Phytoplankton Productivity in an Arctic Fjord (West Greenland): Estimating Electron Requirements for Carbon Fixation and Oxygen Production

Accurate quantification of pelagic primary production is essential for quantifying the marine carbon turnover and the energy supply to the food web. Knowing the electron requirement (Κ) for carbon (C) fixation (Κ C) and oxygen (O2) production (Κ O2), variable fluorescence has the potential to quantify primary production in microalgae, and hereby increasing spatial and temporal resolution of measurements compared to traditional methods. Here we quantify Κ C and Κ O2 through measures of Pulse Amplitude Modulated (PAM) fluorometry, C fixation and O2 production in an Arctic fjord (Godthåbsfjorden, W Greenland). Through short- (2h) and long-term (24h) experiments, rates of electron transfer (ETRPSII), C fixation and/or O2 production were quantified and compared. Absolute rates of ETR were derived by accounting for Photosystem II light absorption and spectral light composition. Two-hour incubations revealed a linear relationship between ETRPSII and gross 14C fixation (R2 = 0.81) during light-limited photosynthesis, giving a Κ C of 7.6 ± 0.6 (mean ± S.E.) mol é (mol C)−1. Diel net rates also demonstrated a linear relationship between ETRPSII and C fixation giving a Κ C of 11.2 ± 1.3 mol é (mol C)−1 (R2 = 0.86). For net O2 production the electron requirement was lower than for net C fixation giving 6.5 ± 0.9 mol é (mol O2)−1 (R2 = 0.94). This, however, still is an electron requirement 1.6 times higher than the theoretical minimum for O2 production [i.e. 4 mol é (mol O2)−1]. The discrepancy is explained by respiratory activity and non-photochemical electron requirements and the variability is discussed. In conclusion, the bio-optical method and derived electron requirement support conversion of ETR to units of C or O2, paving the road for improved spatial and temporal resolution of primary production estimates.


Introduction
Photosynthesis is the fundamental biological process that converts inorganic carbon into living biomass from solar radiant energy. Through photosynthesis, microalgae primary production fuels the marine food web and its efficiency and dynamics influence the energy supply to higher-trophic levels [1]. Understanding the conversion of the photon flux energy from solar radiation towards fixation of inorganic carbon (CO 2 ) and/or production of oxygen (O 2 ), forms the basis for quantifying the primary production.
Conventionally, O 2 production, 14 C and 13 C fixation, techniques have been used to quantify either gross (GPP) or net (NPP) primary production [2][3][4]. It is, however, still debated what the methods really measures and how to arrive at accurate gross or net primary production rates [5][6][7][8]. Traditionally, GPP refers to the fixation of inorganic 14 CO 2 without accounting for any carbon (C) losses to respiration, while NPP refers to the 14 CO 2 fixation after subtracting the respiratory CO 2 'lost' by phytoplankton over a diel cycle [2,9]. Conventionally, it is assumed that short time (1-2h) incubations yield estimates of GPP while NPP is obtained over 24h incubations [10]. However, Williams et al. [7] convincingly showed that 2h incubations can produce NPP estimates, a conclusion supported by Pei and Laws [5]. Productivity can also be measured from a net change in O 2 concentration over a diel cycle (24h). This way the measure includes the respiratory O 2 consumption of the heterotrophic community of the sample including phytoplankton itself and is defined as the Net Community Production (NCP) [10]. In many marine systems, including the Arctic, low phytoplankton biomass limits the application of 14 C and ΔO 2 techniques to longer incubation times, i.e. 24 hours. And as of today, marine primary production estimates are primarily based on discrete bottle measurements of GPP or NPP with a limited spatial and temporal resolution, with an unquantified degree of uncertainty and the risk of bottle effects [6,11].
Pulse Amplitude Modulated (PAM) fluorescence [12] or Fast Repetition Rate fluorometry (FRRf) [13,14] provide a non-invasive and fast assessment of the conversion of the photon flux to a rate of electron transfer (ETR) in Photosystem II (PSII). Such variable fluorometry methods can be applied in situ and represent an alternative measuring approach for photosynthetic activity in phytoplankton. Variable fluorescence can provide a high temporal (seconds) and spatial resolutions compared to traditional bottle incubations. Thus, if ETR can be converted to GPP or NPP based on an adequate understanding of the intermediate processes and on empirical evidence, variable fluorescence can be applied for primary production estimates in absolute terms [15]. Such knowledge enables the assessment of primary productivity with a high temporal resolution, and potentially enables the use of moorings and glider platforms for efficient and large-scale assessment of marine primary productivity.
Conversion of ETR to C fixation or O 2 production is, however, still challenging [15][16][17]. The relationship between ETR and C fixation/O 2 production has been compared in a range of studies on algal cultures and pelagic ecosystems and generally linear correlations are documented between ETR and gross C fixation and/or O 2 production [18][19][20][21]. Deviations are reported under extreme conditions as for instance very high or low light conditions [22,23], extreme temperature [22,24], or nutrient stress [25,26]. Discrepancies have been proposed to be caused by changes in O 2 consumption in the light, cyclic electron transport around PSII and I, Mehler-type reactions, and electron requirements for nutrient uptake and cellular maintenance. In some studies the interrelations between ETR and C fixation/O 2 production have also been shown to be species-specific [20,24,26].
