Photochemical alteration of dissolved organic matter and the subsequent effects on bacterial carbon cycling and diversity

FEMS Microbiology Ecology,92,2016,fiw048 doi: 10.1093/femsec/fiw048 Advance Access Publication Date: 2 March 2016 Research Article RESEARCH ARTICLE Photochemical alteration of dissolved organic matter and the subsequent effects on bacterial carbon cycling and diversity Christian Lønborg1,2,∗,MarNieto-Cid 3, Victor Hernando-Morales4, Marta Hern´ andez-Ruiz4, Eva Teira4and Xos´ eAnt´ on ´ Alvarez-Salgado3 1Australian Institute of Marine Science, PMB 3, Townsville MC, QLD 4810, Australia, 2Centre for Sustainable Aquatic Research, College of Science, Wallace Building, Swansea University, Swansea SA2 8PP, UK, 3CSIC, Instituto de Investigaci´ ons Mari ˜ nas, Eduardo Cabello 6, 36208 Vigo, Spain and 4Departamento de Ecolox´ ıa e Biolox´ ıa Animal, Universidade de Vigo, 36200 Vigo, Spain ∗Corresponding author: Australian Institute of Marine Science, PMB 3, Townsville MC, QLD 4810, Australia. Tel: +61-7-4753-4382; Fax: +61-7-4772-5852; E-mail: [email protected] One sentence summary: Sunlight exposure impacts different organic matter sources differently, which impacts the physiology and community composition of the bacteria degrading it. Editor: Gary King ABSTRACT The impact of solar radiation on dissolved organic matter (DOM) derived from 3 different sources (seawater, eelgrass leaves and river water) and the effect on the bacterial carbon cycling and diversity were investigated. Seawater with DOM from the sources was first either kept in the dark or exposed to sunlight (4 days), after which a bacterial inoculum was added and incubated for 4 additional days. Sunlight exposure reduced the coloured DOM and carbon signals, which was followed by a production of inorganic nutrients. Bacterial carbon cycling was higher in the dark compared with the light treatment in seawater and river samples, while higher levels were found in the sunlight-exposed eelgrass experiment. Sunlight pre-exposure stimulated the bacterial growth efficiency in the seawater experiments, while no impact was found in the other experiments. We suggest that these responses are connected to differences in substrate composition and the production of free radicals. The bacterial community that developed in the dark and sunlight pre-treated samples differed in the seawater and river experiments. Our findings suggest that impact of sunlight exposure on the bacterial carbon transfer and diversity depends on the DOM source and on the sunlight-induced production of inorganic nutrients. Keywords: dissolved organic matter; solar radiation; bacterial diversity; bacterial carbon demand; bacterial growth efficiency INTRODUCTION Microbial activity in aquatic systems is mainly regulated by the energy and nutrients contained within the dissolved organic matter (DOM) pool (Hedges 2002). Coastal waters are the most productive and biogeochemically active marine ecosystems, and therefore play key roles in the production and degradation of DOM (Wollast 1998; Lønborg and ´ Alvarez-Salgado 2012). DOM in coastal waters originates from either autochthonous or allochthonous sources; autochthonous DOM is produced within the system, primarily by macrophytes (Søndergaard 1981)and planktonic organisms (Kawasaki and Benner 2006; Lønborg et al. 2009), whereas allochthonous DOM is mainly of terrestrial origin (Sobczak et al. 2005). The combined effects of both Received: 26 October 2015; Accepted: 28 February 2016 C FEMS 2016. All rights reserved. For permissions, please e-mail: [email protected] 1 by guest on April 12, 2016http://femsec.oxfordjournals.org/Downloaded from 2FEMS Microbiology Ecology, 2016, Vol. 92, No. 5 photochemical and microbial processes have been considered the main factor responsible for the degradation of natural DOM (Tranvik and Bertilson 2001). In addition, both processes are also able to produce recalcitrant DOM with lifetimes of years to millennia (Miller and Moran 1997; Benner and Biddanda 1998;Obernosterer, Reitner and Herndl 1999). Therefore, DOM in coastal waters is a complex mixture of exudates, leachates, and degradation and condensation products that varies widely in elemental composition and molecular structure and, consequently, in reactivity. Given that heterotrophic bacteria are the major biological DOM sink, they regulate whether the degraded compounds are used for biomass or energy production. The ratio between bacterial production (BP) and the sum of BP and bacterial respiration (BR) has been termed the bacterial growth efficiency (BGE). This ratio varies widely (from <1 to 90%) depending on the bacterial community composition (Reinthaler, Winter and Herndl 2005), nutrient and DOM bioavailability (e.g. Rivkin and Anderson 1997; Apple and del Giorgio 2007; Lønborg et al. 2010a), water temperature (Rivkin and Legendre 2001) and UV-light exposure (Lønborg et al. 2013). The heterotrophic bacterial community consists of members from various phylogenetic lineages, with the distribution