Distribution and photo-reactivity of chromophoric and fluorescent dissolved organic matter in the Northeastern North Pacific Ocean

Deep–Sea Research I xxx (xxxx) xxx

Contents lists available at ScienceDirect

Deep-Sea Research Part I journal homepage: http://www.elsevier.com/locate/dsri

Distribution and photo-reactivity of chromophoric and fluorescent dissolved organic matter in the Northeastern North Pacific Ocean Fang Cao a, 1, Yuting Zhu b, 2, David J. Kieber b, William L. Miller a, * a b

Department of Marine Sciences, University of Georgia, Athens, GA, 30602, USA Department of Chemistry, State University of New York College of Environmental Science and Forestry, Syracuse, NY, 13210, USA

A R T I C L E I N F O

A B S T R A C T

Keywords: Gulf of Alaska Northeastern North Pacific Ocean Dissolved organic matter (DOM) Excitation emission matrix Photochemical degradation

Deep waters in the Northeastern North Pacific Ocean store the most aged, largely biologically refractory dis­ solved organic matter (DOM) in the global ocean. We conducted high resolution vertical profiles to survey the optical properties of this DOM in the Gulf of Alaska (GoA) onboard the R/V Melville using ultraviolet/visible (UV/ Vis) absorbance spectra and excitation-emission matrix spectra (EEMs) along with parallel factor analysis (PARAFAC). Chromophoric DOM (CDOM) absorption coefficients in the GoA were similar to the global open ocean waters reported elsewhere outside the Northeastern North Pacific. Three humic-like and two protein-like fluorescent components were identified. Profiles of fluorescent DOM (FDOM) showed a low signal for the humiclike organic matter in the surface that increased in the subsurface and remained fairly constant at depth. In contrast, the protein-like components had elevated levels in surface samples that decreased with depth. Both CDOM and FDOM optical proxies remained spatially uniform within the deep ocean water masses. Photo­ chemical irradiations performed onboard showed that photochemistry of the FDOM from deep seawater trans­ formed the deep ocean fluorescent organic matter signals into a fluorescent component distribution that resembled surface water with a dominance of protein-like fluorescent material. This work contributes to the global ocean survey database of chromophoric and fluorescent DOM and provides evidence of a role for the photochemical alteration of deep, biologically refractory DOM once it is mixed back into the sunlit surface ocean.

1. Introduction Marine dissolved organic matter (DOM) constitutes one of the most abundant active carbon reservoirs, holding 662 Pg C in a global in­ ventory. As a carbon substrate for heterotrophic bacteria and a re­ pository for autotrophically fixed carbon, marine DOM plays a significant role in the global ocean carbon cycle (Hansell, 2002; Hansell et al., 2009). It is largely produced in the euphotic zone and exported into the ocean interior where it is eventually transported via deep ocean circulation to the deep North Pacific Ocean (Hansell et al., 2009). As a consequence, the Northeastern North Pacific Ocean contains the oldest deep water, arguably containing the ocean’s most biologically refractory DOM. This reservoir of unreactive organic matter is of paramount importance to the long-term sequestration of oceanic carbon on centennial to millennial time scales due to its slow turnover rate. Over the past few decades, research on oceanic DOM composition and

transformations using both bulk optical methods including UV/Vis absorbance and fluorescence (e.g., Andrew et al., 2013; Helms et al., 2013; Jørgensen et al., 2011) and molecular techniques including high resolution mass spectroscopy (e.g., Chen et al., 2014; Dittmar and Stubbins, 2014; Lechtenfeld et al., 2014; Repta, 2014) has greatly advanced our understanding of the sources and fates of marine DOM. To date, however, the optical properties of DOM in this oldest reservoir of deep water in the Northeastern North Pacific Ocean have not been examined in any detail. A portion of this biologically refractory DOM is comprised of UVabsorbing chromophoric DOM (CDOM) that is expected to be photo­ reactive. Indeed, numerous lines of optical and molecular evidence indicate that the photochemical breakdown of this bio-refractory DOM into bioavailable molecules or small organic/inorganic carbon com­ pounds is the main removal mechanism for this oceanic DOM (Helms et al., 2013; Medeiros et al., 2015; Mopper et al., 1991; Stubbins et al.,

* Corresponding author. E-mail address: [email protected] (W.L. Miller). 1 Present address: State Key Laboratory of Estuarine and Coastal Research, East China Normal University, Shanghai, 200241, China. 2 Present address: Wadsworth Center, New York State Department of Health, Albany, NY 12201, USA. https://doi.org/10.1016/j.dsr.2019.103168 Received 30 June 2019; Received in revised form 3 November 2019; Accepted 12 November 2019 Available online 15 November 2019 0967-0637/© 2019 Elsevier Ltd. All rights reserved.

Please cite this article as: Fang Cao, Deep–Sea Research I, https://doi.org/10.1016/j.dsr.2019.103168

F. Cao et al.

Deep-Sea Research Part I xxx (xxxx) xxx

performed high-resolution surveys of DOM optical properties, collecting samples from the sea surface down to ~5000 m using a 24 Niskin bottle rosette with a Sea-Bird CTD (conductivity-temperature-depth) sensor. Hydrographic stations are mapped in Fig. 1. Corresponding hydro­ graphic (depth, T, S) and chemical (nitrate, nitrite, ammonia, silicate, phosphate, oxygen) data were collected for each cast and submitted to the U.S. Biological and Chemical Oceanography Data Management Of­ fice (BCO-DMO, http://data.bco-dmo.org/jg/serv/BCO/NorthPacifi c_RDOC/CTD_Profiles.brev0) by the chief scientist (D. Hansell, Univer­ sity of Miami). All labware was acid soaked in 2N HCl overnight with glass bottles subsequently precombusted (450 � C, 3 h minimum), and rinsed thoroughly with Milli-Q water (MQ; >18.2 MΩ⋅cm; Millipore) that was freshly produced on the ship (within 1 day) prior to sample collection. Seawater was gravity filtered directly from the Niskin bottle via silicon tubing through a pre-cleaned, 0.2 μm Whatman Polycap AS 75 nylon cartridge filter into the glass bottles. Filtered samples were analyzed for optical properties within 4 h after sampling to avoid po­ tential long-term storage artifacts.