Lately, focus has been increasingly directed towards deriving the electron requirement for photosynthesis [15,17,20,21]. Lawrenz et al. [15] compiled a large amount of ETR data obtained using FRRf instruments and compared them to available 14 C uptake rates across different regions. They arrived at a mean electron requirement for carbon fixation of 10.9 ± 6.9 mol é (mol C) −1 , overall ranging from 1.2 to 54.2 mol é (mol C) −1 . The large variability partly originates from the multiple experimental approaches included in the study and the varying accuracy in the assessment of the light absorption by PSII. Still only few studies have focused on deriving the electron requirement for carbon fixation and oxygen production applying PAM fluorescence [21], none including both short and long term incubations. Comparisons of PAM versus FRRf measurements have shown a close relationship between the two, but with FRRf overestimating primary production relative to PAM measurements [21]. Essential for the conversion of ETR to absolute rates of primary production is an accurate assessment of the PSII-specific light absorption and of the available spectral irradiance. Only few studies have sufficiently included this when PAM derived quantum yields are converted to absolute units of ETR [17,21].
In the present study, we investigated the relationship between photosynthetic electron transport rate, 14 C and 13 C fixation, and O 2 production of the natural phytoplankton community in the inner and outer part of an Arctic fjord. The aim was to quantify the electron requirement for gross and net carbon fixation and NCP in a natural low-biomass pelagic ecosystem. Through careful assessment of the PSII-specific light absorption and incubator spectral irradiance, absolute rates of ETR were derived and compared to measured rates of C fixation and O 2 production. Variability of the electron requirement and photosynthetic efficiency is discussed along with the potential for applying PAM fluorescence for assessing in situ productivity in marine systems.

Theory
In this section, we present how ETR and GPP can be calculated in absolute terms from PAM measurements when combined with knowledge of the absolute rate of photons absorbed by photosystem II.
Gross photosynthesis from variable fluorescence (P PSII ) can be quantified in absolute units of C (P PSII_C , mmol C L −1 s −1 ) or O 2 (P PSII_O2 , mmol O 2 L −1 s −1 ) from the knowledge of the quantum yield of charge separation in PSII (F PSII ), the spectrally-weighted specific absorption of photons in PSII ( a PSII, m −1 ), the integrated incident irradiance (E PAR ), and the electron requirement for C or O 2 (Κ X ), respectively (Eq 1) [27,28].
Where E PAR is the integrated Photosynthetic Available Radiation between 400 and 700 nm, and Κ X is the electron requirement for carbon fixation (Κ C ) or oxygen production (Κ O2 ), respectively, in units mol electrons (mol C fixed or O 2 produced) −1 . Note, that the spectral distribution of EPAR, i.e. E(λ), is here included in a PSII (see below, Fig 1). By normalizing a PSII to the Chlorophyll a (chl a) concentration [denoted a Ã PSII ] productivity is given in units per chl a, which is convenient for comparing rates across different environments and biomass abundances (Eq 2). photosynthetic research most often F refers to a quantum yield of a process [mol product (mol photons absorbed) −1 ], being the reciprocal of the quantum requirement [mol photons (mol product) −1 ]. Thus, we consider using F to be counterintuitive and instead prefer applying the symbol Κ to describe the electron requirement for photosynthetic carbon fixation or O 2 production.
The quantum yield of charge separation in PSII (F PSII ), also often named the quantum efficiency of photosynthesis, can be measured from variable fluorescence, e.g. using a Pulse-Amplitude-Modulated (PAM) fluorometer that measures the conversion efficiency of photons to electrons in PSII [12]. Measurements of the quantum efficiency rely on measuring the ratio between the minimum and maximum PSII fluorescence before and after a saturation pulse, which is why the quantum efficiency arrives on a biomass independent scale between zero and one; one being the theoretical maximum value given that all photons absorbed by PSII yields an electron transport. Typically the maximum quantum yield for marine algae is 0.5 to 0.8 and species dependent [24,29].
Secondly, the ratio of absorbed quanta in PSII (a Ã PSII : a Ã ) can be obtained from measuring the fluorescence excitation spectrum [34] and scale it to a Ã (λ), by applying the 'non-overshoot' procedure [27,35]. This procedure quantifies the spectral absorption in PSII, which includes the light absorption by light-harvesting pigments associated with PSII but excluding photoprotective carotenoids and light-harvesting pigments associate with PSI. Obtaining the phytoplankton fluorescence excitation spectrum is, however, cumbersome and requires a scanning spectro-fluorometer with is not available to most research laboratories. A simple alternative to determining the PSII-specific absorption is to multiply a Ã with the fraction of quanta absorbed in PSII to the total absorption (a Ã PSII : a Ã ) (Fig 1). This procedure is applied here. It requires the knowledge of the ratio a Ã PSII : a Ã , which is specific to phytoplankton pigment-classes, but can be found in a comprehensive investigation of 33 species of phytoplankton representing 12 pigment classes [36].
Finally, the PSII-specific absorption coefficient (a PSII Ã ) is calculated from Eq 4. a If excluding Κ X from Eq 2 the equation calculates the electron transfer rate through PSII in absolute units (Eq 5, ETR PSII , mol é (mg chl a) −1 time −1 ).