and abundance of these species being controlled by both biological and chemical factors (Fuhrman et al. 2006; Giovannoni and Vergin 2012). However, much less is known about how shifts in community structure may influence the microbial carbon cycling. Photochemical reactions induced by solar radiation, especially in the UV range of the spectrum (UV-B, 280–315 nm; UV-A, 315–400 nm), are particularly important in coastal waters with high loads of coloured allochthonous DOM, where they have been shown to transform DOM into labile inorganic (mainly NH4+and HPO42–) and organic (e.g. amino acids) compounds that can support bacterial respiration and biomass production (e.g. Moran and Zepp 1997; Obernosterer and Benner 2004). In addition, photochemical processes have also been shown to enhance the cross-linking, humification and polymerization of labile biomolecules into more recalcitrant compounds (Kieber et al. 1997; Benner and Biddanda 1998; Obernosterer, Reitner and Herndl 1999). These sunlight-induced reactions can be seen as the abiotic counterpart of the microbial carbon pump, which suggests that the microbial utilization of organic matter is related to the release of recalcitrant compounds that accumulate in the ocean (Jiao et al. 2010). Solar radiation has also been found to mineralize DOM directly to inorganic carbon species (CO2or CO), free radicals and reactive oxygen species (ROS; e.g. H2O2)(e.g.Cooperet al. 1989;MillerandZepp1995). These combined photochemical reactions have been demonstrated to have a complex impact on the microbial community activity resulting in enhanced, negative, mixed or no effect (see Mopper, Kieber and Stubbins 2015 for overview). Some studies hypothesize that these variable responses to sunlight exposure are linked with the DOM origin with recently produced autochthonous DOM getting less and allochthonous DOM more bioavailable to bacteria upon irradiation, while others have found the contrary (e.g. Benner and Ziegler 2000; Sulzberger and Durisch-Kaiser 2009). While numerous studies have addressed the link between DOM photochemistry and bacterial growth, only a few (if any) investigated how changes in DOM sources and sunlight exposure influence the bacterial activity and diversity in coastal waters. Since the impact of photochemistry on DOM and bacterial activity depends on the chemical composition, changes in the exposure to sunlight and DOM source could impact the microbial community differently. In this study, we assessed the impact of photo-alteration on DOM derived from specific aquatic sources (seawater, eelgrass and river water) and the subsequent impact of this exposure on the bacterial carbon cycling and community structure. We hypothesized that the bacterial community response will vary depending on the DOM source and sunlight exposure. This hypothesis was tested using laboratory incubations where sunlight-altered DOM was added to surface seawater from the coastal upwelling area of the R´ ıa de Vigo (N. W. Iberian Peninsula) and changes in coloured dissolved organic matter (CDOM) optical properties and bacterial abundance, activity (production and respiration) and diversity were measured over a period of 4 days. MATERIALS AND METHODS Site description and sample treatments In order to determine the impact of sunlight on specific DOM sources and the subsequent effects of the addition of these materials on the bacterial community of the R´ ıa de Vigo, we collected DOM from 3 different sources: marine surface water, river water and DOM leached from leaves of the eelgrass Zostera marina. Surface seawater was collected in the coastal upwelling system of the R´ ıa de Vigo. This coastal embayment is influenced by wind-driven upwelling and downwelling episodes, with northerly winds resulting in upwelling, which prevailed during our sampling period (late spring). The seawater sample was collected on 30 May 2012 in the middle of the R´ ıa de Vigo, a suitable site for evaluating the influence on the whole embayment (Nogueira, P´ erez and R´ ıos 1997). The water was collected at 5 m depth using a 12-L acid-cleaned Niskin bottle and combined into a 50 L acid-washed and aged polyethylene container. After collection, the sample water was kept in the dark until processed at the base laboratory. Water temperature was measured immediately after collection, while aliquots for the analysis of salinity were collected and measured in the laboratory using an Autosal 8400A. Material for chlorophyll a(Chl a) determination was collected by filtering seawater (200 mL) through a GF/F filter and analysed after 90% acetone extraction with a Turner Designs 10000R fluorometer. Seawater filtrations were started within 1 h after collection. One part was filtered through pre-combusted (450◦C for 4 h) GF/C filters to establish a microbial culture to be used in all experiments. This was kept in the dark at 15◦C until use. These were the same conditions used for the incubation study and ensured that the added microbial community was adapted to these