2012; Timko et al., 2015). However, there are still large uncertainties surrounding the extent to which this recalcitrant organic matter can be photochemically degraded and how long photoproduction rates can be sustained during exposure to solar irradiation. Mopper et al. (1991) propose that the photo-oxidation of DOM largely controls the oceanic residence time for DOM. This hypothesis implicitly assumes that the entire reservoir of DOM is photoreactive, and that it will continue to photodegrade at a constant rate until it is completely removed via mineralization or the production of biological substrates. Recent labo­ ratory studies, however, have shown that photoproduction rates of reactive oxygen species resulting from CDOM photo-degradation are not constant but instead, decrease with increasing photon exposure (Powers et al., 2015), implying that the photo-oxidation of deep bio-refractory DOM may be a function of the accumulated photon dose. These poten­ tially contradictory results call for a closer examination. Studying the changes in the optical properties of DOM over the course of a photo-irradiation experiment may therefore provide a valuable proxy for constraining models of oceanic DOC. In this study, optical measurements (i.e., absorbance and fluores­ cence) were used together with parallel factor analysis (PARAFAC) to assess potential biogeochemical processes controlling the distribution of DOM optical properties in the Gulf of Alaska (GoA). We also evaluated potential impacts of photochemical processing on this deep ocean DOM. The objectives of this study were (1) to reveal the characteristics of DOM absorption and fluorescence in seawater containing perhaps the ocean’s oldest dissolved organic carbon, and (2) to provide data about how photochemistry affects the optical properties of this bio-refractory deepwater DOM once it returns to the surface ocean. This work contributes to the global survey database of DOM optical properties and provides po­ tential optical constraints for modeling marine DOM photochemical decay kinetics.

2.2. Photochemical irradiations The 0.2 μm filtered seawater collected from four stations (Stn. 8, 17, 24, and 29, Fig. 1) was exposed to a constant, spectrally quantified irradiance in 600 mL or 1 L jacketed beakers (Ace Glass) covered with quartz glass using a Suntest CPS (Atlas) solar simulator equipped with a 1.5 kW xenon lamp on the shipboard to simulate sunlight reaching the sea surface. A detailed description of the set up and spectral irradiance entering the beakers is presented in Powers et al. (2015). Surface and deep samples from each station were exposed for up to 48 h with sub-sampling at 24 h for optical analysis. 2.3. Optical measurements and calculation of optical proxies

2. Materials and methods

2.3.1. CDOM absorption measurement and spectral slope calculations CDOM absorption spectra were obtained with filtered samples at room temperature using a single-beam liquid waveguide capillary flow cell (World Precision Instruments Inc., measured at 92.2 cm pathlength), coupled with quartz fiber optic cables to a deuterium-tungsten light

2.1. Sample collection As part of the Deep Ocean Refractory Carbon (DORC) field campaign aboard the R/V Melville in the Gulf of Alaska (August 4–21, 2013) we

Fig. 1. Survey stations occupied during the DORC field campaign in the Gulf of Alaska in August 2013, with bathymetry (water depth (in meters)) and topography (land elevation (in meters)) color coded. Stations with (þ) symbols represent samples collected for photochemical irradiations (Stn. 8, 17, 24, and 29). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) 2

F. Cao et al.

Deep-Sea Research Part I xxx (xxxx) xxx

source (DT-mini GS) and a MAYA2000 PRO spectrophotometer (Ocean Optics, Inc.). To eliminate contamination from pump tubing, cleaning solutions and samples were pulled slowly through the system using Teflon lines and a peristaltic pump (~0.5 ml/min) located downstream from the waveguide. The flow cell was cleaned at regular intervals (~every 10 samples) using a rotating valve that sequentially rinsed with a 20:80 vol/vol acetone:water mixture, 2N HCl, and fresh MQ water. In addition, to alleviate the common problem of microbubbles that accumulate on the interior surfaces of the flow cell when analyzing seawater samples, we dissolved bubbles off the cell walls back into degassed MQ supplied through a Liqui-CelTM MM-0.5x1 Series Membrane Contactor (3MTM) attached to an oilless vacuum pump. This was routine after the cleaning sequence and also used when blanks remained high after cleaning or between seawater samples. Removal of microbubbles was confirmed when no measurable change in absorbance was observed due to bubble compression after pressure was applied to the flow cell with MQ using a separate syringe. While this procedure proved excellent for removing attached microbubbles, the internal epoxy used to attach the membranes to the polycarbonate body of the Liqui-CelTM contributed an observable “CDOM-like” absorbance that could not be eliminated through cleaning the cartridge. Consequently, we flushed the system with MQ water after cleaning, degassing, and between samples until no change was observed in the UV portion of the blank spectrum with repeated “re-blanking” of the MAYA2000 signal. This re-blanked MQ reference accounted for in­ strument drift. Baseline corrections for CDOM absorbance spectra were made by subtracting an offset value that corrects for scattering and refractive index differences between seawater and the MQ blank as described in Reader and Miller (2011). Our capillary waveguide produced a small, reproducible optical resonance (�0.002 m-1) centered close to zero absorbance at the longer wavelengths typically used for blank correc­ tions (e.g., 700–800 nm). We chose to determine our offset value by fitting the raw absorbance spectra between 630 and 640 nm with a nonlinear fitting routine (“nlinfit” function, MATLAB® 2016 Statistics Toolbox; MathWorks, MA) to Eq. (1) (Reader and Miller, 2011): A ¼ Fe

