Corresponding author: K. E. Giesbrecht, School of Earth and Ocean Sciences, University of Victoria, Victoria, BC V8W 3P6, Canada. (firstname.lastname@example.org)
 We compared net community production determined from an in situ O2/Ar mass balance (O2/Ar-NCP) with incubation measurements of new and primary production in the subarctic northeast Pacific. In situ O2/Ar-NCP was strongly correlated to new production from 24-h15NO3− uptake integrated over the mixed layer (15N-NewP), if measurements were separated into high and low-productivity conditions. Under high-productivity conditions, O2/Ar-NCP estimates were similar to15N-NewP, whereas under low productivity conditions O2/Ar-NCP was up to two times higher than15N-NewP. The relationship between O2/Ar-NCP and 24-h13C primary production (13C-PP) was more variable, but with a consistent mean O2/Ar-NCP:13C-PP ratio of 0.52 ± 0.17 when only low-productivity, summer measurements were considered. This relationship with primary production is perturbed by high productivity events such as a late-summer, iron-stimulated bloom observed at the offshore stations. Finally, we show that diapycnal mixing usually dominates the O2/Ar mass balance in winter in the subarctic Pacific, preventing the determination of NCP by the O2/Ar method at that time, except for one unusual stratification event in February 2007.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Biologically produced organic carbon and its export from the surface to the deep ocean partially regulates atmospheric CO2 concentrations and therefore global temperatures. This biological pump results in atmospheric CO2 concentrations that are nearly half the levels predicted if all marine life were extinct [Siegenthaler and Sarmiento, 1993]. The rate of the biological pump may not have changed since pre-industrial times [Siegenthaler and Sarmiento, 1993], because nutrients, light, and zooplankton grazing control carbon export, rather than CO2 concentrations [Raven and Falkowski, 1999]. However, there is potential for significant changes to this export flux in the future as a result of warming-induced stratification [Sarmiento et al., 1998], making quantification and monitoring of the biological pump essential.
 A wide variety of methods have been developed to quantify biological productivity, but methodological differences and the lack of a standard for flux measurements can complicate interpretation [e.g., Cullen, 2001]. With each method susceptible to different errors, productivity may be best constrained by comparing several different methods. Incubation-based methods suffer from isolation of the plankton community within the incubation bottle, and potential mismatches in the physical conditions of incubation versus the original environment, such as light levels and temperature [Platt and Sathyendranath, 1993; Cullen, 2001]. In situ methods are susceptible to other errors, such as ±15% errors in air-sea gas exchange fluxes for the O2 mass balance method [e.g., Juranek and Quay, 2010]. However, comparing several different methods to obtain a constrained measure of productivity is also complicated as different methods rarely directly measure the same process [Falkowski et al., 2003]. In certain cases, this complication can be minimized by comparing two methods that have measured rates that should be equivalent if averaged over large enough temporal and spatial scales or if the system is at steady state. New production and net community production are a common example of this [e.g., Legendre and Gosselin, 1989; Laws, 1991; Falkowski et al., 2003]. Available techniques can integrate over temporal and spatial scales ranging from milliseconds and single cells (Fast Repetition Rate Fluorometry) to hours and specific depths (incubation-based) to weeks and the mixed layer (O2mass balance) to years and global coverage (satellite-based). As a result, the temporal and spatial scales of each method also affect method comparisons.
 In this study, we compare in situ estimates of net community productivity (O2/Ar-NCP), with incubation-based estimates of new (15N-NewP, based on15NO3−) and carbon-based primary (13C-PP) productivity (Table 1) along a coastal-oceanic transect in the subarctic northeast Pacific. There have been a number of studies that compare incubation-based productivity estimates using different tracers [e.g.,Bender et al., 1987, 1999; Dickson et al., 2001], but fewer studies have compared in situ and incubation-based methods [e.g.,Hendricks et al., 2004; Juranek and Quay, 2005]. There are fewer still that directly compare in situ estimates of net community productivity (NCP) with incubation-based estimates [e.g.,Quay et al., 2010; Hamme et al., 2012] and only one that compared measurements in the subarctic northeast Pacific [Emerson et al., 1993].
Table 1. Productivity Methods Used in This Study
Steady State Equivalence
In situ based; photosynthesis minus community respiration.
Incubation based; assimilation rate of new nitrogen sources (primarily NO3−) during photosynthesis.
Incubation based; uptake of labeled dissolved inorganic carbon (24-h incubations fall between Gross Primary (fixation rate of CO2 into organic carbon through photosynthesis) and Net Primary (Gross Primary minus autotrophic respiration) [Marra, 2009]).
 With tri-annual cruises that cover a range of productivity regimes and a comprehensive historical data set with several intensive studies, Line P samples an important region for productivity method comparison. Located in the subarctic North Pacific, Line P extends from the southern tip of Vancouver Island out to one of the longest running deep-ocean time series in the world, P26 (Ocean Station Papa, OSP, Station P, or P, at 50°N 145°W,Figure 1). The time series includes over five decades of measurements at P26, with additional stations added in 1959 and increasing in 1981 to the 27 sampled today. Line P spans a range of physical, chemical and biological regimes [Boyd and Harrison, 1999; Varela and Harrison, 1999; Whitney et al., 2005], shifting from a highly productive near-coastal environment affected by seasonal upwelling [Whitney et al., 1998] out to the High-Nutrient Low-Chlorophyll (HNLC) region of the subarctic gyre where iron limitation and microzooplankton grazing control phytoplankton standing stocks [Miller et al., 1991; Boyd et al., 1996]. Offshore temporal and spatial variations in productivity along Line P are low [Boyd and Harrison, 1999; Varela and Harrison, 1999], simplifying our comparison of productivity methods that integrate over different temporal and spatial scales. However, though productivity along Line P tends to be nearly constant, iron inputs have caused sporadic high productivity events in the offshore region throughout the time series [Parsons and Lalli, 1988; Wong et al., 1995; Lam et al., 2006]. Although several methods have been used to estimate productivity along Line P (Table S1 in the auxiliary material), measurements done by some methods were only conducted sporadically. Identifying relationships between these different methods can help in interpreting historical measurements by allowing conversion between them.
