Rates of net community production (NCP) and air-sea CO2 flux in the Northeast Pacific subarctic, transition zone and subtropical regions (22°N–50°N, 145°W–152°W) were determined on a cruise in August–September 2008 by continuous measurement of surface values of the ratio of dissolved oxygen to argon (O2/Ar) and the partial pressure of CO2 (pCO2). These estimates were compared with simultaneous measurements of sea surface temperature (SST), chlorophyll-a (chl-a), flow cytometry, and discrete surface nutrient concentrations. NCP and CO2 influx were greatest in the subarctic (45°N–50°N, 25.8 ± 4.6 and 4.1 ± 0.9 mmol C m−2 d−1) and northern transition zone (40°N–45°N, 17.1 ± 4.4 and 2.1 ± 0.5 mmol C m−2 d−1), with mean NCP ∼6–8× greater than mean CO2 invasion (error estimates reflect 1 σ confidence intervals). Contrastingly, the southern transition zone (32°N–40°N) and subtropics (22°N–32°N) had lower mean NCP (5.4 ± 1.8 and 8.1 ± 2.1 mmol C m−2 d−1, respectively) and mean CO2 efflux (3.0 ± 0.5 and 0.1 ± 0.0 mmol C m−2 d−1, respectively). In the subarctic and transition zone, NCP was highly correlated with surface chl-a and CO2 influx, indicating strong coupling between the biological pump and CO2 uptake. Meridional trends in our NCP estimates in the transition zone and subtropics were similar to those for integrated summertime NCP along the cruise track determined using an upper ocean climatological carbon budget.
 Surface CO2 measurements yield a mean global net ocean sink for atmospheric CO2 of ∼2 Pg C yr−1, however there is large variation in oceanic CO2 uptake and release in both space and time [Takahashi et al., 2009] that is controlled by biological drawdown through primary productivity (PP) and export to the deep ocean (the biological pump), temperature-driven solubility of CO2 (the solubility pump) and ocean circulation. The biological pump has been estimated to export organic carbon globally at 8–15 Pg C yr−1 [Emerson and Hedges, 2008], but these rates are poorly constrained and vary widely across ocean regions [Boyd and Trull, 2007; Schlitzer, 2000, 2004]. To predict the response of this sink to future changes in Earth's climate, we must understand how the biological and solubility pumps vary spatially, seasonally and interannually, and what factors determine this variability.
 A remarkable feature in the surface CO2of the North Pacific is the band of strong mean annual air-sea CO2 uptake (2–9 mol C m−2 yr−1) at ∼30°N–45°N, roughly corresponding to the transition zone between the subtropical and subarctic gyres along the path of the Kuroshio current and extension [Takahashi et al., 2009]. Two prominent basin-wide oceanographic features occupy this region: (1) the physically defined North Pacific transition zone (∼30°N–45°N), a region of confluence and convergence bounded by temperature and salinity fronts [Roden, 1991], and (2) the transition zone chlorophyll front (TZCF), a biological hot spot defined by the 0.2 mg m−3 chlorophyll isopleth that seasonally migrates from 30°N–35°N in winter to 40°N–45°N in summer [Polovina et al., 2001].
 Given the biological prominence of the TZCF and the enhanced PP found at similar convergent fronts [Longhurst, 2007; Mann and Lazier, 2006], one would expect that PP and carbon export contribute substantially to the strong CO2 uptake in this region, but the magnitude of this contribution is unclear. Takahashi et al. [2002, 2009] attribute the sink to the combined effects of the solubility and biological pumps. In contrast, Ayers and Lozier  recently modeled CO2uptake in the transition zone and concluded that although temperature-driven solubility strongly affects seasonalpCO2 trends, it controls only ∼17% of the annual sink in this region, with the remainder contributed by biological export and geostrophic divergence of DIC.
 In this study, we sought to estimate the contribution of the biological pump to carbon uptake in the North Pacific by making high-resolution estimates of carbon export and uptake. The strength of the biological pump is quantified by estimates ofNCP, which equals gross PP (GPP) minus community respiration and is equivalent at steady state to the rate of organic carbon export and transfer up the food chain. In the North Pacific, NCP is fairly well constrained at two time series stations, Ocean Station P (OSP) in the subarctic NE Pacific (50°N, 145°W) and Station ALOHA in the subtropical NE Pacific (22°45N, 158°W) through surface layer budgets of dissolved gases, nutrients, DIC and δ13C, and through in situ dissolved gas ratios and 234Th-238U disequilibrium (see Table 1). But because there are comparatively fewer estimates in the rest of the basin and only a few snapshot NCP estimates for the transition zone [Wong et al., 2002b; Juranek, 2007; Howard et al., 2010; Juranek et al., 2012], it is unclear how representative these sites are of the basin as a whole. On a basin-wide scale,NCP estimated from climatological seasonal DIC drawdown [Lee, 2001] is high (2–6 mol C m−2 yr−1) in the transition zone.