sλ

þO

sample that maximized emission intensity detection at short excitation wavelengths without saturating the CCD at long excitation wavelengths (Gentry-Shields et al., 2013). Post-processing EEM spectra to correct and calibrate the spectra was conducted following Murphy et al. (2010) and included four steps: (1) manufacturer-provided spectral correction pa­ rameters were applied to each EEM, (2) inner filter effects were cor­ rected for sample EEMs following Lakowicz (2013), (3) spectra were normalized to the MQ Raman peak area at an excitation wavelength of 350 nm, and (4) MQ Raman and Rayleigh scatter signals were removed using an interpolation method following protocols developed by Bahram et al. (2006). Fluorescence intensities (FI) are herein reported in Raman Units (RU). A total of 518 samples were pooled for PARAFAC analysis using the DOMFluor toolbox in MATLAB (Stedmon and Bro, 2008). These included survey samples taken directly from vertical profile casts (N ¼ 504) and from shipboard irradiations (see section 2.2) to capture po­ tential components from photo-degraded EEMs measurements (N ¼ 14). Due to higher fluorescent intensities in surface waters relative to deep samples, all EEM spectra were first normalized to unit intensity (i.e., individual fluorescence intensity maxima for each sample was set to 1.0) prior to the modeling process (Murphy et al., 2008). PARAFAC was then applied to the scaled EEM spectra and the model was constructed and further validated using split-half analysis. Once the modeling process was complete, fluorescence intensities were multiplied by their maximum intensities to obtain actual, non-normalized fluorescence in­ tensities for each component in a given sample. 2.4. Ancillary data Estimated surface chlorophyll-a (Chl a) concentrations were ob­ tained from the NASA Ocean Biology Distributed Active Archive Center (https://oceancolor.gsfc.nasa.gov). Daily, 1 km resolution MODIS-A Level 2 ocean color data (data processing version R2014.0) spanning the entire duration of the field campaign were collected for the GoA and spatially binned at 1 km resolution to generate the chlorophyll data product following the default settings for the SeaWiFS Data Analysis System (SeaDAS, 7.3.1). Chlorophyll-a concentrations were then ob­ tained by calculating the median value within a 3�3 pixel array centered on the in situ sampling locations. All derived hydrographic properties (e.g., potential temperature, potential density anomaly) for each Niskin bottle sample were determined using the Matlab seawater library (version 3.3.1, http://www.cmar.csiro.au/datacentre/ext_docs /seawater.htm). The apparent oxygen utilization (AOU) (i.e. the magnitude of oxygen respired in the ocean’s interior) was determined as the difference between the calculated oxygen solubility and the measured CTD dissolved oxygen concentration (SBE43 dissolved oxygen sensor (DO/43–0275)) calibrated to dissolved oxygen bottle samples (standard deviations were 3.142 μmol/kg for all oxygen and 1.365 μmol/kg for deep oxygen).

(1)

where A is the CDOM absorbance, F is a fitting parameter, S is the spectral slope coefficient, and O is a specific offset value obtained for each sample. We then calculated the corrected CDOM absorbance spectrum by normalizing the raw absorbance spectrum using this offset value and converting to a Napierian absorption coefficient spectrum according to Eq. (2): � ag ðλÞ ¼ 2:303 � AðλÞ L; (2) where ag(λ) (m 1) is the Napierian absorption coefficient of CDOM at wavelength λ, A(λ) (unitless) is the offset-corrected CDOM absorbance at λ, and L (m) is the pathlength (0.922 m). The specific ultraviolet absorbance (SUVA254, L mg 1 m 1) was calculated by dividing ag(254) by the DOC concentration (Weishaar et al., 2003). The CDOM spectral slope coefficient (S) for the spectral range 275–295 nm, hereafter referred to as S275-295, was determined by calculating the slope of the log-transformed linear regression over the wavelength interval of 275–295 nm (Helms et al., 2008).

2.5. Statistical analyses All statistical analyses used to assess correlations between selected variables were performed with the open source statistical computing software “R” (http://www.r-project.org/). Pearson correlations were used in all cases unless otherwise noted. A non-parametric MannWhitney U test was conducted to examine optical changes in the photochemical irradiation experiments. Statistical significance levels are reported as non-significant (p > 0.05), significant (0.001 < p < 0.05), or highly significant (p < 0.001).

2.3.2. Fluorescent measurement of DOM and PARAFAC modeling A fluorescent excitation-emission matrix (EEM) spectrum was ob­ tained for each sample with an Aqualog spectrofluorometer (HORIBA Jobin Yvon Inc., NJ, USA) using a 1-cm quartz cell with MQ as the blank. EEM fluorescence intensities were measured by scanning across an excitation range of 240–450 nm (5 nm intervals) and capturing emission spectra over a wavelengths range of 280–500 nm at 3.2 nm intervals. We optimized data quality by determining the integration time for each

3. Results and discussion 3.1. Oceanographic setting Our study area encompassed four distinct water masses (Fig. 2). A 3

F. Cao et al.

Deep-Sea Research Part I xxx (xxxx) xxx

well-mixed, relatively fresh (salinity < 32) and warm (potential tem­ perature (θ) >10 � C) layer extending through a summer mixed layer of ~50 m defines the North Pacific Surface Water (NPSW), a part of the subarctic gyre bounded eastward by the northward flowing Alaskan Current (Peterson et al., 2005). The Pacific Halocline Water (PHW) forms a layer from roughly 50–150 m and is defined by a rapid salinity increase from 32 to 33.5 (Steele et al., 2004). Beneath the PHW, the North Pacific Intermediate Water (NPIW) is delineated by water with a potential density anomaly (σθ) of 26.6–27.4 kg/m3 (Hansell et al., 2002), occupying depths between ~200 and 1000 m. Below this, Pacific Common Water (PCW) is relatively more saline (salinity > 34) than the other three stratified water masses, and makes up a fairly uniform, well-mixed physical feature with elevated σθ of 27–28.1 kg/m3.

