The ocean is a net source of atmospheric carbon monoxide (CO) (Swinnerton et al. , 1970; Bates et al., 1995), which regulates the oxidizing capacity of the atmosphere (Derwent, 1995) and acts as an indirect green-house-effect gas that is partly responsible for global warming (Zepp et al., 1998). CO in the surface ocean is produced primarily from the photolysis of chromophoric dissolved organic matter (CDOM) (Conrad et al. , 1982; Zafiriou et al., 2003) and is lost by microbial consumption and outgassing (Conrad et al. , 1982; Johnson and Bates, 1996; Zafiriou et al., 2003; Xie et al., 2005). Limited data also show that thermal degradation of DOM (Xie et al., 2005). and certain marine organisms (King, 2001) also pro duce CO.
CUITent estimates of the dissolved organic carbon sink associated with DOM photodegradation are 10-30% of total oceanic primary production (Miller and Moran, 1997; Mopper and Kierber, 2000), and are hence clearly relevant on global scales. CO is quantitatively the second largest identified inorganic carbon product of marine DOM photolysis (Mopper and Kierber, 2000). Therefore, CO photoproduction is, by itself, of biogeochemical significance. CO is also considered a useful proxy for general CDOM photoreactivity and for the difficult-to-measure photoproduction of dissolved inorganic carbon (Miller and Zepp, 1995; Johannessen, 2000; Mopper and Kierber, 2000) and biolabile carbon (Kieber et al. , 1989; Moran and Zepp, 1997; Miller et al. , 2002), which together have been proposed to be one of the major terms in the ocean carbon cycle. Moreover, CO has emerged as a key tracer for use in testing and tuning models of various mixed-Iayer processes, including photochemistry, ocean optics, radiative flux, mlxlllg and air-sea gas ex change (Kettle, 1994; Doney et al. , 1995; Najjar et al., 1995; Gnanadesikan, 1996; Johnson and Bates, 1996; 2005). Finally, CO acts as a supplemental energy source to sorne lithoheterotrophs, which mediate a major fraction of CO oxidation in ocean surface water (Moran and Miller, 2007). Therefore, any significant advances or modifications in our knowledge of oceanic CO would affect our view of other major marine biogeochemical cycles .
Existing problems in CO research
Open-ocean CO photoproduction is reasonably constrained (30-90 Tg CO-C a- I) (Zafiriou et al., 2003; Stubbins et al., 2006a). However, spatial and seasonal variability in the levels, sources and nature of the photoreactant, CDOM, in rivers, estuaries, and terrestrially and upwelling-influenced seas, make photoproduction rates for these aquatic environments hard to predict and, therefore, rates are poorly constrained (Valentine and Zepp, 1993; Zuo and Jones, 1995; Law et al., 2002; Zhang et al., 2006). Estimates of the total marine photoproduction have not advanced in recent years, ranging from 30 to 820 Tg CO-C aI (Valentine and Zepp, 1993; Zuo and Jones, 1995; Moran and Zepp, 1997) and only a tentative estimate of global estuarine CO photoproduction exists (- 2 Tg CO-C aI ) (Stubbins, 2001). The significance of CO photoproduction and the uncertainty in current estimates are best illustrated by comparison with other carbon cycle terms. For example, estimates of the total marine photochemical CO source are equivalent to 8- 200% of global riverine DOM inputs (Prather et al., 2001) and 16-350% of carbon burial In manne sediments (Hedges et al., 1997). These compansons c1early illustrate the importance of CO photoproduction and the requirement to better constrain its potential contribution to aquatic carbon cycling.
To quantitatively assess the role of CDOM photooxidation in the fate of organic carbon in the ocean (Miller and Zepp, 1995; Andrews et al., 2000; Vahatalo and Wetzel, 2004), two approaches have been employed most frequently: in situ incubations (Kieber et al., 1997) and optical-photochemical coupled modeling based on apparent quantum yields (AQYs) (Valentine and Zepp, 1993; Johannessen, 2000; Bélanger et al. , 2006). The former determines water column photochemical fluxes by directly incubating water samples at varying depths in the photic zone; it requires laborious fieldwork, but is thought to c10sely simulate the natural photochemistry and the in situ light field. The latter calculates photochemical rates by combining experimentally determined AQY spectra with CDOM absorption coefficient spectra and underwater irradiance. As CDOM absorption coefficients can be retrieved from satellite ocean color measurements (Siegel et al. , 2002; Bélanger et al. , 2008; Fichot et al., 2008), the modeling approach appears promising for large-scale investigations (Johannessen, 2000; Miller and Fichot, 2006). The reliability of this approach depends, to a large extent, on the reliability of the AQY spectra used in the model. Potentially large uncertainties in published AQY spectra are partly associated with the lack of quantitative knowledge of the influences of CDOM quality and environmental conditions on the related photoprocesses, inc1uding CO photoproduction.
Thermal (dark) production of CO, another potentially important manne CO source, has so far drawn little attention and its regional and global-scale source strengths are unknown. (Xie et al., 2005) observed CO dark formation rates of 0.21 ± 0.21 nmol L- 1 h- 1 in nine cyanide-poisoned Delaware Bay water samples. Significant CO dark production was also inferred from modeling upper-ocean CO cycles (Kettle, 1994, 2005). The dark production term is often critical to rationalize model-data discrepancies and greatly affects the values of CO photoproduction and microbial uptake rates that are derived from inverse modeling approaches (Kettle, 2005). Therefore, the lack of quantitative knowledge of this pathway seriously limits modeling and may add substantial uncertainties to the global marine CO budget.
