t) - CFnCt)

-CF12(t)] (A.18)

CF1lo(t) =kC1lo x C1(t) (AI9)

CFZ1(t) = kCZ1 x Cz(t) (A20)

CFZ4(t) = kC24 x C2(t) (A.2I)

CFzo(t) = kCzo x Cz(t) (A22)

CF31(t) = kC31 x C3(t) (A23)

CF6t3(t) = kC6t3 x C6(t) + M'D(nCaleite) (A24)

CF7(3(t) = kC7t3 x C7(t) + M'D(nAcagonite) (A.25)

CFsp(t) = kCsp x Cs(t) + L'>PD(nMg-calcite) (A26)

CF43(t) = kC43 x C4(t) (A.27)

CF4B(t) = kC4B x C4(t) (A28)

CF40(t) = 0 (A.29)

CFs3Ct) = kCS3 x Cs(t) (A30)

CFSB(t) = kCSB x Cs(t) (A31)

CFso(t) = 0 (A32)

160 CF6B(t) = kC6B x C6(t) (A.33)

CF60(t) = kC6Q x C6(t) (A.34)

CF7B(t) = kC7B x C7(t) (A.35)

CF70(t) = kC70 x C7(t) (A.36)

CFSB(t) = kCSB x CS(t) (A.37)

CFso(t) = kCso x Cs(t) (A.38)

CFOul(t) = kCOuI x CO(t) (A.39)

CFRI (t) = kCRI x CR(t) (A.40)

CFR2, OOC(t) = kCR2, DOC X CR(t) (A.41)

CFR2,POC(t) = kCR2,POC X CR(t) (A.42)

CFR4(t) = kCR4 X CR(t) (A.43)

CFRS(t) = kC RS X CR(t) (A.44)

A.3.3 Flux equations: special cases

A.3.3.1 Atmosphere-surface water CO2 exchange

At the onset ofthe simulation, the shallow-water ocean environment was assumed to be a net source of22x1012 mol C to the atmosphere. The largest fraction

(l5X 1012 mol C) originated from biogenic precipitation ofcalcium carbonate (see equation 1.8; Frankignoulle et aI., 1994; Wollast, 1998) and the remaining flux from the assumption that the shallow-water ocean environment was net heterotrophic (7x 1012 mol

C; Smith and Hollibaugh, 1993). Changes in the CO2 flux between the surface water and the atmosphere were attributed to changes in biogenic calcification, changes in

161 photosynthesis and respirationlremineralization, and changes owing to increasing atmospheric C02. The flux was described by the following equation:

(A.45)

The first term represents the flux ofcarbon between the surface water and the atmosphere owing to changes in the CO2 concentration ofthe atmosphere. It is described by the time rate ofchange ofthe total dissolved inorganic carbon content ofthe surface water as a function ofchanges in atmospheric carbon content (see subsequent discussion). The second term represents the flux ofcarbon owing to imbalances in net ecosystem production (photosynthesis - respiration) and the last term owing to net ecosystem calcification (calcification - dissolution). The term tI> is the fraction ofCO2 released to the atmosphere for every mol CaC03 net precipitated (0.8 in the standard model; Ware,

1992; Frankignoulle et aI., 1994; Wollast, 1998). Negative values imply a net flux to the atmosphere.

Assuming that the chemical equilibrium between the atmosphere and surface water was instantaneous, changes in surface water total dissolved inorganic carbon as a function ofchanges in atmospheric CO2 concentrations can be calculated from the following approximation ofthe Revelle equation (Bacastow and Keeling, 1973; Revelle and Munk, 1977; Ver, 1998):

C -C -C C -C R = A A,'"O IC I 1,'"0 '" R + d A A,I"O o (A.46) C A,'"O CI,,"o C A,I"O

Solving for the surface water concentration ofdissolved inorganic carbon (Cl ) yields:

162 (A47) and taking the time derivative:

CI,t=D xdx(C -CA,l=O) ]x dC A A (A48) (RDX C A,t=D +d(CA-C A,t=D))' dt

The Revelle constant was equal to 4, and the initial Revelle factor (~) was assumed to be equal to 9, i,e, the buffer mechanism ofthe seawater caused a fractional rise ofC02 in the surface water that was one ninth ofthe increase in the atmosphere (Revelle and Mank,

1977),

A.3.3.2 Marine photosynthetic flux (Ver, 1998; Ver et al., 1999)

CFI2(t) ~ kF12, 1'=0 xfN! X/PI (A.49)

(A50)

(A51)

Nx and Px is the nitrogen and phosphorus concentration ofthe surface water,

Note: in TOTEM the surface water reservoir is denoted 4 instead of 1,

A.3.3.3 Marine biogenic calcification

CF1X(t) ~ [kC1X x C1(t)] xfR(Q) x fR(T), X~6,7,8 (A52)

A(m ~ R(Ox), / R(Ox)~o (A53) fR(f) ~ R(T), / R(T)~o (A54)

R(Ox) ~ 21 ,30x + 12, or (A55)

163 R(nx) = 228(I_e-oxlo69) - 128 (A.56)

2 R(T) = 100 - l.32LlT , or (A.57)

R(T) = 28T + 100, (A.58)

A.3.3.4 Pore water-sediment system carbonate precipitation and dissolution

CFx13(t) = kCx13 x Cx(t) + LlPD(nx), X=6,7,8 (A.59)

LlPD(nx) = PDx,,(nx) - PDX,FO(nX) (A.60)

PDx

Fornx > 1,

Dx(nx) =0 (A.62) px(nx) = kx(nx - I)"x x Cppt (A.63)

Fornx

Dx(nx) = kx(nx - l)"x x Cd;" (A.64) px(nx) = 0 (A.65)

164 A.4 Derivation ofinitial reservoir masses in SOCM

TABLE A2. ESTIMATES OF INITIAL RESERVOIR MASSES WITHIN THE SHALLOW-WATER OCEAN ENVIRONMENT AT THE ONSET OF THE MODEL SIMULATION IN THE YEAR 1700. A STAR (*) INDICATES THAT THE MASS IS THE SAME AS ADOPTED IN TOTEM (Ver, 1998; Ver et aI., 1999).

Reservoir Reservoir 1012 mol C Reference/Comments notation 'Surface water 1 6000 Calculated by Ver (1998) from Broecker and Peng (1982). Volume- weighted average TC02 ofoceanic waters with an average temperature> l 16°C = 1983 f!mol L- . TOTEM shallow water volume = 0.3xlO19 L. Thus, total shallow water reservoir = 1983 f!IDol L-l x 0.3xl019 L = 6000x1012 mol C.

'Surface water 2 367 Calculated by Ver (1998). organic matter L(POC+DOC+Biota). POC from Murray (1992); shallow water POC computed by partitioning total organic particulate mass in shallow water between shallow water and open ocean volumes: 920x1012 mol C x (0.3XlO19 L/ 3.6lxlO19 L). DOC from Williams (1975); 83.3xlO·3 mol L-l x 0.3xl019. Biota (42.lxl012 mol) adopted from Lerman et al. (1989). Thus, total organic matter = 75x1012 mol C + 250xlO12 mol C + 42.1xlO12 mol C = 367xlO12 mol C.

Sediment pore 3 54 Pore water volume calculated from water carbonate sediment coverage (28.3xl06 km2) (see subsequent calculation), 1 m sediment thickness and 50% porosity = 14150xl012 L. Average pore water TC02was based on a variety ofenvironments from the Bahamas and elsewhere (Morse, 1985; Morse and Mackenzie, 1990) =

165 Reservoir Reservoir 1012 mol C Reference/Comments notation 3800 /lmol L 1(DOC is not included). Thus, pore water TC02 = 12 1 14150x10 L x 3800 /lmol C = 54x1012 mol C.

Sediment organic 4 134100 Total young (rapidly recycled, see matter Berner, 1987) shallow water C sediments (organic C + inorganic C) calculated by Ver (1998); from Berner (l987) young C reservoir = 0.115 x total C reservoir; from Lerman et al. (1989), total shallow water C sediments = 180x1016 mol. Thus, shallow water young C sediments = 0.115 x 180x1016 mol = 207000x1012 mol C. Shallow water young organic C sediments = total shallow water young C sediments - total shallow water carbonate sediments (see subsequent 12 calculation) = 207000x10 - 72900xlQ12 = 134100x1012 mol C.