3.2.2. Profiles of DOC, CDOM, and FDOM Vertical profiles of DOC in the GoA are similar to those reported in previous studies (e.g., Swan et al., 2009), with greater concentrations in the surface ocean (mean (�SD) of 62 (�6.7 μM)) that decreased with depth (39.5 (�2.4 μM))) in the ocean interior. CDOM optical properties varied widely in the surface mixed layer. ag(325) ranged from 0.12 to 0.47 m–1 with an overall mean (�SD) of 0.24 (�0.092) m 1 and ag(254) ranged from 1.24 to 2.44 m–1 with an overall mean (�SD) of 1.60 (�0.304) m 1. This variation likely reflects the local balance of various processes, including CDOM sources from biological production and coastal runoff and removal processes via solar bleaching, potential loss into the POC pool, or other in situ processes that convert colored to uncolored organic compounds. Values increased gradually in the sub­ surface and upper mesopelagic layer (~150–500 m) and were slightly lower with less variance at depth (e.g., an overall mean ag(325) from 0.22 m 1 at 400 m to 0.15 m 1 at 4500 m) (Fig. 4). Elevated ag(325) in the surface water was occasionally observed associated with lower salinity (e.g., ag(325) ¼ 0.47 m–1, salinity ¼ 31.5 at Stn.15), and most likely reflects terrestrial materials discharged from glacial runoff and Alaskan rivers (Hood et al., 2009; Neal et al., 2010). Terrestrial inputs could also partly explain the greater vertical differences of ag(325) at the more coastal stations compared to the uniform CDOM profiles observed at the remote, open-ocean stations (e.g., Stn. 28) considered to be less influenced by coastal runoff as in Yamashita and Tanoue (2009). Our data often showed a slight increase in ag(325) between ~100 and 200 m, presumably resulting from CDOM release accompanied with degrada­ tion of sinking biogenic particles (Nelson et al., 2010; Swan et al., 2009) in the absence of photodegradation. SUVA254 displayed higher values and slightly larger variability in the surface ocean compared to deeper samples (SUVA254 of 2.13 � 0.23 L mg 1 m 1 in waters less than 50 m, versus values of 1.94 � 0.1 L mg 1 m 1 at 200 m, p < 0.001), with low, relatively constant values in the deep ocean (e.g., SUVA254 of 1.97 � 0.05 L mg 1 m 1 at 1000 m compared to 1.81 � 0.04 L mg 1 m 1 at 4000 m, p > 0.05). Similarly, CDOM spectral slope coefficients decreased from the sea surface down to 200 m (e.g., S275-295 of 0.0315 � 0.003 nm 1 at 50 m compared to 0.0238 � 0.002 nm 1 at 200 m, p < 0.001), and exhibited more variability compared to intermediate and deep waters (e.g., mean S275-295 of 0.0186 nm 1 at 1000 m and 0.0214 � nm 1 at 4000 m). Since SUVA254 has been used as a surrogate for DOC aromaticity (Weishaar et al., 2003) and S275-295 is reported to be related to DOM molecular weight (Helms et al., 2008)-, the fairly con­ stant SUVA254 and S275-295 in deeper waters below the permanent thermocline (SUVA254 of 1.90 � 0.09 L mg 1 m 1 and S275-295 of 0.0196 � 0.001 nm 1, depths > 1000 m) suggest that, at least for average molecular weight and aromaticity, the DOM composition re­ mains stable in the deep GoA. This is understandable given that CDOM in this deep water may have already been significantly modified during horizontal transport via the overturning oceanic circulation, resulting in DOM with a uniform composition. Vertical profiles of the FDOMH and FDOMP displayed two distinct patterns (Fig. 4). The lowest values of FDOMH were observed at the sea surface (e.g., FI (mean � SD) ¼ 0.014 � 0.004, 0.013 � 0.004, 0.012 � 0.003 RU for C1, C2, and C4 in NPSW) with the highest FI values in PHW of 0.021 � 0.005, 0.021 � 0.005, and 0.017 � 0.002 RU for components C1, C2, and C4, respectively. Profiles of FDOMP, however, contrasted markedly with those of FDOMH, having slightly elevated values in sur­ face waters (FI of 0.021 � 0.005 and 0.019 � 0.003 RU for C3 and C5 in NPSW, respectively), with a decrease through the water column (FI of 0.013 � 0.002 and 0.013 � 0.001 RU for C3 and C5 in NPIW, respec­ tively) until reaching a constant value in the deep ocean (FI of 0.012 � 0.002 and 0.012 � 0.001 RU for C3 and C5 at depth > 1000 m, respectively). Our depth profiles of FDOMP are similar to those reported in the Northwestern North Pacific by Yamashita et al. (2010) and the extensive open-water survey made earlier in areas other than the North Pacific by Jørgensen et al. (2011). Thus, the similarity in FDOMP observed across ocean basins is consistent with the proposed

3.2. DOM optical properties in the GoA 3.2.1. Spectral characteristics of fluorescent DOM Five EEM fluorescent components (hereafter referred to as C1 to C5) were resolved and validated by PARAFAC modeling (Fig. 3). C1 (Exmax/ Emmax ¼ 250/476) and C2 (Exmax/Emmax ¼ 250 (355)/404) are similar to previously reported humic-like components having broad emission maxima at wavelengths >400 nm (Jørgensen et al., 2011). These humic-like components have been proposed to be of terrestrial origin and are widely observed in aquatic environments, including open ocean waters (Kowalczuk et al., 2013; Yamashita et al., 2010). Our component C4 (Exmax/Emmax ¼ 320/350) exhibited a spectral signature that most closely resembled a combination of a marine humic-like component produced during organic matter breakdown by microbes and a protein-like component (Dainard and Gu�eguen, 2013; Stedmon and Markager, 2005). The vertical profile of C4 (as shown later in Fig. 4), however, is similar to others reported for humic-like marine components and therefore, we interpret C4 here in the GoA as a humic-like compo­ nent of marine origin. Two protein-like fluorophores (C3 (Exmax/Emmax ¼ 275/330) and C5 (Exmax/Emmax ¼ 270/294)) were also resolved that exhibit maximum emission wavelengths shorter than 400 nm and are proposed to be characteristic of proteinaceous components. Consistent with other oceanic surveys, C3 and C5 are assigned as tryptophan-like and tyrosine-like proteinaceous components, respectively (Jørgensen et al., 2011; Yamashita and Tanoue, 2003). In the text that follows, we refer to the pooled humic-like (C1, C2, and C4) and protein-like (C3 and C5) components as “FDOMH” and “FDOMP”, respectively.