 Here, we present a three-year data set of dissolved O2/Ar measurements and a one-year data set of13C/15N dual tracer incubations along Line P from tri-annual cruises (February, June and August) in 2007 to 2009. We will show that O2/Ar-NCP showed little variability at the offshore stations in June and August 2007–2009 (excluding the anomalous August 2008 event). We also address the impact of diapycnal mixing on estimates of NCP when using an O2/Ar mixed layer mass balance in this region. Rates of 15N-NewP were strongly related to O2/Ar-NCP for all stations sampled in June and August, but comparisons of13C-PP to O2/Ar-NCP or15N-NewP were more variable. Finally, we present observations of two high productivity events along Line P, one in winter (February 2007) and one in summer (August 2008).
2.1. In Situ Estimates of NCP From Dissolved O2/Ar Measurements
 Dissolved gas samples for O2/N2/Ar ratios were collected in duplicate on nine Line P cruises over a 3-year period (2007–2009) at the five major stations (P4, P12, P16, P20 and P26). Samples were collected at 5-m for most stations, with a depth profile at P26 for every cruise and additional stations on the 2009 cruises. Discrete O2 samples were also collected in duplicate (pooled s.d. = 0.68 μmol kg−1) and analyzed on-board using the Carpenter-modified Winkler titration [Carpenter, 1965].
 See Emerson et al.  for a complete description of the O2/N2/Ar sampling and analysis method. Briefly, water was transferred at sea from Niskin bottles to evacuated, HgCl2-poisoned, 180-mL glass flasks until half full through CO2-flushed tubing. Back at the lab, the water was equilibrated with the headspace and then removed. Headspace gases were purified through liquid N2 traps and O2/N2/Ar ratios were measured against a standard with similar ratios on a dual-inlet isotope ratio mass spectrometer (Finnigan Delta X/L at Univ. of Wash. or MAT 253 at Univ. of Victoria). Standards with known O2/N2 ratios were used to determine the effect of differences in the gas ionization efficiencies when the sample and standard gases had different O2 concentrations (the chemical slope) [Emerson et al., 1999; Hamme, 2003]. Corrections to O2/Ar measurements for this study were on the order of 0.05%.
 The O2/Ar ratio in the mixed layer is a tracer of net community O2 production. Argon has a similar solubility and temperature dependence to O2, thus acting as an abiotic analogue to O2. Because physical processes, such as temperature change and bubble-mediated gas exchange, have a nearly equivalent effect on O2 and Ar, the O2/Ar ratio is a measure of biological O2 production and can be quantified as
where (O2/Ar)sample is the measured ratio in seawater, (O2/Ar)eq is the ratio at equilibrium with the atmosphere for the potential temperature and salinity of the water mass [García and Gordon, 1992, 1993; Hamme and Emerson, 2004], and ΔO2/Ar is typically presented in percent.
 At steady state, net production of biological O2 in the mixed layer is balanced by diffusive gas exchange with the atmosphere, vertical mixing, and horizontal advection. When mixing and advection are negligible contributions to the mass balance, net biological O2 production (NCP) can be quantified as [Reuer et al., 2007]:
where k is the gas transfer velocity (m d−1), [O2]eq is the equilibrium concentration of O2 in the mixed layer (μmol kg−1), and ρ is the density of the mixed layer (kg m−3). We estimate the gas transfer velocity, k, using the quadratic wind speed parameterization of Ho et al. and the 6-h NCEP/QuikSCAT blended wind product provided by Colorado Research Associates (http://dss.ucar.edu/datasets/ds744.4/) for all cruises except for August 2009, where the QuikSCAT wind product provided by the Physical Oceanography Distributed Active Archive Center (PO.DAAC) at the NASA Jet Propulsion Laboratory, Pasadena, CA (http://podaac.jpl.nasa.gov) was used. To account for wind speed variability over the weeks preceding the sampling date, we use the wind speed weighting scheme of Reuer et al. , which weights and averages gas transfer velocities over a 60-day period (assuming constant values for net production and mixed layer thickness). We used a photosynthetic quotient (PQ, ΔO2:ΔCO2) of 1.4 [Laws, 1991] to convert O2production into carbon-based units, based on the PQ of nitrate assimilation.
 Errors in NCP estimates from the O2/Ar method are normally dominated by uncertainty in the gas transfer velocity, with contributions from analytical uncertainty, and occasional contributions from vertical mixing and horizontal advection. We estimate the uncertainty in the gas transfer velocity, k, to be ±14%, a conservative estimate representing twice the uncertainty reported in Ho et al. . Following Juranek and Quay , this value also represents twice the percent difference between the parameterization of Sweeney et al. , an update of Wanninkhof  based on global bomb 14C values, and that of Nightingale et al. , based on dual-tracer experiments.Ho et al.  falls midway between these two. Precision of the ΔO2/Ar measurements was ±0.10%, based on the pooled standard deviation of duplicate samples, which contributes a mean uncertainty to our NCP estimates of 3.8% and up to a maximum of 9.0% when ΔO2/Ar measurements are low (i.e., close to equilibrium, or ∼1%). Uncertainties in the equilibrium concentration of O2 (±0.2%) are small compared to the errors associated with the gas transfer velocity and ΔO2/Ar measurements. We estimate the mean combined error in NCP to be ±15%, with a maximum of ±17% when ΔO2/Ar is close to equilibrium.
 The effect of diapycnal mixing on the O2/Ar mass balance depends on the gradients of O2 and Ar below the mixed layer and the eddy diffusion coefficient (Kz). In section 3, we present our finding that diapycnal mixing is a negligible contribution to the O2/Ar mass balance in June and August, but typically overwhelms the mass balance in February. In addition to diapycnal mixing, entrainment of water by a deepening mixed layer is another vertical flux that can affect the O2/Ar mass balance, but is unlikely to affect our summer results. During this period, increasing heat fluxes and decreasing wind speeds cause the mixed layer to shoal, forming the seasonal thermocline and preventing significant entrainment [Whitney and Freeland, 1999]. During the fall and winter, entrainment will affect the mixed layer mass balance, as cooler temperatures and increased wind speeds lead to a deepening of the mixed layer reaching maximum depths in February–April [Whitney and Freeland, 1999].
