The effect of surface irradiance on the
Transcription
The effect of surface irradiance on the
Deep-Sea Research I 63 (2012) 52–64 Contents lists available at SciVerse ScienceDirect Deep-Sea Research I journal homepage: www.elsevier.com/locate/dsri The effect of surface irradiance on the absorption spectrum of chromophoric dissolved organic matter in the global ocean Chantal M. Swan a,n, Norman B. Nelson a, David A. Siegel a,b, Tihomir S. Kostadinov a a b Earth Research Institute, University of California, Santa Barbara, USA Department of Geography, University of California, Santa Barbara, USA a r t i c l e i n f o a b s t r a c t Article history: Received 10 October 2011 Received in revised form 13 January 2012 Accepted 23 January 2012 Available online 1 February 2012 The cycling pathways of chromophoric dissolved organic matter (CDOM) within marine systems must be constrained to better assess the impact of CDOM on surface ocean photochemistry and remote sensing of ocean color. Photobleaching, the loss of absorption by CDOM due to light exposure, is the primary sink for marine CDOM. Herein the susceptibility of CDOM to photobleaching by sea surfacelevel solar radiation was examined in 15 samples collected from wide-ranging open ocean regimes. Samples from the Pacific, Atlantic, Indian and Southern Oceans were irradiated over several days with full-spectrum light under a solar simulator at in situ temperature in order to measure photobleaching rate and derive an empirical matrix, esurf (m 1 mEin 1), which quantifies the effect of surface irradiance on the spectral absorption of CDOM. Irradiation responses among the ocean samples were similar within the ultraviolet (UV) region of the spectrum spanning 300–360 nm, generally exhibiting a decrease in the CDOM absorption coefficient (m 1) and concomitant increase in the CDOM spectral slope parameter, S (nm 1). However, an unexpected irradiation-induced increase in CDOM absorption between approximately 360 and 500 nm was observed for samples from high-nutrient low-chlorophyll (HNLC) environments. This finding was linked to the presence of dissolved nitrate and may explain discrepancies in action spectra for dimethylsulfide (DMS) photobleaching observed between the Equatorial Pacific and Subtropical North Atlantic Oceans. The nitrate-to-phosphate ratio explained 27–70% of observed variability in esurf at observation wavelengths of 330–440 nm, while the initial spectral slope of the samples explained up to 52% of variability in esurf at observation wavelengths of 310–330 nm. These results suggest that the biogeochemical and solar exposure history of the water column, each of which influence the chemical character and thus the spectral quality of CDOM and its photoreactivity, are the main factors regulating the susceptibility of CDOM to photodegradation in the surface ocean. The esurf parameter reported herein may be applied to remote sensing retrievals of CDOM to estimate photobleaching at the surface on regional to global scales. & 2012 Elsevier Ltd. All rights reserved. Keywords: Marine CDOM Solar irradiation Surface Photobleaching Photoproduction 1. Introduction Marine chromophoric dissolved organic matter (CDOM), the light-absorbing portion of total dissolved substances in seawater (filtrateo0.2 mm), is responsible for nearly 90% of ultraviolet (UV) radiation attenuation in the global ocean (Johannessen et al., 2003; Zepp et al., 2011). In open ocean waters, CDOM is produced through autochthonous marine food web processes throughout the water column, and seasonally depleted through solar photobleaching in the surface ocean (Nelson et al., 1998, 2004; Del Castillo and Coble, 2000; Stedmon and Markager, 2001). Photobleaching of CDOM, driven by absorption of solar radiation, n Corresponding author. Tel.: þ41 78 849 4755; fax: þ 1 805 893 2578. E-mail address: [email protected] (C.M. Swan). 0967-0637/$ - see front matter & 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.dsr.2012.01.008 typically results in loss in absorption by the chromophores and subsequent alteration of the spectral properties of CDOM (Del Vecchio and Blough, 2002; Helms et al., 2008; Swan et al., 2009). The satellite-derived 8-year mean climatology of colored dissolved and detrital material from SeaWiFS (1997–2005) (Fig. 1, ‘CDM’, of which CDOM comprises480%; Nelson et al., 1998) illustrates that solar bleaching, as balanced by CDOM inputs from river outflow and in situ production, is a prominent forcing mechanism on the global surface distribution of CDOM on multi-annual time scales. Low CDOM absorption values are found in surface waters with slow renewal rates and high radiation exposure (e.g., subtropical gyres), while elevated CDOM values are seen in regions of lower annual insolation (e.g., subpolar gyres) or regions of significant upwelling of CDOM from subsurface waters (e.g., equatorial Pacific and eastern boundary currents) (Siegel et al., 2002, 2005). The influence of photobleaching C.M. Swan et al. / Deep-Sea Research I 63 (2012) 52–64 Fig. 1. SeaWiFS eight-year mean (1997–2005) global composite of absorption by CDOM and detrital particulates at 443 nm, ‘aCDM (m 1, 443 nm)’ as determined by the GSM algorithm (Siegel et al., 2005; Maritorena et al., 2010). Detrital particulates contribute only a small percentage (5–17%, Nelson et al., 1998; Siegel et al., 2002) to overall absorption in the open ocean, thus aCDM (m 1, 443 nm) is considered as overall representative of CDOM. White triangles represent the sample sites at which laboratory determinations of apparent quantum yield for CDOM photobleaching were conducted for this study. is also observed at depth, such as in Atlantic mode waters, which manifest as minima within the basin-scale CDOM distribution as photobleached waters are entrained subsurface during convective overturn (Nelson et al., 2007, 2010). The solar exposure of CDOM also plays a significant role in balancing the ocean carbon budget with respect to terrestrial inputs of organic material to the sea, as photodegradation of CDOM within rivers and among continental shelf waters serves as an important removal and conversion pathway of terrigenous material into climate-relevant trace gases (Blough et al., 1993; Hedges et al., 1997; Vodacek et al., 1997) and lower molecular weight organics (Benner and Biddanda, 1998; Miller et al., 2002; Stedmon et al., 2007; Tedetti et al., 2008). Parameterization of CDOM photobleaching in the surface ocean is critical for a better understanding of its effect on elemental cycles and satellite detection of ocean color-derived properties. Direct measurements of the rates of CDOM photobleaching within the global open ocean are scarce, as most derive from near-coastal and estuarine waters or are concentrated within the North Atlantic (Vähätalo et al., 2000; Moran et al., 2000; Osburn et al., 2001, 2009; Del Vecchio and Blough, 2002; Tzortziou et al., 2007). In these environments, considerable losses of CDOM (approximately 50%, by most studies) occur on weekly to monthly time scales (Kouassi and Zika, 1992; Vodacek et al., 1997; Nelson et al., 1998). Photobleaching can remove up to 96% of terrestrial CDOM that enters the ocean, and near-zero CDOM absorption coefficients reported for the South Pacific subtropical gyre suggest that solar bleaching may completely degrade autochthonous CDOM in oligotrophic areas of the open ocean (Vähätalo and Wetzel, 2004; Swan et al., 2009). It remains to be determined whether there is a fraction of marine CDOM that is photochemically recalcitrant (Stedmon and Markager, 2001; Twardowski and Donaghay, 2002). The chief objectives of this paper are (1) to describe how CDOM photobleaching proceeds at the surface of the open ocean, particularly in tropical and subtropical areas that have not been previously investigated, (2) to examine the environmental controls on CDOM photobleaching, and (3) to quantify the photochemical susceptibility of CDOM on the seasonal scale for potential application to remote sensed CDOM data in estimating global surface photobleaching rates. The challenge of resolving measurable changes in CDOM under natural sunlight during typical cruise durations ( 5 weeks or less) led to the need for conducting accelerated exposures on shore using a solar simulator. Compounding this challenge, narrow bandwidth or partial spectral exposures, such as those 53 employed in traditional approaches to calculating apparent quantum yields for photobleaching, may not result in light flux sufficient to produce detectable losses in CDOM absorption in open ocean samples, particularly those from highly oligotrophic gyres in which CDOM absorption is lowest. Recent work has indicated that chromophores irradiated by monochromatic irradiation have an indirect (‘‘off-axis’’) effect on other chromophores not directly affected by the irradiation wavelength (attributed to electron transfer or photosensitizing intermediate species), emphasizing that chromophore interdependence must be considered when designing photobleaching studies (Del Vecchio and Blough, 2002, 2004; Goldstone et al., 2004; Tzortziou et al., 2007; Ziolkowski and Miller, 2007). Due to these constraints, full-spectrum simulated solar irradiance was employed in controlled laboratory incubations of filtered water samples collected from a variety of open ocean water types (Fig. 1, white triangles) to explore CDOM photobleaching at the surface of the global ocean. Results were used to develop an empirical model for the wavelength-specific effect of irradiance on spectral CDOM absorption, which encapsulates the net effect of chromophores undergoing simultaneous photobleaching, for potential application with ocean color-derived surface data. Thus, a mathematical description of bleaching is derived using the irradiance and CDOM absorption spectra as ‘inputs’. We characterize a surface photobleaching effect matrix, esurf (m 1 mEin 1), which quantifies the susceptibility of open ocean CDOM to typical illumination conditions at the sea surface. We subsequently examine the relationship of esurf to environmental variables that relate to the autochthonous production and destruction of CDOM to assess whether esurf is a function of the quantity or the quality of open ocean CDOM (Osburn et al., 2001; Twardowski and Donaghay, 2002; Nelson et al., 2004; Biers et al., 2007). We report both typical and novel photochemical transformations of light-exposed CDOM, and further discuss the implications of these within the context of the optical and hydrographic conditions of the sampling regions. These results open a pathway toward a better understanding and characterization of CDOM dynamics in the global surface ocean. 