Fig. 2. Potential temperature-salinity (T–S) diagram of the study area, overlaid with potential density anomaly (σθ, kg/m3) contour lines (isopycnals) in gray. Abbreviations are NPSW for North Pacific Surface Water, PHW for Pacific Halocline Water, NPIW for North Pacific Intermediate Water, and PCW for Pacific Common Water. 4

F. Cao et al.

Deep-Sea Research Part I xxx (xxxx) xxx

Fig. 3. (a–e) Contour plots for the five fluorescent components revealed and validated using PARAFAC modeling. (f–j) Line plots represent split-half validation results for each component, with excitation (blue line) and emission (red line) spectra estimated for the entire dataset. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

biologically and chemically recalcitrant nature of DOC in the deep ocean.

from direct algal inputs, grazing, or fairly rapid production of secondary metabolites via bacterial transformation of algal material in the surface ocean (Kinsey et al., 2018). In contrast, neither the tryptophan-like component C3 nor the tyrosine-like C5 component was correlated with Chl a concentrations (p > 0.05), suggesting that recently produced algal DOM may not be directly responsible for FDOMP enrichment in the sea surface, as suggested by the moderate correlations between protein-like components and surface chlorophyll previously reported throughout the Pacific Ocean (Yamashita et al., 2017). Because protein-like fluorescence components have been attributed to trypto­ phan containing compounds (e.g., peptides) excreted by phytoplankton and/or terrigenous materials containing proteinaceous compounds in

3.3. Correlations between DOM optical properties and other biological indicators The potential influence of algal biomass on the abundance of CDOM and FDOM variations in surface waters of the GoA is suggested by the moderate positive correlations observed between CDOM and humic-like FDOMH to Chl a (e.g., r ¼ 0.7 for ag(325) with Chl a and r ¼ 0.56 for C1 with Chl a, N ¼ 13, p < 0.05, figures not shown). This is consistent with autochthonous production of CDOM and FDOMH in the GoA possibly 5

F. Cao et al.

Deep-Sea Research Part I xxx (xxxx) xxx

Fig. 4. Vertical distributions of DOC (μM), CDOM optical properties (represented as ag(254) (m 1), SUVA254 (L mg 1m 1), ag(325) (m 1), CDOM spectral slope (S275-295) (nm 1)), and the five fluorescent components in Raman Units (RU) resolved from PARAFAC modeling.

previous publications, our lack of correlation between the fluorescent protein-like components versus Chl a indicate that the surface C3 and C5 components may not represent newly released algal compounds at the time of sampling in the GoA. The overall concentrations and variations observed in surface waters may rather reflect non-algal processes (e.g. bacterial decay) or a background signal of biologically recalcitrant C3 and C5 in this area of the Pacific. The reader is reminded too, that the Chl a data used to examine these correlations are average values ob­ tained over a 1 km grid by remote sensing and not direct measurements made from bottle data that would more readily capture smaller scale variability. We also investigated relationships between CDOM and each FDOM component versus AOU in the GoA. No correlations were found between AOU and CDOM abundance in the NPIW water mass (150–1000 m, p > 0.05), implying that dynamics of CDOM cycling in the mesopelagic zone might be driven by processes other than remineralization of biogenic organic carbon. In the deeper ocean below the permanent thermocline at depths >1000 m, however, we observed that CDOM was strongly correlated to AOU (r ¼ 0.84, p < 0.001, N ¼ 315, Fig. 5 (a)), supporting the suggestion that CDOM is produced as a result of remineralization of organic materials in the oldest water mass. Comparatively, the FDOMH components generally showed changes coincident with AOU in the

entire ocean interior at depths greater than 100 m (e.g., r ¼ 0.92 for C1 to AOU, p < 0.001, N ¼ 459, Fig. 5 (b)). This is consistent with previous observations made in the Pacific Ocean (Kim and Kim, 2015; Swan et al., 2009; Yamashita et al., 2010; Yamashita and Tanoue, 2009), suggesting that FDOMH is regenerated during the remineralization of biogenic organic matter throughout the water column. Interestingly, C1, which is most often interpreted as a humic-like fluorophore of terrestrial origin, shows significant positive correlation with AOU (Fig. 5 (b)), indicating a source associated with marine respiration. This result underscores the fact that EEMs and PARAFAC analysis, while potentially useful optical proxies that capture the fluorescent properties of DOM, are not true chemical analyses that can confirm origin or molecular composition. Consequently, FDOM results should be interpreted with care and only in the context of known oceanographic processes. In contrast to FDOMH, no statistically significant relationships were observed between FDOMP and AOU throughout the water column (p > 0.05, Fig. 5 (c)). This suggests the accumulation of FDOMP is largely independent of remineralization activities. Similar results which found no correlations between the protein-like components and AOU were obtained in the global dataset of waters other than the North Pacific by Jørgensen et al. (2011). Moreover, the C3 and C5 components in the GoA were comparable in magnitude to the FDOMP reported across other 6

F. Cao et al.

Deep-Sea Research Part I xxx (xxxx) xxx

Table 1 Changes of DOM fluorescence intensity (mean � SDa, RU) for surface and deep ocean water during photo-irradiation experiments in the GoA. t ¼ 0 (initial) C1 C2 C3 C4 C5

t ¼ 24 h

t ¼ 48 h

Surface

Bottom

Surface

Bottomb

Surface

Bottom

0.014 � 0.001 0.012 � 0.001 0.020 � 0.003 0.011

0.022 � 0.001 0.020 � 0.001 0.012 � 0.002 0.014 0.012

0.009 0.001 0.009 0.001 0.020 0.002 0.011 0.001 0.021 0.003

0.008 � 0.002 0.008 � 0.003 0.023

0.018 � 0.002

0.010 � 0.001 0.010 � 0.001 0.034 � 0.005 0.011 � 0.001 0.030 � 0.003

0.006 0.001 0.007 0.002 0.017 0.006 0.009 0.003 0.017 0.004

± ± ± ± ±

0.009 � 0.002 0.026 � 0.005

� � � � �

a

SD standard deviation was calculated for water at the four stations sampled for photochemical experiments. SD values less than 0.001 are omitted. b Results marked bold were significantly different from the previous time point at a confidence level of 0.05.

