Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Assessing the "ballast" hypothesis for carbon transport in the ocean : global sediment trap data analysis… Izumi, Ryusuke 2010

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-UBC_2010_fall_izumi_ryusuke.PDF [ 2.77MB ]
Metadata
JSON: 24-1.0053295.json
JSON-LD: 24-1.0053295-ld.json
RDF/XML (Pretty): 24-1.0053295-rdf.xml
RDF/JSON: 24-1.0053295-rdf.json
Turtle: 24-1.0053295-turtle.txt
N-Triples: 24-1.0053295-rdf-ntriples.txt
Original Record: 24-1.0053295-source.json
Full Text
24-1.0053295-fulltext.txt
Citation
24-1.0053295.ris

Full Text

ASSESSING THE “BALLAST” HYPOTHESIS FOR CARBON TRANSPORT IN THE OCEAN: GLOBAL SEDIMENT TRAP DATAANALYSIS AND SIMULATION IN AN EARTH SYSTEM MODEL by  RYUSUKE IZUMI B.A., Keio University, 1999 B.A., University of Hawaii at Hilo, 2003  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Oceanography)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  April 2010 © Ryusuke Izumi, 2010  Abstract Sediment trap particle flux data analysis and development of a model representation of mineral “ballast” mechanism for carbon transport in the ocean is presented in this study. The validity of several classical POC remineralization models as well as recently hypothesized “ballast” mechanism based POC remineralization models were tested by analyzing data from selected 79 sediment traps at >1500 m from around the world. POC flux modelled with different variations of model representations at each sediment trap site was statistically compared with corresponding measured POC flux in order to evaluate the overall predictability of each model at the global scale. A CaCO 3 single-mineral-ballast model could explain up to —79% of the global F variability at depth >1500 m and suggests that CaCO 3 may potentially be the mineral type that has dominant control on the vertical transport of Fc,c from sea surface to depth in the open ocean. In addition, ai assessment of the impact of reduced CaCO 3 production (as a result of ocean surface acidification) on the marine carbon cycle and implications for future atmospheric CO 2 concentration under the assumption of mineral ballasting of POC is presented. A CaCO 3 single-mineral-ballast model derived from the data analysis is incorporated into GENIE-i, a computationally efficient carbon-climate Earth System Model of intermediate complexity. Simulation results from a “business as usual” future carbon emissions scenario in GENIE-i suggest that, by year 2300, calcification response of marine calcifying organisms to increased atmospheric CO 2 concentrations in a CaCO -ballasting 3 ocean is —63% weaker compared to that in a non-ballasting ocean. With the “ballast” effect in operation, the net effect of climate feedback and calcification feedback is a global decrease in POC export production, except for in some high latitude regions where enhanced POC production due to decreased sea-ice coverage overrides. If the “ballast” hypothesis is true, a CaCO -ballasting mechanism could completely counter the reduction in 3 atmospheric CO 2 concentration by calcification feedback alone in an ocean where no ballasting mechanism is present.  11  Table of Contents Abstract Table of Contents List of Tables List of Figures List of Abbreviations Acknowledgements Dedication I Introduction 1.1 POC:PIC “rain ratio” and “carbonate compensation 1.2 POC remineralizaion 1.3 CaCO 3 dissolution 1.4 “Ballast” hypothesis 1.5Summary 2 Data Analysis 2.1 Data 2.2 Classical models 2.2.1 PP-based model 2.2.2 EP-based model 2.3 Ballast models 2.3.1 Single-ballast model 2.3.2 Multiple-ballast model 2.4 TE-based analysis 2.4.1 TE-based single-ballast model 2.4.2 TE-based multiple-ballast model 2.5 Summary 3 Modeling in GENIE-I 3.1 GENIE-i 3.1.1 POC flux in the Non-ballasting Ocean -ballasting Ocean 3 3.1.2 CaCO 3 flux in the Non-ballasting Ocean and the CaCO -ballasting Ocean 3 3.1.3 POC flux in the CaCO 3.1.4 Model experiment setup and permutations 3.i.5NBOvsCBO 3.2 Non-ballasting Ocean 3.2.1 NBC ‘no feedback’ 3.2.2 NBC ‘+climate’ 3.2.3 NBC ‘i-calcification’ 3.2.4 NBC ‘i-climate i-calcification’ -ballasting Ocean 3 3.3 CaCO 3.3.1 CBC +climate’ 3.3.2 CBO ‘i-calcification’ 3.3.3 CBC ‘i-climate +calcification’ 3.4 Multiple-ballasting Ocean 3.5 Summary 4 Discussion 4.1 GENIE-i simulation results 4.1.1 NBC ‘i-climate’ 4. 1. 2 NBC ‘+calcifiation’ 4.1.3 CBC ‘i-climate’ 4.1.4 CBO ‘i-calcification’ 4.1.5 CBO ‘+climate +calcification’ 4.1.6 Multiple-ballasting Ocean -ballasting Ocean 3 4.1.7 Non-ballasting Ocean vs CaCC 9 3 dissolution 4.2 Shallower CaCO  .  ii iii V  Vi Viii  ix x 1 1 2 3 4 5 6 6 7 7 7 9 9 12 14 14 15 15 17 17 17 18 19 19 20 21 21 21 21 22 22 22 22 22 23 23 24 24 24 24 25 25 25 26 26 27  111  4.3 Or is it the other way around 7 5 Conclusions References Appendices Appendix A: Exchange of CO 2 between the atmosphere and ocean Appendix B: Carbonate buffer Appendix C: Calcium carbonate precipitation/dissolution Appendix D: Silicate-carbonate carbon cycle  28 29 55 59 59 60 61 62  iv  List of Tables Table 1. Data  30  Table 2. Correlation matrix  31  Table 3. Linear regression equations, regression coefficients, and R 2 against measured data  32  Table 4. GENIE-i simulation results time-series (global)  33  V  List of Figures Figure 1. Data  34  Figure la. 79 sediment trap locations showed on the GENIE-i Longitude-Latitude horizontal grid  35  Figure lb. Sea surface PP field on the GENIE-I grid  35  Figure Ic. Sea surface SST field on the GENIE-i grid  35  Figure Id. Sea surface MLD field on the GENIE-i grid  36  Figure ie. Sea surface f-ratio field on the GENIE-i grid  36  Figure if. Sea surface EP field on the GENIE-I grid  36  Figure 2. Data analysis results  37  Figure 2a. F 00 measured from 79 sediment traps at different depths> 1500 m  38  Figure 2b. TE measured from 79 sediment traps at different depths> i500 m  38  00 modeled with Equation (2a) Figure 2c. Linear regression between F 00 measured and F  38  Figure 2d. Linear regression between b measured and b-modeled with Equation (2c)  38  Figure 2e. Linear regression between F 00 measured and F 00 modeled with Equation (2e)  38  Figure 2f. F 00 and Fm normalized to F  39  Figure 2g. F , F 0 0 and F 1 normalized to Fm  39  Figure 2h. Correlation between F 0 and Fm F , F 0 , F 0 1  39  Figure 2i. Linear regression between F 00 measured and F 00 modeled with Equations (3a-3d)  39  Figure 2j. Linear regression between F 00 measured and Fy 00 modeled with Equations (9a-9d)  39  Figure 2k. Linear regression between Fy 00 measured and modeled 00 with Equations (12k, 13a, 13b) F  40  Figure 21. Correlation between TE and Fm, F , F 0 , F 0 1  40  Figure 2m. Linear regression between TE measured and TEmodeled with Equation (16b, 17a, 17b)  40  Figure 2n. Linear regression between TE measured and TEmodeled with Equation (19, 20a, 20b)  40  Figure 3. GENIE-I simulation results time-series (global)  41  Figure 3a. Global carbon emissions inventory from historical record and prescribed future scenario  42  Figure 3b. Global carbon emissions rate from historical record and prescribed future scenario  42  Figure 3c. Global atmospheric pCC 2 trajectory obtained from model simulations (NBC)  42  Figure 3d. Global atmospheric pCC 2 trajectory obtained from model simulations (CBO)  42  Figure 3e. Difference in atmospheric pCC 2 with respect to ‘no feedback’ (NBC)  42  Figure 3f. Difference in atmospheric pCC 2 with respect to ‘no feedback’ (CBC)  42  Figure 3g. Global mean sea surface [H], [CC ], [HCC] and [CC 2 ] (NBC) 2 3  43  Figure 3h. Global mean sea surface [H], [CC ], [HCC 2 ] and [CC 3 ] (CBC) 2 3  43  Figure 3i. Global mean sea surface calcite saturation state (NBC)  43  Figure 3j. Global mean sea surface calcite saturation state (CBC)  43  Figure 3k. Global CaCC 3 export production (NBC)  43  Figure 31. Global CaCO 3 export production (CBC)  43  vi  Figure 3m. Global POC export production (NBC)  44  Figure 3n. Global POC export production (CBO)  44  Figure 3o. Global mean sea surface [PC ] (NBC) 4  44  Figure 3p. Global mean sea surface [PC ] (CBO) 4  44  Figure 4. GENIE-I simulation results time-slices (zonal mean)  45  (NBC ‘i-climate)  46  Figure 4b. Zonal mean F 0 (CBC i-climate’)  46  Figure 4c. Zonal mean [PC ] (NBC ‘i-climate’) 4  47  Figure 4d. Zonal mean [PC ] (CBC ‘i-climate’) 4  47  Figure 4e. Zonal mean F 0 (CBC ‘i-calcification’)  48  Figure 4f. Zonal mean [PC ] (CBC ‘i-calcification’) 4  48  Figure 5. GENIE-i simulation results time-slices (surface layer)  49  Figure 5a. Sea surface CaCC 3 flux at 1765  50  Figure Sb. Sea surface PCCflux at i76S  50  Figure Sc. Sea surface calcite saturation state (NBC ‘i-climate +calcification’)  51  Figure Sd. Sea surface calcite saturation state (CBC ‘+climate i-calcification’)  Si  Figure Se. Sea surface CaCC flux (CBC ‘i-climate i-calcification’) 3  52  Figure Sf. Sea surface PCP flux (CBC ‘i-climate i-calcification’)  52  Figure 5g. Sea surface PCC flux (CBO ‘i-climate i-calcification’)  53  Figure Sh. Sea surface [PC ] (CBO ‘i-climate i-calcification’) 4  53  Figure Si. Fractional sea-ice extent (NBC ‘i-climate’)  54  Figure 4a. Zonal mean  vii  List of Abbreviations ALK  Alkalinity  BIOGEM CBO DIC  =  Biogeochemical Model  CaCO 3 Ballasting Ocean  =  Dissolved Inorganic Carbon  =  DOP  =  EP  Export Production  =  Dissolved Organic Phosphorus  EMBM EMIC ESM  =  =  Energy-Moisture Balance Model  Earth System Model Of Intermediate Complexity Earth System Model  =  GENIE  =  MBO  Multiple-Ballasting Ocean  =  Grid Enabled Integrated Earth System Model  MLD (or z ) 0 NBC PlC  =  =  Mixed Layer Depth  Non-ballasting Ocean  Particulate Inorganic Carbon  =  POC  =  Particulate Organic Carbon  POP  =  Particulate Organic Phosphorus  PP  Primary Production  =  SST  =  TE  Transfer Efficiency  =  Sea Surface Temperature  viii  Acknowledgements Ryusuke Izumi acknowledges and thanks support from -the Canadian Foundation for Climate and Atmospheric Sciences for research funding. -Dr. Andy Ridgwell for supervising and providing the GENIE-i model. -Dr. Roger Francois for supervising and providing helpful advices on the data analysis. -Dr. Philippe Tortell for supervising and providing overall support throughout the project.  ix  Dedication  /?72  x  I Introduction Carbon-bearing “sinking particles” in the open ocean, in their organic and inorganic forms, are one of the major pathways for transportation of carbon from the surface ocean to the deep ocean. One of the highest concerns for those involved in the study of the ocean’s role in the global carbon cycle is in the deciphering of the mechanisms of these carbon-bearing sinking particles as the processes of production, sinking, remineralization of particulate organic carbon (POC) and particulate inorganic carbon (PlC) control the distribution of dissolved inorganic carbon (DIC) and alkalinity (ALK) throughout the ocean reservoir. The distribution of DIC and ALK in turn regulates the strength of the oceanic uptake of carbon dioxide (CQ) gas from the atmosphere. Despite historical efforts by the oceanography community in revealing their nature, the exact controls that determine the composition and magnitude of the fluxes of these sinking particles throughout the ocean water columns as well as their response to future climate change remains poorly understood, which calls for continuous study. Quantitative understandings of the processes involved in the vertical flux of carbon-bearing sinking particles  in the ocean are essential for modelling the marine carbon cycle and its role in the regulation of the atmospheric 2 concentration. The efficiency of the biological pump depends not only on the absolute magnitude of POC CO and PlC export fluxes that fall down from the euphotic mixed layer of the ocean but also, and more critically, on their vertical remineralization length-scale profiles through the water column. In other words, the transfer efficiency of carbon-bearing particles from the sea surface to the deep ocean. Here, I briefly review the classical paradigms in oceanography on the behaviour of carbon-bearing sinking particles (and discuss the recent shifts in some of those views in later sections). Particular attention is placed here on the PlC component, which plays an important role in determining the distribution of ocean chemical properties and thus the buffering capacity of the ocean against acidification induced by the increasing uptake of atmospheric CO 2 gas at the air-sea interface. PlC in the ocean occurs in the form of biogenic calcium carbonate (CaCO ) and the two terms will be used 3 interchangeably in this paper. See Appendix A-D for more detailed descriptions of the chemical reactions and mechanisms referred to in this paper.  1.1 POC:PIC “rain ratio” and “carbonate compensation” In order to maintain a steady state of ALK in the ocean on timescales of thousand of years, weathering inputs of ALK into the ocean must be balanced by equivalent outputs, and any perturbation that drives a change in the 3 burial rate must be compensated for by some kind of a mechanism which adjusts the carbonate ion CaCO concentration [CO ] in the ocean to restore balance; a process termed “carbonate compensation” (Broecker 2 3 and Peng, 1987). One of the most plausible biogeochemical mechanisms for controlling the level of atmcspheric 2 concentrations over timescales of thousands to hundred of thousands of years, is that of the POC:PIC “rain CO ratio” (i.e. POC to PlC molar ratio of the sinking particles) (Archer and Maier-Reimer, 1994). This rabo has an influence on the preservation of PlC in the ocean water columns and sediments. While the classical estimate for the export rain ratio of POC:PIC is somewhere around 4 (i.e. 1 mol PlC per 4 mol of POC) (Li et al., 1969), a recent WOCE data analysis of phosphate and alkalinity gradients in the 100-200 m by Sarmiento et al. (2002)  1  has suggested 10 or more. CaCO 3 dissolution is not only determined by the saturation state of CaCO 3 (C)) (see Section 3.1.2 for definition of 0) in the ambient environment but also by the amount of metabolic CO 2 (i.e. acidity) released through the respiration of POC. The change in ocean 2 [CC3 ] , in turn, impacts the atmospheric partial pressure of CO ) due to its well established inverse relationship between atmospheric pCO 2 2 (pCO 2 (see Appendix B: Carbonate Buffer).On the assumption that any changes in the POC:PIC export rain ratio at the surface ocean will be transmitted to the deep sea in a direct proportional manner, any change in the magnitude of POC flux relative to CaCO 3 flux will affect the fraction of CaCO 3 that dissolves. In other words, the strength of respiratory dissolution of CaCO 3 can be altered by changing the surface ocean productivity of POC and PlC, relative to each other. Earlier numerical model studies have reported that carbonate compensation only fully operates on a timescale of 5-10k yr(Archeretal., 1997) and that any perturbations to the POC:PIC rain ratio over relatively shorter timescales could affect the strength of oceanic CO uptake by altering the ocean pH before the disturbance can be smoothed out. Recent studies on future projections of surface ocean C) with respect to calcite and aragonite over the next several hundred years strongly suggest that in the course of increasing atmospheric 2 and subsequent ocean acidification, calcification rates of marine organisms would significantly decrease pCO due to a pronounce lowering of [C0 ] (Orr et at., 2005; Caldeira and Wickett, 2005). A reduction in the export 2 3 rate of CaCO 3 relative to POC caused by negative calcification responses of calcifying organisms to increasing 2 invasion (Gattuso et at., 1999; Riebesell et at., 2000; Zondervan et al., 2001) could initiate a faster CO carbonate compensation process by leaving a larger fraction of ALK in the ocean. Carbonate compensation to this effect is an initial decrease in CaCO 3 burial that is eventually compensated for by an increase in the [CC ] 2 3 through enhanced CaCO 3 dissolution that leads to CO 2 uptake by the ocean (i. e. decrease in atmospheric  pCO ) 2 .  