Total shallow water 5,6,7,8 72900 Calculated from Milliman (1974) carbonate sediments assuming I m sediment thickness, 50% porosity and average carbonate density = 2.83 g cm-3. Total carbonate area =28.3 x106km2 (lAx1012 km2 with an average carbonate composition of80 percent by weight. 26.9x106 km2with an average carbonate composition of 15 percent by weight). Thus, total shallow water carbonate sediments = [(lAX10 12 m2 x 0.8) + (26.9x 1012 m2 x 0.15)] x 1 m x 0.5 x 2.83x106 g m-3 x (100 gram!mo1r1= 72900x1012 molC.

Refractive PIC 5 29800 It was assumed that accumulation rates ofriver transported refractive

166 Reservoir Reservoir 1012 mol C Reference/Comments notation PIC (lOX 1012 mol yr ') and in situ produced CaCOJ (14.5xI012 mol yr-1) have been relatively constant during Aragonite 6 27350 the time required to accumulate the carbonate sediment reservoir under consideration (I m thickness, 50% porosity). In that case, refractive PIC constitutes approximately 40% ofthis 12 Calcite 7 5400 reservoir (= 0.4 x 72900x I 0 mol C). The other 60% is produced by accumulation ofin situ produced 12 CaC03 (= 0.6 x 72900x10 mol C). In the latter reservoir, the relative Magnesian calcite 8 10350 proportion ofaragonite, calcite and magnesian calcite was calculated from observed proportions in recent neritic sediments compiled by Land (1967) (Aragonite:Calcite:Mg-calcite = 63.4%:12.6%:24.0%).

167 A.S Derivation of initial carbon fluxes in SOCM

TABLE A3. ESTIMATES OF INITIAL CARBON FLUXES WITHIN THE SHALLOW-WATER OCEAN ENVIRONMENT AT THE ONSET OF THE MODEL SIMULATION IN THE YEAR 1700. A STAR (*) INDICATES THAT THE FLUX IS THE SAME AS ADOPTED IN TOTEM (Ver, 1998; Ver et aI., 1999).

Flux Flux 1012 mol C Reference/Comments notation yr-) CFx Atmosphere ­ 22 (~A) C02 gas exchange between the Surface water atmosphere and the surface water. Assumed net heterotrophy ofthe shallow water ocean environment of 7xl012 mol C yr-I. Assumed release of0.8 mol C02 to the atmosphere for every mol CaC03 precipitated (Wollast, 1998). Net carbonate production (accumulation+export) was 18.5x 1012 mol C yr-I. Thus, the net flux from the surface water to the atmosphere was 7xl012 mol C yr-I + 0.8xI8.5xlO I2 mol C yr-I = 22xlOl2 mol C yr-I .

•Open ocean ­ 477 (~O) Water exchange between the surface Surface water water ofthe shallow-water ocean environment and the open ocean. Estimated by Ver (1998). Assumed that water upwells from intermediate depths ofthe open ocean to the coastal zone, thus the net direction of water transport at the surface is towards the open ocean. Adopted number is based on hydrographic rate constant (0.0833 yr-I) for the net transport ofwater from the coastal ocean to the open ocean (Mackenzie, 1987). Thus, 6000xlOI2 mol C x 0.0833 yr-1 = 500x1012 mol C yr-I. Estimate was refined with respect to the mass balance ofthe shallow water surface water reservoir.

168 Flux Flux 1012 mol C Reference/Comments notation yr-I CFx 'Open ocean Oul 473.5 Upwelling from the deep ocean to the (upwelling) - surface water ofthe shallow water Surface water ocean environment. Based on rate constant (1.619xlO-4 yr-1) of hydrologic cycle (Mackenzie, 1987). Deep sea carbon reservoir = 2.90x1018 mol C (Ver, 1998). Thus, upwelling flux is 2.90xl018 mol ex 1.6l9xlO-4 yr-1 = 470x1012 mol C yr-J. Estimate was refined with respect to the mass balance ofthe oceanic reservoirs in TOTEM.

'Organic matter- 20 18 Flux oforganic matter from the Open ocean coastal zone to the open ocean (Smith and Hollibaugh, 1993).

'Organic matter - 24 32 Settling oforganic matter to the Sediment organic sediments. Estimates range from matter 24x1012 mol yr-l (Lerman et aI., 12 1989) to 59x10 mol yr-l (Smith and Hollibaugh, 1993). Flux calculated to preserve mass balance ofthe organic matter reservoir (Ver, 1998).

'Organic matter- 21 576 Remineralization and respiration of Surface water organic matter (Lerman et aI., 1989).

Pore water- 37 0.25 Abiotic aragonite precipitation within Sediment aragonite the pore water sediment system. Based on global abiotic carbonate 12 precipitation (OAx 10 mol yr-1, see calculations in this study). Aragonite fraction (63A%) was based on proportion in recent neritic carbonate sediments (Land, 1967). Thus, flux is 12 OAx 10 mol C yr-1 x 0.634 = 12 O.25x10 mol C yr-1.

169 Flux Flux 1012 molC Reference/Comments notation yr-! CFx Pore water- 36 0.05 Abiotic calcite precipitation within Sediment calcite the pore water sediment system. Based on global abiotic carbonate precipitation (O.4x I012 mol yr-l, see calculations in this study). Calcite fraction (12.6%) was based on proportion in recent neritic carbonate sediments (Land, 1967). Thus, flux is 0.4x1012 mol C yr-lx 0.126 = 0.05x1012 mol C yr-l.

Pore water- 38 0.1 Abiotic magnesian calcite Sediment Mg- precipitation within the pore water calcite sediment system Based on global abiotic carbonate precipitation (O.4x I012 mol yr-l, see calculations in this study). Magnesian calcite fraction (24.0%) was based on proportion in recent neritic carbonate sediments (Land, 1967). Thus, flux is O.4x 1012 mol C yr-l x 0.24 = 0.1 x1012 mol C yr-l.

Pore water- 3tl 42 (--+1) Exchange ofwater between pore Surface water (net flux water and surface water. Total sum of into remineralization ofsediment organic reservoir 1) matter (Smith and Hollibaugh, 1993) and dissolution ofriver derived PIC and in situ produced carbonate minerals (Milliman, 1993; Wollast, 1994); Thus, flux is (31+5+6) x1012 mol C yr-l= 42xI012 mol C yr-l.

170 Flux Flux 1012 mol C Reference/Comments notation yr-l CFx 'River refractive R4 8 River flux ofrefractive particulate POC- Sediment organic carbon. Based on total flux of organic matter organic carbon via rivers (34xI012 mol C yr-!) (Smith and Hollibaugh, 1993) minus river flux ofdissolved organic carbon (18xlO12 mol C yr-1) (Meybeck, 1982) and separated into reactive and refractive particulate organic carbon (SO% is labile according to Smith and Hollibaugh, 1993). Thus, flux is (34-18)xO.5xI012 mol C yr-1= 8xl012 mol C yr-1.

'River PIC- RS IS River flux ofdetrital particulate Sediment river inorganic carbon (pIC) originating derived PIC entirely from continental erosion. (Meybeck, 1981, 1982). Assumed to be mainly refractive calcite.

'RiverDIC- Rl 32 River flux ofdissolved inorganic Surface water carbon (DIC) (Meybeck, 1982).

'River DOC- R2,DOC 18 River flux ofdissolved organic Surface water carbon (DOC). (Meybeck, 1982). organic matter

'River reactive POC R2,POC 8 River flux ofreactive particulate - Surface water organic carbon (POC). Based on total organic matter flux oforganic carbon via rivers (34x 1012 mol C yr-1) (Smith and Hollibaugh, 1993) minus river flux of dissolved organic carbon (l8x1012 mol C yr-l) (Meybeck, 1982) and separated into reactive and refractive particulate organic carbon (SO% is labile according to Smith and Hollibaugh, 1993). Thus, flux is (34- 18)xO.Sxl012 mol C yr-1= 8xl012 mol C yr.-1

171 Flux Flux 1012 mol C Reference/Comments uotatiou yr-1 CFx Sediment aragonite 7B 9.2 Accumulation ofaragonite within the - Burial sediments. Based on total accumulation (14.5X1012 mol C yr-l) ofin situ produced calcium carbonate (Milliman, 1993; WoUast, 1994). Aragonite fraction (63.4%) was based on proportion in recent neritic carbonate sediments (Land, 1967). Thus, flux is 14.5x1012 mol C yr-l x 12 0.634 = 9.2x10 mol C yr-l.

Sediment aragonite 70 2.5 Export ofaragonite to the continental -Open ocean slope. Based on total carbonate export (4x1012 mol C yr-l) (Milliman, 1993; Wollast, 1994). Aragonite fraction (63.4%) was based on proportion in recent neritic carbonate sediments 12 (Land, 1967). Thus, flux is 4x10 12 mol C yr-l x 0.634 = 2.5x 10 mol C yr-1.