 Finally, we find that the effect of horizontal advection in this region is usually small [e.g., Jackson et al., 2006]. The contribution of the horizontal fluxes to our mass balance can be estimated from
where d[O2]/dx is the horizontal gradient in the O2 concentration in the mixed layer between adjacent stations (μmol kg−1 m−1), vc is the current speed (m d−1), and h is the depth of the mixed layer (m). Horizontal gradients along Line P are generally 10−6 to 10−5μmol O2 kg−1 m−1and variable in sign between stations (based on 5-m O2 concentrations). Average surface current speeds at P26 are ∼5 × 103 m d−1 and variable in direction (based on current speeds at 5 and 35 m depth from the NOAA Station P mooring at http://www.pmel.noaa.gov/stnP/index.html). For an average mixed layer depth of 40 m, the horizontal flux of O2 could be as much as 0.2 to 2 mmol O2 m−2 d−1 or about 1–10% of the usual diffusive gas exchange flux. Variability in both the horizontal O2 concentration gradients and the direction of the surface currents make this estimate an upper limit. In a mass balance study with O2 and N2 measurements at Station P, Emerson and Stump  require a large horizontal advection term (order of 20 mmol O2 m−2 d−1) to balance their N2mass balance (primarily to account for the time rate of change and bubble input fluxes). They do not calculate this flux directly as we have here, but argue that this advective flux required to close their mass balance could be generated by lateral gradients in air-sea heat fluxes. If this were the case, such lateral gradients would affect O2 and Ar in the same manner and result in little impact to the O2/Ar mass balance we use here.
2.2. Dual-Tracer13C/15N Experiments
 Samples for 13C and 15N uptake rate experiments were collected on the same cruises as the dissolved gases starting in August 2008 at station P26, and in February 2009 adding P4 and P16. At depths corresponding to 100%, 55%, 30%, 10% and 1% of the surface photosynthetically available radiation (PAR), water for the dual-tracer incubation experiments was collected into two acid-washed 1-L polycarbonate bottles. Blanks were collected at either the 100% or 1% light depth. Samples for dissolved inorganic carbon (DIC) and NO3− concentrations were collected from the same Niskin. Water samples for dissolved NO3−concentrations were kept at 4°C until analysis on-board shortly after collection using a Technicon II Autoanalyzer® [Barwell-Clark and Whitney, 1996]. DIC samples were collected into 500-mL borosilicate bottles, preserved with 200μL of a saturated HgCl2solution and kept at 4°C until analysis onshore using a SOMMA-Coulometer system following the methods ofDickson and Goyet .
 Nitrate and carbon uptake rates were measured using a dual tracer method with stable isotopes 15N and 13C [Dugdale and Goering, 1967; Slawyk et al., 1977]. Samples were kept under low light conditions at 4°C for no more than 2 h prior to incubation and after removal from the on-deck incubator until termination by gentle filtration. For each depth, both samples were enriched with13C labeled NaHCO3 (Cambridge Isotope Laboratories, 99 atom % 13C) and with a 15N labeled solution of either KNO3 or NH4Cl (Cambridge Isotope Laboratories, 99 atom % 15N). Isotopic additions were made at ∼10% of the ambient concentration of DIC and NO3− or NH4+. When NO3− or NH4+ concentrations were below the detection limit (0.5 μM), 15N additions were 0.05 μM. Seawater blanks were inoculated according to their ambient concentration and filtered immediately after the start of incubation to account for initially adsorbed 13C and 15N on the filter and/or cell membranes.
 Samples were placed in an on-deck incubator for 24 h under a pre-determined amount of neutral density screening that simulated the in situ light conditions of the sampling depth. Temperature was controlled by continuously flowing surface seawater through the incubator. After 24 h, samples were filtered under gentle vacuum through pre-combusted Whatman® GF/F filters (ca. 0.7 μm nominal porosity). Filters were kept at −80°C until the end of the cruise and then dried at 60°C for 3 days. The Stable Isotope Facility at the University of California, Davis analyzed the filters for atom % 13C and 15N, and particulate organic carbon (POC) and nitrogen (PON) concentrations.
 Carbon (13C-PP) and nitrate (15N-NewP) uptake rates were calculated using the POC or PON measured at the end of the incubation and equations (6) and (3) ofDugdale and Wilkerson . Though these equations are based on nitrogen physiology, calculating 13C-PP using equations based on carbon physiology [Hama et al., 1983] resulted in only a mean difference of 0.5%. When the initial concentration of nitrogen was below the limit of detection (0.05 μM), ambient nitrate concentrations were given a value of 0 for the rate calculations. Dissolved organic pools were not measured, so we cannot correct for the potential loss of the isotopic label as DO13C [e.g., Karl et al., 1998] or DO15N [e.g., Bronk et al., 1994]. Neglecting this correction could have potentially biased our rates by up to 30%. Results at individual depths were trapezoidally integrated to the base of the mixed layer, which was chosen to facilitate comparisons with the mixed-layer O2/Ar mass balance results. We used the stoichiometric ratio of POC:PON measured at the end of the incubation to convert our nitrogen-based rates to carbon-based rates in each bottle, though using the Redfield ratio (106 C:16 N) [Hedges et al., 2002] instead changes our integrated rates by a maximum of ±3%. We estimate the errors associated with our incubation-based measurements to be ±16%, which represents the mean difference between duplicate samples for the carbon-based uptake rates.
3. Impact of Diapycnal Mixing on the O2/Ar Mass Balance
 We generalize the impact of diapycnal mixing on our results by month (February, June, and August), characterized by the magnitude and direction of the dissolved O2 and Ar gradients below the mixed layer. To diagnose the importance of diapycnal mixing to the O2/Ar mass balance, we estimated the magnitude of the eddy diffusion coefficient that would be required for the diapycnal mixing flux to balance the surface gas exchange flux with no biological production.
where d[O2]/dz and d[Ar]/dz are the gradients in the O2 and Ar concentrations below the mixed layer (μmol kg−1 m−1), Kz is the eddy diffusion coefficient (m2 d−1 or cm2 s−1), and the right-hand side of the equation is the same asequation (2). Equation (4) is derived by subtracting the mass balance of Ar in the mixed layer from that of O2 (see Text S1). We present the results at P26 in 2007 in detail, but we completed this analysis at all stations and cruises where O2/Ar samples were collected. We calculated the O2 and Ar depth gradients from the bottom of the mixed layer to 5 m below using depth profiles of dissolved O2 from the O2 CTD sensor (calibrated by discrete Winkler O2 samples) and the equilibrium concentration of Ar from potential temperature and salinity profiles. Although the CTD O2 sensor can be biased, the gradient in the O2 concentrations should be more accurate. Argon supersaturations (determined from discrete O2/Ar and O2 measurements) were nearly constant (±0.3%) above and directly below the mixed layer, so the gradient in the equilibrium concentration of Ar approximates the true Ar gradient. Our station nearest the coast (P4) is located at the continental shelf break and does not experience direct upwelling or river input [Whitney et al., 1998]. The diapycnal mixing analysis therefore should be reasonably valid even at this station.