24 h, compared to a FI of 0.008 � 0.002 RU at t ¼ 48 h, p > 0.05), presumably because the photochemically labile fluorescent components in surface waters were already removed by sunlight prior to sampling. The protein-like fluorescent components, C3 and C5, showed a different response to irradiation in the solar simulator than that of FDOMH. Fluorescence intensities of both the C3 and C5 components increased in deep water samples after 24 h exposure (FI of 0.020 � 0.002 and 0.021 � 0.003 RU for C3 and C5, respectively) relative to their initial FI values of 0.012 � 0.002 and 0.012 RU (p < 0.05), but remained constant with further exposure to 48 h (FI of 0.017 � 0.006 and 0.017 � 0.004 RU for C3 and C5, respectively, at t ¼ 48 h, p > 0.05), suggesting that photochemical production of protein-like components seen in the first 24 h was not sustained over longer photon exposure. By contrast, in the surface samples neither of the two protein-like components showed changes in fluorescence intensity through the entire exposure up to 48 h (with a FI of 0.034 � 0.005 and 0.023 RU at t ¼ 24 and 48 h, respec­ tively, relative to an initial FI of 0.020 � 0.003 RU for C3 at t ¼ 0, p > 0.05) as observed for the three humic-like components in the surface water samples. Different responses of the fluorescent components be­ tween the deep and surface waters upon exposure reinforces the fact that fluorescent components derived from PARAFAC are not in themselves identifiable chemical compounds but rather only represent similar op­ tical properties that do not necessarily correlate with environmental reactivity. The increase of the protein-like components in deep water samples with simulated solar irradiation observed here is opposite of the photochemical loss of overall fluorescence that was reported in the Atlantic Ocean deep water under otherwise similar exposure conditions (24 h, 1000 W/m2; Timko et al., 2015) Because our EEM spectra were collected immediately after exposure and therefore, the fluorescent in­ tensity increase of protein-like components was likely not caused by the recovery of photodegraded DOM due to dark storage as demonstrated by Grzybowski et al. (2019). Rather, the increase observed for the protein-like components in the deep samples may be attributed to the different water masses investigated in the two studies. The C3 and C5 components in our bottom water sample are part of the oldest oceanic carbon reservoir, compared to the relatively fresher carbon components in the deep Atlantic, which was proposed to be more photochemically sensitive and perhaps less likely to survive transport to the North Pacific in the GoA (Timko et al., 2015). Despite the similar fluorescence feature that PARAFAC identifies as the same protein-like component in both studies, the protein-like components seem to be intrinsically different in the GoA, at least in terms of their photochemical reactivity, from that in protein-like components in the deep Atlantic Ocean. Again, this re­ inforces the idea that FDOM analysis offers a useful optical proxy for DOM distributions and reactivity but does not represent direct

Fig. 5. Relationships between apparent oxygen utilization (AOU) and (a) ag(325), (b) humic-like component C1, and (c) protein-like component C3. Sampling depth (in meters) is denoted with the color bar on the right. Data collected at depths <100 m were not included. In (a), the regression line was determined using only data collected below the permanent thermocline at depths�1000 m (N ¼ 315). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

ocean basins. Taken together, these results account for more than half of the total fluorescence in the ocean’s interior, suggesting that a consid­ erable fraction of FDOM is ubiquitous in the ocean. 3.4. Photo-irradiation effects on FDOM Changes in sample FDOMH varied as a function of its original collection depth as well as irradiation time in our photochemical irra­ diation experiments (Table 1). Relative to dark controls, all three humiclike components, FDOMH, were extensively bleached in the first 24 h of irradiation (t ¼ 24 h), with deep water samples resulting in 60, 52, and 27% intensity loss for C1, C2, and C4, respectively, relative to their initial intensities at t ¼ 0 (Mann-Whitney U test, p < 0.05). Continued irradiation of deep samples to 48 h, however, did not result in additional discernible changes for FDOMH from those measured at t ¼ 24 h (e.g., with a FI of 0.009 � 0.001 RU and 0.006 � 0.001 RU for C1 at t ¼ 24 and 48 h, respectively, p > 0.05). Compared to bottom waters, no discernible changes were seen for the three humic-like components in the surface water samples in either the first 24 h or after extended exposure to 48 h (e.g., for C1 FI of 0.014� 0.001 RU at t ¼ 0 and 0.01 � 0.001 RU at t ¼ 7

F. Cao et al.

Deep-Sea Research Part I xxx (xxxx) xxx

molecular/structural information regarding specific DOM constituents that comprise the different fluorescent signatures. The proportion of total fluorescent intensities attributed to the five PARAFAC components in the deep samples changed as the fluorescent humic fractions were removed and the protein-like components were enriched during irradiation, making them more similar to those in sur­ face samples. Prior to exposure in the solar simulator (t ¼ 0), the C1 component of FDOMH in the deep samples was 28% of the total fluo­ rescence and comparable to the FDOMP signal that included both the C3þC5 components and which accounted for ~29% of the total fluo­ rescence in deep samples. After 24 h of irradiation, C1 decreased to 13% of the total fluorescence while FDOMP (i.e., C3þC5) increased to more than 58%, becoming comparable to the fluorescent signature in surface samples (i.e., C1 11–18% and C3þC5 >50% in surface waters at t ¼ 0 and t ¼ 24, respectively). The same observations were reported by Helms et al. (2013) who showed that the optical characteristics of deep-ocean DOM were altered to closely resemble those for surface waters after exposure of the deepwater to solar radiation. Further, our observations that the relative contributions of specific PARAFAC com­ ponents within the FDOM reservoir found in deep ocean samples were similar to that of surface water after exposure is consistent with evidence from molecular analysis on DOM in the GoA reported earlier for some of the same water samples collected from the same bottles and analyzed here (Medeiros et al., 2015).

4. Conclusions Here we present a full-depth distribution of absorbance and fluo­ rescence optical properties for the dissolved organic carbon reservoir located in the GoA. PARAFAC modeling of the fluorescent DOM iden­ tified three humic-like and two protein-like fluorescent components that are ubiquitous in the world’s oceans. Depth profiles of fluorescent DOM properties revealed lower humic-like and higher protein-like fluores­ cence in surface waters, while deep samples were characterized by elevated humic-like and lower abundance of protein-like components. In terms of optical characteristics, we found a compositionally stable DOM reservoir within the oldest water masses in the Northeastern subarctic Pacific Ocean. Measurements of the photochemically-induced alterations in the optical properties of oceanic DOM confirmed the initial, general pho­ toreactivity of deep ocean DOM, which resulted in dramatic changes in relative contributions of the PARAFAC-determined FDOM components to the total fluorescent signal. Photolysis led to a more uniform contri­ bution from the different identified fluorophores in the deep ocean water after exposure to simulated sunlight, with the deep water becoming similar in fluorescent composition to that found in the surface ocean. We also found that the protein-like optical components dominate what can be considered the “photo-bio-refractory” FDOM signal after irradiation. We encourage further investigation to determine the chemical composition of the optical components for a better under­ standing of the biogeochemical significance of these protein-like fluo­ rescent materials in the deep-ocean carbon reservoir.