conditions. The changes measured during our experiments in the microbial community was therefore only due to the changes in the organic matter sources and/or sunlight exposure. The other part of the seawater was gravity filtered through a dual-stage (0.8/0.2 µm) filter cartridge (Pall-Acropak Supor membrane), which had been pre-washed with Milli-Q (>10 L). The seawater was thereafter used both as the control treatment and to dilute the DOM obtained from the river water and eelgrass leaves. In total 20 L of river water was collected in the main tributary to the R´ ıa de Vigo, River Oitab´ en-Verdugo. This river is not significantly affected by industrial or sewage waste, has a drainage area of 350 km2and receives a rainfall of 2500 mm per year, which results in an average flow of 15 m3s–1 (Gago et al. 2005). The water samples were taken in the upstream limit of the freshwater–seawater interface. After collection the by guest on April 12, 2016http://femsec.oxfordjournals.org/Downloaded from Lønborg et al.3 sample water was kept in the dark until filtration commenced at the base laboratory, about 2 h later. The river water was gravity filtered through a dual-stage (0.8/0.2 µm) filter cartridge (Pall-Acropak supor Membrane) and thereafter concentrated in a proportion of 1:10 using a metal-free tangential flow ultrafiltration system, provided with a GH2540F30 membrane (GE Power & Water – Water & Process Technologies). Only the DOM material with a molecular weight >1 kDa (representing 72% of the total pool) was used in this study. Fresh leaves of the eelgrass Zostera marina were collected during low tide in the San Simon bay, in the innermost part of the R´ ıa de Vigo. Within an hour of collection, the leaves were brought back to the base laboratory and rinsed thoroughly with the 0.2 µm-filtered seawater collected in the middle of the R´ ıa de Vigo. The eelgrass-derived DOM was thereafter extracted by adding approx. 5 g of wet leaves to a glass bottle containing 1 L of the 0.2 µm-filtered seawater. After extraction (48 h in the dark) the water was filtered first through a pre-combusted (450◦C for 4 h) GF/D filter and then through a dual-stage (0.8/0.2 µm) filter cartridge (Pall-Acropak Supor membrane). Experimental design The DOM derived from river water and eelgrass leaves were added to 10 L polyethylene carboys, containing 7 L of seawater, in order to reach a dissolved organic carbon (DOC) increase of about 40 µmol L–1, while the control seawater did not receive any addition. The samples were thereafter divided into 2 experimental treatments: dark (termed ‘dark’) and full sunlight treatment (termed ‘UV’). The dark treatments were established by placing the sample water into UVR-transparent low density polyethylene incubators (3.5 L in each) that were covered with aluminium foil and dark plastic bags, while the UV samples were distributed into UVR-transparent low density polyethylene incubators that were left uncovered. No headspace was left in either dark or UV treatments. The samples were thereafter placed in a recirculation water bath (water depth: 25 cm) in the terrace of the laboratory and exposed to 100% natural sunlight for 4 days encompassing the natural light–dark cycle. During the sunlight exposure the incubators were almost completely covered in water (around 85%). The temperature was not specifically controlled during the exposure period but a constant flow (approx. 3 L min−1) of cold tap water cooled the samples. This ensured that the temperature was kept constantly low, suggesting that changing temperature did not impact our UV exposure results. Incident irradiance during the UV treatments was taken from the meteorological observatory on the terrace of the base laboratory, showing that over the 4 days the UV samples were exposed to 20 MJ m–2 d–1 of total solar radiation. Before and after sunlight exposure, subsamples were collected for the analysis of DOC, dissolved inorganic nitrogen (DIN: NH4+,NO 2–,NO 3–), dissolved inorganic phosphorus (DIP: HPO42–) and DOM optical properties (absorption and induced fluorescence). After the 4 days of sunlight exposure the water samples were within 15 min combined into different (dark and UV) carboys and the seawater microbial community was added in a ratio of 1 part of microbial culture, established from seawater collected in the R´ ıa de Vigo on 30 May 2012, to 9 parts of exposed water. The water was thereafter transferred into fifteen 500 mL glass bottles per treatment (90 bottles in total) and incubated in the dark at a constant temperature of 15◦C, with 3 replicate bottles being used for sub-sampling at incubation times 0, 1, 2, 3 and 4 days. The processing of the samples at initial time point started approximately 1 h after completion of the sunlight exposure. Unfiltered water from these bottles was used to follow changes in bacterial abundance (BA), diversity (using automated rRNA intergenic spacer analysis (ARISA)), bacterial production (BP) and bacterial respiration (BR). Samples for the analysis of DOC, DIN, DIP