Table 1. Upper Ocean NCP in NE Pacific Subarctic, Subtropics, and Transition Zonea
Annual NCP (mol C m−2 yr−1)
Daily NCP (mmol C m−2 d−1)
See footnotes for seasonality of daily estimates. Values set in bold type are mean ± s.d. of previous observations cited below.
 In this study, we continuously measured surface ocean O2/Ar, pCO2and chl-a and carried out discrete flow cytometry in the subarctic, transition zone and subtropical Northeast Pacific to determine the heterogeneity in the rates of the biological pump and air-seaCO2flux on a cruise (August–September 2008) that crossed the transition zone three times (Figure 1). We compare these results with previous NCP estimates at OSP and Station ALOHA and surface CO2 climatology, and examine the influence of NCP on CO2 uptake in the region.
 The main circulation features of the Northeast Pacific are the cyclonic subarctic gyre centered in the Gulf of Alaska (∼45°N–60°N, 160°W–135°W), the anticyclonic subtropical gyre (∼15°N–35°N, 135°E–135°W), and the transition zone between the two. In the subarctic Alaska gyre, a halocline limits winter mixed layer depths to ∼90–120 m and strong seasonal variation in temperature and wind speed leads to stratification in summer, with mixed layers shoaling to ∼40 m [Harrison et al., 2004]. In this high nitrate, low chlorophyll (HNLC) region, iron generally limits the growth of large microphytoplankton like diatoms [Harrison et al., 1999; Boyd et al., 1996], and the phytoplankton community is dominated by small cells <5 μm.
 The subtropical gyre is a low-nutrient, low-chlorophyll region with lower seasonal variability in temperature, wind speed, productivity and light than in the subarctic gyre; mixed layer depth ranges from 40 m in summer to 100 m in winter [Karl, 1999; Keeling et al., 2004]. Prochlorococcus is the dominant photoautotroph, but N2 fixation by other cyanobacteria may provide up to half the nitrogen responsible for export [Karl et al., 1997, 2001].
 Between these two gyres, the North Pacific transition zone extends across the basin at ∼30°N–45°N. Confluence of the Kuroshio and Oyashio currents and Ekman convergence set up two frontal regions with strong meridional gradients in temperature and salinity in winter and salinity in summer, the subarctic frontal zone (SAFZ, defined by 33‰–33.8‰ isohalines, ∼40°N–43°N) and the subtropical frontal zone (STFZ, defined by 34.8‰–35.2‰ isohalines, ∼31°N–34°N [Roden, 1991]) (Figure 1). On shorter timescales, these frontal zones are associated with highly temporally and spatially variable mesoscale perturbations including meandering fronts, jets and eddies [Roden, 1991; Yuan and Talley, 1996].
2.2. Previous Estimates of NCP in the Northeast Pacific
 Notably, despite different limitations to biological productivity at these two sites, annual NCP is similar at 2.0 ± 0.3 mol C m−2 yr−1 at OSP and 2.7 ± 0.9 mol C m−2 yr−1 at ALOHA (mean ± s.d. of multiple estimates), within the ±50% uncertainty of the measurements (Table 1) [Emerson et al., 2008]. However, at OSP, NCP is primarily confined to spring and summer at high daily rates (14 ± 5 mmol C m−2 d−1), whereas at ALOHA, daily rates are lower (8 ± 4 mmol C m−2 d−1) but NCP takes place throughout the year (mean ± s.d. of multiple studies, Table 1).
 Few studies have examined NCP in the transition zone. Wong et al. [2002b] estimated NCP at 2.6 mol C m−2 yr−1 and 0.9 mol C m−2 yr−1 based on seasonal NO3 drawdown in two regions (40°N–45°N and 35°N–40°N, respectively). Summertime daily NCP for these two provinces was equal at 14–15 mmol C m−2 d−1, but the southern region had a much shorter export season. Juranek  and Juranek et al.  estimated daily NCP in the transition zone mixed layer (∼32°N–42°N) of 21 and 13 mmol C m−2 d−1 in spring and summer, respectively, and Howard et al.  (30°N–45°N) estimated rates of 8 mmol C m−2 d−1 in summer and 3 mmol C m−2 d−1 in fall, based on O2/Ar measurements (Table 1).