2. Methods 2.1. Sample collection Water samples for the present study were collected shipboard during austral or boreal summer cruises from various upper ocean depths around the globe during 2005–2008 as part of the U.S. CO2/CLIVAR Repeat Hydrography Survey (P16S, P16N, I8S, I9N and P18 field campaigns; Feely et al., 2005). A subtropical North Atlantic (Sargasso Sea) sample was collected during a U.S. Bermuda Atlantic Time-series Study (BATS) core cruise in winter 2006, and a sample from the Santa Barbara Channel, CA was collected in winter 2007. Geographical locations of the sample sites are listed within Table 1 and represented by white triangles in Fig. 1. Samples were collected from various depths (from surface—500 m) in an effort to sample the range of CDOM that may be cycled into the mixed layer during seasonal changes. This sampling scheme provided a diverse backdrop of optical and hydrographic regimes for assessing environmental controls on CDOM photobleaching in the surface ocean. Two liters of seawater were drawn from Niskin or PTFE-coated Go-Flo bottles, and either pressure-filtered through acid-leached 0.4 mm Nuclepore polycarbonate membranes, or vacuum-filtered through 0.2 mm Nuclepore polycarbonate membranes (preconditioned with ultrapure water from a Barnstead Nanopure Diamond UV system) in order to remove bacterial cells and other particulates 0.0005 0.0004 0.0005 0.0004 0 0.0001 0.0006 0.0001 0.0005 0.0006 0.0003 0.0008 0.0007 0.0005 0.0009 0.0030 0.0022 0.0028 0.0018 0 0.0025 0.0027 0.0013 0.0007 0.0005 0.0005 0.0006 0.0008 0.0002 0.0013 0.009 0.006 0.009 0.006 0 0.009 0.008 0.006 0.004 0.000 0.007 0.004 0 0 0 0.068 0.038 0.061 0.040 0 0.068 0.042 0.044 0.037 0.021 0.055 0.005 0 0 0 0.217 0.073 0.082 0.072 0.031 0.093 0.073 0.103 0.104 0.153 0.188 0.073 0.091 0.070 0.079 0.14 o0.01 0.05 0.03 0.25 0.15 0.07 1.49 2.42 1.67 2.87 0.90 0.80 0.59 1.6 0.20 o 0.01 0.08 0.07 0.10 0.14 0.06 21.36 33.32 19.07 42.12 10.74 9.34 6.32 24.32 1.4 1.0 1.6 2.3 0.4 0.9 0.9 14.3 13.8 11.4 14.7 11.9 11.7 10.7 15.2 0.023 0.032 0.026 0.028 0.038 0.027 0.029 0.026 0.029 0.028 0.026 0.032 0.028 0.031 0.032 esurf (325;325) (m 1 lEin 1) esurf (310;310) (m 1 lEin 1) P (lmol kg 1) Fig. 2. CDOM absorption coefficient (m 1, 325 nm) of samples after dark storage at 4 1C plotted versus CDOM absorption at time of collection. 1:1 line is plotted. Closed circles (K) represent samples stored for periods of 6–12 months. Open triangles (W) represent samples stored for 13–22 months. Error bars reflect precision in the CDOM absorption measurement at 325 nm, 7 0.013 m 1, at the 95% confidence interval determined from replicate samples in the Pacific (Nelson et al., 2007). (Whitehead and de Mora, 2000; Nelson and Siegel, 2002; Nelson and Coble, 2009). Filtrates were refrigerated in polycarbonate 1-L containers in the dark at 4 1C until their use in shore-based laboratory irradiation experiments conducted within a year of sample collection. Tests show stability in CDOM absorption spectra of open ocean samples for up to 12 months with this storage protocol. Fig. 2 displays CDOM absorption at 325 nm in samples stored for lengths of time ranging from 6 to 22 months versus CDOM absorption at the time of sample collection. No statistically significant variation in CDOM absorption at 325 nm was observed in samples stored for 12 months or less; however, CDOM absorption in filtrates stored for 13–22 months varied in magnitude by up to 0.028 m 1, with a slight bias toward increased CDOM relative to initial (Fig. 2). This indicated that a 1-year time limit for filtrate storage was an appropriate conservative guideline for the experimental work. Z (m) 0 80 140 40 25 40 40 200 500 80 200 100 80 5 50 Santa Barbara Channel, CA Sargasso Sea Subtropical N. Pacific Subtropical N. Pacific Subtropical S. Pacific Subtropical Indian Equatorial Indian Equatorial Pacific Subtropical S. Pacific (deep) Subarctic Pacific Subarctic Pacific Equatorial Pacific Subantarctic Pacific Frontal Zone Subantarctic Pacific Frontal Zone Southern Ocean 341N 321N 291N 291N 211S 311S 1.81N 0.51N 211S 551N 551N 0.51N 461S 461S 621S 1201W 641W 1501W 1501W 1031W 951E 921E 1501W 1031W 1501W 1501W 1501W 1501W 1501W 1031W 16 23 19 20 23 20 30 13 7 3 4 22 10 13 4 2.2. Experimental design Sample Site Lat. Lon. Temp. (1C) N (lmol kg 1) N:P initial S (nm 1) initial aCDOM (325, m 1) esurf (350;350) (m 1 lEin 1) esurf (440;440) (m 1 lEin 1) C.M. Swan et al. / Deep-Sea Research I 63 (2012) 52–64 Table 1 Hydrographic information and surface photobleaching parameters at selected wavelengths, esurf(lo;li), in ocean samples. A value of zero for esurf(lo;li) indicates that CDOM was resistant to solar irradiation over the exposure period. Positive values of esurf(lo;li) indicate that photoproduction of CDOM was observed over the exposure period. 54 CDOM irradiation experiments were conducted using an LS1000 Solar Simulator (Solar Light Co., Glenside, PA) fitted with a high-pressure Xenon arc lamp and filters specialized to closely match the spectrum and intensity of terrestrial irradiance (Mobley, 1994). The practical lower limit of terrestrial irradiance is 290 nm under normal stratospheric ozone levels (Whitehead and de Mora, 2000). An array of twelve 10-mL cylindrical quartz cells with PTFE-lined caps were used as sub-sample compartments with no head space and arranged on a matte black surface within a dark side-walled enclosure to minimize reflection, backscattering and container wall effects among sub-samples (Johannessen and Miller, 2001; Hu et al., 2002). The sample cell array was submersed within a water bath set to in situ temperature at which the original sample was collected. A dark vial in the water bath served as a control for any non light-related alteration of CDOM during the course of irradiation. The sample array was positioned under the collimated light beam of the simulator at the manufacturer-specified distance to ensure uniform spectral quality and intensity over the sample exposure area. The spectral irradiance within the exposure area was periodically quantified and tested for its spatial uniformity using an 11-channel UV–vis C.M. Swan et al. / Deep-Sea Research I 63 (2012) 52–64 aCDOM (λo) Eo (λi) 14 55 (μEin m−2 s−1 nm−1) 12 10 8 6 4 LS1000 Solar Simulator spectrum 2 0 300 400 500 λi (nm) 600 λo (nm) Qa (λi) εsurf (325;λi) Qa (λi) εsurf (325;λi) λi (nm) λi (nm) λi (nm) Fig. 3. Schematic of terms used to model the CDOM surface photobleaching effect using experimental data from irradiation of Sargasso Sea 80 m water. (A) Quantum scalar irradiance, Eo(li) (mEin m 2 s 1 nm 1), is the output of the LS1000 solar simulator. (B) Time course of the CDOM absorption spectrum, aCDOM(lo) (m 1), during irradiation (with inset of spectral slope parameter, S (m 1), versus exposure time in days) is used to determine (C) the rate of change in CDOM absorption, d(aCDOM(lo))/dt (m 1 s 1), at the given observation wavelength (lo ¼ 325 nm). (D) The time course of absorbed quanta, Qa(li) (mEin s 1 nm 1) is the product of Eo(li), āCDOM (li), and the sample cell volume, v. (E) The surface photobleaching effect parameter, esurf(325;li) (m 1 mEin 1), is solved by inversion of Eq. (2) (see text). The modeled value of d(aCDOM(325))/dt is the area underneath the curves of (F) the action spectrum for CDOM photobleaching at 325 nm, which equals the product of Qa(li) and esurf(325;li). MicroPro IITM multispectral radiometer (Satlantic, Inc., Halifax, Nova Scotia), as well as monitored daily for variation in flux intensity using a QSL-2201 VIS wand (Biospherical Instruments, San Diego, CA). The spectrum of the LS1000 Solar Simulator is shown in Fig. 3A. Variation in simulator intensity over the course of any individual experiment was negligible. Gradual, spectrally independent variation of simulator intensity ( 20%) was observed across a longer time scale of a few months. This variation was well constrained by the described radiometric calibrations, and applied to the experimental data. Experiments were conducted for approximately 2 days under the simulator (continuous exposure). This exposure period roughly corresponds to 10 days at the sea surface at mid-latitude, or 2–8 weeks of exposure when circulation within the seasonal mixed layer during summer and spring is taken into account. These approximations were based on data from the North Atlantic, and were achieved by comparing the simulator’s 24-hour output at 325 nm (46.3 kJ m 2; calculated from its irradiance at 325 nm; Fig. 3A) with the mean daily insolation at 325 nm estimated for the BATS site in late summer (9.3 kJ m 2) and early spring (8.2 kJ m 2), respectively (Zafiriou et al., 2008). This showed that a 48-hour exposure period under the simulator was equivalent to 10.0 and 11.3 day of surface irradiance in the mid-Atlantic in summer and spring, respectively. The ratio of the mixed layer-averaged CDOM photobleaching rate (PRMLD) to the surface CDOM photobleaching rate (PRsurf) in the ocean may be approximated given mixed layer depth (MLD) and the extinction coefficients of absorbed quanta, KAQ per season or Z PRMLD =PRsurf ¼ ð1=MLDÞ eK AQ z dz ð1Þ where the integration is from the surface (z ¼0) to the respective seasonal MLD. Typical summer and spring BATS values for MLD, 27 m and 89 m, and KAQ, of 0.040 m 1 and 0.055 m 1, respectively were used following Zafiriou et al. (2008). The calculated PRMLD/PRsurf ratio for each season was 0.61 for summer, and 0.20 for spring. Applying this to a 48-hour exposure under the solar simulator, the experimental incubations achieve a dosage roughly equivalent to 16 and 57 day of circulation within the subtropical mixed layer during summer and spring, respectively. The approximate 2-day