Sediment aragonite 73 3.8 Net dissolution ofaragonite. Based -Pore water on total net carbonate dissolution: 6.7x1012 mol C yr-l (Milliman, 1993; Wollast, 1994). Adjusted with respect to mass balance ofthe sediments = 6.0X1012 mol C yr-l. Aragonite dissolution % (63.4%) was assumed to match proportion in recent neritic carbonate sediments (Land, 1967). Thus, flux is 6.0x1012 mol C yr-l x 0.634 = 3.8x1012 mol C yr-l.

172 Flux Flux 1012 mol C Reference/Comments notation yr-1 CFx Sediment calcite ­ 6B 1.8 Accumulation ofcalcite within the Burial sediments. Based on total accumulation (14.5X1012 mol C yr-1) ofin situ produced calcium carbonate (Milliman, 1993; Wollast, 1994). Calcite fraction (12.6%) was based on proportion in recent neritic carbonate sediments (Land, 1967). Thus, flux is 14.5xlQ12 mol C yr-1 x 0.126 = 1.8xlQ12 mol C yr-l.

Sediment calcite ­ 60 0.5 Export ofcalcite to the continental Open ocean slope. Based on total carbonate export (4x1012 mol C yr-1) (Milliman, 1993; Wollast, 1994). Calcite fraction (12.6%) was based on proportion in recent neritic carbonate sediments (Land, 1967). Thus, flux is 4x1012 12 mol C yr-1 x 0.126 = 0.5X 10 mol C yr.-1

Sediment calcite ­ 63 0.8 Net dissolution ofcalcite. Based on Pore water total net carbonate dissolution: 6.7x1012 mol C yr-1 (Milliman, 1993; Wollast, 1994). Adjusted with respect to mass balance ofthe sediments = 6.0X 1012 mol C yr-1. Calcite dissolution % (12.6%) was assumed to match proportion in recent neritic carbonate sediments (Land, 1967). Thus, flux is 6.0x1012 mol C yr-] x 0.126 = 0.8x1012 mol C yr-1.

173 Flux Flux 1012 mol C Reference/Comments notation yr-I CFx Sediment Mg­ 8B 3.5 Accumulation ofmagnesian calcite calcite - Burial within the sediments. Based on total accumulation (14.5xlO12 mol C yr-l) ofin situ produced calcium carbonate (Milliman, 1993; Wollast, 1994). Magnesian calcite fraction (24.0%) was based on proportion in recent neritic carbonate sediments ~Land, 1967). Thus, flux is 14.5x10 2 mol C yr-I x 0.240 = 3.5xI012 mol C yr-l.

Sediment Mg­ 80 1.0 Export ofmagnesian calcite to the calcite - Open continental slope. Based on total ocean carbonate export (4XlO12 mol C yr-l) (Milliman, 1993; Wollast, 1994). Magnesian calcite fraction (24.0%) was based on proportion in recent neritic carbonate sediments (Land, 1967). Thus, flux is 4x1012 mol C yr­ Ix 0.24 = 1.0x1012 mol C yr-I.

Sediment Mg­ 83 1.4 Net dissolution ofmagnesian calcite. calcite - Pore water Based on total net carbonate dissolution: 6.7x1012 mol C yr-I (Milliman, 1993; Wollast, 1994). Adjusted with respect to mass 12 balance ofthe sediments = 6.0xlO mol C yr-I. Magnesian calcite dissolution % (24.0%) was assumed to match proportion in recent neritic carbonate sediments (Land, 1967). Thus, flux is 6.0x1012 mol C yr-I x 0.24 = 1.6x1012 mol C yr-I.

174 Flux Flux 1012 mol C Reference/Comments notation yr-I CFx •Sediment organic 4B 9 Permanent burial oforganic matter in matter - Burial shallow water sediments. Based on total burial oforganic matter (llX1012 mol C yr-I) partitioned between deep sea burial (2x 1012 mol C yr-]) and burial in the shallow water 12 sediments (9 X 10 mol C yr-I) (Berner, 1982, 1989).

•Sediment organic 40 o Export oforganic matter from the matter - Open sediments to the open ocean. ocean Assumed to be zero. Organic matter within the sediments was either remineralized or permanently buried.

Sediment organic 43 31 Remineralization oforganic matter matter - Pore water within the pore water sediment system. Estimate by Smith and Hollibaugh (1993) (SOxlO l2 mol C yr­ I) was adjusted with respect to mass balance ofsediment reservoirs (Ver, 1998).

Sediment river 5B 10 Accumulation ofriver derived detrital derived detrital PIC particulate inorganic carbon (PIC). It - Burial was assumed that detrital PIC consisted ofrefractive calcite originating from continental erosion. It was assumed that one third ofthe total detrital PIC (lS x 1012 mol C yr-\ Meybeck, 1979, 1982) dissolved due to extensive remineralization of organic matter. The remaining fraction accumulated within the sediments.

175 Flux Flux 1012 mol C Reference/Comments ,I notation yr CFx Sediment river 53 5 Dissolution ofriver derived detrital derived detrital PIC particulate inorganic carbon (PIC). -Pore water Due to extensive amounts oforganic matter and subsequent remineralization in regions ofriver input, one third ofthe total detrital PIC (15XlO 12 mol C yr-\ Meybeck, 1979, 1982) was assumed to dissolve.

'Surface water- 12 600 Marine primary production by Organic matter phytoplankton and benthic autotrophs. Estimated by Ver (1998), based on a compromise ofSmith and Hollibaugh (1993) (160 g C m'2 yr'l) and Wollast (1998) (230 g C m'2r'll' TOTEM coastal area was 36x101 m. Thus, 36xl012 m2 x 200 g C m'2 yr'l = 600X 1012 mol yr'l. In present case, shallow water ocean environment was 28.3xl012 m2(Milliman, 1974). The higher estimate by Wollast (1998) was adopted and adjusted to balance the initial mass balance. Thus, 28.3xlO12 m2 x 254 g C m'2 yr'l = 600x10 12 mol yr'l.

Surface water - 16 3.1 Calcite production. Based on total Sediment calcite calcium carbonate production (24.5XlOI2 mol C yr-l) (Milliman, 1993; Wollast, 1994). Calcite fraction (12.6%) was assumed to match proportion in recent neritic carbonate sediments (Land, 1967). Thus, flux is 24.5xlO12 mol C yr'lx 0.126 = 3.lxI012 mol C yr-l.

176 Flux Flux 1012 mol C Reference/Comments notation yr-1 CFx Surface water ­ 18 5.9 Magnesian calcite production. Based Sediment Mg­ on total calcium carbonate production calcite (24.5XIOI2 mol C yr-l) (Milliman, 1993; Wollast, 1994). Magnesian calcite fraction (24.0%) was assumed to match proportion in recent neritic carbonate sediments (Land, 1967). Thus, flux is 24.5x1012 mol C yr-l x 12 0.24 = 5.9x10 mol C yr-l.

Surface water ­ 17 15.5 Aragonite production. Based on total Sediment aragonite calcium carbonate production (24.5xI012 mol C yr-J) (Milliman, 1993; Wollast, 1994). Aragonite fraction (63.4%) was assumed to match proportion in recent neritic carbonate sediments (Land, 1967). Thus, flux is 24.5xI012 mol C yr-lx 0.634 = 15.5x1012 mol C yr-l.

177 REFERENCES

Agegian, C.R, 1985. The biogeochemical ecology ofPorolithon gardineri (foslie). PhD thesis oceanograpby, University ofHawaii Honolulu, HI, 178 p.

Agegian, C.R., and Mackenzie, P.T., 1989. Calcareous organisms and sediment mineralogy on a mid-depth bank in the Hawaiian archipelago. Pacific Science, 43(1): 56-66.

Agegian, C.R., Mackenzie, P.T., Tribble, J.S. and Sabine, C.L., 1988. Carbonate production and flux from a mid-depth ecosystem, Penguin Bank, Hawaii, in Agegian, C.R, ed., Biogeochemical Cycling and Pluxes Between the Deep Euphotic Zone and other Oceanic Realms, Washington D.C., pp. 5-32.

Andersson, AJ., Mackenzie, P.T., Ver, L.M., 2003. Solution ofshallow-water carbonates: An insignificant buffer against rising atmospheric CO2• Geology, 31: 513-516.

Arrhenius, S., 1896. On the influence ofcarbonic acid in the air upon the temperature of the ground. Phil. Mag. (Ser. 5), 41: 237-276.

Bacastow, RD., and Keeling, C.D., 1973. Atmospheric carbon dioxide and radiocarbon in the natural carbon cycle: II. Changes from A.D. 1700 to 2070 as deduced from a geochemical model, in Woodwell, G.M., Pecan, E.Y., oos., Carbon and the Biosphere, U.S. Atomic Energy Commission, pp. 86-135.