2.3. O2/Ar Method for Estimating NCP
 The saturation level of the O2/Ar gas ratio in the mixed layer coupled with air-sea gas exchange rate yields quantitative estimates ofNCP that are mostly insensitive to nonbiological gas saturation processes (e.g., warming, bubble injection) because argon is an inert analog of oxygen, as described previously [Craig and Hayward, 1987 and Emerson et al., 1991, 1997].
 Briefly, the biological O2 supersaturation, which quantifies the influence of NCP on the O2 budget, is defined as:
where [O2/Ar]msr is the measured dissolved O2/Ar gas ratio, and [O2/Ar]sat is the ratio expected at saturation with air (based on temperature and salinity dependence of O2 and Ar solubility [García and Gordon, 1992; Hamme and Emerson, 2004]). The percent biological O2 supersaturation (%O2bio) equals (ΔO2/Ar)*100.
NCP is calculated using O2 and Ar mass balances for the mixed layer, typically assuming steady state and negligible physical supply, yielding:
where kO2 is the gas transfer velocity of O2 (m d−1) and [O2]sat is the concentration of O2 at saturation (mol m−3) [García and Gordon, 1992]. We determined kO2 using winds from QuikScat, the Ho et al. wind speed parameterization using a 60-day time-weighting technique [Reuer et al., 2007], and the temperature- and salinity-dependent Schmidt number [Wanninkhof, 1992]. To convert O2-basedNCPto C-basedNCP, we used a ratio of 1.4 O2: 1 C (export or new production) [Laws, 1991]. This estimate of NCP integrates over the residence time of O2 in the mixed layer (i.e., (mixed layer depth)/kO2), which was 8–14 days. The strong vertical stratification and low winds observed during the cruise favored a negligible impact of vertical mixing on the mixed layer O2 budget.
3.1. Underway and Discrete Sampling and Cruise Track
 We continuously measured surface O2/Ar, pCO2, temperature, salinity and fluorescence from the underway system of the R/V Thompson (at ∼5 m intake) on a cruise betwen Station P and Honolulu, Hawaii (30 August–15 September 2008; Figure 1). Temperature and salinity were measured using the ship's TSG. Underway chl-a was estimated based on continuous fluorometer measurements calibrated with discrete chl-a samples measured using standard methods on a shipboard Turner fluorometer [Strickland and Parsons, 1972]. Mixed layer depth was determined using CTD profiles along the ship's cruise track at a resolution of ∼1°–2° latitude as a density increase of 0.125 kg m−3 from the surface value. Concentrations of nitrate, phosphate and silicate were determined on discrete surface samples collected by Niskin using a shipboard autoanalyzer (Technicon AutoAnalyzer II) and standard methods [Intergovernmental Oceanography Commission, 1994].
3.2. Continuous O2/Ar Measurements Using EIMS
 We continuously measured surface O2/Ar using an EIMS system similar to that described in Cassar et al. . Water from the ship's underway line was pumped at ∼2 L min−1 through a coarse filter to screen out particulates into a 1 L graduated cylinder on which the equilibrator cartridge (Membrana MicroModule G569, 0.75″ × 1″) was mounted. Water from the graduated cylinder was pumped through a 5 μm filter sock sewed inside a 100 μm filter sock (1.5″ × 12″, McMaster-Carr), and then through the cartridge at ∼100 ml min−1. To prevent biofouling, the coarse filter was cleaned and the sock filter was replaced daily, and the equilibrator cartridge was replaced weekly. All tubing used was Tygon silver (antimicrobial).
 Headspace gas from the equilibrator was delivered through an 0.05 mm fused silica capillary to the quadrupole mass spectrometer (Pfeiffer Prisma QMS), which was kept at a constant temperature (40° ± 2°C). Individual ion currents (O2, Ar) were measured at 1-s intervals, and the ratio of O2/Ar currents was averaged to a 10-min timescale. The EIMS system e-folding response time is 7 min [Cassar et al., 2009], yielding a spatial resolution of ∼2 km at the average ship speed of 10 knots.
 EIMS-based O2/Ar measurements were calibrated using the O2/Ar of air measured by the EIMS, and O2/Ar measured by isotope ratio mass spectrometer (IR-MS) on discrete samples collected from either the underway line or surface (5 m) Niskin bottles approximately every 6–8 h following the collection procedures ofEmerson et al.  and mass spectrometer procedures of Juranek and Quay  in the University of Washington Stable Isotope Lab. The percent error ([s.d./mean]*100) in O2/Ar of discrete air standards used for IR-MS calibration was 0.2% and that of duplicate samples was 0.1%. The O2/Ar of discrete samples measured simultaneously on the ship's underway line and from surface Niskins compared well (0.1% error), showing no evidence of O2 consumption in the underway line.