and CDOM absorption and fluorescence were collected after filtration through 0.2 µmfilters (Pall Supor membrane Disc), which were placed in an acid-cleaned all-glass filtration system under low N2flow pressure. All glassware used was acid washed in 10% HCl and rinsed with Milli-Q and sample water prior to use. Sample analysis Samples for BA were fixed with 25% glutaraldehyde (0.5% final concentration) for 30 min at 4◦C, flash-frozen in liquid nitrogen and stored at –80◦C until analysed. Thawed samples were diluted up to 10-fold in autoclaved 0.2 µmfiltered TE buffer (10:1 Tris–EDTA, pH 8.0) and stained with the nucleic acid-specific dye SYBR Green I (Invitrogen/Molecular Probes) for 15 min in the dark and analysed using a FACSCalibur flow cytometer. Bacterial biomass (BB) was calculated from BA, using a carbon conversion factor of 12 fg C cell–1, which is representative for coastal bacterial assemblages (Fukuda et al. 1998). BP was measured by [3H]thymidine incorporation (Fuhrman and Azam 1980). Three replicate 9.9-mL samples and 2 trichloroacetic acid killed samples were added to an aqueous stock solution of [3H-methyl]thymidine (20 nmol final concentration). The samples were incubated in the dark at 15◦C for 1 h, 10 mL of ice-cold trichloroacetate (TCA) was thereafter added and samples were filtered onto 0.2 µmpolycarbonatefilters(presoaked in non-labelled thymidine) and washed with 95% ethanol and autoclaved Milli-Q water. The filters were thereafter dried at room temperature (24 h) and mixed with 10 mL of scintillation fluid (Sigma-Fluor). The radioactivity incorporated into cells was counted using a Wallac scintillation counter. The disintegrations per minute (DPM) of the TCA-killed blank were subtracted from the DPMs of the samples. Thymidine incorporated into bacterial biomass was converted to carbon production using the theoretical conversion factors, 2 ×1018 cells mol–1 thymidine (Fuhrman and Azam 1980) and using the same cell-to-carbon conversion factor as for BA. The bacterial respiration (BR) was estimated using the reduction of 2-(4-iodophenyl)-3-(4-nitro-phenyl)-5-phenyl tetrazolium chloride (INT) following Mart´ ınez-Garc´ ıa et al.(2009). In brief, the activity was measured using 1 h incubations of 3 replicate samples (10 mL) and 1 formaldehyde-killed control. The incubations were terminated by adding formaldehyde and filtering onto 0.2 µmpolycarbonatefilters.Thefilterswerethereafter stored frozen (–20◦C) until further processing. The respiration rates derived from INT reduction (BR, in µmol O2L–1 h–1)were obtained by multiplying the in vivo INT reduction rate (in µmol INTF L–1 h–1) by an empirically derived conversion factor of 12.8. The instantaneous bacterial carbon demand (BCD) was calculated as the sum of C-converted BP and BR: BCD =BP +BR (1) The instantaneous bacterial growth efficiency (BGE) was calculated as BP divided by the sum of BP and BR: BGE =BP/(BP +BR)(2) The integrated BCD (BCDint) over the incubation period was calculated as the bioavailable DOC (BDOC): BCDint =BDOC (3) by guest on April 12, 2016http://femsec.oxfordjournals.org/Downloaded from 4FEMS Microbiology Ecology, 2016, Vol. 92, No. 5 The integrated BGE over the 4 days (BGEint)wascalculatedas the net growth in bacterial biomass between day 0 and maximum abundance (BG; BG =BBmax – BBmin) divided by BDOC: BGEint =BG/BDOC (4) The DOC samples were collected into pre-combusted (450◦C, 12 h) glass ampoules and preserved with 50 µL25%H 2PO4 to 10 mL sample. DOC was measured using a Shimadzu TOC analyser (Pt catalyst) and 3 to 5 replicate injections of 150 µL were performed per sample. Concentrations were determined by subtracting a Milli-Q blank and dividing by the slope of a daily standard curve of potassium hydrogen phthalate. Using the deep ocean reference provided by Prof. D. A. Hansell, University of Miami (Sargasso Sea deep water 2600 m) we obtained a concentration of 46.0 ±2.0 µmol L–1 (mean ±SD), with the DOC value provided by the reference laboratory being 44.0 ± 1.5 µmol L–1.Thedifferencebetweentheinitial(DOC 0)andminimum DOC (DOCmin) concentration over the 4 days’ incubation was here defined as the bioavailable DOC (BDOC). DIN and DIP samples were collected into 50 mL acid-washed polyethylene bottles and measured using standard segmented flow analysis (SFA) as described in Hansen and Koroleff (1999). Absorption spectra (250–600 nm) of CDOM were measured on a Beckman Coulter DU 800 spectrophotometer using Milli-Q water as a blank. Before analysis samples were warmed to room temperature and absorbance was measured with 1 nm increments using a 10 cm quartz cuvette. The absorption coefficient at any wavelength, aCDOM(λ)(m –1), was calculated as: aCDOM (λ)=2.303·[Abs (λ)−Abs (600 −750)] /0.1(5) Where Abs(λ) is the absorbance at wavelength λ;Abs(600–750) is the average absorbance between 600 and 750 nm, which corrects for the residual scattering by fine size particle fractions, micro-air bubbles or colloidal material present in the sample, or refractive index differences