3.5. Deep-ocean FDOM Published results and those from our study indicate that a significant portion of deep-ocean DOM is photoreactive, at least as reflected by changes in its optical properties. In particular, our irradiation results indicate that the FDOMH fraction in the deep ocean DOM reservoir is highly susceptible to solar bleaching, supporting the supposition that deep-ocean DOM is photochemically reactive. However, photochemical loss of FDOMH fluorescence does not continue indefinitely with extended irradiation, indicating a more general lack of photochemical reactivity of DOM once the optically-active components are removed. This is consistent with the photoproduction results of reactive oxygen species (ROS) observed by Powers et al. (2015) on this same set of samples. They determined that over a 24 h irradiation of the deep-ocean samples, superoxide steady-state concentrations fell off sharply and hydrogen peroxide accumulation plateaued. Thus, it appears that as chromophores or fluorophores are photochemically removed, deep ocean DOM loses its limited capacity to initiate photochemical re­ actions, and becomes less sensitive to sunlight and more photochemi­ cally refractory. If we assume our 48 h of continuous irradiation was long enough to sufficiently remove the photoreactive DOM as the Powers et al. (2015) ROS results suggest, then what remains is the photochemically re­ fractory protein-like fluorescent fraction that dominates the photolyzed FDOM signal in the remaining photo-refractory deep ocean carbon pool (>50% of the total FDOM). However, the quantitative contribution of this protein-like fluorescent fraction to the bio- and/or photo-refractory DOM reservoirs in the deep ocean still remains unclear, warranting further work that combines chemical structures and optical data to better understand the importance of protein-like fluorophores in the ocean. In addition, since the biogeochemical cycles of oceanic DOC and its optically active components are largely decoupled (Swan et al., 2009), the FDOMH photodegradation we observed in the deep ocean during exposure to irradiation in a solar simulator does not necessarily represent the DOC decrease to the same extent. Arai et al. (2018) re­ ported a significant decrease in bacterial-derived FDOMH while the DOC concentration changed very little throughout the entire photo-irradiation. This highlights the need to consider both the total DOC reservoir and DOM optical properties separately in the biogeo­ chemical studies of oceanic DOM.

Declaration of competing interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements This study was funded by NSF-Chemical Oceanography Grant num­ ber OCE-1234388. We gratefully acknowledge the caption and crew of the R/V Melville, D. Hansell, and personnel in his lab for assistance with sampling. We thank C. Hopkinson and D. Di Iorio for helpful discussions. We also thank the two anonymous reviewers for their thoughtful com­ ments on the manuscript. References Andrew, A.A., Del Vecchio, R., Subramaniam, A., Blough, N.V., 2013. Chromophoric dissolved organic matter (CDOM) in the Equatorial Atlantic Ocean: optical properties and their relation to CDOM structure and source. Mar. Chem. 148, 33–43. Arai, K., Wada, S., Shimotori, K., Omori, Y., Hama, T., 2018. Production and degradation of fluorescent dissolved organic matter derived from bacteria. J. Oceanogr. 74 (1), 39–52. Bahram, M., Bro, R., Stedmon, C., Afkhami, A., 2006. Handling of Rayleigh and Raman scatter for PARAFAC modeling of fluorescence data using interpolation. J. Chemom. 20 (3-4), 99–105. Chen, H., Stubbins, A., Perdue, E.M., Green, N.W., Helms, J.R., Mopper, K., Hatcher, P. G., 2014. Ultrahigh resolution mass spectrometric differentiation of dissolved organic matter isolated by coupled reverse osmosis-electrodialysis from various major oceanic water masses. Mar. Chem. 164, 48–59. Dainard, P.G., Gu� eguen, C., 2013. Distribution of PARAFAC modeled CDOM components in the north Pacific Ocean, bering, chukchi and beaufort seas. Mar. Chem. 157, 216–223. Dittmar, T., Stubbins, A., 2014. 12.6—Dissolved Organic Matter in Aquatic Systems. Treatise on Geochemistry, second ed. Elsevier, Oxford, pp. 125–156. Gentry-Shields, J., Wang, A., Cory, R.M., Stewart, J.R., 2013. Determination of specific types and relative levels of QPCR inhibitors in environmental water samples using excitation–emission matrix spectroscopy and PARAFAC. Water Res. 47 (10), 3467–3476. Grzybowski, W., Szewczun, A., Tarasiewicz, P., 2019. Dark recovery of photodegraded dissolved organic matter as a source of a protein-like fluorophore in natural waters. J. Photochem. Photobiol. A Chem. 371, 182–187. Hansell, D.A., 2002. DOC in the global ocean carbon cycle. Biogeochem. Mar. Dissolved Org. Matter 685–715.