1.2 POC remineralizaion Despite the very close specific gravity of POC to that of seawater, observations from sediment traps deployed at different depths around the world ocean have confirmed that POC generated in the surface ocean is sinking rapidly (50-200 m d ) to the deep sea. This has been explained mainly as results of aggregation of small 1 individual particles and fecal pellets into larger particles (>0.5 mm) (Honjo, 1996), often referred to as “marine snow”. Some more recent studies postulate either a “ballast” effect or “protection” against biodegradation by other sinking particles of mineral phases (i.e. CaCO2, biogenic opal (SIC ), lithogenic dust) (Armstrong et al., 2 2002; Klass and Archer, 2002). The ratio of export production (EP) to net primary production (PP) in a certain area is defined as the f-ratio. The production of large rapidly settling cells such as diatoms in regions of high f ratios are known to make a greater contribution to the biological pump than the production of small suspended particles originating from pico- and nano-plankton in low f-ratio regions (Ganeshram, 2002). While the carbonate-dominated low f-ratio regions are considered to contribute less to the transferring of POC to the deep sea, Francois et at. (2002), proposed that the fraction of POC exported in these regions is more efficiently transferred to the deep sea.  2  Perhaps the most influential formulation of POC remineralizaion in the oceano graphy community during the past few decades is that published by Martinet al. (1987). Martin et al., by placing an empirical fit to observed data, described the variation of sinking POC flux with depth as a powerlaw function; F (Z) EP(zIzo)’ 0 EP(z/1 00)0858, where the depth of the surface mixed layer (MLD or z ) from which POC is exported is set to 100 0 m and -0.858 is a dimensionless negative constant to account for unexplainable variables. In other words, the POC flux at any depth z can be predicted from the export flux of POC out of the surface mixed layer. While the depth dependent term is necessary for the algorithm to reflect the gradua l remineralization of POC, the effect of depths becomes rather subtle in the bathypelagic zone (> 1500 m depth) as most of the POC is remineralized at shallower depths. Many ocean biogeochemistry simulation models to date have this power-law curve of Martinet al. or its variation for the purpose of predicting the depth distribution of POC remine ralization. However, it has to be emphasized that the “Martin et al. curve” was based on measurements restricted spatially to the Northeast Pacific and only from floating sediment traps at relativ ely shallow depths (<2000 m), and also temporally to short-term (6-34 days). Its applicability to the global ocean and entire depth range of the water column, therefore, remains highly questionable. It has to be also noted that the assumptions in measurements of sinking particle fluxes by sediment traps, in a simple 1-dime nsion mass balance, are that particles settle vertically between traps at different depths, that all traps collect particles with equal efficiency, and that particles trapped in the collection cup are not disturbed. Such caveat s may result in an over- or undertrapping due to hydrodynamic biases such as current turbulence, horizo ntal advection, and the presence of heterotrophic swimmers. While deep-sea sediment traps moored in the bathypelagic zone have been shown to fairly measure the vertical flux of sinking particles, measurements from sediment traps deployed at depths above the mesopelagic zone (< 1500 m) are observed clearly susceptible to under-trapping (Yu et al., 2001; Scholten et al., 2001).  1.3 CaCO 3 dissolution The major producers of biogenic 3 CaCO in the open ocean are known to be planktonic calcifying organi sms such as coccolithophores and foraminifera, both of which secrete calcite , and pteropods which form shells of aragonite. The production of CaCO 3 by calcifying organisms in the surface ocean, by taking up ALK and DIC from the ambient water in a 2:1 ratio through the reaction Ca 2 + 2HCO —* CaCO 3 + CO 2 (aq) + H 0, raises the 2 2 of seawater and acts as a potential source of CO pCO 2 to the atmosphere. It is important to recognize that, in the real ocean, the production of CaCO and release of CO 2 does not occur in a 1:1 ratio. In the modern ocean, the CO 2 release to CaCO 3 production ratio (‘1’) is found to be approximately 0.6 (Ware et al., 1992). This owes to the buffering capacity of the more abundant bicarbonate (HCO) and carbon ate (C0 ) ions in seawater. The 2 3 decrease in pH by an increase in CO 2 is largely compensated for by the reaction of hydrogen ion (H), the other hydrolysis product when CO 2 reacts with seawater to form carbonic acid 3 C0 2 (H ) , with C0 . This ratio is 2 3 predicted to increase as pCO 2 increases (Frankignoulle et al., 1994), that is, the buffering capaci ty decreases as both H and C0 2 are consumed in the course of ocean acidification. 3 An emerging concern on timescales that are directly relevant to our society is changes in the partitioning of  3  ALK and DIC in the ocean owing to changes in CaCO 3 production and dissolution due to the adverse response of planktonic calcifying organisms to increasing pCO 2 in the surface ocean. The depth range over which seawater is saturated with respect to CaCO , is called the saturation horizon. At depths below the saturation 3 horizon, CaCO 3 will spontaneously dissolve if exposed to seawater for a sufficient period of time. CaCQ 3 dissolution below the saturation horizon is often expressed at a rate proportional to (2  -  i)’, where  i  is the  dissolution rate power (i.e. measure of the strength of the non-linear response of calcification to changes in ambient 3 [C0 ] 2 ) and is very difficult to constrain (Zhong and Mucci, 1993). Historically, it has been strongly held that pelagic CaCO 3 is thermodynamically stable and exhibits a conservative nature in the surface ocean and down to several kilometres depth and that significant dissolution of biogenic CaCQ 3 only occur at great depths below the calcite lysocline; the depth at which shell dissolution starts to have a detectable impact on the preservation of CaCO 3 (Broecker and Peng, 1982). 1.4 “Ballast” hypothesis Although often not very obvious, the distinction between PIC:POC export rain ratio (i.e. at the base of the mixed layer) and its sediment rain ratio (i.e. at the sea floor) is important. Measurements of POC fluxes at different locations and depths around the world have discovered that, although absolute POC fluxes vary by more than an order of magnitude, in the deep sea (>1000 m) they appear relatively constant when normalized to total mass flux, making up —5% of dry weight fluxes (Armstrong et al., 2002). Another remarkable feature found from global sediment trap data set is the constancy of the POC:PIC rain ratio found in the deep sea. While the absolute fluxes of POC and PlC vary widely between different sites, the deep sea rain ratio appears quite constant falling in a narrow range between 0.5 -1.0. In search for a plausible explanation, researchers have investigated the importance of relatively denser sinking particles such as CaCO , biogenic opal and lithogenic 3 dusts (collectively termed “ballast minerals”) on the control of POC sinking flux (Armstrong et al., 2002; Klaas and Archer, 2002; Francois et al. 2002). These minerals have been identified as major constituents of the total sinking particle flux, often comprising more than half the mass of the flux leaving the surface (Honjo, 1996). On a basis that the density of POC is so close to that of seawater that individual POC by itself would not sink efficiently, these studies propose that POC flux may be in direct proportion to the flux of ballast minerals, where these minerals either enhance the settling rate of POC by increasing particle density or by protecting it from biodegradation, and that the POC:PIC rain ratio may be fairly fixed in the bathypelagic deep sea. If valid, the ballast model challenges the classical rain ratio hypothesis since it implies that any surface ocean perturbation that decreases CaCO 3 export will also drive a reduction in POC flux through the water column (i.e. POC flux is coupled to PlC flux), and that the POC:PIC rain ratio in the deep ocean above the sediment surface is rather insensitive to changes in the export rain ratio (Ridgwell, 2003). Remineralization curves of POC derived from classical empirical fitting to data such as in Martinet al. (1987) is fundamentally based on the assumption that all the information necessary for prediction is contained in the POC flux itself and that the fluxes of POC and PlC are independent from each other. From a sediment trap study in the equatorial Pacific and the Arabian Sea, Armstrong et al. (2002) proposed a model where POC flux is subdivided into a physically or chemically ‘protected’ fraction associated with ballast minerals and an 4  ‘unprotected’ fraction that remineralizes exponentially with depth. Their equation states that the protected POC flux decreases asymptotically with depth and is dictated propor tionally by the also asymptotic flux of ballast minerals in terms of its “transport ratio”. This transport ratio is a measure of the carrying capacity of POC by ballast minerals (i.e. POC carried per ballast mineral) and has been suggested to vary with ballast composition. Denser 3 CaCO appears more efficient at transferring POC to the deep sea than less dense opal, and that of lithogenic material varies widely and remains unclear since they rarely dominate the composition of particles in the open ocean (Klaas and Archer, 2002; Francois et al., 2002). Klaas and Archer (2002) extended the analysis of Armstrong et al. to a global sediment trap data set and conclu ded that up to 83% of the POC flux to the deep sea is associated with the ballasting by 3 CaCO However, one puzzle is that density cannot be the . sole controlling factor since Si0 2 is also denser than POC, and yet does not appear to show a strong ballasting capacity (Klaas and Archer, 2002; Francois et al., 2002). Francois et al. (2002) attributes this enigma to differences in the “transfer efficiency” (defined as POC flux normalized to export production; TE = FPOC /EP), illustrating that carbonate-dominated regions with low f-ratios and seasonality may have a higher efficiency in transporting POC compared to opal-dominated regions with high f-ratios and seasonality because systematic differences in the degree of biodegradability and “packaging” factor of sinking POC particles are likely. The ballast hypothesis explains the relative constancy of the rain ratio in the deep sea, which can be described as ‘buffered’ against any POC:PIC rain ratio perturbations in the upper ocean (Ridgwell, 2003). Numerical modelling has shown that the ballast effect can be capable of counte racting the negative calcification response of calcifying organism to increasing pCO , and may instead result in a net positive feedback (Barke 2 r et al., 2003; Heinze, 2004).  1.5 Summary Fluxes of ballast minerals may determine POC fluxes at depth. Thus, a mechanism-based model of POC flux must simultaneously predict fluxes of both POC and PIG as well as other potentially ballast minerals. Since the factors controlling their remineralization are still poorly understood, development of pure ecological or mechanistic models is in its infant stage. Meanwhile, marine carbon cycle models struggle for finding alternative ways to incorporate remineralization algorithms of carbon -bearing sinking particles in the open ocean, often resorting to ‘tuning’ any unknown parameters (e.g. Martin et al., 1987). Clearly, further progress in a better understanding of the interactions between POC and PlC fluxes and the development of a truly global depthintegral algorithm on the remineralization lengths scales of both fluxes requires more accurate and globally distributed particle flux measurements, particularly in the upper ocean. The updated global annually averaged sediment trap flux data available at the U. S. Joint Global Ocean Flux Study data directory, with its richer data set, makes it worth reassessing the global applicability of the different algorithms reviewed here. The main objective of this study is to assess the “ballast” hypoth esis for the transport of POC through the water column and how an alternative POC-PIC coupled model would compare to the conventional uncoupled model predictions of marine carbon cycle and global climate change.  5  2 Data Analysis Empirical models of POC remineralization through the water column formulated in preceding studies were derived from observations made in limited ocean regions and water depths (Suess, 1980; Martinet al., 1987; Armstrong et al., 2002)]. Statistical analyses of sinking particle fluxes have used different data sets where different criteria were used for data selection (Lutzet al., 2002; Francois et al., 2002; Klaas and Archer, 2002). Discrepancies in space and time from which conclusions of these earlier studies were derived make their direct comparison difficult, which leaves ambiguity in their validity. In order to truly evaluate the applicability of these models to the global ocean, we tested the validity of several classical and ballast POC remineralization models under a single context with our globally distributed 79 sediment trap data. We evaluated the different models in their original forms, and as an elaboration, different variations of the models were also tested. In order to evaluate the overall predictability of each model for the global ocean, POC flux modelled at each selected sediment trap site was statistically compared with corresponding measured POC flux.  2.1 Data A larger number of sediment traps have been deployed in the world ocean over the past few decades. We took the online database of global sediment trap flux data compiled in a U.S. JGOFS Synthesis and Modeling Project (http://usjgofs.whoi.edu/mzweb/smppi/honjo.html). The original database consists of 193 sites from around the global ocean and contains annual integrated data from 467 sediment traps. However, the rich database could not be fully utilized in this study. Since sediment traps deployed at depths shallower than 1200 m, above the mesopelagic zone, are known to be subject to poor trapping efficiency (Yu et al., 2001; Scholten et al., 2001), we have chosen a 1500 m depth cut-off and excluded all traps above 1500 m from our study. In order to minimize seasonal bias, only traps with sample collection duration of more than one full year (>365 days) in high latitudes (>30° N and S) and 10 months (>305 days) in low latitudes (<30° N and S) were selected. Since this study focuses on the flux of sinking particles in the open ocean, traps deployed in coastal seas as well as in coordinates where the relatively coarse resolution of GENIE-I (our Earth System Model) recognizes as ‘land’ were excluded. Finally, traps which met the above criteria but had insufficient measurement data for proper analysis were excluded. Accordingly, we obtained a total of 79 sediment traps distributed over 14 biogeochemical provinces as defined by the Longhurst Biogeochemical Pra’ince Map (Longhurst et al., 1995) around the global ocean for analysis [Table 1]. The complied data used in this study is shown in Table 1. POC flux measured at the selected 79 sediment traps are shown in Figure 2a. Transfer efficiency (TE) of the POC flux is shown in Figure 2b. Flux of particulate organic matter was calculated by Fpom = I .87F 0 (Anderson, 1995). CaCO 3 flux (F ) was calculated by F 0 0= 0 (based on the ratio CaCO 8.33F, IC = 100.0872/12.011 = 8.33). Flux of opal was derived based on its 3 conversion factor between biogenic silica; F 0=0 .l4Fsb (Mortlock and Froelich, 1989). The combined mineral 2 fraction (Fm) of the total sinking particle flux (Fr) and the lithogenic (F ) component of the total mineral flux were 1 calculated simply by difference; Fm = F Fpom, F 1 = F (F 0 -  -  +  0 F  +  Fpom). A net primary production (PP) field  6  estimated from the satellite-derived productivity model of Behrenfeld and Falkowski (1997), a sea surface temperature (SST) field from NODC Levitus WOA 98 dataset and a mixed layer depth (MLD or z ) field based on 0 change in potential density (Monterey and Levitus, 1997) was adopted for use in this study. These PP, SST and MLD fields from the source data were re-gridded to a resolution of 36 by 36 equal area grids to match the grids of GENIE-i [Fig. lb-id]. In a seminal paper, Eppley and Peterson (1979) defined the ratio of export production (EP) to PP as f-ratio. f-ratio for each grid were calculated using the algorithm of Laws et al. (2000), where a twodimensional interpolation computes f—ratio as a function of PP and SST [Fig. le]. Export production for each grid was then calculated from its definition EP  =  fPP [Fig. If].  2.2 Classical models The correlation matrix between  TE, PP, EP, f-ratio and SST are shown in [Table 2]. A medium positive  correlation is found between F 0 and PP (R  =  0.500) whereas correlation between F 0 and EP is poor (R  0.139). These statistical numbers suggest that neither PP nor EP alone contains all the information needed to predict F 0 in the water column.  2.2.1 PP-based model We first evaluated the global applicability of the PP-based F 0 prediction by Suess (1980), which has been a commonly applied description of (Lutz 00 et al., 2002). F (z) 0 F  =  PP/(0.0238z  +  0.212) (1)  Equation (1) was applied to the 79 sediment traps selected in this study. A linear regression analysis between 0 modelled using equation (1) and F F 0 measured resulted in a coefficient of determination (R ) of 0.445, 2 meaning only —45% of the F 0 variability in the world ocean could be explained by this PP-based 0 F prediction model.  