Here we aim to quantify Κ C and Κ O2 from measures of P Ã PSII_C and P Ã PSII_O2 . Thus reorganizing Eq 2 and 5, the electron requirement can be calculated (Eq 6): In the following, we estimate ETR Ã PSII in absolute units [mmol é (mg chl a) −1 h −1 ] and the rates of O 2 production and C fixation to derive the electron requirement for O 2 production and C fixation, respectively.

Study area and experimental setup
Water was sampled in the Arctic Godthåbsfjord in West Greenland, during August-September 2013. Samples were taken at two stations, GF3 and GF7 representing the outer and inner part of the fjord system, respectively. The two stations are part of the marine monitoring program MARINBASIS maintained by the Greenland Institute of Natural Resources and the University of Aarhus (www.nuuk-basic.dk). Water was sampled in the euphotic zone at 5 and 20m depth with a 5L Niskin water sampler, screened through a > 280 um mesh to remove large zooplankton and transported to the laboratory within 1 to 2 hours. In total~120L of water was sampled at each station. In situ profiles of conductivity, temperature, depth (CTD), chl a fluorescence, optical turbidity and downwelling spectral irradiance were obtained at each station using a free-falling Optical Profiler II (Satlantic, Halifax, Canada). Geographical position, sampling date, in situ water temperature, salinity and light availability are given in Table 1 In the laboratory water samples were incubated at light-limited (~40 μmol photons m −2 s −1 ) and light-saturated (~500 μmol photons m −2 s −1 ) conditions in two large water bath incubators (100 x 100 x 15cm) kept close to in situ temperature (6°C) by a thermostat controlled heater, both installed inside a cooling container (~2°C). The large surface area of the incubator ensured a homogeneous illumination by halogen light sources of incubated bottles for 13 C and 14 C fixation, O 2 production and ETR PSII . The specific scalar irradiance (E PAR ) of each incubated bottle was measured using a small 4π scalar irradiance sensor connected to a light meter (ULM-500, Walz). The spectral composition of the incubator light was measured using the surface unit from the Satlantic Optical Profiler II. The specific irradiance and the spectral composition of the incubator light were used for further calculations of the light absorption by phytoplankton, see below. Twenty-four hour incubations were performed under an 8:8:8 hour light:dark:light regime to mimic the natural light conditions with a day length of~16 hours (and an~8 hours night period), as incubations were started around noon. In addition, the approach avoided light exposure times of >8 hours (discussed below).

Bio-optics, chl a and light microscopy
Optical densities of total particulate matter (OD t , 300-800 nm) was measured from 1 liter of sea water filtered onto GF/F glass fiber filters (Whatman Inc., Florham Park, NJ, USA) in a spectrophotometer (Shimadzu UV-2401PC UV-Vis) equipped with an integrating sphere (ISR-240A) as described by Staehr and Markager [39]. Triplicate filters were used and each filter was measured three times in order to minimize noise. Values above 750 nm were subtracted. Total spectral absorption of suspended particles was obtained by the scattering correction method [40].
where a t is the total absorption coefficient of particles, S is the clearance area of the GF/F filter (m 2 ), V is the filtered volume (m 3 ) and [0.378 + 0.523 × OD filt (λ)] is the beta-factor correcting for a longer path length in the filter compared to suspension. The in vivo phytoplankton absorption spectrum [a ph (λ), m −1 ] was then determined according to [30]: Where a NAP is the absorption coefficient of non-algae particles, i.e. detritus and the nonpigmented parts of phytoplankton, after extraction in methanol. The chl a-specific absorption was then calculated from [31]: The spectrally weighted chl a-specific absorption ( a Ã , m 2 (mg chl a) −1 ), and the spectrally weighted PSII-specific absorption ( a Ã PSII ) were calculated from Eqs 3 and 4, respectively. The Chl a concentration (mg m −3 ) was measured from 300 mL of sea water filtered onto GF/F glass fiber filters (Whatman) extracted in 10 mL 96% ethanol during 24h (dark, 4°C). The concentration was determined in triplets using a pre-calibrated fluorometer (Turner Designs TD-700). Water samples were collected for phytoplankton cell count and species identification, and were fixed with LUGOL (neutral, 1% final solution). The samples were analyzed in a light microscope using a Palmer Maloney chamber (0.1 mL) and after filtering through a 0.45 um polycarbonate filter (50 mL).

PAM fluorescence
The quantum yield of charge separation in PSII (F PSII ) was measured using a PhytoPAM variable fluorometer (System I, Walz, Effeltrich, Germany, Schreiber et al. [12]), equipped with a sensitive Photomultiplier-Detector (PM-101P, Walz). Minimum (F o ) and steady state (F s ) fluorescence excitation was obtained using a weak and non-actinic modulated light supplied by a LED (light emitting diode, Array-Cone PHYTO-ML, Walz, Germany) during darkness or at the incubation irradiance, respectively. The maximum fluorescence (F m ) was obtained during a red saturating light pulse (0.8s >1800 μmol m −2 s −1 , Actinic LED-Array-Cone PHYTO-AL, Walz) ensuring that all PSII reaction centers were closed. The instrument excites fluorescence at four different wavelengths; however, in the present study we used data only from the red light excitation (665 nm), to exclude potential inter-sample differences in the light-harvesting pigments to chl a ratio. We use the nomenclature by van Kooten and Snel [41]. The maximum quantum yield of charge separation (F PSII_max ) was calculated according to Eq 10 [42] after subtraction of the blank fluorescence, measured from a 0.2 μm filtered water sample: Under actinic illumination, the operational quantum yield (F PSII ) was calculated from the steady-state fluorescence (F s ) and the maximum fluorescence after a saturation pulse (F m '), by replacing F o and F m with F s and F m ', respectively in Eq 10. When F PSII was quantified in samples incubated in the water bath a subsample was transferred to the PAM cuvette and measured within a few seconds. PAR inside the PAM cuvette was adjusted to match PAR of the incubator for each step of the light gradient after measuring the irradiance with a 4π scalar irradiance sensor and light meter (ULM-500, Walz). The spectral composition of the PAM cuvette incubation light was measured using the surface unit of the Satlantic Optical Profiler II.