exposure period under the simulator was chosen in order to resolve at least 50% reduction (i.e., one half-life) of the initial absorption coefficient. As can be seen from the above scaling exercise, this reduction in CDOM absorption approaches the typical rates of decay of CDOM by natural sunlight observed on time scales of weeks to months in the ocean depending on latitude and season (Kouassi and Zika, 1992; Vodacek et al., 1997; Nelson et al., 1998; Vähätalo and Wetzel, 2004). We can infer from the BATS calculation, for example, that the photobleaching half-life of CDOM is 2 weeks or less where annual insolation and residence time of surface waters is high (e.g., low latitudes), and several months or more at high latitudes where annual insolation is low and strong seasonal mixing occurs. An assumption implicit in these estimations is that the rate of daily turnover of the mixed layer occurs on time scales shorter than that of photobleaching (i.e., the mixed layer is well-mixed). Duplicate sub-samples from irradiation time-course experiments were stored in the dark at 4 1C before spectroscopic analysis. The possibility for low molecular weight chromophores to react to form more complex molecules through polycondensation reactions (i.e., ‘‘dark recovery’’) in a dark environment following irradiation has previously been a concern in photochemical experimentation (Gao and Zepp, 1998; Del Vecchio and Blough, 2002). Two deliberate tests for such phenomena conducted in this study found that no significant differences in absorption properties were observed between final sub-samples stored in the dark at 4 1C before analysis and replicate final sub-samples that were analyzed immediately after light exposure. This finding was consistent with prior investigation by Del Vecchio and Blough (2002), suggesting that any dark effects on 56 C.M. Swan et al. / Deep-Sea Research I 63 (2012) 52–64 optical properties in filtered water samples were likely negligible on the time scales of experimentation within the present study. Our experimental design was characterized as an optically thin system in which atotalL51, where atotal is the total absorption by water and other constituents (particulate and dissolved), and L is pathlength of the optical cell, which was 0.085 m (Hu et al., 2002). By assuming particulate absorption is negligible within filtered samples, and using a conservative (upper bound) pure water absorption value at 300 nm (0.038 m 1; Morel et al., 2007) along with the maximum CDOM absorption at 300 nm for the open ocean sites sampled (0.402 m 1), we find that the largest expected value of atotalL in our study is equal to 0.037. These conditions permit the assumption that any ‘‘inner filter effects’’ (light attenuation occurring within the sample vial) were negligible (Hu et al., 2002). 2.3. Analytical approach The surface photobleaching effect matrix, esurf(lo;li), was derived to quantify the effect on CDOM absorption at observation wavelength, lo, by irradiation wavelength, li, as influenced by the combined energy of absorbed photons during full-sun exposure. The rate of CDOM photobleaching at any given observation wavelength for an optically thin water sample, daCDOM(lo)/dt (m 1 s 1), is then defined as the product of esurf(lo;li) (m 1 mEin 1), the quantum scalar irradiance, Eo(li) (mEin m 2 s 1 nm 1), the absorption by CDOM incrementally averaged over the duration of exposure, āCDOM (li) (m 1), and the volume of the optical cell used, V (m3): Z esurf ðlo ; li ÞEo ðli Þa CDOM ðli ÞV dli ð2Þ daCDOM ðlo Þ=dt ¼ where the integration is over all irradiation wavelengths, li (300– 700 nm). Eo(li) is the output of the solar simulator quantified using a multispectral radiometer as described in Section 2.2. The absorbance at each time interval, used to calculate the rate of change in CDOM absorption, daCDOM(lo)/dt, over the exposure period, was determined using an UltraPathTM liquid core waveguide scanning spectrophotometer (World Precision Instruments, Sarasota, FL) with a 1.943 m optical pathlength capable of detecting relatively low levels of absorption in the open ocean. Absorption spectra were analyzed over 300–700 nm. The protocol for measurement of CDOM absorption using UltraPathTM, including time-dependent baseline drift and salinity-dependent refractive index correction functions, has been previously described (Nelson et al., 2007; Swan et al., 2009). Given the above inputs, values of esurf(lo;li) were determined from Eq. (2) by assuming a Rayleigh-like distribution function of the form: ðli lref Þ2 l l ð3Þ esurf ðlo ; li Þ ¼ Aðlo Þ i ref2 e 2Bðlo Þ2 Bðlo Þ where lref is a reference wavelength (300 nm), and A(lo) (m 1 mEin 1 nm) and B(lo) (nm) are constants that are inversely solved using an unconstrained non-linear optimization procedure (fminsearch function in Matlabs), which minimized the mean absolute value difference between measured and modeled daCDOM(lo)/dt. B(lo) affects the spread and skewness of the irradiation wavelength dependency of esurf(lo;li) at each observation wavelength, and works in combination with A(lo) as a scaling factor within Eq. (3) in setting the magnitude of the photochemical effect at the observation wavelength. A(lo) and B(lo) are inherent properties of the CDOM at each sample site, and are applicable in circumstances where the spectral light field remains proportional to that used in the current study. As changes in CDOM absorption and spectral slope are dependent on the spectral quality of incident irradiance, our results therefore only strictly apply to the surface of the open ocean (Osburn et al., 2001; Tzortziou et al., 2007). A schematic of the inverse method is provided in Fig. 3 (panels A–F) for the case of Sargasso Sea water irradiation. The solar simulator spectrum in Fig. 3A represents Eo(li). Fig. 3B displays the CDOM absorption spectrum at ten evenly spaced (8-hour interval) time points during the irradiation experiment, from which estimates of daCDOM(lo)/dt were calculated (Fig. 3C). The value of daCDOM(lo)/dt in Fig. 3C is negative, indicating photobleaching, and decreases in time as the bleaching of CDOM absorption results in less overall absorbed quanta available for photochemical work. The average spectral absorption by CDOM during each experiment, āCDOM ðli Þ, was computed for each of the ten evenly spaced time intervals (dt) over the course of irradiation, and multiplied by Eo(li) and the sample cell volume (v) to calculate the effective dose of absorbed quanta for each sample, Qa(li) (mEin s 1 nm 1; Fig. 3D), which was used to solve for esurf(lo;li) (see Fig. 3E for lo ¼325 nm). The modeled esurf (325;li) spectrum for the Sargasso Sea sample (Fig. 3E), when multiplied by the Qa(li) spectrum, yields the action spectrum for surface CDOM photobleaching at the 325 nm observation wavelength (Fig. 3F). The modeled values for daCDOM(lo)/dt at each time point were computed as the area underneath (integrand of) the action spectrum curve in Fig. 3F. The action spectrum, which illustrates that maximum photobleaching of aCDOM(325) in the Sargasso Sea sample occurs at the coincident wavelength (where lo ¼ li), demonstrates that the current analytical method is consistent with recent findings on the kinetics of spectral CDOM photobleaching in natural waters (Del Vecchio and Blough, 2002; Goldstone et al., 2004). Several other functions for esurf(lo;li) were tested for their ability to reproduce the laboratory results, including linear, exponential, and Gaussian-like shapes. Linear and exponential forms for esurf(lo;li) each implied that the efficiency of photobleaching decreases monotonically with increasing wavelength across the full-spectrum of irradiation regardless of observation wavelength. While shorter wavelengths indeed are more energetic, these models do not adequately describe the interactive effects of chromophores undergoing solar bleaching during broadband exposure (Del Vecchio and Blough, 2002). A Gaussian-like model form for esurf(lo;li), which describes maximum photobleaching near the incident wavelength with an effect that tapers off at flanking regions of the spectrum, is more consistent with use of polychromatic irradiation to investigate photobleaching (Tzortziou et al., 2007); however, we observed unrealistically narrow spread (5–10 nm) in esurf(lo;li) using a Gaussian shape and significant ‘‘off-axis’’ bleaching effects were not well modeled. Use of the Rayleigh-like function (Eq. (3)) provided a better approximation to the weighted cumulative photochemical effect of absorbed energy outside the incident wavelength, and proved to be the best empirical fit to the data. Use of this model form yielded superior matchups between measured and modeled daCDOM(lo)/dt values (r2 498%) compared to the other functions tried. The exponential spectral slope parameter, S (nm 1), plotted at each time point (inset of Fig. 3B) was determined for assessing shifts in spectral quality of CDOM. S was calculated using a non-linear exponential curve fit to the spectral region 300–340 nm. The wavelength region selected for computing spectral slope, while internally consistent, does not facilitate comparison with literature values of spectral slope computed across broader wavebands (Nelson et al., 2007), or across longer or shorter spectral regions (Twardowski et al., 2004; Loiselle et al., 2009). The choice of spectral range was made because of anomalous irradiation-induced changes in the visible wavelength region that did not fit an exponential curve (described below). C.M. Swan et al. / Deep-Sea Research I 63 (2012) 52–64 3. Results The change in the CDOM absorption spectrum for the case of irradiated Sargasso Sea water presented in Section 2.3 (Fig. 3B) was characterized by reduced absorption between 300 and 450 nm and increase in S (nm 1) (Fig. 3B inset), with no significant changes observed from 450 to 700 nm. The same observations were made for approximately half of our irradiated samples, and were consistent with trends in CDOM photobleaching in natural waters as documented by a number of previous studies (e.g., Vodacek et al., 1997; Twardowski and Donaghay, 2002; Del Vecchio and Blough, 2002; Vähätalo and Wetzel, 2004; Osburn