between the sample and the reference (m–1); the factor 2.303 converts from decadic to natural logarithms; and the denominator (0.1) is the cell path-length in metres. The estimated detection limit of this spectrophotometer is 0.001 absorbance units or 0.02 m−1. As the treatments had different DOC concentrations, we standardize the CDOM absorption by calculating the C-specific absorption coefficient CDOM absorption spectra (a∗CDOM(λ)). The C-specific absorption coefficient at 254 nm, known as SUVA (Weishaar et al. 2003), was calculated by dividing the decadic aCDOM(254) by the DOC concentration and expressed in litres per milligram per metre. The CDOM spectral slope (S) of the C-specific absorption spectra was modelled as: a∗CDOM(λ)=a∗CDOM (375)·e−S·(λ−375) (6) where a∗CDOM(λ)isthecarbon-specificabsorptioncoefficientat wavelength λ,a∗CDOM(375) is the C-specific absorption coefficient at the reference wavelength of 375 nm, and Sis the spectral slope coefficient of the absorption curve calculated over the range 250– 600 nm. The CDOM spectral slopes were also calculated over two narrow wavelength ranges, S(275–295) and S(350–400), using linear regressions of the natural log-transformed aCDOM(λ)spectra. These slopes were used to calculate the CDOM spectral slope (S(275–295)/S(350–400)) ratio (SR) (Helms et al. 2008).CDOM fluorescence emission excitation matrices (EEMs) and single-point measurements were performed on a Perkin Elmer LS 55 luminescence spectrophotometer equipped with a xenon discharge lamp, equivalent to 20 kW for 8 µsduration.Measurementswere performed at a constant temperature of 20◦Cina1cmquartz fluorescence cell. The EEMs were generated by combining 22 fluorescence emission spectra from 300 to 560 nm at excitation wavelengths ranging from 240 to 450 nm at 10 nm intervals. Point excitation/emission (Ex/Em) measurements were performed at wavelengths characteristic of peak-A (general humiclike substances, average Ex/Em, 250 nm/435 nm), peak-C (terrestrial humic-like substances, at Ex/Em wavelengths of 340 nm/440 nm), peak-M (marine humic-like substances, average Ex/Em, 320 nm/410 nm) and peak-T (protein-like substances, average Ex/Em, 280 nm/320 nm) (Coble 1996; Lønborg et al. 2010b). The fluorescence measurements were normalized to the Raman area using daily spectra of Milli-Q water (Lawaetz and Stedmon, 2009). As the treatments had different DOC concentrations, we standardize the FDOM values by calculating the carbon-specific FDOM as the FDOM signal divided by the DOC concentration. The limit of detection, calculated as 3 ×the standard deviation of the blank, was 0.03 QSU for peak-A, 0.05 QSU for peak-C and 0.02 QSU for peak-M and peak-T. Automated rRNA intergenic spacer analysis Automated rRNA intergenic spacer analysis (ARISA) was conducted with DNA from the initial community and at the end of each experiment, for characterizing the initial and final microbial community for each treatment. For the initial community, samples were prefiltered through a 1.2 µm pore-size filter (Kleenpak Capsule HDCII), and subsequently 2 L was filtered on a 0.2 µmpore-sizepolycarbonatefilter (Nuclepore Whatmann, 47 mm filter diameter). At the end of the experiments 0.5–1 L from each treatment was filtered on a 0.2 µm pore-size polycarbonate filter (Nuclepore Whatmann, 47 mm filter diameter). Filters were then stored at –80◦C until DNA extraction. Microbial community DNA was extracted using Ultra Clean Soil DNA isolation kit (MoBio Laboratories, Inc.) and quantified in a Nanodrop. Bacterial ARISA was performed using ITSF/ITSReub primer set (Thermo Scientific) previously described by Cardinale et al. (2004). The PCR reaction (25 µL) contained final concentrations of 1x PCR buffer (Genecraft), 2.5 mmol L–1 MgCl2(Genecraft), 250 µmol L–1 of each dNTP (Genecraft), 250 nmol L–1 of universal primer ITSF (5′-GTCGTAACAAGGTAGCCGTA-3′)andeubacterialITSReub (5′-GCCAAGGCATCCACC-3′) (Cardinale et al. 2004), the former being labelled at the 5′end with the fluorescein amidite dye (6FAM), 40 ng µL–1 bovine serum albumin, 3.5 U of BioThermD-TM Taq DNA Polymerase (GeneCraft) and approx. 0.13 ng µL–1 of template DNA. The reaction mixture was held at 94◦C for 2 min followed by 32 cycles of amplification at 94◦Cfor15s,55 ◦C for 30 s and 72◦C for 3 min, with a final extension of 72◦C for 10 min. PCR samples were conducted by duplicates for each DNA extraction (this compensates for any anomalously running fragments both in the samples and in the standards). Amplification products were migrated by capillary electrophoresis on a 50 cm capillary ABI Prism 3730XL DNA analyser (Applied Biosystems) at Genoscreen (www.genoscreen.fr/). The standardized migration cocktail contained 0.5 µL of amplification product, 0.25 µL of internal size standard LIZ 1200 (20–1200 pb, Applied Biosystems) and 8.75 µL of deionized Hi-Di formamide (Applied Biosystems). The mixture was denatured for 5 min at 95◦Cand kept on ice before being further processed by the sequencer. by guest on April 12, 