 In February, dissolved O2 depth gradients beneath the mixed layer (Figure 2a) were typically large and negative (lower O2 concentrations below the mixed layer), a result of cooler temperatures and strong winds mixing the surface layer to the depth of the permanent halocline that exists along Line P [Whitney and Freeland, 1999]. ΔO2/Ar in the mixed layer was predominantly undersaturated at this time (Figure 3). For these circumstances eddy diffusivities of ≤0.2 cm2 s−1would be sufficient to generate a mixing flux that balanced air-sea gas exchange. These Kz values are well within the reasonable range (0.1–10 cm2 s−1) of previous measurements of Kz below the mixed layer [e.g., Large et al., 1986; Rousseau et al., 2010]. This indicates that diapycnal mixing plays an important role in the February O2/Ar mass balance and thus, we cannot constrain O2/Ar-NCP under these conditions. Also, we would not even be able to constrain O2/Ar-NCP under these conditions with simultaneous measurements of Kz, as the uncertainty in Kz would overwhelm the mass balance.
 In June, the mixed layer depth has shoaled, the result of decreasing wind speeds and increasing temperatures creating a shallow seasonal thermocline [Whitney and Freeland, 1999]. Dissolved gas concentration gradients were virtually non-existent below the mixed layer (Figure 2b) and ΔO2/Ar in the mixed layer was always supersaturated (Figure 3). Calculated eddy diffusivities needed to balance the gas exchange flux in these conditions were extremely large and sometimes negative (±103 cm2 s−1), because the diapycnal mixing flux is proportional to the O2 and Ar gradients. Given that estimates of Kz values from microstructure measurements in June 2007 were of order 0.1 cm2 s−1 [Rousseau et al., 2010], the contribution of diapycnal mixing to the O2/Ar mass balance in June must be negligible.
 By August, a subsurface O2 maximum (Figure 2c) resulted in positive O2 gradients (higher O2 concentrations below the mixed layer) across the fully formed seasonal thermocline, with ΔO2/Ar supersaturated in the mixed layer (Figure 3). Kz values of 2–5 cm2 s−1 were needed for vertical mixing to balance gas exchange. Because these values are within the reasonable range determined by previous studies (0.1–10 cm2 s−1), we cannot assume a negligible mixing flux as we do in June. The positive O2 gradients indicate that we overestimate O2/Ar-NCP if we exclude diapycnal mixing from the mass balance. However, the degree of this overestimation is probably small given that recent estimates of Kz below the mixed layer in June 2007 at P26 average 0.05–0.08 cm2 s−1 [Rousseau et al., 2010], though Large et al.  found Kz values can increase to >10 cm2 s−1 during storms. If we assume a value for Kz of 0.1 cm2 s−1 in August, the diapycnal mixing flux accounts for less than 1 mmol O2 m−2 d−1, or ∼6% of the gas exchange flux. Thus, though we cannot accurately quantify the mixing flux in the summer, excluding it from our mass balance likely only overestimates O2/Ar-NCP by ∼6%.
 Though our results can typically be separated into these three seasonal cases, we did observe one case in February 2007 when an unusual wintertime stratification allowed us to constrain the O2/Ar mass balance and calculate O2/Ar-NCP. During this period, mixed layer ΔO2/Ar at stations P16 and P20 were supersaturated (1.1 and 0.8% respectively, Figure 3), with O2/Ar-NCP comparable to measurements from June or August (∼9 mmol C m−2 d−1, Table S2), and about twice the rate of 15N-NewP measured at P16 in February 2009. Mean surface chlorophylla concentrations (Institute of Ocean Sciences data) between P16 and P20 in February 2007 were nearly twice the 1989–2006 wintertime average (0.55 ± 0.13 mg m−3, n = 5 versus 0.31 ± 0.02 mg m−3, n = 70), supporting that the ΔO2/Ar supersaturations we observed were a result of increased productivity. Unlike the typical February case along Line P, the mixed layer in 2007 was very shallow at these stations (15 m) and dissolved O2 gradients below the mixed layer were very small (Figure 4). The positive ΔO2/Ar supersaturations at the surface could not have been caused by vertical mixing because the dissolved O2 gradient was negative at the base of the mixed layer. In this case, vertical mixing could only have caused an underestimate of O2/Ar-NCP by as much as 20% if the eddy diffusivity was as high as 1.0 cm2 s−1. Ancillary data show that these increases in O2/Ar-NCP and chlorophyllawere likely due to a relief of light and possibly iron limitation on the phytoplankton, a result of rain-induced stratification between stations P16 and P20 in the days leading up to sampling. Prior to sampling, the NCEP/NCAR mean daily precipitation rate was approximately 2.5 times higher in the region surrounding stations P16 and P20 compared to the surrounding area. This special February 2007 case shows that, when the winter mixed layer is separated from the strong oxycline, the O2/Ar mass balance can be constrained using only the diffusive gas exchange flux, allowing an accurate estimate of O2/Ar-NCP in winter.
4. Individual Productivity Method Results
4.1. O2/Ar-Based Net Community Production
 We observed evidence of fairly constant summertime production at the offshore stations (P12-P26) along Line P in our estimates of O2/Ar-NCP from 2007 to 2009, but with notable exceptions (Figure 5a). In the subarctic North Pacific, iron limitation causes a slow and relatively steady decrease in nitrate over the spring/summer growing season at the offshore stations along Line P [Peña and Varela, 2007], suggesting low and constant net productivity rates. Summertime measurements of ΔO2/Ar, the biological O2 supersaturation, were also fairly consistent at 2.4 ± 0.9% (Figure 3). Summertime O2/Ar-NCP ranged from 5.1 to 55 mmol C m−2 d−1 along Line P, though the mean June–August O2/Ar-NCP for the offshore stations was 13.1 ± 4.4 mmol C m−2 d−1 (n = 20) for 2007–2009 (excluding August 2008, a special case we discuss in section 5.3). This value is similar to the mean O2/Ar-NCP calculated byEmerson and Stump  for June–September 2007 (17 ± 2.6 mmol C m−2 d−1) using in situ O2 and N2measurements (taken every 3-h) from a surface mooring at P26 and to the mean Aug-Sept O2/Ar-NCP for 2008 (15.5 ± 3.6 mmol C m−2 d−1) calculated by Howard et al.  using a mixed layer mass balance and discrete O2/Ar measurements over the region 42–50°N, 148–152°W.