et al., 2009). Time courses of CDOM absorption during irradiation of equatorial Indian (40 m) and subtropical North Pacific water samples (40 m) are shown, respectively, along with insets of S (nm 1) versus exposure time in Fig. 4A and B, as further examples of these types of observations. Photobleaching in both the equatorial Indian and North Pacific samples was characterized by a significant reduction in absorption ( 0.03 m 1, approximately 50% of the initial signal at 325 nm) and steepening of S, indicative of greater proportional losses in absorption at the longer UV observation wavelengths than at the shorter UV wavelengths. For samples from the equatorial Pacific, subtropical South Pacific, subarctic Pacific, subantarctic Pacific Frontal Zone, and Southern Ocean, we observed decreases in CDOM absorption over 300–360 nm (and increase in S) with simultaneous increases in 57 CDOM absorption over 360–500 nm (Fig. 4C–F). The loss in absorption over 300–360 nm was similar to that observed in samples in which no long-wavelength absorption increases were observed. Time courses of CDOM absorption (m 1) during irradiation, with insets of S (nm 1) versus exposure time, are displayed for the case of subarctic Pacific (80 m) and equatorial Pacific (100 m) waters in Fig. 4D and E, respectively, as examples of these unusual spectral transformations. Mean increase in absorption at 440 nm among samples exhibiting photoproduction was 0.038 m 1 over the experimental time course and corresponds to a two-fold increase in the initial absorption value. Three samples from the subantarctic Pacific in which photoproduction was observed did not demonstrate any measureable losses in absorption at wavelengths 300 nm or longer. Fig. 4F demonstrates this phenomenon in a subantarctic Pacific 80 m sample. Finally, the absorption by one sample from the oligotrophic South Pacific (25 m) remained unaltered by the 48-hour long irradiation over the full (300–700 nm) observation wavelength range (Fig. 4C). Spectra of esurf(lo;li) at observation wavelengths of 300, 325 and 350 nm are displayed in Fig. 5A–C. The absolute magnitude of the esurf spectrum among the study sites is greatest at the shortest observation wavelength (300 nm) and decreases with wavelength energy (Fig. 5A–C; note differences in scale of x and y axes), consistent with prior polychromatic bleaching studies (Del Vecchio and Blough, 2002). The esurf(350;li) parameter for samples Fig. 4. Example irradiation time-courses of CDOM absorption spectra with insets of spectral slope parameter, S (m 1), versus exposure time (days): (A) equatorial Indian 40 m, (B) subtropical North Pacific 40 m, (C) subtropical South Pacific 25 m, (D) subarctic Pacific 80 m, (E) equatorial Pacific 100 m, and (F) subantarctic Pacific 80 m waters. Fig. 5. Surface photobleaching effect parameter, esurf (lo;li) (m 1 mEin 1), versus irradiation wavelength, li (nm), at observation wavelengths (A) 300 nm (B) 325 nm and (C) 350 nm (Note: scales for A–C differ). 58 C.M. Swan et al. / Deep-Sea Research I 63 (2012) 52–64 from the subantarctic frontal zone (5 and 80 m) and Southern Ocean (50 m) have a positive value in Fig. 5C due to photoproduction at the 350 nm observation wavelength. However as photobleaching rate was effectively zero at the 300 and 325 nm observation wavelengths (not resolvable over the exposure period), esurf spectra for these few samples are not visible in Fig. 5A and B. Fig. 5A–C also shows a trend of esurf occurring over a greater bandwidth as observation wavelength increases. The esurf(325;li) spectrum approached zero at approximately 375 nm (Fig. 5B), indicating that photochemical reactivity of CDOM is negligible beyond this point in the spectrum, which agrees with previous indications that energy in the range4400 nm has little impact on photobleaching (Del Vecchio and Blough, 2002). On average, the ‘largest’ (i.e., most negative) esurf values at a given UV observation wavelength were observed in the North Pacific and equatorial Indian Ocean samples, while the shallow South Pacific samples (subtropical, subantarctic and Southern Ocean) consistently exhibited the lowest (i.e., least negative) values. Examples of the retrieved values of A(lo, m 1 mEin 1 nm) (Fig. 6A) and B(lo, nm) (Fig. 6B) from the Sargasso Sea and equatorial Pacific, plotted as a function of observation wavelength, show the spectral dependencies of the parameters that describe esurf(lo;li). (For the purpose of illustration, A and B are only plotted for lo ¼350–400 nm in Fig. 6, as A values at lo ¼ 300–340 nm were several orders of magnitude greater.) A(lo) values were typically negative, reflecting loss in absorption at the observation wavelength, except across lo ¼350–440 nm in some samples where A(lo) acquired a positive value indicating CDOM increases during irradiation. Fig. 6A shows the progression of A(lo) in the equatorial Pacific 100 m sample from a negative value at 350 nm (i.e., photobleaching), to a value of zero at 360 nm (no change), to positive values at wavelengthsZ370 nm, corresponding to the increases in absorption spectrum of this sample as seen in Fig. 4E. This is contrasted with the negative value of the A(lo) spectrum for the Sargasso Sea 80 m sample, in which only loss in absorption across the spectrum was observed during irradiation. Regardless of whether CDOM absorption increased or decreased during irradiation, B(lo) values were of similar magnitude among the samples (e.g., Fig. 6B). The general trend in the B(lo) spectrum for a given sample is a decrease in magnitude as observation wavelength increases. In samples where A(lo) had a value of zero indicating no net change in absorption at the particular observation wavelength, the corresponding B(lo) is irrelevant (not a number) due to its position as a denominator within Eq. (3). This is evident at 360 nm for the B(lo) spectrum of the equatorial Pacific 200 m sample displayed in Fig. 6B. Table 1 displays hydrographic information for each open ocean sample evaluated, and provides the esurf(lo;li) (m 1 mEin 1) value determined for coincident wavelengths of 310 nm, 325 nm, 350 nm and 440 nm. A primary objective of our investigation was to understand the environmental controls on the susceptibility of CDOM to photobleaching making use of the associated water sample data collected as part of the core measurements of the U.S. CO2/CLIVAR Repeat Hydrography Program. Data included depth (Z), salinity, temperature, dissolved oxygen (O2), nitrate (N), phosphate (P), silicate (Si), and fluorometric chlorophyll-a (Chl-a) concentrations. All measurements are made following standard WOCE methods (http://ushydro.ucsd.edu). The N:P ratio, as well as the initial CDOM absorption coefficient (aCDOM(lo), m 1) and spectral slope (S, nm 1) of the samples were also evaluated for their role in explaining the observed photoproduction of CDOM and the natural variability in the surface photobleaching effect parameter. Values of A(lo) and B(lo) were not significantly linearly correlated with any of the hydrographic data, thus we conducted simple regressions of esurf(lo;li) with the hydrographic data mentioned above. All samples presented in Table 1 were included in the analysis, including those in which esurf(lo;li) was effectively zero. Linear regression statistics are summarized within Table 2 for several coincident wavelengths across 310–440 nm. The value of esurf(440;440) had significant positive linear relationships with nitrate (r2 ¼0.35, p-value¼0.02, n ¼15) and phosphate concentrations (r2 ¼0.40, p-value¼0.01, n¼15), and in particular the N:P ratio (r2 ¼0.70, p-value¼0.0001, n¼ 15), and was significantly negatively correlated to temperature (r2 ¼0.45, p-value¼0.006, n¼15). Correspondingly, esurf(350;350) and esurf(375;375) also had significant, albeit weaker, relationships with N:P and temperature (Table 2), as photoproduction was also observed at these wavelengths among several samples. Values of esurf(440;440) assumed a rough bimodal distribution when plotted against N, P, and N:P, suggesting that sign ( þ or ) rather than the magnitude of esurf(440;440) signals the presence or absence of photoproduction at 440 nm. Samples in which decreases in CDOM absorption at 440 nm were observed (i.e., subtropical sites and the equatorial Indian Ocean) had N:P values of less than 2.3, while N:P ratios were higher (10.7–15.2) hence closer to the global mean (Redfield) value of 16 among samples in which photoproduction was observed (Table 1). Fig. 6. Parameters A(lo) and B(lo) describing the surface photobleaching effect (Eq. (3) in text) versus observation wavelength (lo) for (A) Sargasso Sea 80 m water and (B) Equatorial Pacific 100 m water. Negative values of A(lo) indicated photobleaching at the observation wavelength, while positive values indicated an increase in CDOM absorption. C.M. Swan et al. / Deep-Sea Research I 63 (2012) 52–64 59 Table 2 Regression statistics for linear correlation of surface photobleaching parameters at selected wavelengths, esurf(lo;li), with initial values of environmental properties. Z Salinity Temp. O2 N P Si Chl-a N:P Initial aCDOM(ko) Initial S r2 ¼0.06 p ¼0.38 n¼ 15 r2 ¼0.03 p ¼0.55 n¼ 15 r2 ¼0.20 p ¼0.11 n¼ 14 r2 ¼ 0.01 p ¼ 0.79 n ¼15 r2 ¼ 0.01 p¼ 0.70 n¼15 r2 ¼ 0.04 p¼ 0.47 n¼15 r2 ¼0.09 p ¼0.28 n¼ 15 r2 ¼ 0.17 p ¼ 0.13 n ¼15 r2 ¼ 0.27 p ¼ 0.05 n ¼15 r2 ¼ 0.52 p ¼ 0.003n n¼ 15 r2 ¼0.002 p ¼0.89 n¼ 15 r2 ¼0.17 p ¼0.13 n¼ 15 r2 ¼0.26 p ¼0.06 n¼ 14 r2 ¼ 0.04 p ¼ 0.50 n ¼15 r2 ¼ 0.06 p¼ 0.39 n¼15 r2 ¼ 0.01 p¼ 0.80 n¼15 r2 ¼0.08 p ¼0.30 n¼ 15 r2 ¼ 0.22 p ¼ 0.08 n ¼15 r2 ¼ 0.18 p ¼ 0.12 n ¼15 r2 ¼ 0.37 p ¼ 0.02n n¼ 15 r2 ¼0.003 p ¼0.84 n¼ 15 r2 ¼0.27 p ¼0.05 n¼ 15 r2 ¼0.14 p ¼0.18 n¼ 14 r2 ¼ 0.11 p ¼ 0.24 n ¼15 r2 ¼ 0.13 p¼ 0.19 n¼15 r2 o 0.001 p¼ 0.96 n¼15 r2 ¼0.04 p ¼0.49 n¼ 15 r2 ¼ 0.34 p ¼ 0.02n n ¼15 r2 ¼ 0.12 p ¼ 0.21 n ¼15 r2 ¼ 0.27 p ¼ 0.04n n¼ 15 r2 ¼0.001 p ¼0.90 n¼ 15 r2 ¼0.35 p ¼0.02n n¼ 15 r2 ¼0.05 p ¼0.41 n¼ 14 r2 ¼ 0.25 p ¼ 0.06 n ¼15 r2 ¼ 0.28 p¼ 0.04n n¼15 r2 ¼ 0.04 p¼ 0.47 n¼15 r2 ¼0.14 p ¼0.16 n¼ 15 r2 ¼ 0.52 p ¼ 0.003n n ¼15 r2 ¼ 0.09 p ¼ 0.28 n ¼15 r2 ¼ 0.23 p ¼ 0.07 n¼ 15 r2 ¼0.005 p ¼0.79 n¼ 15 r2 ¼0.34 p ¼0.02n n¼ 15 r2 ¼0.09 p ¼0.30 n¼ 14 r2 ¼ 0.19 p ¼ 0.10 n ¼15 r2 ¼ 0.23 