2.2.2 EP-based model Another widely adopted description of F 0 is that formed by Martin et al. (1987). In their original equation, which was derived from observation in the upper 2000 m in the northeast Pacific only, the mixed layer depth  (z)  is fixed to 100 m and the dimensionless parameterb for unknown factors is tuned to -0.858 in order to obtain a best fit to measured  C() 0 F  =  EP(zIzo)”  =  858 EP(z/i00)°  (2)  7  By simply applying Equation (2) as it is to our global data set resulted in a poor regression between modelled and measured F 00 (R 2 0.150). In reality, upon applying equation (2) to the global ocean, it has to be taken into consideration that z 0 in fact vary among different ocean regions [Table 1; Fig. id]. Thus, in order to evaluate the applicability of Equation (2) to the global ocean better, we allowed it to take region specific values of in which the 79 sediment traps are located. F 00 at each sediment trap was predicted using equations with z 0 values specific to each grid. F ( 0 z) = 0 EP(ZIZ8 Ogrid) (2a) With Equation (2a), regression between modeled and measured dropped to an extremely low R 2 value of 0.00350 [Fig. 2c]. These results strongly show that the applicability of Martn et al.’s (1987) original equation to the global ocean is very poor. Figure 2c shows two distinctive cluster s of low latitude regions with less negative b and high latitude regions with more negative b. It also demonstrates that the original parameter exponent b (0.858) systematically overestimates F 00 in the low latitude regions and underestimates in the high latitude regions. Overestimation or underestimation of F 00 by this equation at a certain sediment trap could be because either b is too high or low, or because Law et al’s (2002) algorithm systematically overestimate/underestimate EP in that region. By accepting Laws et al.’s (2000) EP estimate and taking the assumption that b correlates with the flux of biogenic minerals (i.e. ballast hypothesis; CaCO 3 in particular) and other variables such as SST or f ratio (Francois et al., 2002), we calculated ‘b-modeled’ specific to each sediment trap. By re-arranging equation (2), b for the measured at each sediment trap could be calculated. b=) 0 ln(F,,jE P)IIn(z/z (2b) Multi-linear-regressions including F 0 and SST or f-ratio as controlling factors give: e” = kF 0  +  kSST  = 4.78061 0 F 0 +9.53841 Q3. SST +2.119110 (2c) eb = k F 0  +  k,ratiofratio  = 7.250910 F 0 6.6667.10i f-ratio +5.21471Q (2d) When compared against calculated eb from measured data, modele eb d derived from Equation (2c) and (2d) obtained R 2 of 0.755 and 0.574, respectively. Subsequently, eb calculated from Equation (2c) were used to derive modeled values of b. R 2 obtained from regression between b-modeled and b measured obtained 0.762 [Fig. 2d]. A more negative exponent b drives shallower F 00 remineralization (i.e. lower TE) of carbon to the deep sea, and a more positive exponent b drives POC remineralization at deeper depth (i.e. higher TE). We applied the region specific z 0 and the modeled b values to Equation (2):  8  F ( 0 z)  EP.(ZIZo)bmd  (2e)  Equation (2e), with region specific z 0 and b, dramatically improved the predictability of F 00 up to a much more reasonable R 2 of 0.581 [Fig. 2e]. These results confirm that exponent b must vary between different ocean regions rather than globally constant, and also support the speculation that F 0 seen in the ocean may be dependent on F 0 and seasonality under some yet unknown systematic relationship.  2.3 Ballast models POC flux at depths >1500 m around the world ocean differ considerably between different geochemical regions and within each region, ranging between 0.1 and 5.96 (g rn 2 yj [Fig. 2a]. Despite that Fpc measured throughout the world ocean differ by a factor of —60 when POC flux is normalized to total flux, the range of FpOC/Ft becomes strikingly narrow (0.0125-0.205), with a global mean of 0.0529 [Fig. 2f]. This consis tency over the world ocean where JF 00 becomes virtually constant at depths >1500 m along with the fact that most of F is F composed of Fm (mean Fm/Ft = 0.901) and F 0 being the major fraction (mean Fe/Fm = 0.545) [Fig. 2gj, makes it plausible to speculate a deep-water asymptotic behaviour of F tightly coupled with F, particularly F , in the 0 ocean. In order to evaluate the postulated association between POC flux and potentially ‘ballast’ mineral flux in our global data set, we first performed a correlation analysis between Fm and . 00 We also repeated the same F analysis for F , F 0 0 and F to find the strength of correlation between the flux of each mineral type fraction 1 and F 0 [Fig. 2h]. Correlation between F and mineral flux (Fm) resulted in a strong positive correlation (R = 0.863). Correlation between F 00 and CaCO 3 flux (F ) alone (R = 0.860) did not differ from that of Fm. The lack of 0 considerable correlation between opal flux (F ) and F, 0 , (R = 0.387) suggests that despite the commonly 0 observed high F 0 to F 0 ratio in some highly productive ocean regions, F 0 does not play an important role in the control of F. This could be simply due to the comparatively less dense property of biogenic opal (2.1 g crrn ) 3 than biogenic carbonate (2.71 g cmj, or could be attributed to factors other than particle density, such as lower transfer efficiency (TE) of POC in opal-dominated regions due to higher biodegradability of POC in such low SST high f—ratio regions (Francois et al., 2002). The relatively lower correlation betwee 00 and the flux of lithogenic n F minerals (F), R = 0.654, compared to that between F and F 0 is most likely because although the density of F 1 (e.g. 2.65 g cm 3 for quartz) is relatively high, F 1 is known to be quantitatively small in bathypelagic depth of the open sea. The potential ballasting effect of F , if any, may be restricted to coastal oceans and would require 1 evaluations outside the scope of this study.  2.3.1 Single-ballast model From regression analysis, we obtained simple equations for testing the ability of sinking mineral particles in predicting Fpcm, expressed as:  9  (3)  F=kbFb  Fb  is the flux of ballast mineral(s). The coefficient kballast could be interpreted as the ‘carrying capacity’ of each  mineral type as ballast. Regression between modeled F 0 and measured 0 F obtained R 2 of 0.744 for F 0 predicted with all mineral types grouped together as one fraction  Fm  (Table 3. Eq. 3a). F 0 modeled by each  mineral fraction F, F 0 and F 1 alone (Table 3. Eq. 3b-3d) compares with measured F 2 values of 0.739, 0 by R 0.150 and 0.428, respectively [Fig. 2i]. While R 2 obtained between  Fm  and F were both strong and practically did  not differ between each other, R 2 obtained between F 0 and F were distinctively low. F appears to be the mineral flux which may potentially have dominant control on F, while the influences of F 0 and F on F are likely to be negligible, at least at the global scale. Each model was tested with an additional depths dependent term which reflects the possibility of gradual remineralization of ballasted POC flux with depth. Here, we arbitrarily chose a simple inverse function to describe this effect. F  = kt,Fb  +  (4)  1 kz  Regression with a depth term in the equations resulted in slightly improved R 2 values (Table 3. Eq.4a-4d). The additional depth dependent term accounted for up to —4% (mineral ballast) and —2% (CaCQ 3 ballast) of the measured 0 F 0 variability, which suggests that at least a fraction of POC flux still undergoes some gradual remineralization at depth >1500 m while most of the POC in the deep sea is protected by ballast minerals. Armstrong et al. (2002) derived a more mechanism-based single-ballast model from their regional study in the equatorial Pacific and the Arabian Sea. Based on a mechanistic assumption, F is partitioned into two distinctive pools, conceptually written as: 0 (z) = F F 0 QA (z)  +  0 E (z) (5) F  where QA denotes POC flux “quantitatively associated” with ballast minerals and E denotes “excess” POC flux. FpocaA is the ballasted (or protected) POC flux subject to remineralization only when the coupled mineral particles dissolve. F E is the free POC flux which is not coupled with any mineral flux, therefore quickly remineralizes in 0 the upper water column. It is assumed that both types of F 0 are associated with the same aggregates of sinking particles (e.g. flocks, fecal pellets) which provides the increased density needed for both pools of carbon to sink. As described in Armstrong et al. (2002), to incorporate into model experiments, Equation (5) could be expressed as:  F poc (z)  —  —  =  ()  +  r  Fp ( 0 )  +  [EP  poc  I  poc (zO) —  .  —  poc (.)  [-(z-zO)/öE]  FpOc()].e[_2o)EJ  (6)  where F() is the asymptotic POC flux quantitatively associated with ballast minerals. [EP calculates the excess non-coupled free POC flux, where  E  —  remineralization length scale.  (zO( E  EP. 10  When either direct measurements or model predictions of the ballast mineral flux at the mixed layer depth (F)) and asymptotic ballast mineral flux (Fb()) are available, FO() can be evaluated by: FO() = p  Fb()  = p Fbc[Fb(/Fb(Zo)]  (7)  where p  asymptotic transport ratio (FpOC()/Fb()) and [Fb()/Fb(zo)] proportion of ballast mineral dissolution. The constant of proportionality p describes the proportional coupling between (.) 00 and Fb(.). The asymptotic F POC flux Fpc is calculated point-by-point by the surface ballast mineral flux Fb rather than by the surface POC flux Fp C(Zo). Therefore, for a global scale study, unless direct measurement 0 s are available for the entire area of interest (which is often not the case), Equations (6) and (7) must be evaluated with a simulation model that could output Fb > and Fb >. When only Fb are known from limited numbers of sediment traps (assuming Fb () () E Fb(z>1500m)), as in the case of our data set here, we further assume Fb() Fb() as demonstrated in Francois et al. (2002). With these assumptions, Fb()/Fb() yields 1 and Equation (7) reduces to: ,  () = P Fb(z>1500m) (8) 00 Fp Armstrong et al. (2002) had suggested parameter values P 0.05 and E 500 m for testing the implications of ballast mineral association in global scale model simula tions. Assuming a single-ballast of Fb as the sum of all mineral types (i.e. Fb Fm), Equation (6) at depths >1500m becomes: 0 F  0.05Fm  +  [EP - 0.05.Fm].er  ZO)(5OO  (9)  Regression between 0 F predicted with Equation (9) and measured at the 79 sediment traps resulted in a value of 0197, which is considerably high and suggests that mineral fluxes may indeed play an important role in the control of . 00 In our analysis we calculated p at the 79 sediment F traps and obtained a global mean p value of 0.0605. 0 F  0.0605Fm  +  [EP - 0.0605Fm]e  °°°  (9a)  2 obtained from using Equation (9a) was 0.798 [Fig. 2j], R which does not differ from that obtained by equation (9). This suggests that the model is not sensitive to a 0.01 change in the global asymptotic transport ratio.While Armstrong et al’s (2002) original equation simply consid ered Fb Fm, we elaborated the idea by testing the possibility where one particular mineral type may be domin antly responsible for the proposed ballasting effect; Fb F, Fb 0 and Fb F F. The global mean asymptotic transport ratio p betwee n POC and each mineral type, assuming ballasting effect by a single mineral type, was calculated from measured data; p Fpc,c(oo)IFc(), P p (c0)IF(o0). Subsequently, the following equations 0 F were formulated.  11  F 0 0  0 0.13 0F  +  [EP 0.130.Fc].ezoofh (9b)  F 0 0  0 0.29 5F  +  [EP  F 0 0 = 0.531F  +  -  -  (9c)  5 0.53 1.F]. o eIzO J [EP o )/ (9d) -  R b 2 etween 0 F 0 measured and F 0 modelled obtained with equations (9b), (9c) and (9d) was 0.787, 0.173 and 0.451, respectively [Fig. 2j]. F 00 modelled with F 0 alone as ballast mineral matched fairly tightly with meas ured 00 good as that modelled with Fm while predictions mad with F as e 0 or F F 1 alone were poor. These results again suggest that F 0 may potentially be the mineral type that has dominant control on Fpc. As >90% of F 00 is remineralized at shallow waters >1500 m [Fig. 2b; mean FPOCIEP 0.0634], we tested the possibility that ‘excess” free POC flux E 00 at depth >1500 m may be in fact negligible. With this assumption, F Equations (9a-9d) reduces to: 00 = 0.0605Fm (lOa) Fy F 0 0 = 0.13 0 0F (lOb) F 0 0 = 0.29 0 5F (lOc) F 0 0 = 0.531F  (lOd)  R o 2 btained from Equations (lOa), (lob), (lOc) and (lOd) was 0.744, 0.739, 0.150, and 0.428, respectively. The variability of measured F 0 explained by each equation was only up to —5% lower compared to equations with the F E term. This suggests that, at most, only a very small fraction 0 of FpOCE survive to depth >1500m and experience gradual remineralization in the bathypelagic dept hs of the ocean. We find that Armstrong et al.’s [2002] mechanism-based single-ballast model could explain up to —80% of the global F 00 variability at depth >1500 m. For the equations assuming Fm and F 0 as ballast mineral (Equations 1 Oa and lob), it is notewo rthy that the asymptotic transport ratio p calculated from measured data and the statistically derived carrying capacity coefficient kb in the single-ballast models (Equations 3a and 3b) are strikingly proximate (p = 0.0605, O.130,kb = 6.0966.10.2, 1.125110’). On the other hand, p and kb did not match well for equations assuming 0 and F F 1 as ballast. Armstrong et al.’s [2002] ballast model appears to be valid only when F 0 is the dominant mineral flux that is coupled with . 00 F  2.3.2 Multiple-ballast model Next, we evaluated Klaas and Archer’s (2002) multiple-ball ast model. Conceptually following Armstrong et 12  al.’s [2002] formation: (11)  F=kFb+FE  To start from a simple form, F E is first assumed negligible at depths >1500 m so that the ballasted POC fluxes 0 are assumed to sink conservatively. Fb is partitioned into multiple ballast fractions F, F , and F instead of 0 collectively Fm. 0 = kF F  +  F 0 k  +  kF  (12)  The POC carrying capacity k for each mineral fraction can then be statistically determined by multiple-linearregression. We obtained coefficients of 9.0426.102, 9.709810 and 8.0843.10.2 for k, k 0 and k, respectively (Table 3. Eq. 12). Regression between Fpm measured and  modelled with Equation (12) resulted in a  remarkably high R 2 of 0.867, which suggests that a multiple-ballast algorithm is able to explain —87% of the global variability in F 0 at depth >1500 m [Fig. 2k]. We also highlight that the high coefficient for F compared to that of F 0 confirms the importance of CaCO 3 mineral as ballast mineral for POC transportation to the deep water. The high coefficient for F 1 could potentially be important if observed lithogenic fluxes were higher. In order to explore the importance of other potential variables in predicting POC flux, additional SST or f-ratio dependent terms were experimentally added. kF  +  F 0 k  +  F 1 k  +  kvariable  (13)  2 obtained from equations adding SST or f-ratio dependent terms were almost equivalent compared to that R obtained from Equation (12) [Table 3. Eq. 13a, Eq. 13b; Fig. 2k]. Although adding the variable SST orf-ratio into the regression does not improve the overall predictability of  it is worth noting the signs of their coefficients  (positive for SST, negative for f-ratio), which agrees with the correlation matrix in [Table 2]. F 0 seems to increase with SST and decrease with f-ratio. Since the combination of high SST and low f-ratio are generally found in carbonate-dominated regions with low seasonality whereas the combination of low SST and high f-ratio are found in opal-dominated regions with high seasonality (Francois et al., 2002) [Fig. ic, Fig. le], this trend implies a relationship between the remineralization profile of F and seasonality in different regions. A depth dependent term can be added to Equation (12) in order to account for the small remineralization of the ballasted POC flux sinking through the water column at depth >1500m. 00 = kF F  +  F 0 k  +  F 1 k  +  1 kz  (14)  Adding a depth dependent term to the multiple-ballast model improves the Fpc variability explained by only —2% (Table 3. Eq. 14), which again suggests that the ballasted F 0 flux at depths >1500 m may be sinking in an almost conservative manner by either ‘fast sinking’ or ‘protection’ effect, while a fraction may still experiences 13  some decay through their journey to the deep ocean as the associated minerals themselves experience dissolution. Subsequently, as in the preceding analyses, we also tested if the ‘excess’ free POC fraction Fpcc E may still be a considerable component of the flux at depth >1500 m. As we could assumed that FE represents most of the EP sinking down from the MLD that appears to almost entirely remineralized above the mesopelagic depth (< 1500 m) [Fig. 2b], this flux could be considered propor tional to surface estimates of EP that remineralizes with depth. F 0 0  F 0 k  +  F 0 k  +  F 1 k  +  1 kz-  +  kepEP  F 0 =k  +  F 0 k  +  kF  +  kz’  +  k E 0 P  (15)  F 0 0 modeled from Equation (15) did not differ from that obtain ed from Equation (14) where FE is considered negligible (Table 3). This suggests there still remains a possib ility that ‘excess’ free fraction of POC flux may actually be totally remineralized within the upper 1500 of m the ocean before reaching the bathypelagic waters.  2.4 TE-based analysis An alternative approach to analyzing the relationship betwee n POC flux and mineral flux is to evaluate them in terms of ‘transfer efficiency’. By normalizing the POC flux measured at depth to estimates of export production, the unknown dependence between Fpcm and EP is removed to look specifically at the factors that affect the efficiency of the transfer of carbon from the surface to the deep sea. The range of variability of the normalized data (FPOCIEP) is much smaller than that of EP alone [Table 1].  