 We can also compare our results to O2/Ar-NCP measurements made in this area in August 1988. Using the raw measurements reported inEmerson et al. and the 6-h NCEP/NCAR reanalysis wind speeds provided by the NOAA/OAR/ESRL PSD (Boulder Colorado, USA;http://www.esrl.noaa.gov/psd), we calculated O2/Ar-NCP in the same manner as for our own measurements. NCP values at Station R (located 3° north of P26, 53°N, 145°W) were the same within errors to our 2007–2009 summertime mean (14.0 ± 4.2 mmol C m−2 d−1). However, O2/Ar-NCP rates at P26 in August 1988 were much higher (29.1 ± 2.6 mmol C m−2 d−1), more like our August 2008 bloom measurements. Emerson  found clear interannual variability in the O2 saturation state at P26 between 1969 and 1978 that he attributes to interannual variability in NCP. From these studies, we suggest that productivity offshore along Line P shows some limited interannual variability in the summer.
 Both O2/Ar-NCP (Figure 5a) and ΔO2/Ar (Figure 3) at the most coastal station (P4) were far more variable, ranging from 6.7 to 54.9 (mean 27 ± 17) mmol C m−2 d−1, likely due to upwelling effects. Seasonal upwelling along the coast of Vancouver Island and transport of these waters offshore [Whitney et al., 2005] could potentially contribute low O2 waters to the surface and bias O2/Ar-NCP measurements low at P4. However, seasonal upwelling also replenishes surface nutrients, creating the potential for increased productivity. We found a high positive correlation (r > 0.91) between the June and August estimates of O2/Ar-NCP at P4 and a mean of the daily Upwelling Index at 48°N, 125°W over a period from the sampling date to 5–10 days prior. The relationship must be a result of productivity fuelled by the offshore transport of upwelled nutrients, which has had time to erase the low O2signal of the upwelled waters by air-sea gas exchange. The 5–10 day lag in the correlation between O2/Ar-NCP and the Upwelling Index is similar to the residence time of O2 in the mixed layer, the time over which the O2/Ar method integrates.
4.2. Incubation-Based Carbon and Nitrogen Uptake Rates
 Our measurements of new production from nitrate uptake (15N-NewP) and primary production from carbon uptake (13C-PP) in 2009 were of similar magnitude and exhibited similar weak seasonal and spatial trends to previous studies of15N-NewP [Emerson et al., 1993; Wheeler, 1993; Varela and Harrison, 1999; Varela et al., 2000] and 14C-PP [Welschmeyer et al., 1993; Boyd and Harrison, 1999; Boyd, 2000] along Line P. By integrating our incubation estimates to the base of the euphotic zone, we can compare to historical values measured along Line P during the Canadian Joint Global Ocean Fluxes Study (CJGOFS, http://ijgofs.whoi.edu/) (Figures 5b and 5c). This comparison shows that our measured rates are typical for this region, as is the greater variability seen at the most coastal station P4. The high 15N-NewP measured at P4 in February 2009 may be the result of the offshore transport of nutrient rich upwelled waters, as both the February and August 2009 sampling events at P4 took place after an upwelling event (high Upwelling Index at 48°N, 125°W).
 Our mixed layer integrated 15N-NewP and13C-PP summertime offshore rates (at stations P16 and P26 in 2009), which we will compare to mixed layer O2/Ar-NCP in the next section, were also relatively steady with averages of 7.1 ± 2.0 mmol C m−2 d−1 (n = 4) and 27.7 ± 7.6 mmol C m−2 d−1 (n = 4) respectively. Because 13C-PP includes production from all nitrogen sources, we expect it to exceed15N-NewP as observed. These compare well to offshore mixed-layer integrated rates measured at P26/R during the Subarctic Pacific Ecosystem Research program (SUPER) [seeMiller et al., 1991] (15N-NewP: 12.5 ± 9.2 mmol C m−2 d−1, n = 4; 14C-PP: 33.5 ± 15.0 mmol C m−2 d−1, n = 4). We also observed some limited interannual variability in our short data set. In August 2008, rates of 15N-NewP P26 were nearly ten times those measured the following August, though13C-PP did not show a similar difference (Figures 5b and 5c, open points). Like the O2/Ar-NCP results for the same period, we leave this discussion toSection 5.3. There was generally more variability in the historical data for 15N-NewP [Varela and Harrison, 1999; Peña and Varela, 2007] and 14C-PP [Boyd and Harrison, 1999] along Line P compared to our data set. Episodic iron inputs to the HNLC region, climate conditions as typified by the Pacific Decadal Oscillation (PDO), and offshore transport of recently upwelled waters for the near coastal station have all been show to affect productivity levels along Line P [Boyd et al., 1998; Peña and Varela, 2007]. These controlling factors vary significantly from year-to-year, leading to apparently higher or lower interannual variability in short time series.
5. Productivity Method Comparisons
 The three productivity methods that we compare in this study measure fundamentally different types of biological production, complicating expected relationships between them. O2/Ar-NCP is a measure of the net community productivity, taking into account both autotrophic and heterotrophic respiration in the mixed layer.15N-NewP is a measure of the uptake rate of nitrate, or ‘new’ nitrogen, by phytoplankton. If we assume that the only source of ‘new’ nitrogen (in the form of NO3−) to the mixed layer is supplied from the waters below (as a result of deep respiration of exported organic matter), then, averaged over an appropriate time scale, O2/Ar-NCP and15N-NewP should be equivalent [Falkowski et al., 2003]. 13C-PP is a measure of the uptake rate of inorganic carbon. This “primary production” is expected to fall somewhere between gross primary production (GPP), which does not account for any respiration, and net primary production (NPP), which accounts only for autotrophic respiration. Thus, we expect13C-PP to exceed both O2/Ar-NCP and15N-NewP, but by a potentially variable amount due to the fundamental differences between these methods.
5.1. Net Community Production Versus New Production
 Comparison of net community production (O2/Ar-NCP) and new production (15N-NewP) appeared to separate into two regimes depending on if productivity was high or low (Figure 6). This separation is supported by significantly higher surface chlorophyll concentrations for measurements that fell within the high-productivity regime (From this study: 5.24 and 1.38 mg Chl m−3 vs mean 0.34 ± 0.04 mg m−3 based on measurements made at the same stations and time of year). Strong correlations existed between O2/Ar-NCP and15N-NewP for all the summertime data in these two regimes, including August 1988 and 2008 (r = 0.93 for low productivity, r = 0.99 for high). One measurement made at P26 in late August 1988 appeared to fall at the intersection between these two regimes and was included in both regressions. Under the low-productivity conditions, O2/Ar-NCP tended to exceed15N-NewP by up to two times. A neutral regression through the data gives O2/Ar-NCP = 2.8(±0.3)*15N-NewP – 7.0(±3.1) (n = 8, both in carbon units). There is a greater than 98% probability that O2/Ar-NCP and15N-NewP were from different distributions (Mann–Whitney non-parametric test), showing that O2/Ar-NCP is not equivalent to15N-NewP under low productivity conditions. At higher production rates, O2/Ar-NCP was more similar to15N-NewP with a neutral regression of O2/Ar-NCP = 0.43(±0.03)*15N-NewP + 21(±0.9) (n = 4, both in carbon units). There is less than 10% probability that O2/Ar-NCP and15N-NewP in the higher productivity regime are from different distributions (Mann–Whitney non-parametric test), again suggesting closer equivalence.