p¼ 0.07 n¼15 r2 ¼ 0.004 p¼ 0.80 n¼15 r2 ¼0.13 p ¼0.18 n¼ 15 r2 ¼ 0.55 p ¼ 0.001n n ¼15 r2 ¼ 0.07 p ¼ 0.33 n ¼15 r2 ¼ 0.15 p ¼ 0.14 n¼ 15 r2 ¼0.04 p ¼0.48 n¼ 15 r2 ¼0.45 p ¼0.006n n¼ 15 r2 ¼0.02 p ¼0.66 n¼ 14 r2 ¼ 0.35 p ¼ 0.02n n ¼15 r2 ¼ 0.40 p¼ 0.01n n¼15 r2 ¼ 0.06 p¼ 0.36 n¼15 r2 ¼0.13 p ¼0.19 n¼ 15 r2 ¼ 0.70 p ¼ 0.0001n n ¼15 r2 ¼ 0.003 p ¼ 0.84 n ¼15 r2 ¼ 0.07 p ¼ 0.34 n¼ 15 esurf(310;310) r2 ¼ 0.03 p ¼ 0.55 n ¼15 esurf(325;325) r2 ¼ 0.009 p ¼ 0.74 n ¼15 esurf(340;340) r2 o 0.001 p ¼ 0.93 n ¼15 esurf(350;350) r2 ¼ 0.006 p ¼ 0.79 n ¼15 esurf(375;375) r2 ¼ 0.01 p ¼ 0.67 n ¼15 esurf(440;440) r2 ¼ 0.05 p ¼ 0.44 n ¼15 n A p-value less than 0.05 indicated that the corresponding r2 value was significant at the 95% confidence interval. The trend in esurf(440;440), and to a lesser extent, esurf(350;350) and esurf(375;375), versus temperature roughly indicated that positive values of esurf(440;440) occurred at temperatures below 15 1C, and negative values above 15 1C. No significant relationships were observed between esurf(lo;li) and either the initial CDOM absorption coefficient, chlorophyll-a, dissolved O2, Si, salinity or sample depth (Table 2). While temperature was correlated with the absence or presence of photoproduction as indicated by esurf at coincident wavelengths of 350–440 nm, no significant relationship was observed between temperature and esurf(lo;li) at the shorter UV wavelengths (310–340 nm) over which photobleaching occurred in most samples. Across this wavelength range, the initial spectral slope of the samples mainly accounted for variability in esurf(lo;li), with N:P also contributing in explaining variability in esurf(340;340) (Table 2). 4. Discussion 4.1. Implications of CDOM photoproduction in the open ocean The evolution of a pronounced peak in absorption spanning 350–500 nm as observed in half of our irradiated open ocean samples (Fig. 4D–F) has potential to affect remote-sensing estimates of chlorophyll due to the wavelength region of maximum absorption by chlorophyll overlapping with that of the photoproduct(s). However, open ocean CDOM absorption spectra are typically featureless, and peaks of this magnitude have not been observed in the field (e.g., Nelson et al., 1998, 2007; Yamashita and Tanoue, 2009; Swan et al., 2009). However, minimally detectable ‘‘bump’’-like features in the CDOM spectrum between 410 and 420 nm have been reported in southeast Pacific upwelling zones (Bricaud et al., 2010, their Fig. 16). This raises the possibility that the photoproduct may be transient in nature and decrease with further exposure. An irradiation time course of an equatorial Pacific 200 m sample was extended to 7 day of continuous simulated solar irradiation (equivalent to roughly 35 day of mid-latitude irradiance at the sea surface in summer and a few months of exposure in an equatorial Pacific mixed-layer). Fig. 7A displays the spectral transformations (300–550 nm) of the equatorial Pacific 200 m sample during irradiation with an inset of S (nm 1) versus exposure time, and Fig. 7B displays absorption at the 440 nm observation wavelength versus exposure time. Evolution of a photoproduct (between 360 and 500 nm) was observed by day one, maximum increase in absorption at 440 nm was observed by day two, and subsequent bleaching of the absorbing compound and a return to the initial level was observed by the 6.8-day mark (Fig. 7B). The negligible net change in absorption over 360–500 nm by the 6.8-day mark indicated that photoproduction of CDOM within filtered samples under simulated solar irradiance does not proceed indefinitely or past the time scale of months in nature. This is consistent with open ocean CDOM absorption spectra (e.g., Nelson et al., 2007; Yamashita and Tanoue, 2009; Swan et al., 2009; Bricaud et al., 2010) and leads us to postulate that in the water column such a photoproduct is either (a) intermediate and subject to near-simultaneous photodegradation, (b) biologically labile and subject to rapid microbial consumption (Nelson et al., 2004), or (c) simply does not get produced in appreciable concentration due to the diel cycle of solar irradiance and mixing within the water column. In other words, this phenomenon may be observed only within a controlled system of accelerated light exposure such as that used in the current study. For these reasons, significant net effects of the observed photochemical phenomena on satellite estimations of CDOM or chlorophyll are unlikely. Nevertheless, we present several hypotheses on the potential implications for the observed photoproducts as they pertain to photochemistry in the surface ocean. Reported instances of increased DOM absorption due to solar irradiation within aquatic environments are rare, and have been 60 C.M. Swan et al. / Deep-Sea Research I 63 (2012) 52–64 Fig. 7. (A) Seven-day irradiation time-course of CDOM absorption spectra with inset of S (nm 1) versus exposure time (days) and (B) absorption coefficient (m 1) of CDOM at 440 nm versus exposure time (days) for equatorial Pacific 200 m waters. primarily associated with photoreduction of iron (Küpper et al., 2006; Martin et al., 2006) or photohumification of dissolved substances (Kieber et al., 1997; Benner and Biddanda, 1998; Obernosterer et al., 1999; Reche et al., 2001). Photohumification, the light-stimulated polymerization of labile algal exudates such as polyunsaturated fatty acids, has been invoked as one pathway for the origin of marine humics (Kieber et al., 1997; Del Vecchio and Blough, 2004). Photohumification is typically associated with a long tail of absorption across the visible wavelengths, and has been attributed to intramolecular charge or energy transfers between proximate chromophores, as well as to long conjugated and aromatic compounds in terrestrially influenced water samples (Twardowski and Donaghay, 2002; Goldstone et al., 2004; Del Vecchio and Blough, 2004). It is unlikely that the CDOM photoproduction observed during our study is representative of photohumification as filtered water samples from the open ocean do not contain sufficient quantities of labile exudates or long fatty acid chains for such reactions to take place (Whitehead and de Mora, 2000). Furthermore, photoproduction herein was observed as an absorption peak between 360 and 500 nm rather than the long tail of absorption extending to the visible region beyond 500 nm, as is characteristic of photohumification (Reche et al., 2001; Del Vecchio and Blough, 2002). The correlations observed between esurf(440;440) and N:P suggest a connection of CDOM photoproduction with the biogeochemical state of the water column as it influences CDOM reactivity. The sampling regions in which we observed CDOM photoproduction (e.g., equatorial Pacific, subarctic Pacific, subantarctic Pacific and Southern Ocean) classify as iron-limited, high-nutrient low-chlorophyll (HNLC) regions (Martin et al., 1991; Behrenfeld and Kolber, 1999), corroborated by the relatively high nitrate values observed in these regions (6.32–42.12 mmol kg 1, Table 1). (The temperature correspondence of esurf(lo;li) at wavelengths at which photoproduction was observed (350–440 nm) may therefore be an artifact of the regionality of HNLC (not focused within subtropical regions) rather than a photochemical phenomenon. This is supported by the lack of correlation of temperature to instances of photobleaching at the shorter UV wavelengths, suggesting CDOM photoreactivity is not strictly temperature-dependent as other photoreactions in nature are (Toole et al., 2003; Mopper and Kieber, 2002).) Nitrate is active in surface ocean photochemistry by its absorption of UVB radiation, which results in the production of hydroxyl radicals (OH ), the dominant photosensitizers in aqueous solutions (Goldstone et al., 2002; Tedetti et al., 2007, 2008). Toole et al. (2004) provide evidence for a light-mediated role of nitrate in stimulating dimethylsulfide (DMS) photobleaching rates. Action spectra for DMS photobleaching in the North Atlantic reported by Toole et al. (2003) peak in the UVB region, and contrast similar spectra made from the equatorial Pacific which peak in both the UVB and 380–460 nm range (Kieber et al., 1996). Through its absorption of UV light, CDOM is the primary reactant controlling DMS photobleaching in the surface ocean, as DMS does not absorb solar photons directly. It is therefore possible that the HNLC-associated CDOM photoproduct between 360 and 500 nm in the equatorial Pacific may be responsible for the peak in the action spectrum of DMS photobleaching observed by Kieber et al. (1996). The hypothesis is that nitrate sensitizes the production of a reactive CDOM molecule that absorbs light in the 380–460 nm range, greatly increasing DMS photobleaching rates. Although we have speculated that CDOM photoproducts may be subject to near-instantaneous degradation in the water column, the proposed photoreactions of nitrate, CDOM and DMS likely proceed on comparably rapid time scales (Toole et al., 2004). In order to test for the role of nitrate in CDOM photoproduction as suggested by the correlations of esurf(440;440) with N, P and the N:P ratio, an additional sample from the Santa Barbara Channel, CA was collected in winter 2009 from the identical geographic location and depth as indicated in Table 1. A sample volume of 0.75 L was amended with 1.3 mL of 1000 mg L 1 NaNO3 stock solution (prepared using ultrapure water and 99.995þ % sodium nitrate, Sigma-Aldrich) to achieve a nitrate concentration of 20.6 mmol kg 1, which approached the median value of nitrate (19.1 mmol kg 1) amongst open ocean samples in which photoproduction was observed. Replicate NO3 -amended and unamended sub-samples were irradiated at in situ temperature along with dark controls for approximately 2 day following the procedures outlined in Section 2.2. Initial and final absorption spectra from this experiment are plotted in Fig. 8A. Addition of sodium nitrate caused an initial increase in the absorption coefficient for wavelengthso330 nm prior to irradiation, as seen in the pre-irradiation difference spectrum (solid black line) between the NO3 -amended and unamended samples (Fig. 8B). This initial change in absorption may be