2.4.1 TE-based single-ballast model Correlation coefficients between TE and the flux of each minera l type were calculated for our global sediment trap data assuming a single-ballast model, giving R values of 0.551, 0.612, 0.120 and 0.497 for Fm, F , F 0 0 and F , 1 respectively [Fig. 21]. F 0 showed a reasonable correlation between TE, follow ed by Fm. F 0 and F 1 showed no considerable correlation between TE. Assuming single-ballast , TE could be modeled by the following equation. 1 TE= kbFb+kZi  (16)  Regression between TE measured and TE modeled with Equations (16a-16d) showed a highest predictability of transfer efficiency by F 0 alone, accounting for —47% of the global TE variability at depth >1500 m [Table 3. Eq. 16b; Fig. 2m]. This result shows that although the coeffic ient of determination (l) is medium, F 0 indeed appears to have a dominant control over other minerals on the efficiency of POC transport to the deep ocean interior. As in the preceding analyses, additional terms of SST and f-ratio were added to Equation (16)to explore the importance of these variables on the control of POC flux.  14  TE =  kb  Fb +  1 kz-  +  kvariable  (17)  TE modeled with an additional term of SST or f-ratio improved added —10% to the ability in explaining the measured variability of TE [Table 3. Eq.17a, Eq. 17b; Fig. 2m]..Aain, the positive and negative signs of the coefficients of SST and f-ratio, respectively, should be highlighted. 0 F a t each trap location could be predicted by re-arranging Equation (17b). 0 = EP[3.00721O F F 3  +  8.5165.10*l.z  1.7268.1Otfratio  -  +  2.444210j  (18)  0 modeled using Equation (18) obtained a R F 2 = 0.568 against F 0 measured.  2.4.2 TE-based multiple-ballast model Multiple-linear-regression was carried out to obtain the contribution of each mineral type to transfer efficiency of POC flux. TE =  +  F 0 k  +  krFi  +  (19)  1 kz  Equation (19) obtained R 2 = 0.550 when compared against TE measured [Table 3. Eq. 19; Fig. 2n]. Again, additional linear terms of SST or f-ratio were added to the regression to explore the importance of these variables on the control of POC flux. TE =  +  0 kF  +  ktFi  +  1 kz  +  kvahable  (20)  An additional term of SST or f-ratio further improved the predictability of TE at depth >1500 m, with R 2 of 0.646 and 0.688, respectively [Table 3. Eq. 20a, Eq. 20b; Fig 2n]. The same trend is repeatedly seen, where the coefficient for SST and f-ratio is positive and negative, respectively. These results again suggest the importance of these variables when discussing variations between regions of different seasonality. Again, F at each trap location was predicted by re-arranging Equation (20b).  Fpoc = EP[1.197610 F 3  +  0 F 3 2.970310  +  1 F 3 2.176110  +  z 1 1 6.2472i0 3.700810 f -ratio -  +  Regression between F 0 modeled with Equation (21) and F 0 measured was considerable at R 2  6.3772109 (21) 0.633.  2.5 Summary We found that the applicability of classical EP-based model (i.e. Martin etal., 1987) to the global ocean is very poor. Predictability of F 0 improves dramatically by applying a region specific mixed layer depth (zo) and a  15  parameter (b) dependent on F and seasonality. A CaCO 3 single-mineral-ballast model could explain up to —79% of the global F 0 variability at depth >1500 m. Furthermore, a multiple—mineral-ballast model could explain up to —89% of the global variability in 0 F at depth >1500 m. The lower regression coefficients seen from the TE approach is due to the removal of the strong influence of large variations in ER The disadvantage of the TE approach in developing algorithms to predict POC flux is that transfer efficiencies are often low, typically less than 10% [Fig. 2bj, so that small errors on the TE results in large variations in These results suggest that 3 may potentially be the mineral type that has dominant control on the vertical transport of F from CaCO sea surface to depth in the open ocean.  16  3 Modeling in GENIE-I As part of the GENIE project (www.genie.ac.uk), a new coupled carbon climate model called ‘GENIE-I’ was developed. At its core is a 3D frictional geostrophic ocean circulation model, a 2D atmosphere EMBM (Energy Moisture Balance Model) and a dynamic-thermodynamic sea-ice model To this a single nutrient (phosphate) based representation of the marine carbon cycle ‘BIOGEM’ (BlOGEochemical Model) is coupled. With the .  resulting EMIC (Earth system Model of Intermediate Complexity), its computational efficiency enables >2000 yrs integration per hour CPU, which makes it suited for conducting multiple model simulations of short to long timescale variability. See Ridgwell et al. (2007a; 2007b) for a more detailed description of the GENIE-i model. 3.1 GENIE-I In this study, we defined and configured two modes of ocean; a Non-ballasting Ocean (NBC hereafter), where F and F are not coupled but independent of each other; and aCaCO -ballasting Ocean (CBO hereafter), 3 where the two fluxes are coupled and Fpc is dependent on F.  3.1.1 POC flux in the Non-ballasting Ocean In the NBC, F is determined by the default equations adopted in GENIE-i (Ridgwell et al., 2007a). POC export production is dependent on the available surface nutrient concentrations of phosphate (PQ). First, the biologic uptake of PC 4 (F) is calculated as: F  =  uopoPO I 4 ( PC 4  +  )(1 0 K 4  0 A)III  (22)  where 0 u p04 is the maximum uptake rate of phosphate under the assumption of no limitation phytoplankton growth, and K 04 is the half-saturation constant of a Michaelis-menten type kinetic limitation of nutrient uptake. Both u 0 PO (1.91 pmol kg-i) and 0 K 4 (0.21 pmol kg-i) are calibrated in the model for their appropriate values are unknown in the simplified ecosystem function in our model. Modifier I on productivity is applied to represent the effect of sub-optimal ambient light levels, where the strength of local insolation I is normalized to the solar constant 1 to give a linear limitation term. Another modifier A is applied to represent the effect of fractional sea ice coverage of each grid cell (Edwards and Marsh, 2005). The changes in P0 4 and dissolved organic phosphorus (DOP) concentrations in the surface ocean layer are governed by equations below.  aPo i 4 at = 8DOP/öt  -  =  F  +  vF  2. DOP  -  2DOP  (23) (24)  A proportion of PC 4 uptake (v) by the biota is partitioned into DOP where the relatively labile organic phosphorus is remineralized with a time constant of 1/2.. The values of v (0.66) and X (0.5 yr ) are taken from the 1  17  assumptions by Najjar and Orr (1999). The export flux of particulate organic phosphorus (in units of mol PC 4 rn2 ) is determined directly from P0 1 yr 4 uptake.  Fpop(zo)  1O-.zO  p (1  -  v)J’dz  (25)  where p is the seawater density and z 0 is the thickness of the euphotic zone. Export flux of particulate organic phosphorus (F ) is calculated from PC 0 4 uptake. Then F 0 is derived simply from the 106:1 molar ratio between dissolved inorganic carbon (DIC) and PC 4 upon production of organic matter (Redfield et al., 1963).  FPOC(Zo)  =  (26)  lO6Fpop(Zo)  Below the surface layer POC, is partitioned into two distinct fractions; a conservative recalcitrant fraction and a labile fraction. Remineralization of labile POC occurs instantaneously with an exponential decay. POC flux at depth z in the water column is expressed as:  FOC(Z)  C(oy(rC 0 F  =  +  (1  —  rpoc)exp((zo  —  )) 0 z)/I  (27)  POC is partitioned into a recalcitrant fraction (r) and a labile fraction (1  -  ). The recalcitrant fraction (r) and 0 r  the e-folding remineralization depth of labile fraction (I) are both parameters that are calibrated in the model, with values of 0.055 and 556 m respectively.  3.1.2 CaCO 3 flux in the Non-ballasting Ocean and the CaCO -ballasting Ocean 3 In both NBC and CBO, the export flux of CaCO 3 is related to the export flux of POC based on a thermodynamical description of carbonate precipitation rate.  F(o)  =  y rocaco3:pocFpoc(zo)  ‘y =(O-1) , =0.0  0>1.0  fl1.0  (28)  (29)  (30)  where r 0 CaCO3:POC is a spatially-uniform scalar, and ‘y is thermodynamically-based local modifier of the carbonate production rate. 0 is the ambient surface saturation state with respect to calcite, defined: O  =  ]/Ksp [Ca j 2 [C0a  (31)  where [Ca ] and [CC 2 ] are the concentrations of calcium ion and carbonate ion, respectively, and K 2 3 6 is the 18  solubility constant (Zeebe and WoIf-Gladrow, 2001). r is the thermodynamic calcification rate power. The higher the value of  i,  the more responsive calcification is to ambient saturation state. r is an unknown parameter  constant calibrated to 1 .28 in the model. As in the remineralization of POC through the ocean water columns, the 3 flux below the surface layer is described as: CaCO  F(Z)  =  F(y(r  +  (1  -  rc)exp((zo z)/I)) -  (32)  3 is partitioned into a recalcitrant fraction (re) and a labile fraction (1 CaCO  -  re). The recalcitrant fraction (re) and  the e-folding remineralization depth of labile fraction (ia) are both parameters that are calibrated in the model, with values of 0.489 and 1055 m respectively.  3.1.3 POC flux in the CaCO -ballasting Ocean 3 In the CBO, POC flux below the surface layer is dependent on the availability of CaCQ 3 flux and its carrying capacity as ballast mineral, as in Equation (lOb); F= 0.130F. The recalcitrant fraction (r ) is equated to 0 -ballasted POC instead of being prescribed as a set parameter, and is now a function of depth rather than 3 CaCO a fixed value for the ballasted fraction changes as a function of depth. 00 (z) F  00 (ZO)(rC F =  +  (1  —  rpoc)exp((zo  FpOC(Zo)(0.130FC(Z)  +  (1  -  —  )) 00 z)/I  0.130FC())exp((zO z)/I )) 0 -  (33)  Given the observed relationship between POC flux and CaCO3 flux in terms of the mineral serving as ballast controlling the POC flux, we assume that this relationship also holds in all depths of the water column below the surface layer.  3.1.4 Model experiment setup and permutations In both ocean modes, GENIE-I was first spun up for 10,000 years to reach a stable atmosphere/ocean and considered as the ‘pre-industrial’ state at year 1765. From there, lo investigate the range of potential effects of anthropogenic CO emissions on the ocean and its feedback on atmospheric CO 2 concentration, GENIE-i is forced with historical pCO records (ice core and observation records) from year 1765 to 2003 (Enting et al., 1994; Keeling and Whorl, 2005) followed by a prescribed future CO 2 emissions scenario thereafter from year 2004 to 3000, where the “business as usual” (lS92a) future emissions scenario is applied based on an use up of conservative estimates of fossil fuel resources (coal, oil and gas). In this prescribed scenario, CO 2 emissions rate peak at 2100 (20 Gt C yr )followed by a linear decline to zero by the end of year 2331 with a total emission 1 of circa 4000 GtC (Lenton, 2000; Lenton et al., 2006) [Fig. 3a, 3b]. Sediment burial of sinking POC and CaCO 3 particulates are not considered in this study as these processes control the climate-ocean interaction only on time scales> 1000 years (see Appendix D). In this version of GENIE-i, all of the sinking particles are  19  remineralized within the water column or at the ocean floor. In each ocean modes, model integrations of four permutations with different feedback operating are compared. :POC export ratio are fixed to constant. 3 ‘no feedback’ Both climate (including ocean circulation) and CaCC -  Neither carbonate production nor climatology are responsive to changing chemistry in the ocean (solid line). ‘+climate’ Only climate is responsive to increasing atmospheric CO 2 while CaCC :POC export is fixed constant. 3 -  2 while the spatial field of CaCC 3 export production Only the climate responds to the change in atmosphericpCO is fixed constant to pre-industrial (year 1765) state; the rate of calcification in the surface ocean does not respond to CO 2 invasion (dotted line). ‘+calcification’ Climate does not respond and only CaCO :POC export is responsive to increasing atmospheric 3 -  -Calcification feedback by forbidding climate to respond to change in 2 . This experiment extracts the CC 2 CC atmospheric pCC 2 while CaCO 3 export flux varies according to the surface saturation state with respect to calcite (Q) (dashed line). ‘+climate +calcification’ Both climate and CaCO :POC export are responsive to increasing atmospheric CO 3 . 2 -  This experiment represents a more real ocean and is the overall result of both climate and calcification feedback operating (dot-dash line).  3.1.5 NBC vs CBO It is important to note that the NBC and the CBC are in slightly different equilibrium state at year 1765 for they were spun up with different equations governing the flux of POC. The recalibration of GENIE-I inCaCO 3 ballasting mode was outside the scope of this study for we did not have the computing resource and expertise at ths time. As the parameter calibration has been carried out only for the NBC, the initial states (1765) between the NBC and CBO are slightly different. Therefore, the differences we find between the NBC and CBO may be either actually due to ballasting mechanism or due to the differences in the initial states of the two ocean modes. Ideally, we would have equivalently calibrated models of NBC and CBC (and MBC) that are equally good at reproducing the observed biogeochemical cycles. That said, no studies to our knowledge up to date have compared future CO 2 trajectories between equivalently calibrated NBC and CBC models. We highlight that even though our two ocean modes had slightly different starting points, where each model were spun up to a equilibrium state (1765) with different mechanisms governing the flux of PCC, still, the initial global CaCC 3 and POC export productions at pre-industrial equilibrium state (1765) do not differ much between the two ocean modes [Table 4, Fig. 3k-3n]. Furthermore, the spacial variations were also found very small when we compared sea surface CaCC 3 and PCC export production at 1765 between the two ocean modes, with no considerable difference [Fig. 5a, 5b]. The atmospheric concentrations reached at 2300 between two ocean modes (both with ‘no feedback’,) are practically equal at 1301 ppm and 1300 ppm, respectively [Table 4]. 20  Hence, in this study, we considered it safe and reasonable to say that the two ocean modes are directly comparable. i  3.2 Non-ballasting Ocean The results from NBO model runs are shown to first elucidate the carbonate system feedback on increasing atmospheric CO 2 under our more conventional perspective of the ocean; no ballasting mechanism present. We display the experiment results from the four different permutations described above. The trajectory of each major variable is monitored as a time-series and their numbers at major time-slices (1765, 2000, 2100, 2300) are compiled in [Table 4]. We chose these four time-slices as representing pre-industrial state (1765), modern state (2000), time of maximum CO 2 emissions rate (2100) and time when atmospheric pCO 2 reaches maximum from the prescribed CO 2 forcing (2300) [Fig. 3c].  3.2.1 NBO ‘no feedback’ In the baseline ‘no feedback’ run, neither climate feedback nor calcification feedback are in operation. Both POC export production and CaCO 3 export production stays constant over time, maintaining their pre-industrial 2 reaches 1301 ppm by 2300, (1765) value throughout the entire period of CO 2 forcing. Global atmospheric pCO which is more than four times the pre-industrial value [Table 4, Fig. 3c]. The increase in atmosphericpCO 2 here is purely due to the decrease in seawater buffer capacity as a consequent of change in ocean carbonate chemistry.  3.2.2 NBO ‘i-climate’ Allowing only the climate in GENIE-i to respond to CO 2 forcing (‘-i-climate run) adds 116 ppm to the baseline value. This can be attributed to C0 -climate positive feedback from restricted vertical transportation of nutrient 2 supply (PC ) to the surface ocean, thus reduced POC export to the ocean interior [Fig. 3m, 3o; Note that the 4 dotted line of ‘+climate’ feedback in PC 4 is being overlaid by the dot-dash line of ‘+climate +calcification’]. Global POC export decreases by 1.2 Gt C over the course from 1765-2300.  3.2.3 NBO ‘+calcification’ In the ‘+calcification’ run where climate feedback are absent and only calcification (and CaCO2:POC export ratio) responds to increasing oceanic CO 2 uptake, C0 -calcification negative feedback (as described in Section 2 1.1) kick in and depress atmospheric pCO 2 by 38 ppm, partially counter-acting the otherwise 116 ppm enhancement by climate feedback alone [Table 4, Fig. 3e]. Compared to the present day (2000) values, oceanic carbon uptake doubles by year 2100 (when the carbon emissions rate reaches its peak) [Table 4] and is followed by a decrease in line with the linear decrease in global CO 2 emissions rates [Fig. 3b]. At 2100 the climate feedback suppress global oceanic carbon uptake by 0.8 Gt C yr , while the calcification feedback counteract by 1 21  enhancing carbon uptake by 0.4 Gt C yr . Although not strong enough to fully reverse the direction of CQ 1 invasion into the ocean, the global reduction in calcification cancels out -‘50% of the increase in oceanic uptake by climate response at peak CO 2 emission rates (2100).  