2016http://femsec.oxfordjournals.org/Downloaded from Lønborg et al.5 Capillary electrophoresis parameters were as follows: 10 kV (run voltage), 1.6 kV (injection voltage), 22 s (injection time) and 63◦C (oven temperature). Resulting electropherograms were analysed using DAx software (Data Acquisition and Analysis Software, Van Mierlo Software). Internal size standards were built by using a second-order least-squares method and local Southern method. Profiles were double checked manually for perfect internal size standard fit and stable baselines. Baselines were then extracted, and subsequently, peak sizes, heights and absolute areas were determined. The same process was done for the PCR negative sample. From the negative sample, the 95th percentile was calculated for the height measurement, and used as a threshold. Sample with peak heights below the 95th percentile were discarded (the 95th percentile of each duplicated PCR negative presented values of 9 and 8.7 relative fluorescence intensity (RFI) respectively). Profile peaks were binned and rearranged by operational taxonomic units (OTUs) by using R automatic binning and interactive binning scripts (Ramette 2009). Binning was carried out independently of the sample (peaks from all samples together). Only peaks in the range 200 to 1200 bp and with values above 0.09% of total RFI were taken into account. Peaks from duplicates were manually checked using binned-OTU tables, to avoid erroneous OTU divisions due to rearrangement of all samples together. Statistical analyses In this paper t-tests were performed to test the significance of the differences observed in bacterial abundance and activity between dark and UV-irradiation incubations (Sokal and Rohlf 1995). The confidence level was set at 95%, with all statistical analyses conducted in Statistica 6.0. Differences in ARISA fingerpints between initial and final community as well as among treatments were analysed by comparing Bray–Curtis similarities using the package Primer 6 and Permanova+.AlogX+1transformation of ARISA relative abundance was used to reduce the dominant contribution by a small number of highly abundant species to the Bray–Curtis analysis (despite results being very similar with and without transformations). Similarity patterns among samples were then examined using a hierarchical cluster analysis. Dendograms were generated using the group average method, and a similarity profile (SIMPROF) test (999 permutations) was applied to test for significantly similar clusters. RESULTS Effects of sunlight on inorganic nutrients and dissolved organic matter The recorded surface seawater salinity (35.4), temperature (17.4◦C), Chl a(4.3 mg m–3) and nutrient levels (HPO42–:0.04± 0.03 µmol L–1;NO 3–+NO2–:0.14±0.02 µmol L–1;NH 4+:1.21±0.17 µmol L–1) were similar to typical late spring conditions in the R´ ıa de Vigo (Nogueira, P´ erez and R´ ıos 1997). In all incubations, exposure to sunlight resulted in increased NH4+concentrations relative to the dark control, while increased HPO42– concentrations were found in the seawater and eelgrass experiments (Table 1). The DOC concentrations before irradiation were 99 ±1µmol L–1 in the seawater, 131 ±4µmol L–1 in the eelgrass and 154 ±2µmol L–1 in the river samples (Table 1). The light treatment caused a 5±2% decrease in DOC concentrations in the seawater experiments, while 12 ±8% and 9 ±3% lower concentrations were found in the eelgrass and river samples, respectively (Table 1). The UV exposure resulted in a lower concentration of bioavailable DOC (BDOC) over the 4 days’ incubation compared with the dark treatment in the seawater (8 ±1vs17±4µmol L–1)and river experiments (19 ±1vs26±1µmol L–1), while increased BDOC levels were found in the eelgrass experiment (38 ±2vs20 ±5µmol L–1)(Table1). The initial C-normalized CDOM absorption spectra measured from 250 to 600 nm showed that the river DOM was more coloured than the seawater and eelgrass treatments (Fig. 1a–c). The maximum photoproduction of CDOM was found at wavelengths below 260 nm in both the seawater and eelgrass experiments, while the largest loss of carbon-specific absorption for all analysed wavelengths and treatments was found around 310 nm in the river experiments (Fig. 1d). The C-normalized spectral slope determined by fitting the absorption spectra to a single exponential decay function (Eq. 6) showed significantly higher slopes in the sunlight-exposed eelgrass and river experiments (t-test, P<0.05). The C-normalized absorption at 254 nm (SUVA) was here taken as a measure of the aromaticity and abundance of carbon double bonds in the DOM (Weishaar et al. 2003) present in the dark and sunlight-exposed samples. Sunlight exposure did not significantly impact the SUVA in any of the experiments. The spectral slope ratio (SR), which is inversely correlated to the molecular weight, increased in all UV light-exposed samples (Table 1), suggesting a photochemically induced decrease in the average molecular weight (Helms et al. 2008). There were clear differences in the C-specific