 O2/Ar-NCP and15N-NewP measure the rates of different processes (Table 1), but averaged over large enough temporal and spatial scales, or at true steady state, these rates should yield equivalent results [Falkowski et al., 2003]. Biological O2 should only be available for exchange to the atmosphere if there is net production of organic matter. Light inhibits the conversion of respiratory NH4+ to NO3− in the euphotic zone, so uptake of 15NO3 should represent production of organic matter that is available for export. Typically, this equivalence is only expected on seasonal to annual time scales, but we find strong relationships on much shorter timescales.
 A possible explanation for the two relationships observed in this comparison of O2/Ar-NCP and15N-NewP stems from whether conditions were iron-limited or iron-replete. The low-productivity conditions in our data set were dominated by measurements at the offshore stations where low iron availability typically limits productivity [Boyd et al., 1996]. When O2/Ar-NCP and15N-NewP rates are low, O2/Ar-NCP tends to exceed15N-NewP. The uncertainty in O2/Ar-NCP is dominated by uncertainties in the gas transfer velocity, which is unlikely to be particularly biased high during low productivity periods. On the other hand, previous studies have found evidence for underestimates of15N-NewP due to the excretion of labeled15N as dissolved organic nitrogen (DON) during periods of suboptimal growth [Bronk and Ward, 2000]. Bronk and Ward found that phytoplankton under physiological stress and grazing pressure in Monterey Bay, California produced higher rates of DON release. Under the typical, iron-limited conditions at the offshore stations along Line P, phytoplankton consist primarily of small-cell organisms tightly controlled by microzooplankton grazing, rather than iron availability [Miller et al., 1991]. So, these offshore stations may have been more likely to experience enhanced release of labeled-DON. One near-coastal measurement, at P4 in June 2009, also fell within our low-productivity regime. Productivity at P4 is not iron limited, but rather is dependent on seasonal upwelling replenishing surface nutrients [Whitney et al., 2005]. Nitrate concentrations at P4 in June 2009 were below the detection limit and silicate concentrations were an order of magnitude lower than in August 2009. This suggests that phytoplankton at P4 in June 2009 were under nutrient stress, another factor Bronk and Ward  suggested could lead to increased rates of DON release. A compilation of field measurements showed a mean DON release of 35 ± 29% of the total 15NO3 uptake, with some values as high as 85% [Bronk and Ward, 2000]. Given that a 50% DON release would cause 15N-NewP to be half the correct value, our comparison with O2/Ar-NCP falls within this range. In addition, our incubation length (24-h) was longer than most of the incubations in the DON release studies, which should bias our15N-NewP even further as grazing continues at night while uptake has stopped.
 The high-productivity measurements in our data set consisted of near-coastal or bloom conditions that could be expected to be iron-replete. In these cases, O2/Ar-NCP was more similar to15N-NewP. The effects of DON release on15N-NewP might not be as pronounced under high productivity conditions. If the high rates result from a bloom of larger celled phytoplankton (iron fertilization primarily enhances large diatoms in this region [Boyd et al., 1998]), their macro-zooplankton predators, which have much slower growth rates, may not be abundant, decreasing the effects of grazing, especially when the bloom first initiates.
 Other potential biases that may affect the relationship between O2/Ar-NCP and15N-NewP are the differences in the integration times of the methods, heterotrophic bacterial uptake of15NO3−, and mixed layer nitrification. The O2/Ar-NCP method integrates over the residence time of O2 in the mixed layer (∼2 weeks) for O2/Ar-NCP, while the15N-NewP method integrates over the 24-h period of the incubation. Where growth rates are rapidly changing, O2/Ar-NCP should tend to lag behind15N-NewP. Bacterial growth rates along Line P are lower than those of phytoplankton [Varela and Harrison, 1999; Sherry et al., 1999]. However, if heterotrophic bacteria are iron-limited in this region, as some evidence suggests [Tortell et al., 1996], greater bacterial 15NO3− uptake during blooms could result in 15N-NewP being biased high. Finally, euphotic zone nitrification (oxidation of NH4+ into NO3−) can cause 15N-NewP to reflect other sources of nitrate than from the deep ocean, leading to an overestimate of new production. However, this bias was likely small given that our discrete measurements of15N-NewP were highest at the surface where nitrification rates range from very low to undetectable due to photoinhibition [Ward et al., 1982; Dore and Karl, 1996; Yool et al., 2007].
 It should be noted that for three of the four O2/Ar-NCP estimates from 1988, our recalculated values are much higher than those reported byEmerson et al. . This results primarily from differences in estimates of the gas transfer velocity. Emerson et al. used independent measurements of the gas transfer velocity based on radon profiles, while we used a 60-day weighted average of the gas transfer velocity based on the wind speed parameterization ofHo et al. and the 6-h NCEP/NCAR Reanalysis wind speeds. Gas transfer velocities from radon measurements can be biased by analytical uncertainties, complex mixing histories, internal waves affecting the apparent depth of the mixed layer, and lateral processes, while NCEP Reanalysis winds also contain significant uncertainties. However, reanalysis of historical radon-derived gas transfer velocities generally support theHo et al.  wind speed parameterization [Bender et al., 2011]. We have no particular reason to discount the 1988 radon measurements, so we also present the original O2/Ar-NCP calculated with these values (Figure 6). Use of these values instead of our recalculated ones lowers the correlation of O2/Ar-NCP to15N-NewP (r = 0.74 for low productivity, r = 0.92 for high), but does not change our conclusion that O2/Ar-NCP tends to exceed15N-NewP under low productivity conditions and be more similar to each other under high productivity conditions.