ETR versus irradiance (P-E) curves were calculated from the instantaneous quantum yield in samples after 2 hours of incubation in 100 mL Winkler bottles, in a light gradient. The quantum yield was measured in the exact same bottles from which 14 C fixation was measured (see below). The P-E curves were fitted from Eq 11 [43], as no photoinhibition was observed. The maximum photosynthetic rate (P max ), the light utilization coefficient (α) and the light saturation index (E k ) was calculated; E k = P max /α. Curve fitting was carried out using ordinary leastsquares criterion in Origin 8.5 (OriginLab).

C fixation
Gross 14 C fixation rates were measured in 100 mL Winkler bottles in a light gradient incubated for 2 hours (exact same as used for PAM P-E curves). We added 200 μl of Na 2 14 CO 3 with an activity of 20 μCi mL −1 [2]. The initial DIC concentration was assumed to be 2 mM according to the robust regional relationship between DIC and the salinity [44]. Nine bottles were incubated at different light intensities and 2 bottles in the dark. The content of all bottles was filtered onto GF/F filters that were transferred to glass scintillation vials, thereafter 100 μl 1 M HCl was added, and the filters were fumed for 8 hours. After addition of scintillation fluid (UltimaGold+) the samples were measured on a PerkinElmer scintillation counter. The dissolved fraction of fixed 14 C was measured by collecting two duplicate 5 mL samples of filtrate from each Winkler bottle following the protocol by Moran et al. (2001). The dissolved 14 C fixation was finally added to the particulate pool. The dark bottle activity was subtracted from the light bottles.

C fixation
Diel rates of carbon fixation was measured from the 13 C incorporation over a 24h time period according to Yun et al. [45], and normalized to the chl a concentration (P Ã C ). Briefly, 13 C bicarbonate was added to the sea water before incubation to a concentration of 200 μM in triplicate 500 mL square Nalgene polycarbonate bottles and incubated as described above. The initial DIC concentration was assumed to be 2 mM (see above). After the incubation the algae were filtered onto pre-combusted GF/F filters (Whatman). Filters were dried at 50°C for 48 hours before analysed for the content of 13 C on a Thermo Elemental Analyser Flash EA 1112HT in line with a Thermo Delta Plus V isotope ratio mass spectrometer.

O 2 production
Diel rates of the net community production (P Ã O2 , equal to NCP) were measured from the change of O 2 concentration over a 24h time period under an 8:8:8 hour light:dark:light regime in the incubator. Incubations were done in five 100 mL glass Winkler bottles for each treatment (high and low light). For each incubation, five Winkler bottles were preserved with 1 mL 7 M ZnCl 2 at the start and the incubated bottles were preserved after 24 hours incubation. The preserved bottles were stored in a water bath at ca. 5°C until analysis which was done within 24 hours of ZnCl 2 addition.
Oxygen concentration was determined using a Clark type O 2 microsensor [46] with a micro flow cell mounted on the tip. Water was drawn through the flow cell by gravity by keeping the surface of the source water 45 cm above the outlet. Once the flow path was filled with sample water the difference in pressure between inlet and outlet maintained a steady flow of 4 mL/ min. The sample temperature was adjusted before entering the flow cell by passing through a 75cm coil of 1/16" steel tubing, that was positioned with the sensor and flow cell in a temperature controlled water bath (5.0°C ± 0.1°C). The O 2 sensor was calibrated before and after the analysis of each sample. The calibration water was prepared as follows: Two liter of tap water was adjusted to within 50 μM of the expected sample O 2 concentration and transferred to a gas-tight plastic bag [47], where after all bubbles were removed and 20 mL 7 M ZnCl 2 was added to stop biological activity. The bag water was mixed and placed in the water bath to secure a stable temperature. The plastic bag was connected via a three way valve to the flow path upstream from the steel tubing coil. 'Tygon' tubing was used throughout this setup. During analysis the three way valve was turned every 120 seconds, alternating between drawing sample and calibration water though the flow cell on the O 2 sensor, and the signal was read as the average of the last five seconds of each interval. The microsensor was connected to a Unisense PA2000 picoammeter and the signal was recorded using Unisense Sensortrace Basic and a Unisense ADC-216USB A/D converter.