attributed to the collective absorption properties of sodium nitrate and CDOM in the water sample as influenced by nitrate’s absorption properties in the UVB region (e.g., Johnson and Coletti, 2002). At the end of the 2-day exposure period, while net photoproduction was not explicitly observed, less net absorption loss was observed between approximately 340 and 500 nm in the NO3 -amended C.M. Swan et al. / Deep-Sea Research I 63 (2012) 52–64 61 Fig. 8. Absorption spectra of CDOM from the Santa Barbara Channel, CA amended with NO3 . (A) Solid black line ¼NO3 -amended CDOM before irradiation. Dotted black line¼ NO3 -amended CDOM after 2-day irradiation. Solid gray line ¼unamended CDOM (control) before irradiation. Dotted gray line ¼unamended CDOM (control) after 2-day irradiation. (B) Difference spectra between NO3 -amended CDOM and unamended CDOM (control) before irradiation (solid black line), and after irradiation (dotted black line). sample than in the control sample. This wavelength region corresponded to that of the absorption increases observed in our high N samples. Fig. 8B (dotted line) displays the postirradiation difference spectrum between NO3 -amended and unamended samples. We hypothesize that the significantly less change in the absorption coefficient at 440 nm (0.0159 m 1) post-irradiation in the NO3 -amended than control sample suggests that underlying transient CDOM photoproduction may be responsible for the less net overall photobleaching. While a mechanism cannot be defined with certainty without further study, these observations in conjunction with the elevated nitrate levels observed in natural samples exhibiting photoproduction imply a role for nitrate in CDOM photoproduction observed herein. It should be noted that the strong increase in absorption at 300 nm due to the experimental addition of nitrate may suggest that nitrate drives much of the actual absorption of CDOM in the UV region; however, this increase is likely an isolated incidence pertaining to the artificial amendment with laboratory-grade sodium nitrate. This is supported by the lack of correlation between dissolved nitrate and CDOM absorption within a large field data set from the North Atlantic Ocean (Nelson et al., 2007), as well as the lack of correlation of nitrate with aCDOM at 300 nm in the current study (r2 ¼0.17, p-value¼0.13, n ¼15). Biers et al. (2007) observed simultaneous photochemical formation of CDOM and fluorescent DOM upon irradiation of a coastal seawater amended with N-containing tryptophan, raising the possibility that this amino acid influenced photoproduction within the current study, as tryptophan occurs in measureable concentration within ocean samples (Coble, 2007). However, Excitation Emission Matrix spectra (EEMs) analysis using a FluoroMaxs-4 spectrofluorometer (HORIBA Jobin Yvon, Inc.) on equatorial Pacific samples did not reveal evolution of a fluorescent photoproduct coincident with the observed CDOM formation (not shown). In fact, only fading of fluorescence was observed in these samples during irradiation, which is inconsistent with a role for tryptophan in mediating the absorption increases we observed (Biers et al., 2007). There is another possibility that biologically mediated iron ligands, such as marine siderophores involved in the photoreduction of Fe3 þ to Fe2 þ , may play a role alongside nitrate in the observed photoproduction of CDOM (Barbeau et al., 2001; Martin et al., 2006; Vraspir and Butler, 2009). Fe3 þ has been associated with humic-type fluorescence intensity in the water column (Tani et al., 2003; Kitayama et al., 2009), which itself is strongly related to CDOM absorption (Coble, 2008; Yamashita and Tanoue, 2008). For many of the samples taken from the CLIVAR cruises, dissolved [Fe2 þ ] determinations were made; however, no correspondence was found between dissolved [Fe2 þ ] concentrations and esurf(440;440) (r2 ¼0.002, p-value¼0.97, n ¼11), thus identifying a role for iron-chelating molecules in the observed CDOM photoreactions requires further experimentation. 4.2. Global variability in the surface photobleaching effect Changes in CDOM at the UV observation wavelengths could not be resolved within a 2-day exposure period among samples collected from the subtropical South Pacific (25 m), subantarctic Pacific frontal zone (5 m and 80 m), and Southern Ocean (50 m) (see Table 1, e.g., esurf(325;325) values ¼0). These samples were collected during austral summer in regions that experience relatively high UV irradiation doses, partly due to stratospheric ozone depletion (e.g., Smith et al., 1992). The South Pacific gyre is characterized by the most optically clear waters in the global ocean (Morel et al., 2007; Swan et al., 2009; Bricaud et al., 2010), and the 25 m sample from this region had the lowest CDOM in our study (Table 1). The low but detectable absorption in these samples suggests that a photochemically resistant fraction of CDOM persists past seasonal time scales in the open ocean and potentially contributes to the ‘‘background’’ pool of CDOM in the global ocean (Nelson et al., 2002, 2010). The lack of correlation between depth and esurf at all coincident wavelengths suggested that photochemically labile CDOM may be found in deep or surface waters depending on insolation and circulation dynamics of a region. Further, the lack of correlation between esurf and the initial absorption coefficient of CDOM suggests the quantity of CDOM does not have a strong effect on susceptibility of CDOM to photobleaching at the surface. Marine and freshwaters that are heavily influenced by terrestrially derived materials comprise a chromophore pool with a highly variable chemical make-up and photochemical reactivity (Coble, 2007). Even though localized influences of riverine input, such as the Amazon and Orinoco River plumes, are detected in the multi-annual global CDOM distribution, the chromophore pool in the open ocean is largely remote from terrestrial effects on seasonal time scales and chiefly driven by autochthonous in situ processes (Siegel et al., 2002; Nelson et al., 2004, 2010). This leads to an explanation for why the initial S and N:P values explain 62 C.M. Swan et al. / Deep-Sea Research I 63 (2012) 52–64 much of the variance in the photobleaching effect parameter at the shorter UV wavelengths. Initial S values of all samples within this study, including those from the Santa Barbara Channel, were in the range representative of marine sources of CDOM (i.e., 4 0.02 nm 1, Nelson and Siegel, 2002; Bricaud et al., 2010). S values have been associated with compositional variation in CDOM relating to source (e.g., terrestrial, marine) and to photobleaching (Twardowski and Donaghay, 2002; Weishaar et al., 2003; Helms et al., 2008). It is plausible that chromophores in the open ocean have a relatively similar chemical character and spectral quality because CDOM is autochthonously produced and S variability in the open ocean is relatively low (Nelson et al., 2010). High S values are generally indicative of a chromophore pool that has had considerable light exposure (Osburn et al., 2001; Twardowski and Donaghay, 2002; Loiselle et al., 2009). This is further supported by the increases in S observed during irradiation experiments, as well as by our observation that the photochemically resistant subtropical South Pacific 25 m sample had the highest initial S value in this study. The positive linear correlation between S and esurf (325;325) suggests that higher spectral slopes lead to a weaker (less negative) surface photobleaching effect. As an indicator of the biogeochemical history of a water parcel and time since last surface contact, S explains a mechanism for the photochemical susceptibility of CDOM. In the open ocean, CDOM with low S values relative to surface waters (such as in deep aphotic waters) is either relatively newly produced or has accumulated over decades as a result of deep ocean remineralization processes (Rochelle-Newall and Fisher, 2002; Nelson et al., 2004, 2010; Kitidis et al., 2006). In either case, the lower S values reflect a pool of CDOM that has remained relatively unexposed to sunlight, which may confer a higher probability of photochemically labile chromophores. On average, the strongest photobleaching effect was observed in the North Pacific and Indian Ocean samples, while the weakest photobleaching effect was observed among shallow South Pacific samples (subtropical, subantarctic and Southern Ocean). Accordingly, North Pacific and Indian Ocean samples had lower initial S values on average than the shallow South Pacific samples. Regions of the North Pacific and equatorial Indian Oceans experience greater degrees of mixed layer renewal than the South Pacific, attributed in part to North Pacific Intermediate Water (NPIW) circulation (You et al., 2003) and equatorial Indian upwelling patterns (Xie et al., 2002), respectively. CDOM spectral slopes in the mixed layers of the North Pacific and equatorial Indian are lower due to renewal with less photoexposed (and higher AOU) subthermocline waters (Swan et al., 2009; Nelson et al., 2010). Elucidating a mechanism for the effect of N:P on CDOM photobleaching however, as suggested by the positive linear relation between N:P and esurf(325;325) for example, poses more of a challenge as nutrient ratios, particularly N:P, vary widely in open ocean regimes due to many processes including denitrification, nitrogen fixation, and nutrient or iron limitation (Klausmeier et al., 2004; Falkowski et al., 1998). The N:P ratio of the surrounding water column has been shown to regulate phytoplankton species selection (Quigg et al., 2003). One hypothesis that follows is that global variability in N:P leads to variation in phytoplankton assemblages, which are subsequently decomposed by microbes producing CDOM of variable chemical composition thus photochemical susceptibility. On the other hand since we demonstrated earlier an example of a light-stimulated role for nitrate in CDOM photoreactions, we cannot rule out the hypothesis that N:P, regardless of its relevance to the biogeochemical state of the water column, affects CDOM photobleaching directly along a photochemical pathway. It is possible, for example, that varying levels of reactive oxygen species generated during UV light absorption by nitrate and phosphate play a role in determining CDOM photoreactivity as N:P was naturally highly correlated to N within samples from our study (r2 ¼0.73, p ¼0.0001, n¼15) (Mopper and Kieber, 2002; Goldstone et al., 2002). On the other hand, esurf is more highly correlated with N:P than N, which lends stronger support to the idea that N:P may affect CDOM quality and thus photoreactivity through a biogeochemical pathway. Although the photobiogeochemical parameters described above account for a significant portion of the variance in esurf within the spectral ranges examined and allow us to generate several hypotheses as to the drivers of open ocean CDOM photobleaching, a unified explanation for the effect of N:P on CDOM photobleaching may not be possible yet given the small sample size of our study. The overall interpretation of the results is that both photo-oxidative and biogeochemical conditions are important synergistic drivers of the susceptibility of marine CDOM to photochemical transformations. 