 The variability and the range of O2/Ar-NCP:15N-NewP ratios we observe is consistent with other regions of the oceans.Reuer et al.  compared their measurements of O2/Ar-NCP with previous historical measurements of euphotic zone integrated15N-NewP from 24-h incubations in the Southern Ocean, another major HNLC region. Their mean rate of O2/Ar-NCP measured in the subantarctic and polar frontal zones was higher than in this study (39.2 mmol O2 m−2 d−1 for 46°S–60°S versus 18.9 mmol O2 m−2 d−1 for 48°N–50°N), while their mean O2/Ar-NCP:15N-NewP ratio (1.4 ± 0.3, n = 2) fell between the low and high-productivity ratios we observed.Hamme et al.  found more variable O2/Ar-NCP:15N-NewP ratios from 1 to as high as 4 in the Southern Ocean following a very unusual period of net heterotrophy in the surface ocean, indicating that the relationship is probably just as variable in the Southern Ocean.
5.2. Primary Production Versus Net Community Production/New Production
 Net Community Production, equivalent to carbon export, is a key flux in the global carbon cycle. However, productivity estimates in most regions are dominated by 14C- or13C-based primary production (14C-PP or13C-PP), with relatively few measurements of NCP. Identifying a consistent relationship in this region between NCP and C-PP would allow estimates of NCP to be derived from the more common C-PP measurements. To extend our comparisons, we combine our new data with O2/Ar-NCP, mixed-layer-integrated15N-NewP, and mixed-layer-integrated14C-PP estimates made at concurrent stations and times during summer 1988 for Project SUPER [Welschmeyer et al., 1993; Emerson et al., 1993]. We treat estimates of carbon-based productivity as equivalent for 24-h13C and 14C incubations [Slawyk et al., 1984]. In making these comparisons, we must also consider that NCP and C-PP are fundamentally different measures of productivity, with O2/Ar-NCP accounting for both autotrophic and hetereotrophic respiration and13C-PP accounting for some proportion of autotrophic respiration. Thus, a weak relationship between these different measures of productivity might be expected as a result of differences in respiration or methodological issues.
 Late spring through early fall rates of O2/Ar-NCP and C-PP, excluding the August 2008 event, were strongly correlated (r = 0.85), as were rates of15N-NewP to C-PP (r = 0.95) (Figure 7). The mean ratio of O2/Ar-NCP to C-PP during this time was 0.52 ± 0.17 (n = 10). This is higher than the mean ratio of15N-NewP to C-PP of 0.30 ± 0.08 for this period, which we expect given the O2/Ar-NCP to15N-NewP ratio of greater than one for many of these data. These comparisons reflect incubation data integrated only to the base of the mixed layer, because O2/Ar-NCP is a mixed layer estimate. If we compare15N-NewP to C-PP integrated to the base of the euphotic zone (and including available data for this integration from CJOGFS but not SUPER), we find a ratio of 0.27 ± 0.09, similar to the mixed-layer estimate.
 Strong correlations among these rates are expected at the offshore stations considering that C-PP, like O2/Ar-NCP and15N-NewP exhibits steady rates at the offshore stations during May through September (Figure 5c). However, we also find that our limited coastal data (two sampling events at station P4) follow a similar trend to the offshore stations in terms of their O2/Ar-NCP to C-PP and15N-NewP to C-PP relationships. The summertime O2/Ar-NCP:C-PP ratio is 0.52 with or without inclusion of the coastal data, and the15N-NewP:C-PP ratio is 0.30 including coastal data versus 0.32 without. On the other hand, for the August 2008 bloom event at P26, we find a higher ratio of O2/Ar-NCP to C-PP (near one) and a significant decoupling of15N-NewP from C-PP (Figure 7 and next section). We also find that February measurements at the coastal and offshore stations had a somewhat higher ratio of 15N-NewP to13C-PP, though the small number of data points limits our conclusions.
 Based on the available data, we conclude that 0.52 ± 0.17 is the most appropriate value to convert summertime measurements of mixed-layer-integrated C-PP to O2/Ar-NCP in the subarctic NE Pacific. The May–Sept15N-NewP:C-PP ratio is lower than the May–Sept O2/Ar-NCP:C-PP ratio that we are recommending, but as discussed in the previous section, it appears likely that15N-NewP was underestimated due to DON release under the lower productivity conditions. However, as the comparison to the February data and special August 2008 event illustrates, the ratio of both O2/Ar-NCP and15N-NewP to C-PP is subject to significant variability, so this conversion ratio should be used with care.
 At Station ALOHA (23°N 158°W), in situ mixed layer measurements of O2/Ar-NCP were compared with 24-h, incubation-based, 0–100 m integrated estimates of14C-PP to yield a mean annual O2/Ar-NCP to14C-PP ratio of 0.28 ± 0.09 [Juranek and Quay, 2005; Quay et al., 2010]. Note that we converted from 12-h (dawn-dusk)14C-PP measurements to 24-h using a factor of 1/1.15 followingKarl et al. . Our observed O2/Ar-NCP:C-PP ratio along Line P was nearly twice this value, but the difference is comparable to the difference between the typical f-ratios (including urea-based production) observed in each region (∼0.1 at ALOHA [Karl, 1999] and ∼0.2 along Line P [Varela and Harrison, 1999]).
5.3. Decoupling of Relationships During Bloom Conditions in August 2008
 In 2008, ash deposition across the subarctic NE Pacific from the 8–9 August eruption of Kasatochi volcano in the Aleutian Islands fueled a wide-scale iron fertilization and anomalous plankton bloom [Hamme et al., 2010]. Briefly, particle transport models based on prevailing winds indicate that the ash was likely deposited along Line P on 11–12 August 2008. Satellite chlorophyll levels were the highest ever recorded by satellite in the wider region, while changes in pCO2 and pH of the surface water measured by the mooring at P26 established that the bloom began about two days after ash deposition.
 Our measurements of O2/Ar-NCP, included inHamme et al. , were 2–3 times greater at P20 and P26 than the mean August 2007/09 measurements for these stations and 1.6 times greater (45.9 vs 29.1 mmol C m−2 d−1) than the highest measurement made at P26 in 1988 (Figure 5a). Both the incubation based 13C-PP and15N-NewP at P26 in August 2008 were greater than those measured in August 2009 (Figures 5b and 5c); however, the intensification in 15N-NewP was much greater than13C-PP, with rates of15N-NewP nearly 10-times greater than in August 2009. Rates of13C-PP were higher, but not significantly different from those measured the following August (Figure 5c). During this period, all three measures of productivity were very similar. While we might expect such a similarity between O2/Ar-NCP and15N-NewP (Section 5.1), it is surprising that 13C-PP also gives a similar rate of production.Dickson et al.  observed a similar disparity between rates of total production (15N-NewP + 15NH4+based regenerated production) and 24-h14C-PP in the Arabian Sea that they attribute to heterotrophic bacterial uptake of NO3− and NH4+. However, we observed higher rates of O2/Ar-NCP comparable to15N-NewP during this event. Given the fundamental differences between these methods, the similarity we observe between our three measures of productivity suggests that13C-PP is biased low, as we would expect13C-PP to exceed both O2/Ar-NCP and15N-NewP.