8

F. Cao et al.

Deep-Sea Research Part I xxx (xxxx) xxx 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 (3). Peterson, T.D., Whitney, F.A., Harrison, P.J., 2005. Macronutrient dynamics in an anticyclonic mesoscale eddy in the Gulf of Alaska. Deep Sea Res. Part II Top. Stud. Oceanogr. 52 (7), 909–932. Powers, L.C., Babcock-Adams, L.C., Enright, J.K., Miller, W.L., 2015. Probing the photochemical reactivity of deep ocean refractory carbon (DORC): lessons from hydrogen peroxide and superoxide kinetics. Mar. Chem. 177, 306–317. Reader, H.E., Miller, W.L., 2011. Effect of estimations of ultraviolet absorption spectra of chromophoric dissolved organic matter on the uncertainty of photochemical production calculations. J. Geophys. Res.: Oceans 1978–2012 (C8), 116. Repta, D.J., 2014. Chemical characterization and cycling of dissolved organic matter. In: Hansell, D.A., Carlson, C.A. (Eds.), Biogeochemistry of Dissolved Organic Matter, pp. 21–63. Stedmon, C.A., Bro, R., 2008. Characterizing dissolved organic matter fluorescence with parallel factor analysis: a tutorial. Limnol Oceanogr. Methods 6, 572–579. Stedmon, C.A., Markager, S., 2005. Tracing the production and degradation of autochthonous fractions of dissolved organic matter by fluorescence analysis. Limnol. Oceanogr. 50 (5), 1415–1426. Steele, M., Morison, J., Ermold, W., Rigor, I., Ortmeyer, M., Shimada, K., 2004. Circulation of summer pacific halocline water in the arctic ocean. J. Geophys. Res.: Oceans 1978–2012 (C2), 109. Stubbins, A., Niggemann, J., Dittmar, T., 2012. Photo-lability of deep ocean dissolved black carbon. Biogeosciences 9 (5), 1661–1670. 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. Oceanogr. Res. Pap. 56 (12), 2175–2192. Timko, S.A., Maydanov, A., Pittelli, S.L., Conte, M.H., Cooper, W.J., Koch, B.P., SchmittKopplin, P., Gonsior, M., 2015. Depth-dependent photodegradation of marine dissolved organic matter. Front. Mar. Sci. 2, 66. 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 (20), 4702–4708. Yamashita, Y., Cory, R.M., Nishioka, J., Kuma, K., Tanoue, E., Jaff� e, R., 2010. Fluorescence characteristics of dissolved organic matter in the deep waters of the Okhotsk Sea and the northwestern North Pacific Ocean. Deep Sea Res. Part II Top. Stud. Oceanogr. 57 (16), 1478–1485. Yamashita, Y., Hashihama, F., Saito, H., Fukuda, H., Ogawa, H., 2017. Factors controlling the geographical distribution of fluorescent dissolved organic matter in the surface waters of the Pacific Ocean. Limnol. Oceanogr. 62, 2360–2374. Yamashita, Y., Tanoue, E., 2003. Chemical characterization of protein-like fluorophores in DOM in relation to aromatic amino acids. Mar. Chem. 82 (3), 255–271. Yamashita, Y., Tanoue, E., 2009. Basin scale distribution of chromophoric dissolved organic matter in the Pacific Ocean. Limnol. Oceanogr. 54 (2), 598–609.

Hansell, D.A., Carlson, C.A., Repeta, D.J., Schlitzer, R., 2009. Dissolved Organic Matter in the Ocean: a Controversy Stimulates New Insights. Hansell, D.A., Carlson, C.A., Suzuki, Y., 2002. Dissolved organic carbon export with North Pacific intermediate water formation. Glob. Biogeochem. Cycles 16 (1), 7-1-78. Helms, J.R., Stubbins, A., Perdue, E.M., Green, N.W., Chen, H., Mopper, K., 2013. Photochemical bleaching of oceanic dissolved organic matter and its effect on absorption spectral slope and fluorescence. Mar. Chem. 155, 81–91. 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 (3), 955–969. Hood, E., Fellman, J., Spencer, R.G., Hernes, P.J., Edwards, R., D’Amore, D., Scott, D., 2009. Glaciers as a source of ancient and labile organic matter to the marine environment. Nature 462 (7276), 1044–1047. Jørgensen, L., Stedmon, C.A., Kragh, T., Markager, S., Middelboe, M., Søndergaard, M., 2011. Global trends in the fluorescence characteristics and distribution of marine dissolved organic matter. Mar. Chem. 126 (1), 139–148. Kim, J., Kim, G., 2015. Importance of colored dissolved organic matter (CDOM) inputs from the deep sea to the euphotic zone: results from the East (Japan) Sea. Mar. Chem. 169, 33–40. Kinsey, J.D., Corradino, G., Ziervogel, K., Schnetzer, A., Osburn, C.L., 2018. Formation of chromophoric dissolved organic matter by bacterial degradation of phytoplanktonderived aggregates. Front. Mar. Sci. 4, 1–16. Kowalczuk, P., Tilstone, G.H., Zabłocka, M., R€ ottgers, R., Thomas, R., 2013. Composition of dissolved organic matter along an Atlantic Meridional Transect from fluorescence spectroscopy and Parallel Factor Analysis. Mar. Chem. 157, 170–184. Lakowicz, J.R., 2013. Principles of Fluorescence Spectroscopy. Springer Science & Business Media. Lechtenfeld, O.J., Kattner, G., Flerus, R., McCallister, S.L., Schmitt-Kopplin, P., Koch, B. P., 2014. Molecular transformation and degradation of refractory dissolved organic matter in the Atlantic and Southern Ocean. Geochem. Cosmochim. Acta 126, 321–337. Medeiros, P., Seidel, M., Powers, L.C., Dittmar, T., Hansell, D.A., Miller, W.L., 2015. Dissolved organic matter composition and photochemical transformations in the northern North Pacific Ocean. Geophys. Res. Lett. 42 (3), 863–870. Mopper, K., Zhou, X., Kieber, R.J., Kieber, D.J., Sikorski, R.J., Jones, R.D., 1991. Photochemical degradation of dissolved organic carbon and its impact on the oceanic carbon cycle. Nature 353 (6339), 60–62. Murphy, K.R., Butler, K.D., Spencer, R.G., Stedmon, C.A., Boehme, J.R., Aiken, G.R., 2010. Measurement of dissolved organic matter fluorescence in aquatic environments: an interlaboratory comparison. Environ. Sci. Technol. 44 (24), 9405–9412. Murphy, K.R., Stedmon, C.A., Waite, T.D., Ruiz, G.M., 2008. Distinguishing between terrestrial and autochthonous organic matter sources in marine environments using fluorescence spectroscopy. Mar. Chem. 108 (1), 40–58. Neal, E.G., Hood, E., Smikrud, K., 2010. Contribution of glacier runoff to freshwater discharge into the Gulf of Alaska. Geophys. Res. Lett. 37 (6).

9