Barnes, DJ., and Cuff, C., 2000. Solution ofreefrock buffers seawater against rising atmospheric CO2, in Hopley, D., Hopley, M., Tamelander, J., and Done, T., Proceedings ofthe Ninth International Coral ReefSymposium Abstracts, State Ministry for the Enviromnent, Indonesia, pp. 248.

Barnola, J.-M., D. Raynaud, C. Lorius, and N.r. Barkov, 2003. Historical CO2 record from the Vostok ice core. In Trends: A Compendium ofData on Global Change. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department ofEnergy, Oak Ridge, Tenn., U.S.A

Bates, N.R, 2001. Interannual variability ofoceanic C02 and biogeochemical properties in the Western North Atlantic subtropical gyre. Deep-Sea Research II, 48: 1507­ 1528.

Bates, N.R., Michaels, AP., and Knap, AH., 1996. Seasonal and interannual variability ofthe oceanic carbon dioxide system at the U.S. JGOPS Bermuda Atlantic Time­ series site. Deep-Sea Research II, 43: 347-383.

178 Bathurst, RG.C., 1975. Carbonate sediments and their diagenesis: Developments in sedimentology, vol 12, Elseiver, New York, 620 p.

Ben-Yaakov, S., 1973. pH buffering ofpore water ofrecent anoxic marine sediments. Limnology and Oceanography, 18: 86-94.

Berger, W. H., 1976. Biogenous deep sea sediments: production, preservation, and interpretation, in Riley, J.P., and Chester, R., eds., Chemical Oceanography, vol 5, 2nd edition, Academic Press, London, pp. 266-388.

Bermuda Atlantic Time-Series (BATS) data archive. http://www.bbsr.edulcintoolbatslbats.html

Berner, RA., 1971. Principles ofChemical Sedimentology. McGraw-Hili, New York, 240p.

Berner, R.A., 1975. The role ofmagnesium in the crystal growth ofcalcite and aragonite from seawater. Geochimica et Cosmochimica Acta, 39: 489-504.

Berner, RA., 1980. Early diagenesis: A theoretical approach: Princeton, New Jersey, Princeton University Press, 241 p.

Berner, RA., 1982. Burial oforganic carbon and pyrite sulfur in the modem ocean: Its geochemical significance. American Journal ofScience, 282: 451-473.

Berner, R.A., 1987. Models for carbon and sulfur cycles and atmospheric oxygen: Application to Paleozoic geologic history. American Journal ofScience, 287: 177-196.

Berner, RA., 1989. Biogeochemical cycles ofcarbon and sulfur and their effect on atmospheric oxygen over Phanerozoic time. Palaeogeography Palaeoclimatology Palaeoecology, 75: 97-122.

Berner, RA., 1991. A model for atmospheric C02 over Phanerozoic time. American Journal ofScience, 291: 339-376.

Berner, RA., 1994. Geocarb II: A revised model ofatmospheric CO2over Phanerozoic time. American Journal ofScience, 294: 56-91.

Berner, R.A., and Kothavala, Z., 2001. Geocarb III: A revised model ofatmospheric CO2 over Phanerozoic time. American Journal ofScience, 301: 182-204.

Berner, RA., and Lasaga, A.C., 1989. Modeling the geochemical carbon cycle. Scientific American, 260: 74-81.

179 Berner, R.A., Lasaga, A.C., and Garrels, RM., 1983. The carbonate-silicate geochemical cycle and its effect on atmospheric carbon dioxide over the past 100 million years. American Journal ofScience, 283: 641-683.

Berner, RA, Westrich, J.T., Graber, R, Smith, J., and Martens, C.S., 1978. Inhibition of aragonite precipitation from supersaturated seawater: A laboratory and field study. American Journal ofScience, 278: 816-837.

Bertram, M.A., Mackenzie, F.T., Bishop, F.e., and Bischoff, W.D., 1991. Influence of temperature on the stability ofmagnesian calcite. American Mineralogy, 76: 1889-1896.

Bijma, J., Spero, H.I., and Lea, D.W., 1999. Reassessing foraminiferal stable isotope geochemistry: Impacts ofthe oceanic carbonate system (experimental results), in Fisher, G., and Wefer, G., eds., Use ofProxies in Paleoceanography: Example from the South Atlantic, Springer-Verlag, New York, pp. 489-512.

Bischoff, W.D., Bertram, M.A., Mackenzie, F.T., and Bishop, F.C., 1993. Diagenetic stabilization pathways ofmagnesian calcites. Carbonates and Evaporites, 8: 82­ 89.

Bischoff, W.D., Mackenzie, F.T., and Bishop, F.C., 1987. Stabilities ofsynthetic magnesian calcites in aqueous solution: comparison with biogenic materials. Geochimica et Cosmochimica Acta, 51: 1413-1423.

Boehme, S.E., Sabine, C.L., and Reimers, C.E., 1998. CO2 fluxes from a coastal transect: a time-series approach. Marine Chemistry, 63: 49-67.

Borowitzka, M.A., 1981. Photosynthesis and calcification in the articulates coralline red algae Amphiroa anceps and A. foliacea. Marine Biology, 62: 17-23.

Boss, S.K., and Neumann, AC., 1993. Physical versus chemical processes of"whiting" formation in the Bahamas. Carbonate and Evaporites, 8: 135-148.

Boucher, G., Clavier, J., Hily, C., and Gattuso, J.-P., 1998. Contribution ofsoft-bottoms to the community metabolism (primary production and calcification) ofa barrier reefflat (Moorea, French Polynesia). Journal ofExperimental Marine Biology and Ecology, 225: 269-283.

Brink, K.H., Abrantes, F.F.G., Bernal, P.A, Dugdale, RC., Estrada, M., Hutchings, L., Jahnke, R.A., Milller, P.J., and Smith, RL., 1995. Group report: How do coastal upwelling systems operate as integrated physical, chemical, and biological systems and influence the geological record? The role ofphysical processes in defining the spatial structures ofbiological and chemical variables, in

180 Summerhayes, C.P., Emeis, K.-C., Angel, M.V., Smith, R.L., and Zeitzschel, B., eds., Upwelling in the ocean: Modem processes and ancient records, John Wiley and Sons, England, pp. 103-124.

Broecker, W.S., and Peng, T.H., 1982. Tracers in the sea. Lamont-Doherty Geological Observatory, Columbia University.

Broecker, W.S., Sanyal, A., and Takahashi, T., 2000. The origin ofBahamian whitings revisited. Geophysical Research Letters, 27(22): 3759-3760.

Brown, B.E., Dunne, R.P., Goodson, M.S., and Douglas, A.E., 2002. Coral Reefs, 21: 119-126.

Buddemeier, R.W., and Fautin, D.G., 1993. Coral bleaching as an adaptive mechanisms: a testable hypothesis. Bioscience, 43: 320-326.

Burton, E.A., and Walter, L.M., 1990. The role ofpH in phosphate inhibition ofcalcite and aragonite precipitation rates in seawater. Geochimica et Cosmochirnica Acta, 54: 797-808.

Busenberg, E., and Plummer, N.L., 1989. Thermodynamics ofmagnesian calcite solid solution-solutions at 25°C and 1 atm total pressure. Geochimica et Cosmochimca Acta, 53: 1189-1208.

Caillon, N., Severinghaus, J.P., Jouzel, J., Barnola, J.-M., Kang, J., and Lipenkov, V.Y., 2003. Timing ofatmospheric CO2 and Antarctic temperature changes across termination III. Science, 299: 1728-1731.

Chave, K.E., 1954. Aspects ofthe biogeochemistry ofmagnesium 1. Calcareous sediments and rocks. The Journal ofGeology, 62: 266-283.

Chave, K.E., 1954. Aspects ofthe biogeochemistry ofmagnesium 2. Calcareous sediments and rocks. The Journal ofGeology, 62: 587-599.

Chave, K.E., 1962. Factors influencing the mineralogy ofcarbonate sediments. Limnology and Oceanography, 7: 218-223.

Chave, K.E., 1967. Recent carbonate sediments: An unconventional view. Journal of Geolocial Education, 15: 200-204.

Chave, K.E., and Suess, E., 1967. Suspended minerals in seawater. Trans. N.Y. Acad. Sci. Ser. II, 29: 991-1000.

181 Chavez, F.P., and Barber, RT., 1987. An estimate ofnew production in the equatorial Pacific. Deep-Sea Research, 34: 1229-1243.