FDOM signals between treatments both before and after 4 days’ sunlight exposure especially in the river experiment, suggesting differences in the CDOM chemical composition (Fig. 2and Table 1). The initial seawater experiments had the generally lowest FDOM signals, eelgrass had intermediate and river samples the highest values (Fig. 2and Table 1). For the case of the protein-like fluorescence, the eelgrass showed the highest C-specificfluorescence, while the river samples had a generally higher humic contribution (Fig. 2and Table 1). The C-normalized fluorescence signals of the humicand protein-like substances decreased in all sunlight-exposed samples showing the largest impact on the humic substances in the river experiment (Fig. 2and Table 1). Bacterial response to DOM of different origin and the effect of solar radiation Initial bacterial abundances (BA) were the same (∼2×105cells mL–1)inalltreatments(Fig.3). Bacterial abundances increased in all experiments (Fig. 3) following the consumption of DOC (Table 1), reaching maximum abundances of 23 ×105(dark) and 27 ×105cells mL–1 (UV) in the seawater, 24 ×105(dark) and 35 × 105cells mL–1 (UV) in the eelgrass, and 17 ×105(UV) and 19 ×105 cells mL–1 (dark) in the river incubations (Fig. 3). These increases in BA corresponded to an average bacterial biomass growth (BG) of 2.07 ±0.05 (dark) and 2.44 ±0.03 µmol L–1 (UV) in the seawater, 2.22 ±0.04 (dark) and 3.27 ±0.04 µmol L–1 (UV) in the eelgrass and 1.45 ±0.04 (dark) and 1.72 ±0.04 µmol L–l (UV) in the river experiments (Table 2). The BA was not significantly different in the dark and UV treatments in the 3 experiments. The initial BP varied between 0.14 ±0.09 (river–UV) and 1.47 ±0.02 µmol L–1 (seawater–UV), with higher initial levels in UV-exposed seawater and eelgrass samples, whereas higher levels were found in the dark samples in the river incubations (Fig. 3and Table 2). The integrated BP was larger in the UV seawater (2.47 ±0.55 (dark) vs 2.89 ±0.31 µmol L–1 (UV)) and eelgrass incubations (1.61 ± 0.24 (dark) vs 3.25 ±0.55 µmol L–1 (UV)), while lower levels were found in the river experiments (2.18 ±0.42 (dark) vs 1.29 ±0.25 by guest on April 12, 2016http://femsec.oxfordjournals.org/Downloaded from 6FEMS Microbiology Ecology, 2016, Vol. 92, No. 5 Table 1. Concentrations of phosphate (HPO42–), nitrate (NO3–), nitrite (NO2–) and ammonium (NH4+), the C-specific CDOM absorption coefficient at 254 nm (SUVA) and spectral slope in the 250– 600 nm wavelength range (S), ratio of CDOM spectral slopes (S(275–295)/S(350–400) ratio (SR)), C-specific general (peak-A∗), terrestrial (peak-C∗) and marine humic-like substances (peak-M∗)and protein-like substances (peak-T∗), initial concentrations of dissolved organic carbon (DOC) before and after 4 days in darkness (dark) or exposed to natural sunlight (UV). The DOC concentration at incubation day 0 (DOC0)andtheminimumconcentration(DOC Min)arealsoshowntogetherwiththebioavailablefraction(BDOC)beforeandafterUV-lightexposure.Valuesaremeans±standard deviation; —, not measured. Seawater Eelgrass River Before UV Dark UV Before UV Dark UV Before UV Dark UV HPO42– (µmol L–1)0.06 ±0.02 0.18 ±0.08 0.37 ±0.11 0.21 ±0.08 0.21 ±0.12 0.30 ±0.08 0.25 ±0.01 0.25 ±0.05 0.24 ±0.20 NO3–(µmol L–1)0.09 ±0.01 0.08 ±0.02 0.30 ±0.01 0.08 ±0.02 0.10 ±0.02 0.09 ±0.02 0.07 ±0.08 0.09 ±0.01 0.13 ±0.01 NO2–(µmol L–1)0.06 ±0.01 0.13 ±0.10 0.22 ±0.02 0.08 ±0.01 0.14 ±0.01 0.11 ±0.01 0.09 ±0.02 0.11 ±0.01 0.14 ±0.01 NH4+(µmol L–1)1.60 ±0.17 1.28 ±0.26 2.17 ±0.33 0.62 ±0.06 0.46 ±0.30 2.61 ±0.76 1.63 ±0.34 1.12 ±0.35 2.31 ±0.31 SUVA (L mg−1m−1)0.95 ±0.04 0.96 ±0.04 1.03 ±0.03 1.16 ±0.03 1.17 ±0.04 1.38 ±0.04 1.99 ±0.06 1.99 ±0.06 1.92 ±0.06 S(nm−1)0.0171 ±0.0001 0.0171 ±0.0001 0.0167 ±0.0001 0.0092 ±0.0001 0.0092 ±0.0001 0.0110 ±0.0001 0.0125 ±0.0001 0.0124 ±0.0001 0.0156 ±0.0001 SR1.59 ±0.04 1.59 ±0.05 1.98 ±0.05 1.38 ±0.04 1.38 ±0.06 1.74 ±0.12 0.84 ±0.01 0.84 ±0.02 1.34 ±0.03 Peak-A∗(×108Lmg −1m−1)4.02±0.09 4.97 ±0.15 4.13 ±0.09 3.66 ±0.07 3.77 ±0.08 3.81 ±0.06 10.90 ±0.03 12.00 ±0.07 6.38 ±0.09 Peak-C∗(×108Lmg −1m−1)1.66±0.05 1.81 ±0.01 1.50 ±0.06 2.06 ±0.07 1.92 ±0.03 1.68 ±0.04 5.50 ±0.08 5.75 ±0.03 2.50 ±0.04 Peak-M∗(×108Lmg −1m−1)1.75±0.11 1.82 ±0.03 1.52 ±0.03 1.93 ±0.01 1.62 ±0.03 1.50 ±0.06 4.33 ±0.08 4.48 ±0.02 2.39 ±0.04 Peak-T∗(×108Lmg −1m−1)2.36±0.24 3.17 ±0.10 2.66 ±0.08 3.49 ±0.03 3.16 ±0.06 3.44 ±0.18 1.78 ±0.01 2.78 ±0.12 2.19 ±0.05 DOC (µmol L–1)99±198±294±2131 ±4130 ±1115 ±1154 ±2149 ±1140 ±1 DOC0(µmol L–1)—96±193±1 — 117 ±2111 ±1 — 140 ±1139 ±1 DOCMin (µmol L–1)—80±185±1—83±373±2 — 115 ±1121 ±1 BDOC (µmol L–1)—17±48±1—20±538±2—26±119±1 by guest on April 12, 2016http://femsec.oxfordjournals.org/Downloaded from Lønborg et al.7 Figure 1. The carbon-specific absorption spectra of coloured dissolved organic matter (CDOM) after 4 days in the dark (dark) or exposed to natural sunlight (UV) in the (a)seawater,(b)eelgrassand(c) river experiments. The difference (dark minus sunlight samples) in the carbon-specific absorption spectra of CDOM in