Algae composition and bio-optics
Light-microscopic analyses showed that the outer fjord station, GF3, was dominated by diatoms (Chaetoceros sp.) and smaller flagellates (Cryptophyceae sp. <15μm). The distributions between the two were 75:25% diatoms:flagellates in the surface (5m), and 15:85% in the lower euphotic zone (20m). The inner fjord station, GF7, was by far dominated by diatoms (>99%, Chaetoceros sp. and Thalassiosira sp.) with a minor presence of dinoflagellates (<1%, Gymnodinium sp. and Scripsiella sp.) at both 5 and 20 m. By classifying the phytoplankton in pigment groups based on their light-harvesting pigment signature, the fraction of absorbed quanta in PSII (a Ã PSII :a Ã ) was derived, using the work by Johnsen and Sakshaug [36]. For diatoms, a Ã PSII :a Ã equals 0.76 and 0.74 for low and high light growth, respectively, whereas cryptophytes have a a Ã PSII :a Ã of 0.66 and 0.45 under low and high light, respectively [36]. Using these ratios we quantified a Ã PSII :a Ã for the sampled phytoplankton population ( Table 2). The calculated a Ã PSII :a Ã ranged from 0.67 to 0.76 (dimensionless) and was used in the calculation of the PSII-specific light absorption (Eq 4).
Phytoplankton chl a-specific in vivo absorption coefficients (a Ã ) ranged from 0.0104 to 0.0150 m 2 (mg chl a) −1 demonstrating the natural variability of the light harvesting properties of the algae community, with chl a concentrations ranging from 1.2 to 2.4 mg m −3 ( Table 2). The in vivo absorption spectra, along with the spectral scalar irradiance of the incubator light source (E(λ)/E PAR ), and the derived spectrally weighted absorption spectra [ a Ã (λ)] are shown in Fig 2. Irradiance and weighted-absorption spectra are shown for both the incubator light source and the internal actinic light source of the PhytoPAM instrument ( Fig 2B+2C). The absorption weighted to the internal PAM light source was used to calculate P PSII , while the weighted absorption from the incubator light was used to correct production rates comparing P Ã PSII with P Ã C and P Ã O2 . The weighted PSII-specific absorption coefficients ( a Ã PSII ) ranged from 0.0052 and 0.0108 m 2 (mg chl a) −1 between samples and incubator light sources and are given in Table 2.

Short-term incubations (2h)
The electron requirement for gross 14 C fixation was measured from simultaneous rates of gross electron transfer (ETR PSII , mmol é (mg chl a) −1 h −1 ) and gross 14 C fixation (mmol C (mg chl a) −1 h −1 ) from the same bottles, incubated with surface water (5m) for 2 hours in an irradiance gradient (0 to 600 μmol photons m −2 s −1 , Fig 3). The experiment was repeated two times for each of the two stations (no replicates). The P-E curves demonstrated minor differences between stations, but a somewhat different curvature of the relationship between methods, with a relatively steeper α for 14 C fixation. The difference in curvature resulted in 2-3 times higher E k for ETR PSII than for 14 C fixation. The photosynthetic parameters E k , P max and α are given in Table 3. Chlorophyll a (chl a) concentrations (mean ± S.D.), chl a-specific absorption coefficient (a*), ratio of absorbed quanta in PSII (a* PSII :a*), spectrally weighted absorption coefficients (a Ã ), and the PSII-specific weighted absorption coefficient (a Ã PSII ) for the incubated samples and applied incubators (i.e. the water bath incubator and internal cuvette of the PAM instrument). a High Light The short-term (2h) incubations revealed a significant linear relationship between C fixation and ETR PSII during light-limited condition (E PAR <E k , corresponding to an ETR PSII <2 mmol é (mg chl a) −1 h −1 ), giving a slope coefficient of 0.13 ± 0.014 mol C (mol é) −1 . This corresponded to an electron requirement for gross carbon fixation of 7.6 ± 0.6 mol é (mol C) −1 (R 2 = 0.85, P <0.001, Fig 4). Inclusion of light-saturated samples lead to a non-linear relationship with ETR PSII exceeding the 14 C fixation under high light conditions (E>E k ), that could be described applying a simple inverted exponential decay function (Fig 4). The equation parameters are given in the figure. The dissolved fraction of fixed 14 C amounted to 18 ± 26% (data not shown) of the total and is accounted for in the assessments of the gross 14 C fixation. In the following section we apply the electron requirement for the light-limited carbon fixation (i.e. 7.6 mol é (mol C) −1 ) to investigate the relationship between P Ã PSII , P Ã C and P Ã O2 during 24h incubations. , and C) the spectrally-weighted chl a-specific absorption of phytoplankton at GF7 (5m), corrected for E(λ) in the water bath (green) and for the internal light source of the PhytoPAM (blue). Integrated values for a* and a Ã are given in Table 2. Variability of Φ PSII and diel primary production (24h) Electron requirement for net C fixation and NCP was investigated by applying long-term (24h) incubation experiments with a natural light-dark cycle. Triplicate samples from surface water  (5m) and from the lower euphotic zone (20m) were incubated simultaneously, the former under light-saturated (E PAR~5 00 μmol m −2 s −1 ) conditions, and the latter under light-saturated and light-limited (E PAR = 40 μmol m −2 s −1 ) conditions. These incubator irradiances correspond to the natural light intensities of a clear-sky day at the sampled depths (Table 1). Photosynthesis versus irradiance (PE) parameters for 2h simultanous measurements of ETR PSII and 14 C fixation, derived from least square regression of the Webb equation (data in Fig 3, Eq 11). a Units of ETR* PSII in mmol é (mg chl a) −1 h −1 and of 14   First, the temporal variability of F PSII with incubation time was investigated over the lightdark cycle. The dark acclimated maximum F PSII ranged from 0.55 to 0.65 while F PSII was~0.3 under high-light conditions (Fig 5A). Under low-light conditions (40 μmol photons m −2 s −1 ) F PSII was~0.6 and the maximum F PSII was similar to the values at high-light (data not shown). Thus, F PSII show the same trend and temporal variability during high and low light conditions. The response of F PSII to a change from darkness to light, and verse versa, showed a fast acclimation response (<0.5h) and little variability during light hours (8h). The corresponding relative ETR (rETR) showed a steady electron generation in the light and obviously none during darkness (Fig 5B). The result demonstrated a stable ETR over time within the incubation period. As bottles for ETR, C fixation and O 2 production were incubated simultaneously under the same conditions, it is reasonable to assume a linear relationship also for C fixation and O 2 production rates during incubations [6,48].