Chavez, F.P., and Toggweiler, J.R, 1995. Physical estimates ofglobal new production: The upwelling contribution, in Summerhayes, C.P., Emeis, K.-c., Angel, M.V., Smith, R.L., and Zeitzschel, B., eds., Upwelling in the ocean: Modem processes and ancient records, John Wiley and Sons, England, pp. 313-320.

Chen, C.-T. A., Rodman, M.R, Wei, C.-L., Olson, EJ., Feely, RA., and Gendron, J.F., 1986. Carbonate chemistry ofthe North Pacific Ocean, U.S. Department of Energy, Washington D.C., 176 p.

Clausen, C.D., and Roth, A.A., 1975. Effect oftemperature and temperature adaptation on calcification rate in the hermatypic coral Pocillopora damicornis. Marine Biology, 33: 93-100.

Cloud Jr., P.E., 1962. Environment ofcalcium carbonate deposition west ofAndros Island, Bahamas. U.S. Geological Survey Professional Papers, 350, 138 p.

Cook, C.B., Logan, A., Ward, J., Luckhurst, B., and Berg, C.J. Jr., 1990. Elevated temperatures and bleaching on a high latitude coral reef: the 1988 Bermuda event. Coral Reefs, 9: 45-49.

Crick, M.J., Hill, M.D., and Charles, D., 1987. The role ofsensitivity analysis in assessing uncertainty, in Proceedings ofan NEA Workshop on Uncertainty Analysis for Performance Assessment ofRadioactive Waste Disposal Systems, Paris, OECD, pp. 1-258.

Crowley, TJ., and Berner, RA., 2001. CO2 and climate change. Science, 292: 870-872.

De Groot, K., 1965. Inorganic precipitation ofcalcium carbonate from sea-water. Nature, 207: 404-405.

Devol, A.H., and Ahmed, S.L, 1981. Are high rates ofsulphate reduction associated with anaerobic oxidation ofmethane? Nature, 291: 407-408.

Dove, P.M., and Hochella Jr., M.F., 1993. Calcite precipitation mechanisms and inhibition by orthophosphate: In situ observations by Scanning Force Microscopy. Geochimica et Cosmochimica Acta, 57: 705-714.

Dubus, LG., and Brown, C.D., 2002. Organic compounds in the environment: sensitivity and first-step uncertainty analyses for the preferential flow model MACRO. Journal ofEnvironmental Quality, 31: 227-240.

182 Fernandez, E., Maraft6n, E., Harbour, D.S., Kristiansen, S., and Heimdal, B.R, 1996. Patterns ofcarbon and nitrogen uptake during blooms ofEmiliania huxleyi in two Norwegian fjords. Journal ofPlankton Research, 18(12): 2349-2366.

Frankignoulle, M., Abril, G., Borges, A., Bourge, I., Canon, C., DeLille, B., Libert, E., and Theate, J.-M., 1998. Carbon dioxide emission from European estuaries. Science, 282; 434-436.

Frankignoulle, M., and Borges, A.V., 2001. European continental shelfas a significant sink for atmospheric carbon dioxide. Global Biogeochemical Cycles, 15: 569-576.

Frankignoulle, M., Canon, C., and Gattuso, J.-P., 1994. Marine calcification as a source ofcarbon dioxide: Positive feedback ofincreasing atmospheric CO2. Limnology and Oceanography, 39: 458-462.

Frankignoulle, M., Gattuso, J.-P., Biondo, R., Bourge, I., Copin-Montegut, G., and Pichon, M., 1996. Carbon fluxes in coral reefs. II. Eulerian study ofinorganic carbon dynamics and measurements ofair-sea CO2 exchange. Marine Ecology Progress Series, 145; 123-132.

Friis-Christensen, E., and Lassen, K., 1991. Length ofthe solar cycle; an indicator of solar activity closely associated with climate. Science, 254; 698-700.

Gao, K, Aruga, Y., Asada, K, and Kiyohara, M., 1993. Influence ofenhanced CO2on growth and photosynthesis ofthe red algae Graciliaria sp. and G. chilensis. Journal ofApplied Phycology, 5: 563-571.

Gao, K, Aruga, Y., Asada, K, Ishihara, T., Akano, T., and Kiyohara, M., 1993. Calcification in the articulated coralline alga Coralline pilulifera. with special reference to the effect ofelevated C02 concentration. Marine Biology, 117: 129­ 132.

Gardner, RH., Huff, D.D., O'Neill, RV., Mankin, J.B., Carney, J., and Jones, J., 1980. Application oferror analysis to a marsh hydrology model. Water Resources Research, 16: 659·664.

Gardner, T.A., Cote, LM., Gill, J.A., Gran, A., and Watkinson, A.R., 2003. Long-term region-wide declines in Caribbean Corals. Science, 301: 958-960.

Garrels, R, and Wollast, R., 1978. Equilibrium criteria for two-component solids reacting with fixed composition in an aqueous phase - Example: the magnesian calcite. Discussion. American Journal ofScience, 278: 1469-1474.

183 Gates, R.D., 1990. Seawater temperature and sublethal coral bleaching in Jamaica. Coral Reefs, 8: 193-197.

Gattuso, J.-P., Allemand, D., and Frankignoulle, M., 1999a. Photosynthesis and calcification at cellular, organismal and community levels in coral reefs: A review on interactions and control by carbonate chemistry. American Zoology, 39: 160­ 183.

Gattuso, J.-P., Frankignoulle, M., and Smith, S.V., 1999b. Measurement ofcommunity metabolism and significance in the coral reefC02 source-sink debate. Proceedings ofthe National Academy ofSciences, 96: 13017-13022.

Gattuso, J.-P., Frankignoulle, M., Bourge, 1., Romaine, S., and Buddemeier, R.W., 1998. Effect ofcalcium carbonate saturation ofseawater on coral calcification. Global Planetary Change, 18: 37-46.

Gattuso, J.-P., Payri, C.E., Pichon, M., Delesalle, B., and Frankignoulle, M., 1997. Primary production, calcification, and air-sea C02 fluxes ofa macroalgal­ dominated coral reefcommunity (Moorea, French Polynesia). Journal of Phycology, 33: 729-738.

Gattuso, J.-P., Pichon, M., Delesalle, B., Canon, C., and Frankignoulle, M., 1996. Carbon fluxes in coral reefs. 1. Lagrangian measurement ofcommunity metabolism and resulting air-sea CO2disequilibrium. Marine Ecology Progress Series, 145: 109­ 121.

Glynn, P.W., 1993. Coral reefbleaching: ecological perspectives. Coral Reefs, 12: 1-17.

Grigg, R.W., 1981. Coral reefdevelopment at high latitudes in Hawaii, in Proceedings of the Fourth International Coral ReefSymposium, Manila, vol. 1, pp. 687-693.

Grigg, R.W., 1997. Paleoceanography ofcoral reefs in the Hawaiian-Emperor Chain­ revisited. Coral Reefs, 16: S33-S38.

Hales, B., and Emerson, S., 1997. Evidence in support offirst-order dissolution kinetics ofcalcite in seawater. Earth and Planetary Science Letters, 148: 317-327.

Halley, R.B., and Yates, K.K., 2000, Will reefsediments buffer corals from increased global C02, in Hopley, D., Hopley, M., Tamelander, J., and Done, T., Proceedings ofthe Ninth International Coral ReefSymposium Abstracts, State Ministry for the Environment, Indonesia, p. 248.

Hamby, D.M., 1994. A review oftechniques for parameter sensitivity analysis of environmental models. Environmental Monitoring and Assessment, 32: 135-151.

184 Hawaiian Ocean Time series (HOT) data archive. http://hahana.soest.hawaii.edulhotlhotjgofs.html

Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, x., Maskell, K., and Johnson, C.A., eds., 2001, Climate change 2001: The scientific basis. Contribution ofWorking Group I to the Third Assessment Report ofthe Intergovernmental Panel on Climate Change: Cambridge, Cambridge University Press, 881 p.

Houghton, J.T., Meiro Filho, L.G., Callander, B.A., Harris, N., Kattenberg, A., Maskell, K., eds., 1996. Climate change 1995: The science ofclimate change. Contribution ofworking group I to the Second Assessment Report ofthe intergovernmental Panel on Climate Change, Cambridge, Cambridge University Press, 572 p.

Iglesias-Rodriguez, M. D., Armstrong, R, Feely, R, Hood, R., Kleypas, J., Milliman, J.D., Sabine, C., and Sarmiento, J., 2002. Progress made in study ofocean's calcium carbonate budget. EOS, 83(34), 365: 374-375.

Jansen, H., Zeebe, RE., and Wolf-Gladrow, D.A., 2002. Modeling the dissolution of settling CaC03 in the ocean. Global Biogeochemical Cycles, 16: 11-1 -11-16.