3.2.4 N BC ‘+climate +calcification’ The overall combined effect of the two counter-acting feedback +climate’ (+116 ppm) and ‘+calcification’ (-38 ppm) is an increase of 77 ppm atmospheric pCO 2 by year 2300 (Table 4, Fig. 3e); a net positive feedback on rising atmospheric pCO . 2  3.3 CaCO -ballasting ocean 3 The effects of F 0 ‘ballasting’ by F are explored in a CBC with the exact same four permutations as in the NBC experiments. While the general behavior of carbonate chemistry in the CBC is the same as in the NBC, there are some noticeable differences seen in the behaviour of F 4 concentrations between the CBC 0 and PC and the NBC, largely due to the coupled ‘ballasting relationship between F 0 and F.  3.3.1 CBC ‘+climate’ With climate feedback only (‘+climate’), PCC flux decreases over time in most part of the global ocean as in the NBC, except the penetration of PCC in the CBC is much deeper in the equatorial water column than in the NBC [Fig. 4b].  3.3.2 CBC ‘+calcification’ We next look at the calcification feedback only (‘+calcification) run in the CBC. Due to the dependence of Fpc on F in the ballasting model, one may simply expect to see reduced PCC export production asCaCC 3 export is depressed from the calcification response to ‘ocean acidification’. The picture is a little more complicated than that. As in the NBC, the progressive decrease of sea surface calcite saturation state due to invading acidity leads to a large decrease in global CaCC 3 export production over time [Fig. 3k, 31] and thus one might expect PCC export production to decrease in line in the FpOD-FC coupled model. However, the PCC export production time-series shows a global increase over time [Fig. 3n].  3.3.3 CBO ‘+climate +calcification’ The end results of both climate feedback and calcification feedback operating in the CBC are shown in the global time-series [Fig. 3f] and sea surface maps [Fig. 5d-5h]. At the global scale, we find that while the net effect of ‘+chmate +calcification’ with respect to ‘no feedback’ in the CBC by 2300 is an atmosphericpCC 2 increase of 96 ppm (7.4%) [Fig. 3f]., whereas the pCC 2 increase with respect to ‘no feedback’ is lower in the NBC at +77  22  ppm (5.9%) [Fig. 3e]. Although the increase in global POC export production fuelled by shallower PQ recycling due to calcification feedback (+calcification) counteract against the diminishing effects by climate feedback (+climate), strong climate feedback (e.g. increased stratification) ultimately dominates over the calcification effects.  3.4 Multiple-ballasting Ocean In addition to the CaCO -ballasting Ocean, we have also experimentally configured a Multiple-ballasting 3 Ocean (MBO hereafter). In the MBO, Fy 00 is dependent on F , F 0 0 and F 1 as in Equation (12); 0 F 0=k F 0  +  F 0 k  +  krFi. Since there is no physical transport process which corresponds to a negative flux of carbon, the small offset negative constant term (not shown) is removed for implementing into GENIE-i, obtaining regression coefficients of 8.4648.i02, 6.4016i0 and 7.9564.102 for F , F 0 0 and F , respectively. The resulting R 1 2 between F 00 measured and F 00 modelled remained exactly the same as before at 0.867. As in the CBO, the recalcitrant fraction (r ) is equated to multiple-mineral-ballasted POC instead of being prescribed as a set parameter, and is 00 now a function of depth rather than a fixed value for the ballasted fraction changes as function of depth. 00 FOC(Z) = (Zo) 00 (r F .  +  (1  —  ) exp((zo 0 r .  = () 00 2 F FC( ((8.4648i0 ) +  (1  —  Fc(z) 2 (8.464810  +  +  —  )) 00 z)/I  6.40i6i0FO(Z)  6.40i6i0 ( 0 F 3 Z)  +  +  F(Z)) 2 7.9564i0  )) 00 FI(Z))) exp((zo z)/I 2 7.956410 .  -  The exact same four simulation permutations as in the CBO were run. We found very liltIe differences between the end results obtained from the CBO experiments and the MBO experiments, where the global POC and 3 export productions between the two ocean modes did not show a considerable difference [Table 4]. CaCO  3.5 Summary With the ‘ballast’ effect in operation, the net effect of climate feedback and calcification feedback is a global decrease in POC export production, except for in some high latitude regions where decreased sea-ice coverage fuels the production of POC. Simulation in GENIE-i showed that calcification response in the CBO is —63% weaker compared to that in the non-ballasting ocean. With both climate and calcification feedback operating, the CBO resulted in a slightly larger net increase of atmospheric pCO2 over time compared to the NBO. The difference in the net increase of atmospheric pCO2 from 1765 to 2300 between the two ocean modes was 12 ppm.  23  4 Discussion 4.1 GENIE-I simulation results 4.1.1 NBO ‘+climate’ A better visualization can be obtained from zonal average cross-sectional plots. The maximum of PC 4 concentrations deepen and surface concentrations deplete over time, leading to a net transfer of P0 4 from the surface ocean to the deeper ocean [Fig. 4c]. The depth of POC flux shallows in line and is most prominent in the equatorial surface waters where strong stratification dampens nutrient supply [Fig. 4a]. It is also interesting to note that an increase in POC flux is seen in the Northern high latitudes just above  >  60N [Fig. 4a]. We attribute  this to the progress decrease of sea-ice coverage over time in the high latitudes as seen in [Fig. 5i]. A decrease in sea-ice coverage, hence increased sea surface area for biological production to occur, would lead to increased flux of POC and POP. We find that the decrease in the extent of sea-ice coverage is a climatesensitive factor that dominates over the effect of increased stratification.  4. 1. 2 NBC ‘+calcification’ The magnitude of marine carbonate chemistry change is greatest at around 2300 for this is when atmospheric pCO 2 reaches maximum. The increase in carbon uptake by the ocean leads to a notable increase in H ions, and decrease in C0 ions, in the global sea surface [Table 4, Fig. 3g]. The consequence is aO.5 units sea surface pH drop, from 8.1 (1765-2000) to 7.6 by 2300 [Table 4], pushing the surface ocean towards more acidic’, hence commonly referred to as ocean acidification’. Predictions of possible pH shifts by a more complex OGCM (Ocean General Circulation Model) fall within comparable range where a maximum reduction of 0.77 pH units is reported (Caldeira and Wickett, 2003). Perhaps one of the most important trend that has to be highlighted here is, due to the drop in surface concentrations of C0 , calcite (and aragonite) saturation state (0) in the surface ocean drops by more than 2 3 60% by 2300 compared to the 1765 pre-industrial state [Table 4, Fig. 3i]. Omega progressively approaches closer to 1, which is the thermodynamical threshold for CaCO 3 precipitation to occur. A number of studies have suggested that reduced pH and decreased degrees of C) could substantially lower the precipitation rates of calcareous plankton even if C)> 1 (still supersaturated with respect to CaCO ) [Bijma et al., 1999; Riebesell et 3 al., 2000; Zondervan et al., 2001]. In line with the decrease in biogenic carbonate precipitation, in the permutations with calcification response operating, global CaCO 3 export production from 1765 to 2300 decreases from 1.3 to 0.3 GtC yr-i; a —77% reduction [Table 4, Fig. 3k]. This number is considerably larger than the —50% reduction of CaCO 3 export predicted in a similar experiment (CO 2 forcing under no climate feedback) by Heinze (2004) using a more complex general circulation model. Spatial variability should also be highlighted. The evolution of global distributions of 0 and CaCO 3 export per surface area from pre-industrial state (1765) through 2300 are shown in Figure Sc. Sea surface saturation state  24  of calcite are not uniform but heterogeneously distributed over different regions of the ocean. Although the drop in 0 is largest in the equatorial regions, the high latitudes are most subject to progression towards 0  <  1  (undersaturated with respect to CaCO ) for degree of calcite saturation in high latitudes are much lower by 3 default. By 2300, 0 in the oceans at high latitudes  >600  drops below the threshold value 1, turning the surface  ocean into undersaturation with respect to calcite. The same trend is also seen in other 3D model studies (Caldeira and Wickett, 2005).  4.1.3 CBO ‘+climate’ The deeper penetration of POC seen in the equatorial water compared to the NBO may be due to the ‘ballasting’ effect in the CBO. The equatorial waters may be a high POC transfer efficiency region, derived by the ballasting by dominating CaCO 3 minerals (Francois et al., 2002). A noticeable increase in F 0 is seen in the high latitudes just above >60° N as in the NBO [Fig. 4bj. We also found PO concentrations in the CBO behave in the same way as in the NBO (Fig. 4c, 4d].  4.1.4 CBO ‘i-calcification’ The increase in POC export in the CBO ‘i-calcification’ run is explainable. In the CBO, the penetration of carbon and thus nutrients into the ocean interior depends on the CaCQ 3 flux. Hence, due to the reduction in 3 flux, POC penetrates less deeply in the ocean and nutrients are released back into the water column at CaCO shallower depths. Surface ocean nutrient supply to plankton is increased at the surface ocean, fuelling POC export production [Fig. 3p]. However, increased POC export at the surface ocean does not necessarily lead to POC penetration into the deep water columns for remineralization depth shallows. This is better illustrated in zonal sections of F 4 [Fig. 4e, 4f]. POC penetrates less deeply into the ocean while showing an 0 and PC increase in flux per surface area only in the surface waters. PC 4 recycling maximum move to shallower depths and the concentrations at depth <1000 m increase up to 1.5  106  mol kg . A vertical re-partitioning of nutrients 1  occur due to shallower remineralization of coupled F 0 and F in a more acidic and ballasted ocean.  4.1.5 CBO ‘i-climate +calcification’ Spatial variations of F,  0 and P0 F 4 concentrations are presented in sea surface maps [Fig. 5e-5h].  Both CaCO 3 and POC export productions per surface area diminish through almost the entire ocean and experience the largest magnitude of drop in the equatorial regions [Fig. 5f, 5g] where CaCO 3 fluxes are dominant by default (pre-industrial state). We have also noticed a net increase in POC export production in two distinctive northern high latitude regions >60 N [Fig. 5gj. In these two areas, sea surface P0 4 concentrations deplete over time and POP export increases [Fig. 5h, 5g]. These regions are identified as opal-dominating regions with relatively high f-ratios [Fig. le]. Calcite saturation state and CaCO 3 export productions in these high latitude regions were low to begin with [Fig. 5d, Se] at 1765 and hence the effect of reduced calcification, hence  25  weakening of ballasting, due to carbonate chemistry change is minimal. Furthermore, as discussed above, POC production driven from the progressive decrease in sea-ice coverage appears to override the calcification feedback.  4.1.6 Multiple-ballasting Ocean With both climate and calcification feedback in operation, the atmosphericpCO 2 at 2300 between the two ocean modes differed by only 4 ppm. We thus do not consider opal-ballasting and lithogenic mineral-ballasting of POC flux important, at least at the global scale, and continue to speculate that CaCQ -ballasting is indeed the 3 strongest candidate as being the major ‘ballast’ mineral for POC flux in the open ocean.  4.1.7 Non-ballasting Ocean vs CaCO -ballasting Ocean 3 Several previous studies have evaluated the feedback strength of ballast effect on atmospheric C concentration in model simulations with no climate feedback considered (i.e. equivalent to ‘-i-calcification’ in our study). Barker et al. (2003), by incorporating the CaCO -ballasting equations of Armstrong et al. (2002) into an 3 eight-box ocean-atmospheric model, reported a complete reverse of the negative feedback of reduced calcification on atmospheric CC 2 by the positive feedback of ballast effect. This was by assuming a 20% increase in the remineralizations rates of F 00 and F 0 (i.e. shortening of remineralization depth) as consequence of predicted reduction of CaCO 3 saturation state of the surface ocean. Heinze (2004) have also reported that the negative feedback effect on increasing atmospheric pCO 2 due to the weakening of CaCO 3 export could be fully compensated if CaCO -ballasting mechanism is introduced. It has to be noted that, in the studies, the ballast 3 mechanism was introduced to an model ocean that previously did not have F 00 coupled with F . In other words, 0 the model ocean was neither calibrated nor spun up to an equilibrium state with a ballast mechanism in operation. When we compare our simulation results from GENIE-i NBC and CBO, we see a similar result as the studies mentioned above. In each ocean mode, ‘+climate’ alone increases atmosphericpCO 2 by 116 ppm (NBC) and 115 ppm (CBO) with no noticeable difference. With CaCO -ballasting mechanism operating (‘CBO 3 +calcification’), the reduction in atmospheric pCO 2 by calcification feedback alone with no ballasting (i.e. NBC ‘+calcification’) are completely countered; 1263 ppm (NBC) vs 1286 ppm (CBC) [Table 4]. We highlight on the difference seen in the strength of calcification feedback between the two ocean modes. In the ‘+calcification’ runs, in the NBC, calcification feedback alone enhance ocean CO 2 uptake by 0.2 Gt C and reduces atmospheric  2 by 38 ppm (2.9%) by 2300, whereas in the CBC, enhancement of ocean CQ uptake by calcification pCC feedback is largely eliminated, and the decrease of atmosphericpCC 2 is by only 14 ppm (1.1%), demonstrating much weaker negative feedback on increasing atmosphericpCC 2 [Table 4; Fig. 3e, Fig. 3f]. By comparison, the negative calcification feedback on atmospheric pCC 2 is —63% weaker in the CBC than in the NBC. The atmospheric pCO 2 trajectory predicted by the model between NBC and CBC with both responses ‘on’ (i.e. both climate and calcification feedback fully operating), reaches 1378 ppm at 2300 in the NBC, and 1396 ppm in the  26  CBO (both ocean modes starts from a pre-industrial atmospheric pCO 2 of 278 ppm) [Table 4]. The strength of ballasting’ depends on the change in magnitude of CaCQ 3 export production, in other words, the sensitivity of calcifying organisms to ocean acidification. One important factor which was not considered in this study is that the response to change in carbonate chemistry is very different between different calcifying species. The response of calcifying organisms to lowered saturation states differ from species to species and therefore certain species will be affected more seriously than others (Bijma et al., 1999; Riebesell et al., 2000; Zondervan et al., 2001). Thus, simply taking the sensitivity of one single species of calcifying plankton to change in saturation state (the power parameter n) as representative for all marine biogenic calcification is not the best assumption for global scale modelling experiments (Ridgwell et al., 2007).  4.2 Shallower CaCO 3 dissolution? A reduction in carbonate production and deposition in the shallow water by corals as predicted by Kleypas et al. (1999) implies an increase to our estimates for the strength of the C0 -calcification feedback. However, some 2 modeling studies on solution of shallow-water carbonates have concluded that although metastable carbonates could dissolve under increased invasion of atmospheric C0 , the shallow-water ocean will not accumulate 2 enough alkalinity to serve as buffer against change in pH and carbonate saturation state (Andersson et al., 2003; Andersson and Mackenzie, 2004). Shallow water calcification is not considered to affect POC export and remineralization and our open ocean model in this study excludes the effect on atmospheric CO 2 from a reduction in shallow water calcification. The conventional view of a relatively conservative PlC flux compared to that of POC is often supported by observations that CaCO 3 shells collected in sediment traps have almost always found to be intact with no trace of partial corrosion, showing no visual evidence of significant CaCO 3 dissolution with depth in the water column above the lysocline (Honjo, 1976; Tsunogai and Noriki, 1991). It has also been observed that CaCO2 in the form of aragonite appears to substantially dissolve in the upper water column (Betzer et al., 1984). Dissolution rates measured by sediment traps in the deep water are generally low, ranging from 0.003 to 0.O3pmol kg 1 yr 1 (Feely et al., 2004). Since sediment traps measure dissolution in the water column only and can not account for sedimentary dissolution, one should also not overlook that considerable CaCO2 dissolution may also occur in the pore waters of sediments and not on the seafloor (Elderfield, 2002). Further more, results from recent studies of ALK distribution and CaCO 3 budgets in the ocean started to indicate considerable dissolution of CaCO 3 in the water column well above the chemical calcite saturation horizon. The discrepancies found between alkalinity budget derived CaCO 3 export production in the surface ocean and sediment trap based measure of CaCO 3 flux to the sediments suggests that as much as up to —6080% of the CaCO 3 produced in the surface ocean may be in fact lost in the upper 1000 m of the water column, presumably owing to biologically mediated dissolution such as in acidic digestion within zooplankton guts and feces or involving microbial oxidation of organic matter (Milliman et al., 1999; Jansen and WoIf-Gladrow, 2001). The existence of such microenvironments may be one critical control on the degree of dissolution that takes place in the water column. The remaining CaCO 3 that survives to further depth would experience further  27  dissolution in the deep water column and at the sediment- water interface, thus it may be that only a small fraction than previously thought of CaCO 3 produced in the euphotic zone is sequestered in the sediments. While such calculation could be an artefact resulting from the different methods of determination between surface and deep fluxes, a more uniform measurement method using floating sediment traps also demonstrates a tight exponential decay curve in which 50-60% of the CaCO 3 export flux is lost in the upper 1000 m of the water column (Martin et al., 1993). 