Diel primary production rates (μmol L −1 d −1 ) derived using the three methods are shown in Fig 6. The methods agreed well with one another at both 5 and 20m depths under both lightsaturated and light-limited conditions, and between stations, with a minor suppression of P C relative to P PSII and P O2 at the 20m_HL treatment. The difference between methods was tested using two-sided paired t-tests between each method, and showed no significant difference between P PSII , P C nor P O2 (P >0.05, performed using the build-in statistical routines in Origin 8.5, OriginLab). In detail, the difference between P PSII and P C were not significantly related to neither light intensity (P = 0.42, two-side t-test), water depth (P = 0.08) or station (P = 0.20). Neither was the difference between P PSII and P O2 significantly related to light intensity (P = 0.21, two-side t-test), water depth (P = 0.45) or station (P = 0.20). Consequently, data were pooled across light intensity, depth and stations in order to quantify the relationship between P Ã PSII , P Ã C and P Ã O2 (Fig 7). For this application the productivity was normalized to chl a to correct for the difference in biomass between depths and stations. Fig 7A shows a linear regression between P Ã PSII and P Ã C with a slope coefficient of 1.4 ± 0.15 (mean ± SE, R 2 = 0.86, P<0.001), which demonstrated a 1.4 times higher electron requirement for net carbon fixation than for gross carbon fixation. This implies a mean electron requirement for net C fixation of 10.9 ± 1.1 mol é (mol C) −1 . Comparing P Ã PSII and P Ã O2 demonstrated a slope coefficient of 0.86 ± 0.12 (mean ± SE, R 2 = 0.94, P<0.001) corresponding to an electron requirement for net O 2 production of 6.5 ± 0.9 mol é (mol O 2 ) −1 (Fig 7B). This is an electron requirement~14% lower than for the gross C fixation Plotting P Ã O2 versus P Ã C yielded a slope coefficient of 1.6 ± 0.53 (mean ± SE, R 2 = 0.75, P = 0.02) that demonstrated higher net O 2 production rates than net carbon fixation rates across all samples (Fig 7C), i.e. the Photosynthetic Quotient (PQ).

Discussion
In this paper, we estimate the electron requirement for C fixation and O 2 production in phytoplankton in an Arctic fjord under post bloom conditions. Crucial to the calculation of ETR PSII in absolute units is to quantify the amount of photons absorbed in PSII accurately. Hancke et al. [27] demonstrated a bio-optical approach to correct standard phytoplankton absorption measurements for the fraction of absorbed quanta in PSII [a Ã PSII :a Ã ]. By weighting the absorption spectrum to the spectral quality of the incubator light source [35] they furthermore calculated and accounted for the PSII-specific absorption. Here we apply a simplified approach without using a sophisticated spectrofluorometer, but by applying published values for the fraction of PSII absorption determined for representative taxonomic groups [36]. By accounting for the phytoplankton light absorption and spectral irradiance of the incubator light (both of the waterbath incubator and inside the PAM cuvette), we calculate the PSII-specific absorption coefficient for the investigated communities at the applied conditions (Figs 1 and  2). This enables to correctly express ETR PSII in absolute units and compare these to measured rates of C fixation and O 2 production.

Electron requirement for gross carbon fixation
Empirical evidence from a range of aquatic systems demonstrates a linear relation relationship between ETR and rates of C fixation and/or O 2 production [20,21,49], however, deviation from linearity have also been reported [17].The derived electron requirement for primary production seem to express considerable variability and there still is considerable uncertainty about what causes this variability and how it relates to 'true' rates of primary production.
We found a consistent relationship between ETR PSII and gross C fixation (Fig 4) yielding an electron requirement for gross C fixation of 7.6 ± 0.6 (mean ± SE, mol é (mol C) −1 ) under light-limited conditions (E PAR <E k ). There exists no exact theoretically defined requirement of electrons for C fixation. However, absolute ETR PSII is considered a proxy for the gross photosynthetic rate, and the electrons generated at PSII are closely coupled to the gross O 2 evolution rate from the water splitting complex in PSII [28]. Thus, it is theoretically given that the minimum electron requirement for gross O 2 production, according to the standard Z-scheme of photosynthesis, is 4 electrons per O 2 produced [28,49]. By multiplying this with a PQ of 1.4 produced O 2 per fixed C [50], the minimum electron requirement for gross C fixation accounts Comparison of diel rates of primary production derived from Φ PSII , 13 C fixation and O 2 production. Calculated rates of gross carbon fixation from Φ PSII (P PSII ), and rates of measured net 13 C fixation (P C ) and net O 2 production (P O2 ) during 24h incubations with 16 hours of light, at (A) station GF3 and (B) GF7. P PSII rates were calculated using the electron requirement for gross 14 C fixation of 7.6 mol é (mol C) −1 (Fig 4). Error bars for P PSII are triplicate samples times 4 measurements across 24h (as shown in Fig 5), for P C triplicate bottles, and for P O2 5 replicate bottle incubations.  Relationships between diel rates of primary production derived from Φ PSII , 13 C fixation, and O 2 production. A) Gross carbon fixation from Φ PSII (P* PSII ) versus 13 C fixation (P* C ), B) P* PSII versus net O 2 production (P* O2 ), and C) P* O2 versus P* C derived from 24h incubations with 16 hours of light (Fig 5). Data are pooled across depth, light intensity and stations (Fig 6). Rates are normalized to chl a and dashed lines are linear regressions, with A) forced through origo.