Jones, RJ., Berkelman, R, and Oliver, J.K., 1997. Recurrent bleaching ofcorals at Magnetic Island (Australia) relative to air and seawater temperature. Marine Ecology Progress Series, 158: 289-292.

Jordan, T.E., and Weller, D.E., 1996. Human contributions to terrestrial nitrogen flux. BioScience, 46: 655-664.

Kasting, J.F., Toon, O.B., and Pollack, J.B., 1988. How climate evolved on the terrestrial planets. Scientific American, 258: 90-97.

Keeling, C.D. and T.P. Whorf. 2002. Atmospheric C02 records from sites in the SIO air sampling network. In Trends: A Compendium ofData on Global Change. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department ofEnergy, Oak Ridge, Tenn., U.S.A.

Keir, RS., 1980. The dissolution kinetics ofbiogenic calcium carbonates in seawater. Geochimica et Cosmochimica Acta, 44: 241-252.

Kleypas, J.A., Buddemeier, RW., and Gattuso, J.-P., 2001. The future ofcoral reefs in an age ofglobal change. International Journal ofEarth Sciences (Geol Rundsch), 90: 426-437.

185 Kleypas, J.A., Buddemeier, R.W., Archer, D., Gattuso, J.-P., Langdon, c., and Opdyke, B.N., 1999. Geochemical consequences ofincreased atmospheric carbon dioxide on coral reefs. Science, 284: 118-120.

Land, L.S., 1967. Diagenesis ofskeletal carbonates. Journal ofSedimentary Petrology, 37: 914-930.

Langdon, c., Broecker, W.S., Hannnond, D.E., Glenn, E., Fitzsimmons, K., Nelson, S.G., Peng, T.H., Hajdas, 1., and Bonani, G. Effect ofelevated COz on the community metabolism ofan experimental coral reef. Submitted to Global Biogeochemical Cycles, 2002.

Langdon, C., Takahashi, T., Sweeney, c., Chipman, D., Goddard, J., Marubini, F., Aceves, H., Barnett, H., and Atkinson, M., 2000. Effect ofcalcium carbonate saturation state on the calcification rate ofan experimental coral reef: Global Biogeochemical Cycles, 14: 639--654.

Laws, E., 1997. Mathematical methods for oceanographers: An introduction. John Wiley and Sons, Inc., New York, 343 p.

Lebron, 1., and Suarez, D.L., 1996. Calcite nucleation and precipitation kinetics as affected by dissolved organic matter at 25°C and pH > 7.5. Geochimica et Cosmochimica Acta, 60: 2765-2776.

Lebron, 1., and Suarez, D.L., 1998. Kinetics and mechanisms ofprecipitation ofcalcite as affected by pCOz and organic ligands at 25°C. Geochimica et Cosmochimica Acta, 62: 405-416.

Leclercq, N., Gattuso, J.-P., and Jaubert, J., 2000, COz partial pressure controls the calcification rate ofa coral community. Global Change Biology, 6: 329-334.

Leclercq, N., Gattuso, J.-P., and Jaubert, J., 2002. Primary production, respiration, and calcification ofa coral reefmesocosm under increased COz partial pressure. Limnology and Oceanography, 47: 558-564.

Lerman, A., Mackenzie, F.T., and Geiger, RJ., 1989. Environmental chemical stress effects associated with the carbon and phosphorus biogeochemical cycles, in Levin, S.A., Harwell, l.R., Kelly, and Kimball, K.D., eds., Ecotoxicology: Problems and Approaches, Springer-Verlag, p. 315-350.

Lesser, M.P., 1997. Oxidative stress causes coral bleaching during exposure to elevated temperatures. Coral Reefs, 16: 187-192.

186 Lesser, M.P., Stochaj, W.R., Tapley, D.W., and Shick, J.M., 1990. Bleaching in coral reef anthozoans: effects ofirradiance, ultraviolet radiation, and temperature on the activities ofprotective enzymes against active oxygen. Coral Reefs, 8: 225-232.

Lewis, E., and Wallace, D.W.R., 1998. Program Developed for C02 System Calculations. ORNUCDIAC-I05. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department ofEnergy, Oak Ridge, Tennessee.

Likens, G.E., Mackenzie, F.T., Richey, J.E., Sedell, J.R, and Turekian, K.K., 1981. Flux oforganic carbon by rivers to the oceans. U.S. Department ofenergy, Washington D.C., 397p.

Lisitzin, A.P., 1996. Oceanic sedimentation: lithology and geochemistry. American Geophysical Union, Washington, p. 400.

Lough, J.M., and Barnes, DJ., 1997. Several centuries ofvariation in skeletal extension, density and calcification in massive Porites colonies from the Great Barrier Reef: A proxy for seawater temperature and a background ofvariability against which to identify unnatural change. Journal ofExperimental Marine Biology, 211: 29­ 67.

Lough, J.M., and Barnes, DJ., 2000. Environmental control on the massive coral Porites. Journal ofExperimental Marine Biology and Ecology, 245: 225-243.

Lough, J.M., Barnes, D.J., Devereux, M.l, Tobin, BJ., and Tobin, S., 1999. Variability in growth characteristics ofmassive Porites on the Great Barrier Reef, Technical Report No. 28, CRC ReefResearch Centre, Townsville, 95 p.

Mackenzie, F. T., Wollast, R., Schoonmaker, J., and Garrels, RM., 1980. Fast reactions ofmagnesian calcite in surface seawater. Unpublished article.

Mackenzie, F.T., 1987. Global mixed-layer natural and anthropogenic fluxes, in Dynamics ofthe Ocean Surface Mixed layer, 291-310.

Mackenzie, F.T., and Agegian, C.R, 1989, Biomineralization and tentative links to plate tectonics, in Crick, RE., ed., Origin, evolution, and modem aspects of biomineralization in plants and animals: New York, Plenum Press, p. 11-27.

Mackenzie, F.T., Bischoff, W.D., Bishop, F.C., Loijens, M., Schoonmaker, J., and Wollast, R, 1983. Magnesian calcites: Low temperature occurrence, solubility and solid-solution behavior, in Reeder, RJ., ed., Carbonates: mineralogy and chemistry, Mineralogical Society ofAmerica, Reviews in Mineralogy, p. 97-143.

187 Mackenzie, F.T., Lennan, A., and Ver, L.M., 1998a. Role ofcontinental margin in the global carbon balance during the past three centuries. Geology, 26: 423-426.

Mackenzie, F.T., Lennan, A., Ver, L.M.B., 2001. Recent past and future ofthe global carbon cycle, in Gerhard, L.C., Harrison, W.E., Hanson, B.M., eds., Geological Perspectives ofGlobal Climate Change, p. 51-82.

Mackenzie, F.T., Ver, L.M., and Lennan, A., 1998b. Coupled biogeochemical cycles of carbon, nitrogen, phosphorus, and sulfur in the land-ocean-atmosphere system, in Galloway, J.N., and Melillo, J.M., Eds., Asian Change in the Context ofGlobal Change, Cambridge University Press, Cambridge, p. 42-100.

Mackenzie, F.T., Ver, L.M., and Lennan, A., 2000. Coastal zone biogeochemical dynamics under global warming: International Geology Review, 42: 193-206.

Mackenzie, F.T., Ver, L.M., and Lennan, A., 2000. Coastal-zone biogeochemical dynamics under global warming. International Geology Review, 42: 193-206.

Mackenzie, F. T., Ver, L.M., and Lennan, A., 2002. Century-scale nitrogen and phosphorus controls ofthe carbon cycle. Chemical Geology, 190: 13-32.

Marland, G., T.A. Boden, and R. J. Andres. 2002. Global, Regional, and National Fossil Fuel CO2 Emissions. In Trends: A Compendium ofData on Global Change. Carbon Dioxide Infonnation Analysis Center, Oak Ridge National Laboratory, U.S. Department ofEnergy, Oak Ridge, Tenn., U.S.A.

Marubini, F., and Thake, B., 1999. Bicarbonate addition promotes coral growth. Limnology and Oceanography, 44: 716-720.

McConnaughey, T.A., Adey, W.H., and Small, A.M., 2000. Community and environmental influences on reefcoral calcification. Limnology and Oceanography, 45(7): 1667-1671.

McConnaughey, T.A., and Whelan, J.F., 1997. Calcification generates protons for nutrient and bicarbonate uptake. Earth-Science Reviews, 42: 95-117.

Meybeck, M., 1979. Concentration des eaux fluviales en elements majeurs et apports en solution aux oceans. Revue Geo!. Dynamique Geog. Phys., 21(3): 215-246.