5. Conclusion We have modeled the effect of surface irradiance on CDOM absorption as a function of the combined energetic effect of all UV–vis wavelengths. A methodology and hypotheses relating to the susceptibility of CDOM to photobleaching in the major ocean basins was described, further characterizing and quantifying aspects of this important removal pathway for CDOM in the open ocean. We conclude that the N:P ratio and spectral slope of CDOM as proxies for the biogeochemical and solar exposure history of a water sample, respectively, appear to be the primary influence on the susceptibility of CDOM to solar bleaching in the open ocean. Overall, our results suggest that the quality of CDOM is a more important regulator of CDOM photoreactivity than quantity. The course of experimentation revealed unusual transient photoproduction in the visible region of the CDOM absorption spectrum within HNLC regimes, which we attributed to nitrate photochemistry and proposed as a causal mechanism for the discrepant action spectra for DMS photobleaching between equatorial Pacific and North Atlantic waters (Kieber et al., 1996; Toole et al., 2003). Further investigation is needed to determine if this phenomenon is exceptional to high-intensity laboratory exposures, but the potential for transient light-induced CDOM increases in the open ocean may not be ruled out. Future application of esurf rests on the assumption that that there is no significant alternate sink in the open ocean (e.g., diagenetic, microbial) for chromophores across the time scales of photobleaching rate assessments as they correspond to those in nature (seasonal to interannual). This is a valid assumption as slight trends in open ocean CDOM spectral characteristics attributed to diagenesis thus far only have been observed on temporal scales of thermohaline circulation (decadal to millennial) in deep water masses (Nelson et al., 2007, 2010). Secondly, microbial consumption is considered insignificant in the open ocean on the time scale of CDOM photobleaching, with the overall net activity of bacterioplankton on CDOM in the water column resulting in production of CDOM (Rochelle-Newall and Fisher, 2002; Vähätalo and Wetzel, 2004; Nelson et al., 2007, 2010). There are clear limitations to the current method, not least of which is that the data may only be applied at the sea surface and on a regional basis given that biogeochemical parameters contribute to regional variability in esurf. However, ocean colorderived monthly climatology of CDOM and irradiance may be used in conjunction with esurf spectra to ultimately generate a synoptic view of photobleaching rate in the surface ocean. Surface rates then could be incorporated using a method similar to Eq. (1) to generate a mixed-layer integrated photobleaching rate, as C.M. Swan et al. / Deep-Sea Research I 63 (2012) 52–64 previous investigations suggest that attenuation of downwelling UV irradiance roughly imitates the attenuation of photodegradation of organic material (Vähätalo et al., 2000; Vähätalo and Wetzel, 2004). Combining estimates of specific turnover rate of CDOM in the mixed layer due to photobleaching with simple vertical mixing rates will ultimately enable calculation of the net in situ biological production rate of CDOM in the ocean, a process that has been difficult to parameterize in laboratory and field settings. Constraining the open ocean CDOM cycle in this manner will be a valuable contribution to future remote-sensing and biogeochemical studies. Acknowledgments Support from NSF Chemical Oceanography and NASA Ocean Biology and Biogeochemistry Programs to N. Nelson, D. Siegel and C. Carlson, as well as the NASA Earth System Science Graduate Fellowship Program to C. Swan, is gratefully acknowledged. We thank the CO2/CLIVAR Repeat Hydrography Program chief scientists, the captains and crew of the R/Vs Revelle, Thompson and Brown, as well as C. Carlson, D. Menzies, N. Guillocheau, M. Meyers, E. Wallner (UCSB), N. MacDonald (Bermuda-BIOS), and the U. Hawaii/FSU trace metal posse (W. Landing, C. Measures, C. Buck, M. Brown, W. Hiscock, P. Hansard, M. Hatta, M. Grand) for field assistance and collection, and to Barbara Prezelin for the loan of the BSI scalar PAR meter. The satellite CDM imagery shown in Fig. 1 were obtained from the Ocean Color MEaSUREs Project at UCSB (http://wiki.icess.ucsb.edu/measures/ Products). Finally, we thank M. Brzezinski, C. Carlson and several anonymous reviewers for insightful comments on the manuscript. References Barbeau, K., Rue, E.L., Bruland, K.W., Butler, A., 2001. Photochemical cycling of iron in the surface ocean mediated by microbial iron(iii)-binding ligands. Nature 413, 409–413. Behrenfeld, M.J., Kolber, Z.S., 1999. Widespread iron limitation of phytoplankton in the South Pacific ocean. Science 283, 840–843. Benner, R., Biddanda, B., 1998. Photochemical transformations of surface and deep marine dissolved organic matter: effects on bacterial growth. Limnol. Oceanogr. 43, 1373–1378. Biers, E.J., Zepp, R.G., Moran, M.A., 2007. The role of nitrogen in chromophoric and fluorescent dissolved organic matter formation. Mar. Chem. 103, 46–60. Blough, N.V., Zafiriou, O.C., Bonilla, J., 1993. Optical absorption spectra of waters from the Orinoco River outflow: Input of colored organic matter to the Caribbean. J. Geophys. Res. 98, 2245–2257. Bricaud, A., Babin, M., Claustre, H., Ras, J., Tieche, F., 2010. Light absorption properties and absorption budget of Southeast Pacific waters. J. Geophys. Res., 115. doi:10.1029/2009JC005517. Coble, P.G., 2007. Marine optical biogeochemistry: the chemistry of ocean color. Chem. Rev. 107, 402–418. Coble, P., 2008. Cycling coloured carbon. Nat. Geosci. 1, 575–576. Del Castillo, C.E., Coble, P.G., 2000. Seasonal variability of the colored dissolved organic matter during the 1994–95 NE and SW Monsoons in the Arabian Sea. Deep-Sea Res. I 47, 1563–1579. Del Vecchio, R., Blough, N.V., 2002. Photobleaching of chromophoric dissolved organic matter in natural waters: kinetics and modeling. Mar. Chem. 78, 231–253. Del Vecchio, R., Blough, N.V., 2004. On the origin of the optical properties of humic substances. Environ. Sci. Technol. 38, 3885–3891. Falkowski, P.G., Barber, R.T., Smetacek, V., 1998. Biogeochemical controls and feedbacks on ocean primary production. Science 281, 200–206. Feely, R.A., Talley, L.D., Johnson, G.C., Sabine, C.L., Wanninkhof, R., 2005. Repeat hydrography cruises reveal chemical changes in the North Atlantic. Eos Trans. AGU 86, 399. doi:10.1029/2005EO420003. Gao, H.Z., Zepp, R.G., 1998. Factors influencing photoreactions of dissolved organic matter in a coastal river of the southern United States. Environ. Sci. Technol. 32, 2940–2946. Goldstone, J.V., Del Vecchio, R., Blough, N.V., Voelker, B.M., 2004. A multicomponent model of chromophoric dissolved organic matter photobleaching. Photochem. Photobiol. 80, 52–60. Goldstone, J.V., Pullin, M.J., Bertilsson, S., Voelker, B., 2002. Reactions of hydroxyl radical with humic substances: bleaching, mineralization and production of bioavailable carbon substrates. Environ. Sci. Technol. 36, 364–372. 63 Hedges, J.I., Keil, R., Benner, R., 1997. What happens to terrestrially-derived organic matter in the sea? Org. Geochem. 27, 195–212. Helms, J.R., Stubbins, A., Ritchie, J.D., Minor, E.C., Kieber, D.J., Mopper, K., 2008. Absorption spectral slopes and slope ratios as indicators of molecular weight, source and photobleaching of chromophoric dissolved organic matter. Limnol. Oceanogr. 53, 955–969. Hu, C., Muller-Karger, F.E., Zepp, R.G., 2002. Absorbance, absorption coefficient, and apparent quantum yield: a comment on common ambiguity in the use of these optical concepts. Limnol. Oceanogr. 47, 1261–1267. Johannessen, S.C., Miller, W.L., 2001. Quantum yield for the photochemical production of dissolved inorganic carbon in seawater. Mar. Chem. 76, 271–283. Johannessen, S.C., Miller, W.L., Cullen, J.J., 2003. Calculation of UV attenuation and colored dissolved organic matter absorption spectra from measurements of ocean color. J. Geophys. Res., 108. doi:10.1029/2000JC000514. Johnson, K.S., Coletti, L.J., 2002. In situ ultraviolet spectrophotometry for high resolution and long-term monitoring of nitrate, bromide and bisulfide in the ocean. Deep-Sea Res. I 49, 1291–1305. Kieber, D.J., Jiao, J., Kiene, R.P., Bates, T.S., 1996. Impact of dimethylsulfide photochemistry on methyl sulfur cycling in the equatorial Pacific Ocean. J. Geophys. Res. 101, 3715–3722. Kieber, R., Hydro, L., Seaton, P., 1997. Photooxidation of triglycerides and fatty acids in seawater: implication toward the formation of marine humic substances. Limnol. Oceanogr. 