doi:10.1371/journal.pone.0133275.g007 to 5.6 mol é (mol C) −1 . Here we found an electron requirement 1.4 times higher than this, which indicate an offset between ETR and C fixation. We speculate that this discrepancy is caused by alternative electron 'requirements' along the pathway from PSII to the fixation of carbon in the Calvin Cycle. Likely, these electrons are utilized to cover the energy requirement for nutrient uptake and cellular maintenance, and in Mehler-type reactions, especially at high light conditions [22,49].
Further, it is still debated whether short-term 14 C fixation measurements represent NPP or GPP. Current consensus is that short term incubations (1 to 3h) quantify something intermediary to the strictly defined GPP and NPP. The uncertainties are mainly related to respiration of photosynthetically fixed 14 C, recycling of respired 14 CO 2 and its preferential use over external CO 2 , but also growth rates have shown to impact the balance of NPP versus GPP [5][6][7][8]. Longer incubation periods and lower growth rates seem to bias short-term 14 C fixation measurements towards representing NPP rather than GPP. For instance, rates measured during 2 hour incubations have shown to underestimate rates obtained during 60 min, and even more during 30 min [51,52]. In the current study we included the dissolved fraction of 14 C labeled carbon to the gross C fixation measure, thus it is included in Κ C for the gross C fixation. The fraction of dissolved to particular 14 C uptake we here report are typical for pelagic C fixation [53,54].
For comparison to studies excluding a spectral correction for the incubator light spectrum, we recalculated the data of the present study excluding the spectral correction, which gave a mean Κ C for gross C fixation of 11.2 mol é (mol C) −1 , relative to 7.6 mol é (mol C) −1 when spectrally corrected (Table 4). This gave an overestimation of Κ C by 1.5 times and a 2-fold increase in the standard error, when excluding the spectral correction. The determined error imposed on Κ C depends obviously on the spectral quality of the incubator light source and thus varies between studies. The closer the incubator light source is to a 'perfect white light' spectrum (also named a 'flat' spectrum) the lower is the error introduced in Κ C. In a field study, Kromkamp et al. [21] derived an electron requirement of 16.8 mol é (mol C) −1 across seasons from fresh water lakes (Table 4). In this estimate, they included the light absorption jointly for both PSII and PSI but did not correct for the spectral quality of the incubation irradiance. Assuming an equal distribution of the light absorption between PSII and PSI the comparable number to this study would be (16.8 × 0.5) 8.4 mol é (mol C) −1 . This number is in close agreement with what found here, however theoretically it is overestimating Κ C by the ratio of [a Ã PSII :a Ã ] to 0.5. Kromkamp et al. estimated that correcting for the spectra difference between the red LEDs of the waterPAM actinic light and the natural underwater light field would reduce the estimated C fixation rate by 35%. Thus, the non-spectral corrected electron requirement [8.4 mol é (mol C) −1 ] would be overestimated by 35%.
Comparing Κ C to previous studies is hampered by difference in measuring procedures and protocols that are used to quantify the absorption of photons in PSII (Eq 1). Table 4 provides values of Κ C for the minimum electron requirement for C fixation and O 2 production derived from the current literature. The table includes only studies that account for the PSII specific absorption, and attempt to express ETR in absolute units. Hancke et al. [24] found a mean Κ C ranging from 3.9 to 6.2 mol é (mol C) −1 in mono cultures of different phytoplankton species, while accounting for the applied spectral irradiance and [a Ã PSII :a Ã ] ( Table 4). Somewhat higher values of Κ C, ranging from 9.2 to 32.3 mol é (mol C) −1 are derived from a similar study of Napoleon et al. [26], that accounted for [a Ã PSII :a Ã ] but ignored the spectral quality of the incubator light.
Under high light intensities (E>>E k ), we observed a non-linear relationship between ETR P-SII and the C fixation, (Fig 4). Even though many studies demonstrate a linear relationship, Napoleon and Claquin [17] report of a similar non-linear relationship between C and ETR at E PAR >E k , in a field study from the English Channel. They ascribed the non-linear relationship Electron Requirement for Primary Production to alternative electron sinks caused by high irradiance, photo inhibition, Mehler-type reactions and nutrient uptake. In fact, Napoleon and Claquin [17] proposed an empirical algorithm to compensate for this. While the derived parameters apply well (R 2 = 0.77) and the study appear thoroughly performed the derived parameterization has little theoretical foundation and cannot a priory be assumed to apply across seasons or regions.