Meybeck, M., 1981. River transport oforganic carbon to the ocean, in Flux ofOrganic Carbon by Rivers to the Ocean, U.S. Department ofEnergy, Office Energy Research, Washington D.C., p. 219-269.

188 Meybeck, M., 1982. Carbon, nitrogen and phosphorus transport by world rivers. American Journal ofScience, 282: 401-450.

Milliman, J.D., 1974, Recent sedimentary carbonates, Part 1: Marine carbonates: Berlin, Springer-Verlag, 375 p.

Milliman, J.D., 1993. Production and accumulation ofcalcium carbonate in the ocean: budget ofa nonsteady state. Global Biogeochemical Cycles, 7: 927-957.

Milliman, J.D., and Droxler, A.W., 1996. Neritic and Pelagic Carbonate Sedimentation in the Marine Environment: Ignorance is not Bliss. Geologische Rundschau 85: 496­ 504.

Milliman, J.D., Troy, P.I., Balch, W.M., Adams, A.K., Li, Y.-H., and Mackenzie, F.T., 1999. Biologically mediated dissolution ofcalcium carbonate above the chemical lysocline? Deep-Sea Research I, 46: 1653-1669.

Morse, J.W., 1983. The kinetics ofcalcium carbonate dissolution and precipitation, in Reeder, RJ., ed., Carbonates: mineralogy and chemistry, Mineralogical Society of America, Reviews in Mineralogy, p. 227-264.

Morse, J.W., and Arvidson, RS., 2001. The dissolution kinetics ofmajor sedimentary carbonate minerals. Earth-Science Reviews, 58: 51-84.

Morse, J.W., and Mackenzie, F.T., 1990, Geochemistry ofsedimentary carbonates: Amsterdam, Elsevier, 707 p.

Morse, lW., Gledhill, D.K., and Millero, F.J., 2003. CaC03 precipitation kinetics in waters from the Great Bahama Bank: Implications for the relationships between Bank hydrochemistry and whitings. Geochimica et Cosmochimica Acta, 67(15): 2819-2826.

Morse, lW., Millero, FJ., Thurmond, V., Brown, E., and Ostlund, H.G., 1984. The chemistry ofGrand Bahama Bank waters: After 18 years another look. Joumal of Geophysical Research, 89: 3604-3614.

Morse, J.W., Zullig, J.J., Bernstein, L.D., Millero, FJ., Milne, P., Mucci, A., and Choppin, G.R., 1985. Chemistry ofcalcium carbonate-rich shallow water sediments in the Bahamas. American Journal ofScience, 285: 147-185.

Moulin, E., Iordens, A., and Wollast, R, 1985. Influence ofthe aerobic bacterial respiration on the early dissolution ofcarbonates in coastal sediments. Proceedings Progress in Belgium Oceanographic Research, Brussels, p. 196-208.

189 Mucci, A., 1986. Growth kinetics and composition ofmagnesian calcite overgrowths precipitated from seawater. Quantitative influence oforthophosphate ions: Geochimica et Cosmochimica Acta, 50: 2255-2265.

Murray, J.W., 1992. The oceans, in Butcher, S.S., Charlson, R.I., Orians, G.H., and Wolfe, G.V., eds., Global Biogeochemical Cycles, Academic Press, p. 175-211.

Nagy, K.L., 1988. The solubility ofcalcite in NaCl and Na-Ca-CI brines. Ph.D. dissertation, Texas A&M University.

Neumann, A.C., 1965. Processes ofrecent carbonate sedimentation in Harrington sound, Bermuda. Bulletin ofMarine Science, 15(4): 987-1035.

Ohde, S., and van Woesik, R., 1999. Carbon dioxide flux and metabolic processes ofa coral reef, Okinawa. Bulletin ofMarine Science, 65: 559-576.

Pelegri, lL., and Csanady, G.T., 1991. Nutrient transport and mixing in the GulfStream. Journal ofGeophysical Research, 96: 2577-2583.

Petit, J.R., Jouzel, J., Raynaud, D., Barkov, N.L, Bamola, J.-M., Basile, L, Benders, M., Chappellaz, J., Davis, M., Delaygue, G., Delmotte, M., Kotlyakov, V.M., Legrand, M., Lipenkov, Y.Y., Lorius, C., Pepin, L., Ritz, C., Saltzman, E., and Stievenard, M., 1999. Climate and atmospheric history ofthe past 420,000 years from the Vostok ice core, Antarctica. Nature, 399: 429-436.

Pilson, M.E., 1998. An introduction to the chemistry ofthe sea. Prentice Hall, Upper Saddle River, NJ, 431 p.

Plummer, L.N., and Busenberg, E., 1982. The solubilities ofcalcite, aragonite and vaterite in CO2-H20 solutions between o-90°C and an evaluation ofthe aqueous model for the system CaC03-C02-H20. Geochimica et Cosmochimica Acta, 46: 1011-1040.

Plummer, L.N., and Mackenzie, F.T., 1974. Predicting mineral solubility from rate data: Application to the dissolution ofmagnesian calcites. American Journal of Science, 274: 61-83.

Pytkowicz, R.M., 1965. Rates ofinorganic calcium carbonate nucleation. l Geo!. 73: 196-199.

Pytkowicz, R.M., 1965. Rates ofinorganic calcium carbonate nucleation. Journal of Geology, 73: 196-199.

190 Rabouille, C., Mackenzie, F.T., Ver, L.M., 2001. Influence ofthe human perturbation on carbon, nitrogen, and oxygen biogeochemical cycles in the global coastal ocean. Geochemica et Cosmochimica Acta, 65(21): 3615-3641.

Rao, C.P., 1996. Modem carbonates: tropical, temperate, polar. Introduction to sedimentology and geochemistry, University ofTasmania, Hobart, Tasmania, Australia, p. 206.

Raven, J.A., 1993. Limits on growth rates. Nature, 361: 209-210.

Raven, J.A., 1997. Inorganic carbon acquisition by marine autotrophs. Advances of Botanical Research, 27: 85-209.

Reeburgh, W.S., 1980. Anaerobic methane oxidation: Rate depth distribution in Skan Bay sediments. Earth and Planetary Science Letters, 47: 345-352.

Reeder, RJ., 1983. Crystal chemistry ofthe rhombohedral carbonates, in Reeder, RJ., Carbonates: Mineralogy and Chemistry, Mineralogical Society ofAmerica, Reviews in Mineralogy, 11: 1-47.

Revelle, R., and Munk, W., 1977. The carbon dioxide cycle and the biosphere, in Energy and Climate (eds. N.G.S. committee), National Academy Press, p. 140-158.

Revelle, R., and Suess, H.E., 1957. Carbon dioxide exchange between atmosphere and ocean and the question ofan increase ofatmospheric CO2 during the past decades. Tellus, 9: 18-27.

Riebesell, D., Wolf-Gladrow, D., and Smetack, V., 1993. Carbon dioxide limitation of phytoplankton growth rates. Nature, 361: 249-251.

Riebesell, D., Zondervan, I., Rost, B., Tortell, P.D., Zeebe, R.E., and Morel, F.M.M., 2000. Reduced calcification ofmarine plankton in response to increased atmospheric CO2• Nature, 407: 364-367.

Ritger, S., Carson, B., and Suess, E., 1987. Methane-derived authigenic carbonates formed by subduction-induced pore-water expulsion along the OregonlWashington margin. Geol. Soc. Amer. Bull., 98: 147-156.

Robbins, L.L., Tao, Y., and Evans, C.A., 1997. Temporal and spatial distribution of whitings: A biologically induced calcium carbonate precipitation mechanisms. Geology, 25: 947-950.

Rowan, R., Knowlton, N., Baker, A.C., and Jara, J., 1997. Landscape ecology ofalgal symbionts creates variation in episodes ofcoral bleaching. Nature, 388: 265-269.

191 Roy, RN., Roy, L.N., Vogel, K.M., Porter-Moore, C., Pearson, T., Good, C.E., Millero, FJ., and Campbell, D.M., 1993. The dissociation constants ofcarbonic acid in seawater at salinities 5 to 45 and temperatures 0 to 45°C. Marine Chemistry, 44: 249-267.

Sabine, C.L., and Mackenzie, F.T., 1995. Bank-Derived Carbonate Sediment Transport and Dissolution in the Hawaiian Archipelago. Aquatic Geochemistry, 1: 189-230.

Schlesinger, W.H., 1997. Biogeochemistry: An analysis ofglobal change, 2nd ed. Academic Press, San Diego, Ca, 588 p.

Schmalz, RF., and Chave, K.E., 1963. Calcium carbonate: Affecting saturation in ocean waters ofBermuda. Science, 139:1206-1207.