42, 1454–1462. Kitayama, S., Kuma, K., Manabe, E., Sugie, K., 2009. Controls on the iron distribution in the deep-water column of the North Pacific Ocean: iron (III) hydroxide solubility and marine humic-type dissolved organic matter. J. Geophys. Res., 114. doi:10.1029/2008JC004754. Kitidis, V., Stubbins, A.P., Uher, G., Upstill-Goddard, R.C., Law, C.S., Woodward, E.M.S., 2006. Variability of chromophoric dissolved organic matter in surface waters of the Atlantic Ocean. Deep-Sea Res. II 53, 1666–1684. Klausmeier, C.A., Litchman, E., Daufresne, T., Levin, S.A., 2004. Optimal nitrogen-tophosphorus stoichiometry of phytoplankton. Nature 429, 171–174. Kouassi, A.M., Zika, R.G., 1992. Light-induced destruction of the absorbance property of dissolved organic matter in seawater. Toxicol. Environ. Chem. 35, 195–211. Küpper, F.C., Carrano, C.J., Kuhn, J., Butler, A., 2006. Photoreactivity of iron(III)– aerobactin: photoproduct structure and iron(III) coordination. Inorg. Chem. 45, 6028–6033. Loiselle, S.A., Bracchini, L., Dattilo, A.M., Ricci, M., Tognazzi, A., Cozar, A., Rossi, C., 2009. Optical characterization of chromophoric dissolved organic matter using wavelength distribution of absorption spectral slope. Limnol. Oceanogr. 54, 590–597. Maritorena, S., Hembise Fanton d’Andon, O., Mangin, A., Siegel, D.A., 2010. Merged satellite ocean color data products using a bio-optical model: characteristics, benefits and issues. Remote Sens. Environ. 114, 1791. Martin, J.H., Gordon, R.M., Fitzwater, S.E., 1991. The case for iron. Limnol. Oceanogr. 36, 1793–1802. Martin, J.D., Ito, Y., Homann, V.V., Haygood, M.G., Butler, A., 2006. Structure and membrane affinity of new amphiphilic siderophores produced by Ochrobactrum sp. SP18. J. Biol. Inorg. Chem. 11, 633–641. Miller, W.L., Moran, M.A., Sheldon, W.M., Zepp, R.G., Opsahl, S., 2002. Determination of apparent quantum yield spectra for the formation of biologically labile photoproducts. Limnol. Oceanogr. 47, 343–352. Mobley, C.D., 1994. Light and Water: Radiative Transfer in Natural Waters. Academic Press, San Diego, CA. (pp. 7–11). Mopper, K., Kieber, D.J., 2002. Photochemistry and the cycling of carbon, sulfur, nitrogen and phosphorus. In: Matter, D.A., Hansell, Carlson, C.A. (Eds.), Biogeochemistry of Marine Dissolved Organic. Academic Press, San Diego, CA, pp. 456–508. Moran, M.A., Sheldon, W.M., Zepp, R.G., 2000. Carbon loss and optical property changes during long-term photochemical and biological degradation of estuarine dissolved organic matter. Limnol. Oceanogr. 45, 1254–1264. Morel, A., Gentili, B., Claustre, H., Babin, M., Bricaud, A., Ras, J., Tie che, F., 2007. Optical properties of the ‘‘clearest’’ natural waters. Limnol. Oceanogr. 52, 217–229. Nelson, N.B., Siegel, D.A., Michaels, A.F., 1998. Seasonal dynamics of colored dissolved organic material in the Sargasso Sea. Deep-Sea Res. I 45, 931–957. Nelson, N.B., Siegel, D.A., 2002. Chromophoric DOM in the Open Ocean. In: Matter, D.A., Hansell, Carlson, C.A. (Eds.), Biogeochemistry of Marine Dissolved Organic. Academic Press, San Diego, CA, pp. 547–578. Nelson, N.B., Carlson, C.A., Steinberg, D.K., 2004. Production of chromophoric dissolved organic matter by Sargasso Sea microbes. Mar. Chem. 89, 273–287. Nelson, N.B., Siegel, D.A., Carlson, C.A., Swan, C.M., Smethie Jr., W.M., Khatiwala, S., 2007. Hydrography of chromophoric dissolved organic matter in the North Atlantic. Deep-Sea Res. I 54, 710–731. Nelson, N.B., Coble, P.G., 2009. Optical analysis of chromophoric dissolved organic matter. In: Wurl, O. (Ed.), Practical Guidelines for the Analysis of Seawater. CRC Press, pp. 401. Nelson, N.B., Siegel, D.A., Carlson, C.A., Swan, C.M., 2010. Tracing global biogeochemical cycles and meridional overturning circulation using chromophoric dissolved organic matter. Geophys. Res. Lett., 37. doi:10.1029/2009GL042325. Obernosterer, I., Reitner, B., Herndl, G.J., 1999. Contrasting effects of solar radiation on dissolved organic matter and its bioavailability to marine bacterioplankton. Limnol. Oceanogr. 44, 1645–1654. Osburn, C.L., Zagarese, H.E., Morris, D.P., Hargreaves, B.R., Cravero, W.E., 2001. Calculation of spectral weighting functions for the solar photobleaching of 64 C.M. Swan et al. / Deep-Sea Research I 63 (2012) 52–64 chromophoric dissolved organic matter in temperate lakes. Limnol. Oceanogr. 46, 1455–1467. Osburn, C.L., Retamal, L., Vincent, W.F., 2009. Photoreactivity of chromophoric dissolved organic matter transported by the Mackenzie River to the Beaufort Sea. Mar. Chem. 115, 10–20. Quigg, A., Finkel, Z.V., Irwin, A.J., Rosenthal, Y., Ho, T., Reinfelder, J.R., Schofield, O., Morel, F.M., Falkowski, P.G., 2003. The evolutionary inheritance of elemental stoichiometry in marine phytoplankton. Nature 425, 291–294. Reche, I., Pulido-Villena, E., Conde-Porcuna, J.M., Carrillo, P., 2001. Photoreactivity of dissolved organic matter from high-mountain lakes of Sierra Nevada, Spain. Arct. Antarct. Alp. Res. 33, 426–434. Rochelle-Newall, E.J., Fisher, T.R., 2002. Production of chromophoric dissolved organic matter fluorescence in marine and estuarine environments: an investigation into the role of phytoplankton. Mar. Chem. 77, 7–21. Siegel, D.A., Maritorena, S., Nelson, N.B., Hansell, D.A., Lorenzi-Kayser, M., 2002. Global ocean distribution and dynamics of colored dissolved and detrital organic materials. J. Geophys. Res., 107. doi:10.1029/2001JC000965. Siegel, D.A., Maritorena, S., Nelson, N.B., Behrenfeld, M.J., McClain, C.R., 2005. Colored dissolved organic matter and the satellite-based characterization of the ocean biosphere. Geophys. Res. Lett., 32. doi:10.1029/2005GL024310. Smith, R.C., Prezelin, B.B., Baker, K.S., Bidigare, R.R., Boucher, N.P., Coley, T., et al., 1992. Ozone depletion: ultraviolet radiation and phytoplankton biology in Antarctic waters. Science 255, 952–959. Stedmon, C.A., Markager, S., 2001. The optics of chromophoric dissolved organic matter (CDOM) in the Greenland Sea: an algorithm for differentiation between marine and terrestrially derived organic matter. Limnol. Oceanogr. 46, 2087–2093. x. Stedmon, C.A., Markager, S., Tranvik, L., Kronberg, L., Slätis, T., Martinsen, W., 2007. Photochemical production of ammonium and transformation of dissolved organic matter in the Baltic Sea. Mar. Chem. 104, 227–240. Swan, C.M., Siegel, D.A., Nelson, N.B., Carlson, C.A., Nasir, E., 2009. Biogeochemical and hydrographic controls on chromophoric dissolved organic matter distribution in the Pacific Ocean. Deep-Sea Res. I 56, 2175–2192. Tani, H., Nishioka, J., Kuma, K., Takata, H., Yamashita, Y., Tanoue, E., Midorikawa, T., 2003. Iron(III) hydroxide solubility and humic-type fluorescent organic matter in the deep water column of the Okhotsk Sea and the northwestern North Pacific Ocean. Deep-Sea Res. I 50, 1063–1078. Tedetti, M., Kawamura, K., Narukawa, M., Joux, F., Charrie re, B., Sempéré, R., 2007. Hydroxyl radical-induced photochemical formation of dicarboxylic acids from unsaturated fatty acid (oleic acid) in aqueous solution. Photochem. Photobiol. 188, 135–139. Tedetti, M., Joux, F., Charrie re, B., Mopper, K., Sempéré, R., 2008. Contrasting effects of solar radiation and nitrates on the bioavailability of dissolved organic matter to marine bacteria. Photochem. Photobiol. 201, 243–247. Toole, D.A., Kieber, D.J., Kiene, R.P., Siegel, D.A., Nelson, N.B., 2003. Photobleaching and the dimethylsulfide (DMS) summer paradox in the Sargasso Sea. Limnol. Oceanogr. 48, 1088–1100. Toole, D.A., Kieber, D.J., Kiene, R.P., White, E.M., Bisgrove, J., del Valle, D.A., Slezak, D., 2004. High dimethylsulfide photobleaching rates in nitrate-rich Antarctic waters. Geophys. Res. Lett., 31. doi:10.1029/2004GL019863. Twardowski, M.S., Donaghay, P.L., 2002. Photobleaching of aquatic dissolved materials: absorption removal, spectral alteration, and their interrelationship. J. Geophys. Res., 107. doi:10.1029/1999JC000281. Twardowski, M.S., Boss, E., Sullivan, J.M., Donaghay, P.L., 2004. Modeling the spectral shape of absorption by chromophoric dissolved organic matter. Mar. Chem. 89, 69–88. Tzortziou, M., Osburn, C.L., Neale, P.J., 2007. Photobleaching of dissolved organic material from a tidal marsh–estuarine system of the Chesapeake Bay. Photochem. Photobiol. 83, 782–792. Vähätalo, A.V., Salkinoja-Salonen, M., Taalas, P., Salonen, K., 2000. Spectrum of the quantum yield for photochemical mineralization of dissolved organic carbon in a humic lake. Limnol. Oceanogr. 45, 664–676. Vähätalo, A.V., Wetzel, R.G., 2004. Photochemical and microbial decomposition of chromophoric dissolved organic matter during long (months–years) exposures. Mar. Chem. 89, 313–326. Vodacek, A., Blough, N.V., DeGrandpre, M.D., Peltzer, E.T., Nelson, R.K., 1997. Seasonal variation of CDOM and DOC in the Middle Atlantic Bight: terrestrial inputs and photooxidation. Limnol. Oceanogr. 42, 674–686. Vraspir, J.M., Butler, A., 2009. Chemistry of marine ligands and siderophores. Annu. Rev. Mar. Sci. 1, 43–63. Weishaar, J.L., Aiken, G.R., Bergamaschi, B.A., Fram, M.S., Fujii, R., Mopper, K., 2003. Evaluation of specific ultraviolet absorbance as an indicator of the chemical composition and reactivity of dissolved organic carbon. Environ. Sci. Technol. 37, 4702–4708. Whitehead, R.F., de Mora, S., 2000. Marine photochemistry and UV radiation. Issues Environ. Sci. Technol. 14, 37–60. Xie, S., Annamalai, H., Schott, F.A., McCreary Jr., J.P., 2002. Structure and mechanisms of South Indian Ocean climate variability. J. Clim. 15, 864–878. Yamashita, Y., Tanoue, E., 2009. Basin-scale distribution of chromophoric dissolved organic matter in the Pacific Ocean. Limnol. Oceanogr. 54, 598–609. Yamashita, Y., Tanoue, E., 2008. Production of bio-refractory fluorescent dissolved organic matter in the ocean interior. Nat. Geosci. 1, 579–582. You, Y., Suginohara, N., Fukasawa, M., Yoritaka, H., Mizuno, K., Kashino, Y., Hartoyo, D., 2003. Transport of North Pacific Intermediate Water across Japanese WOCE sections. J. Geophys. Res., 108. doi:10.1029/2002JC001662. Zafiriou, O.C., Xie, H., Nelson, N.B., Najjar, R.G., Wang, W., 2008. Diel carbon monoxide cycling in the upper Sargasso Sea near Bermuda at the onset of spring and in midsummer. Limnol. Oceanogr. 53, 835–850. Zepp, R.G., Erickson, D.J., Paul, N.D., Sulzberger, B., 2011. Effects of solar UV radiation and climate change on biogeochemical cycling: interactions and feedbacks. Photochem. Photobiol. Sci. 10, 261. Ziolkowski, L.A., Miller, W.L., 2007. Variability of the apparent quantum efficiency of CO photoproduction in the Gulf of Maine and Northwest Atlantic. Mar. Chem. 105, 258–270.