Electron requirement for net carbon fixation and O 2 production
A challenge when comparing rates of ETR PSII and NPP/NCP is to constrain the temporal variations in ETR during the incubation, as ETR is sampled instantaneously and C fixation or O 2 production are integrated over the entire incubation. From monitoring ETR (or really F PSII ) during a diel cycle with an 8:8:8 hour light:dark:light period we documented that ETR was stable under constant light, both at high and low light (Fig 5). This suggests that a single measurement after 1 to 2 hours under the desired light regime is sufficient to calculate the productivity for a 24h time period, under laboratory steady state incubations. Consequently, NPP and NCP can be assessed for any desired day-length and irradiance from single F PSII measurements (assuming state-state conditions of the cellular light-acclimation processes). This conclusion is in accordance with current understanding of the light acclimation in microalgae, which is categorized in sequential 'time windows'. The initial light acclimation processes happens within seconds to minutes (e.g. photochemical and non-photochemical quenching), whereas protein synthesis and pigment metabolism occurs on time scales of hours, while metabolic changes occurs on the time scales of generations [7,12,55]. Nymark et al. [55] have shown that despite complex short-term changes on levels of gene transcription, protein synthesis, and pigment metabolism, microalgae are able to secure a relative stable electron transfer rate through the PSII reaction centre for time period <12 hours. This has been explained as an evolutionary mechanism to secure an efficient photosynthetic capacity even under fluctuating light conditions. Changes in ETR, however, occurred after prolonged exposure to high light of >12 hours [51,55], which is why we recommend that light:dark shift incubations spanning 24 hours are carried out in a way to avoid light periods longer than 12h. It is beyond the scope of this work to elaborate on the photo acclimation processes in microalgae. Comparing diel integrated rates of NPP showed a linear relationship between P Ã PSII and P Ã C with a slope coefficient of 1.4 ( Fig 7A). Here we have applied the electron requirement for gross C fixation [7.6 mol é (mol C) −1 ] in the calculation of P Ã PSII to compare with net C fixation rates. The resulting electron requirement for net C fixation of 1.4 times that of gross C fixation [Κ C = 10.9 ± 1.1 mol é (mol C) −1 (R 2 = 0.86)] accounts for the 'costs' for metabolic activity and cell maintenance, and includes the fraction of labeled 14 C that recirculates between states of fixed (biomass) and respired (CO 2 ) carbon. Estimates of NPP using variable fluorescence rely on the precision and the potential variability of Κ C . Here we estimated P PSII from triplicated bottles; and in each bottle F PSII was measured 4 to 6 times over the 24h incubation. Across the entire data set the coefficient of variance was 4.2 and 16.3% for low and high light, respectively. The 4 times higher variance at high light is related to the inherent decrease in the signal to noise ratio of F PSII with increasing irradiance. The variability of a Ã PSII is on the same order, while the precision of the irradiance measurements are much better (<1%) assuming careful measurements with a calibrated sensor. Thus, we conclude that at the applied settings the NPP was estimated from absolute PAM (F PSII ) measurements with a~20% accuracy.
As for the 2 hour incubation we recalculated Κ C for the NPP without a spectral correction. This resulted in a value of 15.9 ± 1.6 mol é (mol C) −1 , which corresponds to an overestimation of a factor 1.5 (Table 4). To our knowledge, no previous studies have compared rates of ETR PSII and NPP obtained over 24h, thus the result has to be evaluated in context of the studies listed in Table 4.
Nutrients were not limiting in the current study (N > 1 μM in all samples, Table 1) and we rule out F PSII to be depressed by nutrient limitations. The effect of nutrient limitations on F PSII is, nevertheless, not consistent, but evidence of F PSII depression due to N limitation seem convincing in some cases [56], however many studies point to only limited depressing of F PSII during nutrient starvation, partly due to acclimation processes [22,26,[57][58][59].
Like C fixation, NCP in O 2 units (P Ã O2 ) resulted in a linear relationship with P Ã PSII , with a slope coefficient corresponding to an electron requirement of 6.5 ± 0.9 mol é (mol O 2 ) −1 . This was~40% lower than for C fixation during 24 hours and supports the assumption of an considerable electron consumption related to the 'dark' reactions of C fixation, including cell maintenance and nutrient uptake. The respiratory O 2 consumption explain the majority of the discrepancy between the empirical Κ C and the theoretical one [4 é (mol O 2 ) −1 ], and accounted to 10.1 ± 5.2% and 44.9 ± 19.3% of the NPP during high and low light, respectively. The rates of respiration are consistent with typical ratios published for pelagic ecosystems [60,61]. The slope coefficient of P O2 versus P C was 1.6 ± 0.5, which complies with the resolved PQ of 1.4 in the current study [50].
The Arctic marine environment is experiencing dramatic changes in sea ice cover, terrestrial run-off and light attenuation [62]. It is expected that these changes will affect primary production. Yet most routine monitoring programs of marine ecosystems are limited to few measurements of primary production, as the benchmark 14 C method is time consuming and expensive. Application of variable fluorescence techniques has the potential to expand routine measurements to larger regions and with higher temporal resolution. This, however, will require detailed and reproducible assessment of the electron requirement for carbon fixation under natural variable conditions [16,63]. Not least in the under-sampled Arctic [62]. Future studies should ideally compare in situ measured variable fluorescence with in situ bottle incubations of C fixation and O 2 production to preserve the natural optical properties of phytoplankton absorption and available spectral irradiance.