Scoffin, T.P., Tudhope, A.W., Brown, RE., Chansang, H., and Cheeney, RF., 1992. Patterns and possible environmental controls ofskeletogenesis ofPorites lutea, South Thailand. Coral Reefs 11: 1-11.

Shaviv, N.J. and Veizer, J., 2003. Celestial driver ofPhanerozoic climate? GSA Today, 13(7): 4-10.

Shinn, E.A., Steinen, RP., Lidz, B.H., and Swart, P.K., 1989. Whitings, a sedimentological dilemma. Journal of Sedimentary Petrology, 59: 147-161.

Sikes, C.S., and Fabry, J., 1994. Photosynthesis, CaC03 deposition, coccolithophorids, and the global carbon cycle, in Tolbert, N.E., and Preiss, J., eds., Regulation of Atmospheric CO2 and 02 by Photosynthetic Carbon Metabolism, Oxford University Press, New York, p. 217-233.

Sikes, C.S., and Wilbur, K.M., 1982. Functions ofcoccolith formation. Limnology and Oceanography, 27(1): 18-26.

Sikes, C.S., Roer, RD., and Wilbur, K.M., 1980. Photosynthesis and coccolith formation: Inorganic carbon sources and net inorganic reaction ofdeposition. Limnology and Oceanography, 25(2): 248-261.

Smith, A.D., and Roth, A.A., 1979. Effect ofcarbon dioxide concentration on calcification in the red coralline alga Bossiella orbigniana: Marine Biology, 52: 217-225.

Smith, S.V., 1972. Production ofcalcium carbonate on the mainland shelfofsouthern California. Limnology and Oceanography, 17: 28-41.

192 Smith, S.Y., and Buddemeier, R.W., 1992. Global change and coral reefecosystems. Annual Review ofEcological Systems, 23: 89-118.

Smith, S.Y., and Hollibaugh, J.T., 1993. Coastal metabolism and the oceanic organic carbon balance. Reviews ofGeophysics 31(1): 75-89.

Smith, S.Y., and Mackenzie, F.T., 1987. The ocean as a net heterotrophic system: Implications from the carbon biogeochemical cycle. Global Biogeochemical Cycles I: 187-198.

Solanki, S.K., 2002. Solar variability and climate change: is there a link? Astron. Geophys., 43: 5.9-5.13.

Stanley, S.M., Ries, J.B., and Hardie, L.A., 2002. Low-magnesium calcite produced by coralline algae in seawater ofLate Cretaceous composition. Proceedings ofthe National Academy ofSciences, 99: 15323-15326.

Starke, L. (Ed.), 2003. The Worldwatch Institute, State ofthe World 2003. Norton and Company, NewYork, NY, 241p.

Svensmark, H., 1998. Influence ofcosmic rays on Earth's climate. Physical Review Letters, 81: 5027-5030.

Takahashi, T., Broecker, W.S., Brainbridge, A.E., and Weiss, R.F., 1980. Carbonate chemistry ofthe Atlantic, Pacific and Indian Oceans: The results ofthe GEOSECS expeditions, 1972-1978, Technical Report No.1., Lamont-Doherty Geological Observatory, Palisades, N.Y.

Thorstenson, D.C., and Mackenzie, F.T., 1974. Time variability ofpore water chemistry in recent carbonate sediments, Devil's Hole, Harrington Sound, Bermuda. Geochimica et Cosmochimica Acta, 38: 1-19.

Tomczak, M., and Godfrey, J.S., 1994. Regional Oceanography: An introduction. Elseiver science, Oxford, England, 422 p.

Tribble, J.S., and Mackenzie, F.T., 1998. Recrystallization ofmagnesian calcite overgrowths on calcite seeds suspended in seawater. Aquatic Geochemistry, 4: 337-360.

Tribble, J.S., Arvidson, R.S., Lane, M., III, and Mackenzie, F.T., 1995. Crystal chemistry, and thermodynamic and kinetic properties ofcalcite, dolomite, apatite and biogenic silica: Applications to petrologic problems. Sedimentary Geology, 95: 11-37.

193 Ver, L.M., 1998. Global kinetic models ofthe coupled C,N,P, and S biogeochemical cycles: Implications for global environmental change. PhD thesis oceanography, University ofHawaii.

Ver, L.M., Mackenzie, F.r., and Lennan, A., 1994. Modeling pre-industrial C-N-P-S biogeochemical cycling in the land-coastal margin system. Chemosphere, 29: 855-887.

Ver, L.M., Mackenzie, F.r., and Lennan, A., 1998. Partitioning ofanthropogenic CO2 by C-N-P-S cycle coupling on land and in coastal oceanic environment. Mineralogical Magazine, 62A: 1589-1590.

Ver, L.M., Mackenzie, F.r., and Lerman, A., 1999a. Biogeochemical responses ofthe carbon cycle to natural and human perturbations: Past, present, and future. American Journal ofScience, 299: 762-801.

Ver, L.M., Mackenzie, F.r., and Lennan, A., 1999b. Carbon cycle in the coastal zone: Effects ofglobal perturbations and change in the past three centuries. Chemical Geology, 159: 289-304.

Vitousek, P.M., 1994. Beyond global warming: Ecology and global change. Ecology, 75: 1861-1876.

Walsh, J.J., 1991. Importance ofcontinental margins in the marine biogeochemical cycling ofcarbon and nitrogen. Nature, 350: 53-55.

Walter, L.M., and Morse, J.W., 1984. Reactive surface area ofskeletal carbonate during dissolution: Effect ofgrain size. Journal ofSedimentary Petrology, 54: 1081­ 1090.

Walter, L.M., and Morse, J.W., 1985. The dissolution kinetics ofshallow marine carbonates in seawater: A laboratory study. Geochimica et Cosmochimica Acta, 49: 1503-1513.

Ware, J.R., Smith, S.V., and Reaka-Kudla, M.L., 1992. Coral reefs: Sources or sinks of atmospheric CO2? Coral reefs, 11: 127-130.

Wells, AJ., and Illing, L.V., 1964. Present-day precipitation ofcalcium carbonate in the Persian Gulf, in van Staaten, L.M., ed., Deltaic and Shallow Marine Deposits: Developments in sedimentology, vol 1, Elseiver, New York, p. 429-435.

Williams, PJ.L., 1975. Biological and chemical aspects ofdissolved organic matter in seawater, in Riley, J.P. and Skirrow, G., eds., Chemical Oceanography, Vol. 2, Academic Press, p. 301-364.

194 Winn, C.D., Li, Y.-H., Mackenzie, F.T., and Karl, D.M., 1998. Rising surface ocean dissolved inorganic carbon at the Hawaii Ocean Time-series site, Marine Chemistry, 60: 33-47.

Wollast, R, 1994, The relative importance ofbiomineralization and dissolution ofCaC03 in the global carbon cycle, in Doumenge, F., ed., Past and present biomineralization processes: Considerations about the carbonate cycle: Monaco, Bulletin de l'Institut oceanographique, numero special 13, p. 13-34.

Wollast, R, 1998. Evaluation and comparison ofthe global carbon cycle in the coastal zone and in the open ocean, in Brink, K.H., and Robinson, A.R., eds., The Sea, The global coastal ocean, processes and methods, Wiley, New York, p. 213-252.

Wollast, R, and Mackenzie, F.T., 1989. Global biogeochemical cycles and climate, in Berger, A., Schneider, S., and Duplessy, J.C., Eds., Climate and Geo-Sciences, Kluwer Academic Publishing, Dordrecht, The Netherlands, p. 453-473.

Wollast, R, Garrels, RM., and Mackenzie, F.T., 1980. Calcite-seawater reactions in ocean surface waters. American Journal ofScience, 280: 831-848.

Wyrtki, K., 1963. The horizontal and vertical field ofmotion in the Peru current. Bulletin of Scripps Institution ofOceanography, 8: 313-346.

Zeebe, RE., and Wolf-Gladrow, D., 2002. C02 in seawater: equilibrium, kinetics, isotopes, Elseiver, Amsterdam, 346 p.

2 Zhang, Y., and Dawe, R.A., 2000. Influence ofMg + on the kinetics ofcalcite precipitation and calcite crystal morphology. Chemical geology, 163: 129-138.

Zhong, S., and Mucci, A., 1989. Calcite and aragonite precipitation from seawater solutions ofvarious salinities: Precipitation rates and overgrowth compositions. Chemical Geology, 78: 283-299.

Zondervan, 1., Zeebe, RE., Rost, B., Riebesell, D., 2001. Decreasing marine biogenic calcification: A negative feedback on rising atmospheric pC02• Global Biogeochemical Cycles, 15(2): 507-516.

195