3 from deep sea cores indicate a global accumulation rate of —0.1 Pg C yr Accumulation rates of CaCO 1 (Catubig et al., 1998), which implies that —90% of the surface production dissolves before reaching the sediments or at the sediment-water interface. More recently, Feely et al. (2004) estimated the distribution of ALK in the water column by using a water-mass tracer method and concluded that up to 65% (—0.5 Pg C yr ) of total 1 3 export production dissolves back in the water column before reaching the seafloor, and that up to 60% of CaCO the dissolution occurs in the upper water column (<2000 m). While it is convenient to explain such dissolution flux to shallower dissolution of the more pressure susceptible aragonite polymorph, it may be difficult to attribute such huge amount of dissolution solely to aragonite minerals since pteropods probably account for only —10% of the total planktonic carbonate production (Fabry, 1990). A compilation of global sediment trap data through 2000 m depth indicate a global average flux of —0.4 Pg C yr , also suggesting that at least 50% of the CaCO 1 3 export is dissolved in the upper water column (lglesias-Rodriguez et al., 2002). Although these individual studies each contain some degree of ambiguity, if not flaw, such corroborating data serves as collective evidence for shallow water dissolution of CaCO 3 above the thermodynamically predicted lysocline.  4.3 Or is it the other way around? Correlations on its own do not identify cause and effect. The “ballast” and “protection” mechanisms discussed above are still poorly constrained and therefore must be carefully evaluated before accepted as true. It has been claimed that the slow sinking velocity (—0.1 m d ) of small fragments of CaCO 1 3 particles, like individual POC particles, is not sufficient to settle significantly except in forms of aggregates or fecal pellets (Steinmetz, 1994), and that large aggregates of POC (e.g. diatoms) which sink fast (50-200 m d ) can, on its way down the water 1 columns, collect large amounts of inorganic minerals including CaCO , opal and lithogenic dusts that are each 3 too small to sink on their own (Hamm, 2002). On the contrary to the idea that ballast minerals determine POC fluxes, some studies suggest that it is the other way around; it is the sinking POC particles that scavenge inorganic mineral particles to its carrying capacity as they travel through the water column (Passow et al., 1994; Passow, 2004; Passow & De La Rocha, 2006). Aggregation of sinking particles is shown to be a function of extracellular polymers to some degree, which supports the idea that POC may be in fact determining the flux rates of both itself and other inorganic minerals. The apparent importance of biologically mediated CaCOj dissolution also leads to speculate that the flux of ballast minerals are not independent variables but depend on the flux of POC.  28  5 Conclusions We have found that the flux of CaCO 3 mineral could potentially be the dominant controlling factor on the transport of POC from the surface ocean to the deep ocean. Statistical analyses suggest that up to —79% of the global variability seen in  below 1500 m depth could be dependent on F. The differences between the initial  states of our two ocean modes are less important than the anthropogenic impact seen from the model simulations. Cur model simulations have shown that, if the “ballast hypothesis” is valid, the larger net increase of atmospheric pCO 2 in the CBO compared to the NBC over the prescribed CO 2 forcing period could be attributed to the net positive feedback loop between atmospheric pCO , F and 2  With the presence of ballast  mechanism for POC transport in the open ocean, calcification feedback alone in the CBO is —63% weaker than in a conventional NBC. Moreover, the positive feedback of a potential ballasting mechanism on atmospheric  2 could overturn the negative feedback of calcification response on atmosphericpCC pCO . In an acidifying 2 ocean, on top of the negative feedback of marine calcifying organisms on 2 atmosphericpCO the ballast , hypothesis adds additional complication to the direction of the net feedback. Regional applicability of the ballast equations is not investigated in this study. A deeper understanding of the inter-dependence between PCC and PlC at both global and regional scale should be sought and further investigated.  29  Table 1. Data L It d  LG6C Proc cc  L  Ld 5  -________  Traptpr  MLD(z  SOT  PP  rn  m  C  1 9royr  OP  F F F F (rate TO IF grn y ri gm y 4 rr OP/PP Jjoryrigmyri FjEP 4  F/F  F/F  F/F  FJF  North Atlantic Dnitt  49.1  -13.4  3220  201  13.9  292.7  56.3  019  41.3  3.0  0.05  2.0  1.49  35.7  SAT  0.21  0.32  2.18  NorthAtleshoDritt  49.0  -13.9  4000  201  13.9  292.7  56.3  0.19  40.0  2.6  0.04  2.7  0.92  30.4  0.63  0.06  0.31  11.39  NorthAfianscDrifi  47.8  -19.5  3100  201  13.9  292.7  56.3  0.16  22.3  1.9  0.03  IA  1.36  18.7  0.62  0.22  0.16  2.65  North Atanbc Subtropical Gyro  33.0  -220  4150  65  21.2  121.2  10_i  0.16  12.5  CA  0.02  1.1  0.36  11.3  0.76  0.06  0.13  9.77  North Atlantic Subtropical Gyro  31.6  -64.2  1500  Ti  23.2  117.9  19.2  019  hA  09  0.05  0.8  iii  9.8  0.67  0.19  014  3.56  North Atlanbc Subtropical Gyro  31.8  -64.2  3200  71  23.2  117.9  19.2  019  12.8  8.6  0.03  0.9  0.70  11.6  0.64  0.18  0.19  3.63  North Atlantic Subtropical Gyro  29.1  -15.4  3075  59  20.6  196.9  36.5  0_is  16.1  0.8  0.02  1.0  0.8/  is_S  o.oi  0_OS  0A4  10.73  North Atlantic Subtropical Gyro  24.6  -22.8  3870  64  22.5  126.3  25.2  0.16  15.1  0.7  0.04  0.9  0.82  13.7  0.54  0.08  0.39  7.18  Pastors Tropical AUanUc  1.8  -iii  3921  22  27.0  146.9  24.1  0.16  34.2  2.0  0.08  2.2  0.90  30.5  0.61  0.16  0.23  3.63  SouthAtlanscTropioal Gyro  -20.1  92  1648  52  20.2  192.7  37.6  0.20  25.6  2.9  0.08  1.7  1.70  20.1  0.70  0.16  0.14  4.23  NorthwastArabian Uywalling  17.4  58.8  3141  20  26.3  396.3  64.0  0.21  79A  4.8  0.06  4.7  1.03  ISA  0.50  0.26  0.19  2/5  NorthwestArabian Upwelling  172  59.6  1657  28  27.3  240.3  39.0  0.16  63.0  6.0  0.15  4.7  1.26  71.8  0.55  0.25  0.21  2.22  NorthwootArabias Upwaltiog  17.2  59.6  2871  28  27.3  240.3  39.0  0.16  81.2  4.9  0.12  4.7  1.05  72.1  0.54  0.23  0.23  2.34  NorthwostArabiao Uywalling  16.0  60.0  3020  34  172.4  28.0  0.16  510  2.6  5.09  3A  0.77  466  0.60  0.25  0.15  2A0  NorthwastArabias Uywellisg  16.0  60.0  3020  34  27.7 37.y  i72A  28.0  0.16  43.4  2.5  0.09  3.0  0.83  38.7  0.65  0.28  0.07  2.36  NorthwostArabiao Uywalling  16.0  60.0  3020  34  27.7  172.4  26.0  0.16  03.2  3.3  0.12  3.9  0.84  47.0  0.69  0.21  0.10  3.33  NorthwastArabian lipmalling  16.0  60.0  3020  34  27.7  1724  260  0.16  75.6  4.5  0.16  4.3  1.05  67.2  0.53  0.36  0.1/  I AT  loden Ocean Monooon Gyros  10.0  68.0  2800  34  27.7  172A  28.0  0.16  41.5  2.6  0.10  2.5  1.11  36.3  0.58  0/5  0.27  3.82  Indian Ooean Monsoon Gyros  15.0  Indian Ooaan Monsoon Gyros  68.0  2800  34  27.7  172.4  28.0  0/6  22A  1A  0.05  1.3  1.05  16.8  0.06  0.17  027  325  15.0  66.0  2800  34  27.7  172A  28.0  0.16  40.3  2.2  0.08  2A  0.92  362  0.55  0.22  0.24  2.54  Indian Ocean Monsoon Gyros  15.0  68.0  2800  34  27.7  172A  28.0  0.16  29.3  1.9  0.07  2.0  0.96  207  0.64  0.22  0.15  2.93  Indon Ocean Monsoon Gyrao  14.0  64.0  2900  34  27.7  i72A  28.0  0.16  34.3  1.9  0.07  2.7  0.70  30.7  0.74  0.13  0.14  Indian Ocean Monsoon Gyros  14.0  64.0  2900  34  27.7  172.4  28.0  9/6  18.4  1.1  0.04  1A  0.76  16.3  0.73  0.12  0.15  6.33  Indian Ocoon Monsoon Gyros  14.0  64.0  2900  34  27.7  172A  29.0  0.16  41.4  2.6  0.09  2.9  0.90  36.9  0.66  0.14  0.20  4.82  Indian Ocean Monsoon Gyros  10.0  65.0  2363  36  26.2  i32A  21.7  0.16  24A  1A  0.06  1.8  016  21.8  0.70  0.14  0.17  5.05  Indian Ocean Monsoon Gyros  10.0  65.0  3915  36  28.2  i32A  21.7  0.16  21.3  12  0.06  1.0  019  16.0  0.67  Old  0./S  d.78  Indian Ooaan Monsoon Gyros  15.2  892  1717  U  28A  i27A  22.0  0.18  31.7  2.0  0.09  1.8  1.39  27.9  0.44  0.25  0.3/  1.76  Indian Ocean Monsoon Gyras  152  892  1717  U  28.4  i27A  22.0  0.18  32.2  2.2  0.10  1.3  1.73  26.1  0.37  028  0.35  133  Indian Ocean Monsoon Gyros  132  8dA  2282  U  28A  i27A  22.5  5.18  43.2  2.5  0.11  2.0  129  38.5  0A2  021  0.37  2.00  Indian Ocean Monsoon Gyros  132  844  2282  U  284  127A  22.0  0.18  49.8  29  0.13  1.9  1.53  443  0.36  0.24  040  1.52  Indian Ocean Monsoon Gyros  132  8dA  2282  U  29A  i27A  22.5  0.18  63.5  3.1  0.14  1.7  1.82  57.6  0.25  0.24  0.5/  1.05  Indian Ocean Monsoon Gyroo  4.5  87.3  3006  25  26.6  103.1  16.8  0.16  38A  2.1  012  22  0.94  34.5  0.04  021  0.25  2.08  5.78  Indian Ocean Moosoon Gyros  4.0  87.3  3006  25  28.6  103.1  16.8  0.16  31.0  1.9  0.U  1.0  1.02  27.5  0.50  031  0.14  116  Pacific Subarctic  50.0  -190.0  3260  64  0.7  202.3  96.7  CAB  42.0  1.0  0.01  1.4  0.71  40.7  0.28  0.69  0.03  SAl  Pauiflo Subarobu  50.0  -105.0  5090  64  5.7  202.3  96.7  0.48  34.6  0.8  0.0/  0.9  0.65  33.1  0.21  0.73  0.06  0.29  Pacific Subaroflo  44.0  -205.0  2960  60  7.9  231.5  101.7  0.44  57.5  2.2  0.02  1.9  1.22  53.3  0.20  5.63  0.09  0A6  PaoffioSubarotio  44.0  -205.0  4099  60  7.9  231.8  101.7  0.44  43.5  1.6  0.02  1.2  1.30  40.5  0.26  0.61  0.14  042  Paoiflooubarotio  44.0  -205.0  4990  60  7.9  231.9  101.7  0.44  48.6  1.0  0.00  1.1  1.33  45.8  0.20  0.67  0.13  0.30  KuroshioCurnant  40.0  -195.0  2986  72  14.2  204.2  04.5  0.27  35.0  1.3  0.02  1.6  0.81  32.6  SAl  0.54  0.54  016  KuroshioCurront  40.0  -195.0  5025  72  14.2  204.2  54.5  0.27  28.3  1.1  0.02  1.2  0.94  26.3  0.37  0.57  0.06  0.65  Kurouhio Currant  34.2  -216.0  3423  45  20.0  220.7  46.8  0.23  35.5  0.8  0.02  1.9  0A2  34.3  0.46  0.27  0.27  1.74  Kuroshlo Current  34.2  -218.0  5429  45  20.0  2207  49.8  023  26.7  0.6  0.0/  1.0  0.67  25.5  0.31  0.28  GAO  1.12  Koroshio Current  30.0  -1850  3873  36  238  937  15.3  0.16  17.3  1.0  0.06  I 3  0.73  15.5  0.71  0.10  0.19  7.16  Wast Paofio Warm Pool  120  -2257  4300  37  288  526  9A  018  21  02  0.02  0.2  0.99  2A  0.61  0.20  0.19  3.12  Wast Pacoc Warrrr Pool  5.0  -2212  3130  25  29.1  712  12.0  0.17  4.1  0.3  0.02  0.3  0.79  3.6  0.74  5.19  5.57  3.83  West Pauou Worm Pool  4.1  -2231  4574  25  29.1  71.2  120  0.17  34.6  2.2  0.18  hA  1.56  305  0.39  0A2  0.19  0.92  West PacSo Warm Pool  3.5  -225.0  1590  23  29.3  93.6  13.8  0.17  58.7  3A  025  3.1  ISO  52 3  0.50  0.29  0.22  1.75  Wost Pauifio Worm Pool  3.0  -225.0  3900  23  29.3  83.6  13.8  0.17  56.7  2.9  02/  2.8  1.02  51.3  0A6  0.29  0.25  1.58  West PauSu Warm Pool  0.0  -184.8  4363  36  28.8  912  15.6  0.17  13.3  0.6  0.04  1.1  0A9  12.3  017  0.19  0.04  4.09  Wastarn Pacific Arohipalago Deep Basin  -i 7.8  -2052  2304  35  26.6  62.3  10.5  0.17  3.0  0.2  0.02  0.3  0.77  2.5  0.94  0.07  0.09  1128  Pacific Sobarutic  50.0  -145.0  3900  59  9.1  195A  79.7  041  48.2  2.5  0.03  2.2  1.10  436  0A3  0.48  0.09  0.99  Pacific Suborc6c  50.0  -145.0  3800  59  9.1  195A  79.7  0.41  3SA  0.8  0.01  1.7  GAO  28.9  0.50  043  0.07  1.10  Pacific Saburotic  50.0  -145.0  3800  59  9.1  /95.4  79.7  0.41  44.1  1.1  001  2.7  0.41  42.0  0.54  0.45  0.01  1.19  Puo6oSubarcfic  50.0  -145.0  3800  59  9.1  195.4  79.7  SAl  30.7  0.8  0.01  1.5  057  29.1  SAl  0.57  0.01  013  Paoifiuoubarctic  50.0  -145.0  3800  59  9.1  190A  79.7  0.41  32A  1.1  0.01  1.7  0.65  30.3  0.48  0.47  0.06  1.01  North Paurto Equatorial Counleruurmnt  U.S  -140.0  1600  25  26.6  93.4  10.2  0.16  15.2  0.9  0.06  1.0  0.85  13.6  0.62  0.34  0.54  1.82  North Pacific Equatorial Coontorcurrant  U.S  -140.0  3400  25  26.6  93.4  15.2  0.16  13.2  0.6  0.04  0.9  0.68  12.1  060  0.36  034  1.67  North PauBo Equatorial Courtarcurrnrt  9.0  -140.0  2250  23  27.3  049  155  0.16  8.3  0.6  0.04  0.6  0.93  72  0.69  0.26  0.04  2.64  North Pacific Equatorial Countercurrant  9.0  -140.0  2150  23  27.3  94.9  15.5  0.16  5.7  0.6  0.04  0.7  0.93  7.6  0.72  0.23  0.05  3.10  North PauSo Equatorial Countorcurrant  5.0  -139.8  2100  32  27.1  / 07.5  17.9  0.16  27.3  1.6  0.09  2.2  0.76  243  074  0.24  0.02  3.08  North PacificEquatorial Countarcurrant  5.0  -139.5  3800  32  27.1  107.5  17.6  0.16  24.6  IA  0.08  1.9  0.74  21.9  072  0.26  0.02  2.72  North Pacific Equatorial Couotorcurrant  2.0  -140.1  2203  34  26.2  /21.8  16.6  0.10  26.8  1.3  0.06  2.2  0.59  244  074  0.28  -0.02  2.65  North Pacific Equatorial Countercurrent  1.0  -140.5  1895  3d  262  121.8  199  0.16  26.9  1.6  0.08  1.0  0.85  24.0  0.63  0A3  -0.06  1A7  North Pacific Equatorial Couotorcorrent  1.0  -140.0  3495  3d  26.2  /21.8  19 9  0.16  21.8  1.2  0.06  1.5  0.80  19.6  0.63  OAO  -0.03  1.58  North Paoifio Equatorial Coustorcurrant  1.0  -140.0  /983  3d  262  /21.8  19.9  0.16  54A  2.5  012  3.6  0.69  49.9  0.60  0.46  -0.06  129  North Pacific Equatortal Countercurrent  1.0  140.0  2905  34  26.2  121.8  10.5  0.16  41.7  1.9  0.10  2.8  0.69  382  0.60  0.46  -0.06  1.3/  Pacoc Equatooal Dioer5ance  0.0  -140.0  2284  31  26.2  /26A  20.6  0.16  35.2  1.7  0.08  2.8  0.60  32.0  013  0.24  0.04  3.10  Pacific Equatonal Dioargnnce  0.0  -140.0  3618  31  26.2  126.4  20.6  0.16  34.8  1.6  0.08  2.8  0.58  31.8  013  0.31  -0.03  2.37  Pao0o Equalorial Divnrganca  -2.0  -139.8  3593  31  26.2  /26A  20.6  0.16  31.2  1.3  0.06  2.4  0.55  281  010  0.32  -0.02  2.16  South Pacific Subtropical Gym  -5.0  -140.0  2099  45  26.8  104.8  17.4  0.17  22.4  1.0  0.06  1.7  0.58  20.6  0.70  035  0.05  2.75  South PauSo Subtropical Gyra  -5.0  -140.0  2209  45  26.8  1048  1TA  0.17  23.6  1.0  0.06  1.6  0.56  21.7  0.68  028  004  2.38  South PaooooubtmpioalGyre  -5.0  -140.0  2316  45  26.8  104.8  17A  0.17  23A  1.0  0.06  1.8  0.57  21.5  0.70  027  0.04  2.62  Sooth Pacoc Subtropical Gyra  -12.0  -135.0  3584  38  27.6  88.8  14.9  0.17  7.0  0.3  0.02  0.7  0.35  TA  0.84  0.16  0.01  5.35  Antarctic  -50.2  5.9  3U0  122  4.6  90.9  21.9  0.24  6A  0.5  0.02  0.3  1.70  0.5  0.46  CAb  0.10  1.02  Antarctic  -54.3  -3A  2i9d  /18  0.9  59.1  13.8  0.23  285  0.5  0.04  0.2  2A3  27.9  0.06  0.71  0.23  0.09  Anlarofic  -54.3  -3.3  2201  /18  0.9  59.1  13.8  0.23  3.1  0.6  0.05  0.1  7.88  1.9  035  028  037  126  Antaroso  -54.3  -3.3  2251  119  0.9  59.1  13.8  0.23  7A  0.1  0.01  5.0  6.50  0.02  019  0.19  0.03  Antarotlo  -57.0  -37.0  2000  93  CA  58.6  16.6  0.28  8.0  0.1  0.01  0.1  0.83  7.1 y.5  0/3  1.01  -0.14  0.13  -56.9  -1702  4224  154  2A  97.9  25.0  026  24.6  0.7  0.03  1.4  0.51  23.2  0.51  OAO  0.09  1.29 30  Table 2 Correlatixn matrix  F  F.  1.000  PP  EP  jxjy  gmy  0.711  0.500  TE  f.,ax  SST  0.139  -0.172  0.325  -  TE  0.711  3 500  -0 142  -0.389  -0483  0.575  PP  0.500  -0.142  1.000  0.780  0.362  -0.185  OP  0.139  -5.389  0.760  1.000  0.871  -0.567  f.e  -0.172  -0483  0.362  0.871  1.000  -0.706  SST  0.325  0.575  -0.185  -0.567  -0.766  1.000  --  31  Table 3. Linear regression equations, regression nostfinients, and  against measured dote  essian Cseftinient 0 7 fpg  — —  Modal Type Single-ballast  Eq# 3a  F  3b  F= h,F, F,,  3 CaCO  Enpptinn  -  Opal  0.739  h,F,  5.541910’  0.108 1.3948.101  0428  Fr,,  =  h,-F,+ b,-r-’  4b  F,,,  =  k,F,” 6,2’  40  F,,,  =  6,-F,,” Ic-i’  44  F,,,  =  ” 6,-i’ 1 lçF  16a  TE  =  6,F,,” 6,z’  160  TO  =  6,-F,. h,-z’  lee  TEk-F,+6,-z”  104  TE  =  Ic-F,n h,z’  t7a  TE  =  6F° ki’° beSOT  2.0943-10’  175  TE  =  h,F,+ 9,-i’-” 6,,,,,-Lrada  3.007210’  12  F,,,  =  b,F,  +  h,-F,+ kF  9.042610’  9.709510’  5.0943102  13a  F  =  SF  +  6F+ h,F,  8.8753.102  1.170910’  50447.102  2.847310’  135  F,,,  =  6,-F,  +  b,F,+ 0,-F, = lç,,;f’raeo  8.1969-10’  2.7798.102  7.8132182  -1.0335  14  F,,,  =  6,F,  +  6,F,+ k,F,  +  6,i’  8.8710102  1.4100-10’  7.9258-10’  is  F  =  6,-F, = 6,-F,. 