It is expected that the distribution and biogeochemical cycling of CO in coastal (including estuarine) waters would be different from those in the open ocean due at least to 1) coastal waters are highly enriched with DOM relative to blue waters, causing the photochemical depth scale (e.g., e-folding depth oflight at 320 nm) in coastal waters to be smaller than that in the open ocean; 2) coastal DOM is largely of terrestrial origin while DOM in remote-ocean areas is dominantly of marine origin, which could result in different efficiencies of CO production; 3) the far more complex hydrological, physical, chemical and biological dynamics in coastal zones should give rise to more complicated influences on CO cycling in these areas.
The estuary and Gulf of St. Lawrence, referred to as the St. Lawrence estuarine system (SLES) herein, is a semi-enclosed water body with connections to the Atlantic Ocean through the Cabot and Belle-Isle strait. It receives the second largest freshwater discharge (600 km3 a1 ) in North America (Koutitonsky and Bugden 1991; Strain, 1990). Over a relatively small horizontal scale (~1200 km), the SLES provides various hydrological, geographical and oceanographie features. Surface water in the SLES transitions from freshwater-dominated CASE 2 water in the estuary to oceanic water-dominated CASE 1 water in the Gulf (Nieke et al., 1997). The water column is fairly weil mixed in the upper estuary (Quebec City to the mouth of the Saguenay Fjord) but highly stratified, except in winter, in the lower estuary (the mouth of Saguenay Fjord to Pointe-des-Monts) and the Gulf. An average depth of ~60 m in the upper estuary drops abruptly to >200 m over a few kilometers near the mouth of the Saguenay Fjord (Dickie and Trites, 1983). The typical two-layer estuarine circulation creates a maximum turbidity zone near Île d’Orléans, slightly downstream of Quebec City (d’Anglejan and Smith, 1973). Among the other facets of the SLES are runoff plumes, gyres, fronts, and upwelling areas (Koutitonsky and Bugden, 1991). These features make the SLES an ideal natural laboratory to study the transition of biogeochemical processes from freshwater to estuarine to oceanic water systems.
Method for CO photoproduction
Sam pie collection and treatment
Sampling stations were dispersed along a salinity gradient from the upstream limit of the St. Lawrence estuary near Quebec City through the Gulf of St. Lawrence and to the open Atlantic off Cabot Strait. Thirteen stations (Stns. 1-13) were sampled for absorbance and DOC measurements and six (Stns. 1,3, 8, Il, 12, 13) for the AQY study .Water samples (2 m deep) were taken in late July 2004 for Stns. 1-12 and in mid-June 2005 for Stn. 13 using 12-L Niskin bottles attached to a CTD rosette. Samples were gravity-filtered upon collection through Pail AcroPak 1000 capsules sequentially containing 0.8 f.1IT1 and 0.2 f.1IT1 polyethersulfone membrane filters. The filtered water was transferred in darkness into acid-cleaned, 4 L clear glass bottles, stored in darkness at 4 oC, and brought back to the laboratory at Rimouski. Samples were re-filtered with 0.22 f.1IT1 polycarbonate membranes (Millipore) and purged with CO-free air immediately prior to irradiations, which were carried out within 2 months of sample collection.
ln order to evaluate the effect of the CDOM’s light history on <Dco (i.e., dose dependence), filtered samples, placed in a clear glass container covered with a quartz plate, kept at 15 oC and continuously stirred, were irradiated with a SUNTEST XLS+ solar simulator equipped with a 1.5KW xenon lamp. Radiations emitted from the xenon lamp were screened by a Suprax long band-pass cutoff filter to minimize radiations <290 nm, and the spectral composition of the solar simulator closely matched that of natural sunlight reaching the earth’s surface. The output of the lamp was adjusted to 765 W m-2 (280-800nm) at the irradiation surface as determined with an OL-754 UV -vis spectroradiometer (Optronics Laboratories) fitted with an OL IS-270 2 in. integrating sphere. Irradiation time varied from 20 min to 175.0 h to obtain various photobleaching reglmes.
Table des matières
Chapter 1. Introduction
1.1. Significance of CO study
1.2. Existing problems in CO research
Chapter 2. Methods
2.1. Study area
2.2. Method for CO photoproduction
2.2.1 . Sample collection and treatment
2.2.3 . Irradiation for <Dco determination
2.2.4. Calculation of <Dco
2.3. Methods for CO dark production
2.3.2. Contamination assessment
2.3.3. Shipboard incubations
2.3.4. Land-based laboratory incubations
Chapter 3. Photoproduction
3.1. DOM mixing dynamics
3.2. Method evaluation for modeling <Dco(/ »‘)
3.2.1. Reproducibility of <Dco(À) determination
3.2.2. Performance of the curve fit method
3.3. <Dco(À) spectra
3.4. Response spectra of CO photo production
3.5. <Dco ofterrestrial vs. marine CDOM
3.6. Temperature dependence
3.7. Dose dependence
3.8. Implication for modeling
3.9. Calculation ofCQ photoproduction in the SLES
Chapter 4. Dark production
4.1. Incubation results
4.2. Spatial distribution of Qco
4.3. Factors affecting CO dark production
4.3.1. Qco vs. [CDOM]
4.3.2. Temperature dependence
4.3.3. Effect ofpH
4.3.4. Effects of sample storage, ionic strength, iron, and particles
4.3.5 Multiple linear regression analysis
4.4. Fluxes of CO dark production in the SLES and global oceans
4.4.1. CO dark production in the SLES
4.4.2. CO dark production in global oceans
Chapter 5. Conclusion