 Both methodological (e.g., DO13C release) and fundamental (e.g., decreased heterotrophic respiration) effects may have biased rates of 13C-PP low. The production and excretion of labeled DOC may lead to underestimates of C-PP, though this bias will be reduced by the proportion taken up by any bacteria retained on the filter [Bjørnsen, 1988]. One of the inherent assumptions in the 13C (or 14C) based method is that DO13C (DO14C) production is negligible in the range of 5–15% of measured C-PP [Williams and Lefévre, 2008]. Considering that Kirchman et al. found evidence of DOC limitation on bacterial communities at P26, it seems unlikely that production of labeled DOC would significantly affect our measurements of C-PP under normal conditions. However, that we do not see significantly higher rates of C-PP in August 2008 compared to the following year could suggest an increased loss of the13C-label as DO13C in this special case. In an experimental mesocosm diatom bloom, DOC and DO13C concentrations increased throughout the experiment, though these changes were minimal until the bloom began to decline [Norrman et al., 1995]. Significantly higher surface ammonium concentrations and oxidation rates measured at P26 in August 2008 [Hamme et al., 2010] indicate that P26 was sampled during at least mid-bloom conditions and possibly later. If the bloom had already begun to decline when sampling occurred, the small difference we see when comparing13C-PP in August 2008 to 2009 could have resulted from an increased loss of13C as DO13C in 2008, rather than primary productivity actually being anomalously low in comparison to our other August 2008 measurements. It is also possible that shifts in the amount of respiration or reduction in the dominance of heterotrophic respiration relative to autotrophic respiration may have caused 13C-PP to be similar to O2/Ar-NCP and15N-NewP, especially considering that diatoms dominated the bloom [Hamme et al., 2010], being too large to be efficiently grazed by the microzooplankton normally found near P26 [Miller et al., 1991].
 It is also possible that methodological issues with the 15N-NewP and O2/Ar-NCP measurements may have biased our results. The extremely high15N-NewP rates may partially result from overestimation of15N-NewP due to increased euphotic zone nitrification and heterotrophic bacterial uptake of15NO3−. Higher surface ammonium concentrations and oxidation rates at P26 in August 2008 [Hamme et al., 2010] do give an indication of increased euphotic zone nitrification during this high productivity event. It is also possible that the bloom may have led to increased heterotrophic bacterial uptake, though Kirchman et al.  found evidence that heterotrophic bacteria are limited by the supply of dissolved organic carbon at P26. Due to the integration time of the O2/Ar-NCP method, these measurements may have been biased low if we had sampled the bloom before a full residence time of O2 in the mixed layer had passed; however, given that we sampled P26 ten days after ashfall, about the residence time of O2 in the mixed layer, we assume our measurements of O2/Ar-NCP are representative of the bloom conditions.
 We identified strong, consistent relationships between our measurements of in situ O2/Ar-NCP and our incubation based measurements of15NO3− based NewP under all conditions sampled and with 13C based C-PP under more restricted conditions of summertime, offshore, non-bloom conditions. In situ estimates of O2/Ar-NCP and incubation-based 24-h15N-NewP from June to August exhibited different relationships that appeared to depend on whether waters were experiencing high or low productivity conditions. Under low productivity conditions, O2/Ar-NCP tended to exceed15N-NewP, which we attribute to an underestimation of15N-NewP resulting from DON release. These methods were more similar under high productivity conditions. The variability in this ratio is similar to that observed byHamme et al.  in the Southern Ocean, suggesting that relationships between new production and NCP can be complicated by methodological biases dependent on the conditions. We found a mean O2/Ar-NCP:C-PP ratio for May–Sept at offshore stations along Line P including historical data of 0.52 ± 0.17. The greater variability in this relationship suggests that C-PP can only be related to O2/Ar-NCP or15N-NewP under particular conditions. The ratio we calculate is likely only appropriate in non-bloom conditions for the May–Sept period.
 Our comparisons between in situ and incubation based methods used to quantify biological carbon cycling in the surface ocean demonstrate the importance of evaluating the relationships between productivity methods. These results show that relationships can be found over a range of productivity regimes and environmental conditions, even between methods that integrate on very different temporal and spatial scales. Better understanding of these relationships and the processes which affect them will improve our understanding of the mechanisms responsible for biological carbon export to the deep ocean, help us interpret newer methods for measuring productivity such as satellite-based measurements, and aid in predicting the response of the ocean biological pump to future climate change.
 Our measurements of O2/Ar-NCP along Line P also illustrated the importance of evaluating the contribution of diapycnal mixing to the O2/Ar mass balance when calculating NCP. The effect of diapycnal mixing is typically neglected to allow for a simple calculation of in situ NCP from the diffusive gas exchange flux. We found that the depth of the oxycline in relation to the depth of the mixed layer could be used to diagnose the importance of diapycnal mixing to the O2 mass balance. Ideally, accurate estimates of the eddy diffusion coefficient (Kz) based on microstructure measurements would be used to quantify mixing. However, when these measurements are not available, we suggest that the potential importance of diapycnal mixing to the O2 mass balance can be evaluated by calculating the eddy diffusivity needed to balance the diffusive gas exchange flux. If the value calculated falls within a reasonable range for the region studied, diapycnal mixing should not be excluded from the O2 mass balance, and NCP may not be constrainable from O2/Ar measurements.
 We thank the scientists and crew, especially Chief Scientist Marie Robert, on the Line P cruises for their assistance with sample collection. We also thank Charles Stump and Seth Bushinsky for dissolved gas sample collection and analysis. DIC, NO3−and chlorophyll sample collection and analysis was carried out by scientists at the Institute of Ocean Sciences (Fisheries and Oceans Canada). Historical data along Line P is maintained and provided online by Fisheries and Oceans Canada. The daily upwelling index is provided by the Pacific Fisheries and Environmental Laboratory (NOAA). Funding for this work was provided by NSERC Discovery Grant 328290-2006 to R.C.H. and by NSF grant OCE-1029299 to S.R.E.