0,-F,  +  6,-i’ = 5 n EP  8.9113.10.2  1.1619.10.2  7.9060.10.2  19  TO  6,i-’  2.8372-10°  5.0911-10’  2.6612-10’  1.5214-10’  1.0631.10.2  2.3596-10-’  1.1976-10’  2.9703-10’  2.1761-10’  20a 205  6,F,  +  6.1521 tO’ 1 1180W’ 64th-b’ 1.3795-by’ I .6691-10’ 3 2916-10’ 1.1077 10-’ 42938.10.2  6,F,+ 6,-F,  ‘“  6,F,” 6,F,+  TO  6,-F, • 6,-F,. t -F, 1  =  OP 0 744  4a  =  f-rabe  SST  1.1201-tO’  F,,,b,-F,  MuIt,pln-ballaet  Lithogenin  0 096610’  =  =  ——  +  +  h,SST  Icr, + 5,i’ 5,,00T +  6,-z”+ heauf-ratm  1.559910 -  -  I 7268-10.’  2.1079-10.’  0.787  1.471010+2  0.760  2.4498-10”  0.207  1.4391-10”  0.445  1 4534.10+2  0427  t.3t37-1O”  0470  1.519410”  0.135  1.2906.10+2  8.340  1.0099.10+2  6566  8.5165-10”  0.055 0.597 0.967 0574  I .1142-10’  2.3169-10’ -3.7008-10.’  1.4909-10”  0.696  I .5624-10”  0.888  1.2079-10.’  0.550  1.0143.10+2  0.646  6.2472-10”  0.968  32  Table 4. GENIE-I simxlafian resalls lime-series (glabal)  -  Experiment  Ocean Made  cpLbxn Emissians Inxenfx)y (G))  980  C8O  Sea Sxrface pH  0  8  20  3  0  371  780  1301  981 1080  +clamate  278  372  778  1417  +calcjfjxafiax  278  370  702  1263  908  -cclimafe +calcificafixn  278  371  770  1378  1008  xx feedbacb  278  371  760  1300  961  +xlimafe  278  378  777  1415  1072  370  756  1286  949  278  371  773  1396  1050  M8O  +cbmate +xalcificafiax  278  371  773  1392  1044  NBO  nx feedback  8.1  8.1  7.8  7.6  7.7  xxl,mate  8 1  8.1  7.8  7.5  7.6  -cxalcihcalixx  8.1  8.1  7.8  7.8  7.7  -cclimafe +calcificabxx  8.1  8.1  7.8  7.6  77  xx feedbacb  8.1  8 1  7.8  7.6  7.7  xclimafe  8.1  9.1  7.8  7.5  7.7  +calcifcabxx  8.1  8 1  7.8  7.6  7.7  +xlimafe +xalcifxabxx  8.1  8 1  7.8  7.6  7.7  M8O  +climate xcalcifxa5ax  8.1  8.1  7.8  7.8  7.7  N80  xx feedback  5.20  4.45  2.70  1.71  2.17  +climafe  520  4.51  2.80  I 88  2.29  xcelcdixafixx  520  449  2.79  1.85  239  520  4.54  2.93  202  2.54  xx feedback  521  443  2.71  1.72  2.18  +xlimata  5.21  448  2.86  1.89  2.31  xcalxipcabxx  5.21  4.49  2.78  1.82  2.31  xxlimafe +calciflxa5xx  5.21  4.54  2.92  2.00  247  M80  i-climate +calcifxa9xx  5.20  4.53  2.92  2.00  248  990  xx feedback  0.0  34  7.3  3.8  04  xclimafe  0.0  32  6.5  2.8  0.5  4-calxifixafixx  0.0  3.4  7.7  3.8  0.4  +climafe 4-calcihca9ax  0.0  3.2  6.8  3.1  0.5  xx feedback  0.0  3.6  7.3  3.8  0.4  i-climate  0.0  3.5  6.8  2.9  06  --  —— —  -ccalxificefiax  0.0  3.5  7.4  3.6  0.2  +climata xcalcifca5xx  0.0  3.2  6.7  2.9  0.5  MBO  xxlimete xcalxifxesxx  0.0  32  8.7  2.9  0.5  980  xx feedback  1.3  1.3  1.3  1.3  1.3  i-climate  1.3  1.3  1.2  11  11  xcalcifcafixx  1.3  1.1  0.6  0.3  0.5  4-climafe 4-celcoxabxx  1.3  1.1  0.6  0.3  0.5  xx feedback  12  1.2  1.2  1.2  1.2  xclimafe  12  1.2  1.2  1.0  1.1  -ccalxifixatixx  1.2  1.1  0.6  0.3  0.5  06  0.3  05  0.6  0.4  0.5  -  C8O  i-climate 4-calxibcabxx  POC Expxrt Pmdaxhxx (Of C pr)  4193  278  CBO  —  3000  4140  +calcifxabxx  CBO  3 Expxo Prxdacbxx (Of C yr’) CaCO  2300  1937  278  +climate +calcificabpn  Atm-Ocx lCO )Gf C yri)  2160  399  +xlimafe -cxalxifxa9xx  CBO  Sea Sxrface 0 cx  2605  4  xx faedbacb  Carban Emisaienx Rate (GI C pr’) 2 (ppm) Afmxsphekc pCO  1765  1.2  1.1  14  1.1  xx feedback  9.0  9.0  9.0  9.0  9.0  i-climate  9.0  5.9  9.5  7.8  9.0  +calcificafixx  9.0  9.0  9.0  90  9.0  +xlimate .calc,ficaoax  9.0  8.9  9.5  7.8  8.1  xx feedback  8.8  88  88  8.8  88  xclimate  8.8  8.7  8.3  7.5  7.7  i-cafcificafixx  8.8  88  89  9.2  92  i-climafe +cafcifcafixx  8.8  8.7  84  8.0  8.3  i-climate +calcifcatxx  9.0  8.9  8.6  8.1  8.5  M80  i-climate +xalcificabxx  N80  C80  M80  -  —  —  33  Figure 1. Data Figure la. 79 sediment trap locations showed on the GENIE-i Longitude-Latitude horizontal grid Figure lb. Sea surface PP field on the GENIE-i grid Figure ic. Sea surface SST field on the GENIE-i grid Figure Id. Sea surface MLD field on the GENIE-i grid Figure le. Sea surface f-ratio field on the GENIE-i grid Figure if. Sea surface EP field on the GENIE-i grid  34  a  Sediment Trap Locations •  o x  ÷ *  0)  0  D  V  <1 > * 0  •  —260  —200  b  —140  —20 —80 Longitude  40  North Atlantic Drift North Atlantic Subtropical Gyre Eastern Tropical Atlantic South Atlantic Tropical Gyre Northwest Arabian Upwelling Indian Ocean Monsoon Gyres Pacific Subarctic Kuroshio Current West Pacific Warm Pool Western Pacific Archipelagic Deep Basin North Pacific Equatonal Counteivurrent Pacific Equatorial Divergence South Pacific Subtropical Gyre Antarctic  100  2y ) 1 Primary Production (g C m above 300  — 280 300 260 280 — 240 260 —  —  220 240 200 220 180 200 160 180 140 160 120— 140 100 —120 80 — 100 60 80 40 60 20—40 0 20 below 0 —  —  0)  —  -o  —  4-. 4-  —  Cu -J  — — — —  —  —  —  Longitude  Sea Surface Temperature CC)  C  — — — — -o 4-I  -J  — — — — — —  —260  —200  —140  —20 —80 Longitude  40  bcye  .t,—  .5— —2 .5— —2 .5— -1 .5.16 .5—i -1% .5— —10 .5— —7. .5— —4. .5— -1. elowO  100  Figure 1 a-c 35  Mixed Layer Depth (m)  —1 elow 0  Longitude  f—ratio  —6O1 —90 —260  —200  —140  —20 —80 Longitude  40  100  Export Production (g C m ) 1 y 2 ov  :  elow 0  —260  Figure 1  —200  —140  —20 —80 Longitude  40  100  d-f 36  Figure 2. Data analysis results Figure 2a. F 00 measured from 79 sediment traps at different depths> 1500 m Figure 2b. TE measured from 79 sediment traps at different depths> 1500 m . modeled with Equation (2a) 0 . measured and F 0 Figure 2c. Linear regression between F Figure 2d. Linear regression between b measured and b-modeled with Equation (2c) . modeled with Equation (2e) 0 Figure 2e. Linear regression between F . measured and F 0 Figure 2f.  and Fm normalized to F,  Figure 2g. F , F 0 0 and F normalized to Fm . and Fm, F 0 , F 0 , F, 0 Figure 2h. Correlation between F 00 modeled with Equations (3a-3d) Figure 2i. Linear regression between F . measured and F 0 . modeled with Equations (9a-9d) 0 Figure 2j. Linear regression between Fpm, measured and F Figure 2k. Linear regression between F . measured and F 0 . modeled with Equations (12k, 13a, 13b) 0 Figure 21. Correlation between TE and Fm, F , F 0 , F 0 Figure 2m. Linear regression between TE measured and TE modeled with Equation (16b, 17a, 17b) Figure 2n. Linear regression between TE measured and TE modeled with Equation (19, 20a, 20b)  37  Transfer Efficiency (Foc/EP)  Fpoo (g m 2y ) 1 15000  2  .  4  3  5  A  a  0.1  Q05  6  +  b 2000  2000  0.25  00 *  *  0  DO  2500  2500  ,  3000  0.2  I  ..1  ODD  0.15  *  A•  3000  *  E E 3500 a, 0 4000  :>v  E3500  x  4000  A  *  * *L  *  *  V **  3c  Vx  0. A 4500  4500  5000  5000 V  5500  d —0.4 —0.6  j,.  —0.8  E 0)  0) •0  (0  a) :3  a) E  a) E  .0  /vO —1  •0  •:. —1.2  .  L  —1.4  /..  —1.6  —i.e  —1.8  F modeled  =  —1.6  EP•(zzo)°  o x  + *  V ‘  E  1  0) 0  * 0  •  a) E  =0.762 2 R —1.4  —1.2  —1 —0.8 b modeled  —0.6  —0.4  —0.2  North Atlantic Drift North Atlantic Subtropical Gyre Eastern Tropical Atlantic South Atlantic Tropical Gyre Northwest Arabian Upwelling Indian Ocean Monsoon Gyres Pacific Subarctic Kuroshio Current West Pacific Warm Pool Western Pacific Archipelagic Deep Basin North Pacific Equatorial Countercurrent Pacific Equatorial Divergence South Pacific Subtropical Gyre Antarctic  LL  0  1  2  4 3 5 6 7 8 F modeled = 8t EP.(zIza)°  9  10  Figure 2 a-e 38  Flux/Ft  ,,0  0.1  0.2  0.4  0.3  Flux/Fm  0.5  0.7  0.6  0.9  0.8  0.1, 0,2 ‘•x  ,.,0  0.4  0.3  0.5  0.6  0.7  0.8  *  *  0.9  *• * **Kx  2000  **  2000 +  ** *Z xao< K-”  3000  ‘a  +  2  *a,  K  2500  2500 3000  *x  *KKSt  +  ++ + 4+  3500  C  -r  4000  2 3500 e  C  5000  •  *  *  +  ** **  * *flxxtx  *xx  *  K *  K  4500  . 1 •t  Fpoc/Ft FSFt  * *  X)OO  t  *  *  >4  -*  K K  4000  +  * *  ++  4500  x  K  Fc/Fm 5000  **  *  x  5500  4,-nfl,  *.  I  Fo/Fm  Fu’Fm  K  -  h 5  ++ K  +  *  4  I>,  2 I>  x  *  x  *  23 0)  --  I  østx  <4* * 1  +Hi * *  *±++ ++ *+ ++  r  4 X  0  M-  10  + + * ++ *  20  a) 0) to  +++  a)  2  +  ++  2#  •0  +  ++  +++  **  0)  +  +  ft  +  U-  + +  Fm, R = 0.863 ,R=0.860 0 F F, R = 0.387  x  Fi, R  ++  40 30 S (g m  50  60  =  0.654  70  3 F modeled  o  x + *  D  O V  I,.,  2  <1 >  0)  I: 0 1 2 3 4 5 F modeled = F..i  * O  •  +  North Atlantic Drift North Atlantic Subtropical Gyre Eastern Tropical Atlantic South Atlantic Tropical Gyre Northwest Arabian Upwelling Indian Ocean Monsoon Gyres Pacitic Subarctic Kuroshio Current West Pacific Warm Pool Western Pacific Archipelagic Deep Basin North Pacific Equatorial Countercurrent Pacific Equatorial Divergence South Pacific Subtropical Gyre Antarctic  6 7 8 9 10 11 12 13 14 15 16 [EP — Fp]•e ZOl/500] F,,l.) = p•Fb(..) -  Figure 2 f-j 39  -x  +  Fm,R=0.551 Fc, R0.612 F,R=0.120 Fi, R = 0.497  *  x  0.2 *  x  0  +  x  +  *  0.15 x  * x  *  *  La  *%%  +  +  +  0.1 ci,  +  +  x  XX  +  +  *  -Ii--  +  +:!:  +  +  +  +* + -c++ ++ ++  I-  +  0.05  fr.  _+_i+  r*:-  “0  Fpcc modeled  10  20  +  30  +  +  40  +  50  60  70  Flux (g m 2y ) 1  0 Lu LL 0 La Ca ID  E Lu I—  V  0.05  0  o x + *  <) V  <1 * r  •  , R TE(Fc,z ) 1 2 = 0.470 2 = 0.566 TE(F,z,SST), R TE(Fc,Z,f—ratio), R 2=  0.1 0.15 TE modeled  0.2  0.25  0.25  North Atlantic Drift North Atlantic Subtropical Gyre Eastern Tropical Atlantic South Atlantic Tropical Gyre Northwest Arabian Upwelling Indian Ocean Monsoon Gyres Pacific Subarctic Kuroshio Current West Pacific Warm Pool Western Pacific Archipelagic Deep Basin North Pacific Equatorial Countercurrent Pacific Equatorial Divergence South Pacific Subtropical Gyre Antarctic  Figure 2 k-n  40  Figure 3. GENIE-i simulation results time-series (global) Figure 3a. Global carbon emissions inventory from historical record and prescribed future scenario Figure 3b. Global carbon emissions rate from historical record and prescribed future scenario Figure 3c. Global atmospheric pCO 2 trajectory obtained from model simulations (NBC) Figure 3d. Global atmospheric pCC 2 trajectory obtained from model simulations (CBO) Figure 3e. Difference in atmospheric pCO 2 with respect to ‘no feedback’ (NBC) Figure 3f. Difference in atmospheric pCC 2 with respect to ‘no feedback’ (CBO) ] and [CC 3 Figure 3g. Global mean sea surface [H], [CC ], [HCC 2 ] (NBC) 2 3 Figure 3h. Global mean sea surface [H], [CC ], [HCC 2 ] and [CC 3 ] (CBC) 2 3 Figure 3i. Global mean sea surface calcite saturation state (NBC) Figure 3j. Global mean sea surface calcite saturation state (CBC) Figure 3k. Global CaCC 3 export production (NBC) Figure 31. Global CaCO 3 export production (CBC) Figure 3m. Global POC export production (NBC) Figure 3n. Global PCC export production (CBC) Figure 3o. Global mean sea surface [PC ] (NBC) 4 Figure 3p. Global mean sea surface [PC ] (CBC) 4  41  21 20 19 18 17 16 0 15 14 13 12 11 10 89  4000  r 3500  2500 2000  8 w  C 0 .0 ‘5  b  Lii  1500  7 6 5 4 3 2  0 1000 500  1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2600 2900 30 Year  1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year  CBO  NBO 1400  C  1400  1300  1300  1200  1200 1100  1100 E a a 1000  8  900  900  1000  800  800 -C  -C  a  a  700  700 8 <  600  600  500  500  400  400  300  —  300 +climate i-calcification  1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year  1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year  NBO  >  8  a a  0  160 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 —10 -20 -30 -41) —50 —60  — —  no feedbacks ÷climate i-calcification i-climate i-calcification  CBO  .  e  .....  —  —  // 1 1  .  /  :1  .  +climate +calcification  —  : -  1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year  150 140 130 120 110 100 90 80 > 70 8 a 60 a 50 0 40 30 20 10 —10 -20 -30 -40 —50  — —  no feedbacks i-climate i-calcification i-climate i-calcification  ....,  —  .-  —-  —  —  .  —  —  —  -  —  : :  /  —  /  -,  -,  -/  .  -, / -  -  ——--  1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year  Figure 3 a-f 42  CBO  NBO 1 0_2  P  • •:•  h  .VVH  g  -  ——C02  -  _.;____._._____._ V  -  V  I  —C03  V  H C02 -—HCO3 —C03  0  ------------------  E 0  I 8  6 io  io  I  a 10  I  0)  Co  10  a  B)  -7  10_B  1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year  1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year  CBO  NBO -  V  V  —.-----•  :  5  —  :  -  —  —  no feedbacks -i-climate -i-calcification -i-climate i-calcification  4.5 a  8  -  4  -  V  8  3.5  -  z  0  Cl)  0)  a a  C,)  3  -  a SI  V  (0  2.5  -  2  1.5  —  -  V  V  1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 Year  1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3( Do Year  CBO  NBO 1.3 V V••  V•  no feedbacks  :  : :  V.:  ‘  -  -  -  1.2  V  V  \  V  —  V  •>. 1.1 0  V  —  V  —  V  i-climate -i-calcification +climate +calcification  -  1.2  V V V  +cllmate  V  :  V •  —  V  —  V -  i)_  V  1.1  0  —  V  i-calcification i-climate i-calcification  \  a  a0 C 0.9 C 0  V  0.9  \\  C 0  0.8 2 0.7  -  t  0  0.6 0.5  0.5 0.4  0.7  8.  \\  w 0.6  0.8  V  k  V  V  -  -  :  1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year  0.4  -  VI  ___  —  -  1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year  Figure3g-I 43  CBO  NBO 9.’ 9.3 9.2 9.1  no feedbacks  rn  .  —  —  —  —  — —  —  —  —  —  ‘  -  . —  +calcification +ciirnate icalcification  58.7 8.6 0  .  83 E 082 —  —  .  87.9 7.8 7.7 7.6 7.5 1800 1900 20002100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year  -.  ....  I .1  1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year  NBO 66 6.7 6.6 6.5 6.4 6.3 6.2 6.1 6 5.9 5.8 5.7 5.6 5.5 5.4  .  —  -  7.9 7.8 7.7 7.6 7.5 7.4  .  —  —,  81 8  8  +UThite  .  92 9.1  +calcification +climate ÷calcification  —9 8.9 8.8 o 8.7 8.6 8.5 0 8.4 8.3 0 8.2  no feedbacks  — —  .  CBO  7 x10  0  -  — — —  no feedbacks +climate ÷calcification ÷climate +calcification  6.7 6.6 6.5 6.4 6.3 6.2 6.1 6 5.9 -r 5.8 .?‘ 5.7 5.6  no feedbacks  p — —  • -  -  .  — — i-calcification — i-climate +calcification  /  /  .g .s  0  a 5.2 5.1 .t 5 c 4.9 a 4.8 4.7 4.6 4.5 4.4 4.3 4.2 4.1 4 3.9 3.8  N  _._-.  —  1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year  -s 5.4 5.3 5.2 5.1 5 4.9 a 4.8 4.7 4.6 4.5 4.4 4.3 4.2 4.1 4 3.9 3.8 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000 Year  Figure 3 rn-p  44  Figure 4. GENIE-i simulation results time-slices (zonal mean) Figure 4a. Zonal mean F . at 1765, 2000, 2100, 2300 and 2300-1765 difference (NBC ‘+climate’) 0 Figure 4b. Zonal mean F . at i765, 2000, 2100, 2300 and 2300-i 765 difference (CBO ‘+cliniate’) 0 ] at 1765, 2000, 2100, 2300 and 2300-1 765 difference (NBC ‘+climate’) 4 Figure 4c. Zonal mean [PC Figure 4d. Zonal mean [PC ] at 1765, 2000, 2i00, 2300 and 2300-i 765 difference (CBO ‘+climate’) 4 Figure 4e. Zonal mean F at 1765, 2000, 2100, 2300 and 2300-i 765 difference (CBO ‘+calcification’) 0 Figure 4f. Zonal mean [PC ] at 1765, 2000, 2100, 2300 and 2300-i 765 difference (CBC ‘+calcification’) 4  45  a C.  -n Co C  a  -I  -n Co C  A  8  8  id &pth (‘I  8  C,  8  I’ll  c.  C  cc  ,,  C)  -n  0)  0  0)  a  0)  CD  -  (0  a  0  C  0)  a  0  ‘0  a  a a  a  ‘0  a  C  a  0 0  0  0  C  C,  0  a  a  (0  a  C  0  0)  0  a  a  (0  C  0  4 Oath  0 0)  a  0  a  a  (0  0  0  0  0  0  a  t  0  a  a  0  ±_11 id pth  —1..e1 üd d.pth  (a  a  0  0)  a  ‘0  a  a  0  C  (0  a  0  C  0  0  C  a  Co C  a CD  Co C  ,,  -v  -  A  8  a  a a  a  8  8  S  a  8  a  8  hi  I  Tht 22 [[.  a4 aa  \ t :l  S  a  a  8  a  8  I -—  Figure 5. GENIE-i simulation results time-slices (surface layer) 3 flux at 1765 (NBC ‘+climate +calcification’), 1765 (CBO ‘+climate -i-calcification) Figure 5a. Sea surface CaCO difference CBO1765-NB01765 and Figure 5b. Sea surface POC flux at 1765 (NBC ‘+climate +calcification’), i765 (CBC ‘-i-climate -i-calcification’) and CBO1 765-NBO1 765 difference Figure 5c. Sea surface calcite saturation state at 1765, 2000, 2100, 2300 and 2300-1765 difference (NBC ë-’-climate +calcification) Figure 5d. Sea surface calcite saturation state at 1765, 2000, 2100, 2300 and 2300-1765 difference (CBO ‘+climate +calcification’) Figure 5e. Sea surface CaCO 3 flux at 1765, 2000, 2100, 2300 and 2300-1765 difference (CBO ‘+climate +calcification) Figure 5f. Sea surface POP flux at 1765, 2000, 2100, 2300 and 2300-1 765 difference (CBO ‘+climate -i-calcification’) Figure 5g. Sea surface POC flux at 1765, 2000, 2100, 2300 and 2300-1 765 difference (CBO +climate +calcification) ] at 1765, 2000, 2100, 2300 and 2300-i 765 difference (CBO ‘-i-climate +calcification’) 4 Figure 5h. Sea surface [PC Figure 5i. Fractional sea-ice extent at 1765, 2000, 2100, 2300 and 2300-i 765 difference (NBC ‘-i-climate’)  49  01  m  1  ‘1  01  ‘1  -v  -v  8  8  0  Ct Ct  0  0  Ct  8  C,  Ct  0  Ct  8 8 8  +  0  C.  1-TI  Co C  -n  1-fl C,  =  -Il  LA  B 0)  a  0)  0’  B  a  0’  0)  B  B  0)  0)  ‘1 Co = -I CD 01  CD  (31  Co C  ,,  8  8  8  CD  8  CD  8  0  0  0  8  CC  8  8  8  C,  0  8  C,  8  8  8  0  0  C,  +  8  8  8  +  8  8  8  C  (B)  C,’  -v  m  -I  C  ,1  a’  =  ,,  B  t  I -  B  t  B  t  a t  e  C,  *olinate’ 1765  fractional xex—joe extent NBC  fractional sex—ice extent NBC ‘+climxte  2000  fractional sea—ice extent NBC  +climxte’ 2100  fractional sex—ice extent NBC  *climate  2300  fctinna1sna—1cn nxtent  fractional sea—ice extent NBC  +olinate’ 2300—1765  2300—1765  Figure 51 54  References Archer, D. & Maier-Reimer, E. Effect of deep-sea sedimentary calcite preservation on atmospheric CQ concentration. Nature 367, 260-263 (1994). . Geophys. 