the seawater, eelgrass and river experiments are shown in (d). Note that in the case of the eelgrass and river samples, the blank subtracted was the seawater used to dilute the extracts. µmol L–1 (UV)) (Table 2). The BP was significantly higher in the UV treatment in the eelgrass and in the dark for the river experiments (t-test, P<0.05, n =4), while no significant difference was found in the seawater experiment. The instantaneous specific growth rate (µ), calculated as the ratio between BP and BB (BP/BB), showed that the UV pre-exposure had contrasting effects on the initial µ, with an increase in the seawater (3.90 ± 0.01 (dark) vs 5.41 ±0.35 d–1 (UV)) and eelgrass experiments (1.96 ±0.70 (dark) vs 3.46 ±1.26 d–1 (UV)) and a decrease in the river incubations (3.50 ±1.40 (dark) vs 0.65 ±0.47 d–1 (UV)) (Table 2). The µdecreased over the incubation period in all experiments reaching levels between 0.40 ±0.09 (eelgrass–dark) and 0.88 ± 0.16 d–1 (river–dark) at day 4 (Table 2). The initial BR was higher in the dark treatments in all incubations (Fig. 3and Table 2). In the seawater experiments, the BR increased after day 0 reaching maximum levels of 6.52 ±0.41 µmol L–1 d–1 in the dark and 3.49 ±0.15 µmol L–1 d–1 in the UV experiments (Fig. 3and Table 2). In the eelgrass UV incubations a dramatic increase was seen after the initial day, reaching a maximum value of 19.50 ±2.25 µmol L–1 d–1, while more stable levels were found in the dark incubations with a maximum of 5.39 ±0.18 µmol L–1 d–1 (Fig. 3and Table 2). In the river dark treatment the BR remained at a constant higher level (max. 8.95 ± 0.50 µmol L–1 d–1)thantheUVpre-treatedsamples(max.7.86 ±0.61 µmol L–1 d–1) until incubation day 3, whereafter equal BR were found in both experiments (Fig. 3and Table 2). The BR and integrated BR were significantly higher in the dark treatment in the seawater and river experiments, and significantly higher in the UV treated samples in the eelgrass experiment (t-test, P< 0.05, n=4). In all experiments the initial cell-specific BR was lower in the UV pre-treated samples (Fig. 3j–l and Table 2). After the initial day in the seawater and river incubations the cellspecific BR was generally higher in the dark compared with the UV treatments, while a higher activity was found in the UV exposed samples in the eelgrass experiments (Fig. 3j–l and Table 2). The initial BCD was higher in the dark compared with the UV treatments in all experiments (Fig. 4a–c and Table 2). Thereafter a continued higher BCD was found in the dark samples in the seawater and river experiments, while in the eelgrass experiment higher levels were found in the UV pre-treated samples (Fig. 4a–c and Table 2). Overall the average BCD was significantly higher in the dark samples in the seawater and river experiments, while a significantly higher BCD was found in the UV pre-exposed samples in the eelgrass experiments (t-test, P< 0.04, n=4). The UV pre-treatment also decreased the integrated BCD in the seawater (23.7 ±1.7 (dark) vs 11.7 ±1.1 µmol L–1 d–1 (UV)) and river (32.3 ±3.4 (dark) vs 23.5 ±1.9 µmol L–1 d–1 (UV)) experiments, while an increase was found in the eelgrass (16.4 ±1.0 (dark) vs 62.8 ±10.7 µmol L–1 d–1 (UV)) experiment (Table 2). The instantaneous BGE varied initially between 3 ±1 (river– UV) and 58 ±7% (seawater–UV) decreasing thereafter to values between 4 ±1(eelgrass–dark)and14±1% (seawater–UV) at incubation day 4 (Fig. 4d–f and Table 2). The initial BGE was significantly higher (t-test, P<0.01) in the UV pre-exposed seawater experiments, while equal levels for both treatments were found in the eelgrass and river incubations (Fig. 4d–f and Table 2). The integrated BGE (BGEint)showedhighestvaluesintheUV– seawater samples (25 ±6%) and lowest in the dark–river samples (5 ±1%), with no significant differences between the dark and UV pre-treatments in any of the experiments (Table 2). The instantaneous BGE averaged over the 4 days’ incubation showed similar levels as the BGEint (Table 2). The bacterial community composition at time zero and at the end of each experiment was compared based on ARISA by guest on April 12, 2016http://femsec.oxfordjournals.org/Downloaded from 8FEMS Microbiology Ecology, 2016, Vol. 92, No. 5 Figure 2. C-specificfluorescence excitation emission matrix of (a) seawater dark, (b)seawaterUV,(c) eelgrass dark, (d) eelgrass UV, (e) river dark, (f) river UV, (g)seawater dark minus UV, (h) eelgrass dark minus UV and (i) river dark minus UV samples. Fluorescence units are 106Lmg −1m−1. by guest on April 12, 2016http://femsec.oxfordjournals.org/Downloaded from Lønborg et al.9 Figure 3. Time course changes in bacterial abundance (BA; a,b,c), production (BP; d,e,f), respiration (BR; g,h,i) and cell-specific respiration (Cell resp; j,k,l)during the 4-day incubations in the different experiments (seawater, eelgrass and river). Error bars represent standard deviations; where not visible error bars are within the symbol. by guest on April 12, 2016http://femsec.oxfordjournals.org/Downloaded from