2 Archer, D., Kheshgi, H. & Maier-Reimer, E. Multiple timescales for neutralization of fossil fuel CO Res. Left. 24, 405-408 (1997). Anderson, L. On the hydrogen and oxygen content of marine phytoplankton.Deep-Sea Research 42, 1675-1680 (1995). Andersson, A. J., Mackenzie, F. T. & Ver, L. M. Solution of shallow-water carbonates: An insignificant buffer against rising atmospheric CO . Geology 31, 513-516 (2003). 2 ? Front. Ecol. 2 Andersson A. J. & Mackenzie, F. T. Shallow-water oceans: a source or sink of atmospheric CO Environ. 2, 348-353 (2004). Armstrong, R. A., Lee, C., Hedges, J. I., Honjo, S. & Wakeham, S. G. A new, mechanistic model for organic carbon fluxes in the ocean based on the quantitative association of POC with ballast minerals. Deep Sea Res. Part II 49, 219-236 (2002). Barker, S., Higgins, J. A. & Elderlield, H. The future of the carbon cycle: review, calcification response, ballast and feedback on atmospheric CO . Philosophical Trans. Royal SocietyA 36, doi:10.1098/rsta2003.1238 2 (2003). Behrenfeld, M. J. & Falkowski, P. G. Photosynthetic rates derived from satellite-based chlorophyll concentration. Limnol. Oceanogr. 42, 1-20 (1997). Betzer, P.R., Byrne, R. H., Acker, J. G., Lewis, C. S. & Jolley, R. R. The oceanic carbonate system: A reassessment of biogenic controls. Science 226, 1074-1 077 (1984). Bijma, J., Spero, H. J. & Lea, D. W. Reassessing the foraminiferal stable isotope geochemistry: Impact of the oceanic carbonate system. In Use of Proxies in paleoceanography: Examples from the Atlantic (eds Fisher, G. & Wefer, G.) 489-512 (1339 Springer-Verlag Berlin Heidelberg, 1999). Broecker, W. S. & Peng, T. H. Tracers in the Sea. (Palisades, New York, 1982).  2 Broecker, W. S. & Peng, T.-H. The role of CaCO 3 compensation in the glacial to interglacial atmospheric CO change. Glob. Biogeochem. Cycles 1, 15-29 (1987). Caldeira, K. & Wickett, M. E. Anthropogenic carbon and ocean pH. Nature 425, 365 (2003). Caldeira, K. & Wickett, M. E. Ocean model predictions of chemistry changes from carbon dioxide emissions to the atmosphere and ocean. J. Geophys. Res. 110, doi:10.1029/2004JC002671 (2005). Catubig, N. R. et al. Global deep-sea burial rate of calcium carbonate during the Last Glacial Maximum. Paleoceanography 13, 298-310 (1998). Elderfield, H. Carbonate mysteries. Science 296, 161 8-1 621 (2002). Eppley, R. W. & Peterson, B. J. Particulate organic matter flux and planktonic new production in the deep ocean. Nature 282, 677-680 (1979). Fabry, V. J. Shell growth rates of pteropod and heteropod mollusks and aragonite production in the open ocean: 55  Implications for marine carbonate system. J. Marine Res. 48, 209-222 (1990). 3 System in the Oceans. Science 305, 362-366 2 on the CaCO Feely, R.A. et al. Impact of Anthropogenic CO (2004). Francois, R., Honjo, S., Krishfield, R. & Manganini S. Factors controlling the flux of organic carbon to the bathypelagic zone of the ocean. Glob. Biogeochem. Cycles 16, doi: 10.1029/2001 GBOO 1722 (2002). Frankignoulle, M., Canon, C. & Gattuso, J. P. Marine calcification as a source of carbon dioxide: positive . Limnol. Oceanogr. 39, 458-462 (1994). 2 feedback of increasing atmospheric CO Ganeshram, R. S. Oceanic action at a distance. Nature 419, 123-125 (2002). Gattuso, J-P. et a!. Photosynthesis and calcification at cellular, organismal and community levels in coral reefs: a review on interactions and control by carbonate chemistry. American Zoologist 39, 160-183 (1999). Hamm, C. E. Interactive aggregation and sedimentation of diatoms and clay-sized lithogenic material. Limno!. Oceanogr. 47, 1790-1795 (2002). 3 export production in the greenhouse. Geophys. Res. Lett. 13, Heinze, C. Simulating oceanic CaCO doi:10.1020/2004G02061 3 (2004). Honjo, S. Coccoliths: production, transportation, and sedimentation. Mar. Micropaleo. 1, 65-79 (1976). Honjo, S. Fluxes of particles to the interior of the open oceans. In Particle Flux in the Ocean, SCOPE 57 (eds Itekkot, V.,Aschauffer, P., Honjo, S. & Depetris, P.) 91-154 (Wiley, New York, 1996). lglesias-Rodriguez et al. Progress made in study of ocean’s calcium carbonate budget. EQS Amer. Geophys. Un. Trans. 83, 365, 374-375 (2002). Jansen, H., & WoIf-Gladrow, D. A. Carbonate dissolution in copepod guts: A numerical model. Mar. Ecol. Prog. Ser., 221, 197— 207 (2001). Klaas, C. & Archer D. Association of sinking organic matter with various types of mineral ballast in the deep sea: Implications for the rain ratio. Glob. Biogeochem. Cycles 16, doi:10.1029/200IGBOO1 765 (2002). Kleypas, J. A. et al. Geochemical Consequences of increased atmospheric carbon dioxide on coral reefs. Science 284, 118-1 20 (1999). Laws, E. A. et a!. Temperature effects on export production in the open ocean. Glob. Biogeochem. Cycles 14, 1231-1246 (2000). Lenton, T. M. Land and ocean carbon cycle feedback effects on global warming in a simple Earth system model. Tellus Ser. B 52, 1159-1188 (2000). Lenton, T. M. et a!. Millennial timescale carbon cycle and climate change in an efficient Earth system model. Clim. Dynam. 26, 687-711, doi: 10.1 007/s00382-006-01 09-9 (2006). 3 in the oceans. J. Geophys. Res. 74, Li, Y.-H., Takahashi, T., & Broecker, W. S. Degree of saturation of CaCO 5507 (1969). Longhurst, A., Sathyendranath, S., Platt, T. & Caverhill, C. An estimate of global primary production in the ocean from satellite radiometer data. Jour. Plankton Res. 17, 1245-1277 (1995). Lutz, M., Dunbar, R., & Caldeira, K. Regional variability in the vertical flux of particulate organic carbon in the ocean interior. Glob. Giogeochem. Cycles 16, doi:10.1029/2000GB001383 (2002) Martin, J. H., Knauer, G. A., Karl, D. & Broenknow, W. W. VERTEX: Carbon cycling in the northeast Pacific. 56  Deep Sea Res. Part 134, 267-285 (1987). Martin J. H., Fitzwater, S. E., Gordon, R. M., Hunter, C. N. & Tanner, S. J. Iron, primary production and carbonnitrogen flux studies during the JGOFS North Atlantic Bloom Experiment. Deep Sea Res. Part 1140, 115-134 (1993). Milliman, J. D. eta!. Biologically mediated dissolution of calcium carbonate above the chemical lysocline? Deep Sea Res. PartI46, 1653-1669 (1999). Monterey, C. & Levitus, S. Seasonal variability of mixed layer depth for the world ocean. NOAA Atlas NESDIS 14, 100 (U.S. Gov. Print. Off., Washington D. C., 1997).  Mortlock, R. A. & Froelich, P. N. A simple method for the rapid determination of biogenic opal in pelagic marine sediments, Deep-Sea Research 36, 1415-1426 (1989). Orr, J. C. et a!. Anthropogenic ocean acidification over the twenty-first century and its impact on calcifying organisms. Nature 437, 681-686 (2005). Passow, U., Alldredge, A. L. & Logan, B. E. The role of particulate carbohydrate exudates in the flocculation of diatom blooms. Deep Sea Res. Part 141, 335-357 (1994). Passow, U. Switching perspectives: Do mineral fluxes determine particulate organic carbon fluxes of vice versa? Geochemistry Geophysics Geosystems 5, doi: 10.1 029/2003GC000670 (2004). Passow, U. & De La Rocha, C. Accumulation of mineral ballast on organic aggregates. Glob. Biogeochem. Cycles 20, doi: 10.1 029/2005GB002579 (2006). Ridgwell, A. J. An end to the “rain ratio” reign? Geochemistry Geophysics Geosystems 4, doi:10.1 029/2003GC000512 (2003). Ridgwell, A. J. et a!. Marine geochemical data assimilation in an efficient Earth System Model of global geochemical cycling. Biogeosciences 4, 87-104 (2007). Ridgwell, A. J. et a!. Assessing the potential long-term increase of oceanic fossil fuel CO 2 uptake due to CC 2 calcification feedback. Biogeosciences 4, 481-492 (2007). Riebesell, U. et al. Reduced calcification of manne plankton in response to increased atmospheric CO . Nature 2 407, 364-367 (2000). Sarmiento, J. L. & Bender, M. Carbon biogeochemistry and climate change. Photosynthesis Research 39, 209234 (1994). Sarmiento, J. L. et al. A new estimate of the CaCO 3 to organic carbon export ratio. Glob. Biogeochem. Cycles 16, doi:10.1029/2002GB001919 (2002). Scholten, J. C. et al. Trapping efficiency of sediment traps from the deep Eastern North Atlantic: The 230 Th calibration. Deep Sea Res. Part 1148, 2383-2408 (2001). Sigman D.M. & Boyle E.A. Glacial/interglacial variations in atmospheric carbon dioxide. Nature 407, 859-869 (2000). Steinmetz, J. C. Sedimentation of coccolithophores. In Coccolithophores (eds Winter, A. & Siesser, W.) 179-1 97 (Cambridge Univ. Press, New York, 1994). Suess, E. Particulate organic carbon flux in the oceans-surface productivity and oxygen utilization. Nature 288, 260-263 (1980). 57  2 and seasonal biological 2 flux based on climatological surface ocean pCO Takahashi, T. et a!. Global sea-air CO and temperature effects. Deep Sea Research Part l149, 1601-1622 (2002). Takahashi, T. The Fate of Industrial Carbon Dioxide. Science 305, 352-353 (2004). Tsunogai, S. & Noriki, S. Particulate fluxes of carbonate and organic carbon in the ocean: Is the marine biological activity working as a sink of the atmospheric carbon? Tellus Ser. B 43, 256-266(1991). ? Coral Reefs 2 Ware, J. R., Smith, S. V. & Reakakudla, M. L. Coral reefs: sources or sinks of atmospheric CC 11, 127-130(1992). Yu, E.-F. et al. Trapping efficiency of bottom-tethered sediment traps estimated from the intercepted fluxes of Pa. Deep Sea Res. Part 148, 865-889 (2001). Th and 231 230 Zhong, S.J. & Mucci, A. Calcite precipitation in seawater using a constant addition technique-a new overall reaction kinetic expression. Geochimica et Cosmochimca Acta 57, 1409-1417 (1993). Zondervan, I., Zeebe, R. E., Rost, B. & Riebesell, U. Decreasing marine biogenic calcification: A negative feedback on rising atmospheric pCC . Glob. Biogeochem. Cycles 15, 507-516 (2001). 2  58  Appendices 2 between the atmosphere and ocean Appendix A: Exchange of CO Carbon dioxide has a relatively high solubility and chemical reactivity in seawater. Its solubility has been determined experimentally as a function of temperature and salinity (both inverse relationship), where  2 than warmer and more temperature plays a larger role. Colder and less saline water could take up more CO saline water. Henry’s law describes the relationship between solubility (S) of gas in liquid and sea water properties. _l*  C’  solubility constant  k P  =  overlying pressure of the gas in the atmosphere  2 between the atmosphere and the ocean takes place in the surface layer of the ocean, The exchange of CO above the thermocline. Air-sea exchange occurs in the form of gas exchange and is mainly driven by the ). For every 1°C temperature increase in 2 2 between the atmosphere and the ocean (A pCO difference in the pCO  2 ocean is also influenced by the seawater, pCO 2 ocean increases by —4% (Sarmiento and Bender, 1994). pCO thermodynamic relationships between different carbon species (i.e. distribution of carbon species within the DIC 2 will diffuse into the ocean. Conversely, fPCO2 atmosphere 2 ocean conditions, CO pool). Under PCO2 atmosphere> pC0 2 to the atmosphere. The gas transfer rate depends on the magnitude of 2 ocean, the ocean would outgas CO pCO A pCO 2 and other factors such as solubility and turbulence. I  * _‘*C’ F\ 0C02 —  A -°  f’I5  P’-”-’  K = piston velocity SCO2 =  solubility of CO 2 in sea water at a given temperature and salinity  A pCO 2  2 atmosphere 2 ocean pCO pCO -  Since gas exchange across the air-sea is slower than the oceanic processes that affect the partial pressure of 2 in the surface waters, most surface waters are not in gaseous equilibrium with the atmosphere at one CO instantaneous time. The rate of gas exchange varies as warming and cooling processes of surface waters . Cold undersaturated and warm 2 changes the saturation state of the surface seawater with respect to CO supersaturated regions of the ocean are identified as places of carbon ‘sinks’ and ‘sources’, respectively (Takahashi et al., 2002; Takahashi, 2004). The time required for CO2 between the atmosphere and surface ocean to achieve equilibrium is about 6 to 12 months (Sigman and Boyle, 2000) and on this time scale, and on a global spatial scale, the atmosphere and ocean can be considered as in a steady-state with respect to C exchange if there were no anthropogenic disturbances.  59  Appendix B: Carbonate buffer 2 upon its entering the ocean. Most of the dissolved It is important to recognize the rapid speciation of CO 2 is taken up -). When gaseous CO 3 inorganic carbon (DIC) in the ocean is in the form of bicarbonate ion (HCO 2 quickly undergoes dissociation by from the atmosphere and dissolve in seawater, a large portion of the CO -) to form bicarbonate ions. 2 3 reacting with water and carbonate ion (C0 Acid-base pH equilibrium reactions: 2 CO  +  0 2 H  C 2 H < 3 —* 0W H  +  2 3 C0  C0 2 H 3  —*  +  3 HC0  3 HCO  Overall reaction: 2 CO  +  HO +C0 2 3  3 2HCO  3 by 2. DIC increases by 1. TA (total ALK) 2 by 1 unit and increases HC0 3 Adding 1 CO 2 unit decreases C0  CA  (carbonate ALK) remains the same. pH decreases (more acidic) since H is also produced (in an instantaneous sense), but doesn’t lower pH very much (well buffered). Such buffering reactions of the carbonate system maintain the pH of seawater to almost constant at —8, where deviations are largely due to kinetic issues that inhibit equilibrium to be achieved. Through this process of carbonate buffering, the carbonate ion concentration is reduced and the bicarbonate ion concentration is increased. The pH equilibrium reaction maintains an inverse 2 uptake capacity of the ocean is strongly limited ] and pCO 2 3 2 and suggests that the CO relationship between [CO 2 into seawater from biological respiration initiates 2 in the ocean. The release of CO 3 by the concentration of C0 the same buffering reaction. DIC TA  =  2 [CO  +  C0 2 [H ] 3  CA = 2[C0 ] 2 3  +  +  ] 3 [HC0  +  j 2 3 [CO  ] 3 [HC0  +  ] 2 3 [C0  ) 1 j in milliequivalents of charge present in 1 L of solution (meq L 3 [HCO  ] 2 3 Carbonate ion concentration could be approximated from DIC and CA [CO  =  CA- DIC  60  Appendix C: Calcium carbonate precipitationldissolution  3 equilibrium reactions: CaCO 2 Ca  +  —* CaCO 2 3 CO 3  3 HC0 3 HCO C0 2 H 3  + +  H  H  C0 2 H 3  0 2 2 (aq) + H CO  Overall reaction: 2 Ca  +  3 2HCO  3 CaCO  +  2 (aq) + H CO 0 2  2 Precipitation of CaCO 3 by marine organisms (Ca  +  3 2HC0  —*  3 CaCO  +  0) in the surface ocean 2 2 (aq) + H CO  2 decreases the ambient seawater DIG and ALK by 1 and 2 unit(s), respectively. The result is an increase inpCO ocean  2 uptake. It is estimated that, in (causes seawater to become more acidic), hence decrease in oceanic CO.  the modern ocean, for each molecule of CO 2 produced by calcification a fraction of about 0.6 is potentially released to the atmosphere, while the rest is taken up by the carbomte buffer (Ware et al., 1992) Here, we find a positive coupling, where the increase in CaCO 3 production ‘amplifies’ the accumulation of atmospheric CO . This 2 result may be somewhat counter-intuitive and is often easily confused.. The effect of dissolution (CaCO 3  +  2 (aq) + H CO 0 2  —  2 Ca  +  j in the deep ocean is the increase in DIC and 3 2HCO  ALK of the ambient seawater by 1 and 2 unit(s), respectively, thus partially neutralizing the accumulation of metabolic CO 2 uptake. Here, the weakening of the positive coupling 2 with depth, creating potential for future CO ‘depresses’ the accumulation of atmospheric CO . An increase in dissolution (or decrease in precipitation) in the 2 surface ocean would create a more alkaline profile water column that weakens the positive coupling.  61  Appendix D: Silicate-carbonate carbon cycle Weathering-metamorphism reactions: 3 CaSiO 2 Ca  +  +  2 2C0  +  3 2HC0  2 Ca  3 CaCO  +  0 2 2 (aq) + H CO  3 CaCO  +  2 Si0  +  2 Si0  3 2HC0  0 2 H  +  Overall reaction: 3 CaSiO  +  2 CO  2 is removed from the atmosphere by chemical weathering on land, On the time scale of millions of years, CO deposited in the ocean, subducted, and returned to the atmosphere by volcanic activity. The weathering process ) combine in soils and rock crevices to form carbonic (H 0 2 and rain 2 depends on the fact that atmospheric CO ) chemically. Rivers carry off the 3 0 , a weak acid that slowly attacks the silicate rocks (CaSiO acid ) 3 C 2 (H dissolved ions to the ocean, then some of the dissolved ions are taken up by benthic and planktonic calcifying . (SiC ) organisms that form CaCO 3 and silica opals 2 A fraction of the biogenic CaCO 3 would eventually be buried to the seafloor and be incorporated back to the geological reservoir. CaCO 3 sediments are transported downward in the subduction process and are either 3 reacts with melted or transformed by high temperature and pressure. Organic matter is decomposed and CaCQ ). This process termed 3 the silica (SIC ) found in the subducted rocks and forms calcium silicate (CaSiO. 2 metamorphism is the reverse of weathenng and is often also referred to as ‘reversed weathering’. These processes return to the atmosphere and complete (close) the tectonic-scale cycle. The reactions that summarize the chemical changes involved in the weathering and metamorphism phases of this long-term carbon cycle are mirror processes.  62  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0053295/manifest

Comment

Related Items