 We estimated error in NCPusing a Monte-Carlo approach assigning uncertainty in the following terms:kO2 ± 15% (twice the spread in kO2 values predicted by three parameterizations recently validated by Ho et al. : Nightingale et al. , Ho et al. , and Sweeney et al. ) and O2/Ar ± 0.06 (s.d. of mean in offset between EIMS and discrete samples). Mean NCP percentage error ([s.d./mean]*100) was 18% in the subarctic, 27% in the transition zone and 25% in the subtropics.
3.3. Measurements of pCO2 and Estimate of CO2 Flux
 We continuously measured the pCO2in surface water from the ship's underway line and air from the bow with an automated IR-detection-based underwaypCO2 measuring system that has been described in detail elsewhere [Feely et al., 1998; Pierrot et al., 2009]. We calculated the air-seapCO2 gradient (ΔpCO2) as ΔpCO2 = pCO2SW − pCO2atm, where pCO2SW is the calculated pCO2 of the seawater (after correction for water temperature difference between in situ and the pCO2 equilibrator) and pCO2atm is the measured atmospheric pCO2. After correction, the accuracy of the data is within 0.1 μatm for pCO2atm and 2 μatm for pCO2SW [Pierrot et al., 2009], and the equilibrator integration time is 30–45 s. pCO2data were averaged to match the 10-min integration time of the EIMS data.
 The air-sea flux of CO2 is calculated based on ΔpCO2 and the solubility (kH) [Stumm and Morgan, 1996] and gas transfer velocity of CO2 (kCO2):
Defined in this way, CO2flux from the atmosphere into the ocean (influx) is negative, and efflux is positive. kCO2 was calculated as described above for kO2 using the Ho et al. wind speed parameterization with time-weighting [Reuer et al., 2007], and the Schmidt number for CO2 [Wanninkhof, 1992].
3.4. Flow Cytometry Measurements
 Cell counts of phytoplankton groups were collected from the flow-through system of the R/V Thompson at various intervals along the ship track or from surface (5 m) Niskins. Samples were preserved in liquid nitrogen and processed later in the lab. Cells were enumerated and classified using a FACS-Caliber flow cytometer [Sherr et al., 2005]. Following the protocol of Sherr et al. , prokaryotes were stained with SYBR-green;Prochlorococcus counts were subtracted from heterotrophic bacteria counts. We binned data into microphytoplankton (10–60 μm), nanophytoplankton (2–10 μm) and picophytoplankton (1–2 μm); the picophytoplankton fraction was dominated by Synechococcus north of the TZCF and Prochlorococcus south of it.
3.5. Climatological pCO2, CO2 Flux, Alkalinity and DIC
 We used a recent climatological surface pCO2 data set [Takahashi et al., 2009] to compare our pCO2 measurements with the seasonal cycle of pCO2 and DIC over our cruise track (section 5.3). This data set includes monthly pCO2SW, pCO2atm, SST, salinity and wind speed data at resolution of 4° latitude × 5° longitude, with pCO2 data normalized to reference year 2000. We calculated climatological monthly CO2flux using the climatological pCO2SW, pCO2atm, temperature, salinity, wind speed and Ho et al.  parameterization of gas transfer velocity and compared it to our continuous CO2flux estimates in September 2008 (normalized to year 2000).
 To estimate the seasonal climatological DICcycle for the region covered on our cruise, we used the climatological temperature and salinity data to calculate monthly alkalinity with the latitude-dependent parameterization ofLee et al. , and used these alkalinity values and pCO2 to calculate monthly DIC with the program CO2sys [Lewis and Wallace, 1998]. We used monthly climatological mixed layer depth from Monterey and Levitus at 1° resolution, bin-averaged to the 4° latitude × 5° longitude boxes of theTakahashi et al.  data set.
4.1. Regional Setting
 We categorize our results into three regions: the subarctic (45°N–50°N), transition zone (32°N–45°N) and subtropics (22°N–32°N). Sea surface temperature and salinity generally increased and density decreased from the subarctic to the subtropical gyre, with a local density minimum at ∼35°N within the transition zone (Figure 2). The strong gradients in surface salinity (and weaker SST gradients) of the SAFZ and STFZ are apparent (Figure 2) at ∼38°N–43°N, and ∼32°N–37°N, respectively, extending slightly farther than their climatological positions (40°N–43°N and 31°N–34°N, respectively). The strongest surface density gradients were in the northern transition zone (40°N–45°N) and subtropics (22°N–24°N) (Figure 2). The TZCF (location of the 0.2 mg m−3chl-a isopleth) ranged from 42°N to 44°N within the northern transition zone (Figure 3), consistent with its climatological late summer position (40°N–45°N) [Polovina et al., 2001].
 Mixed layer was relatively shallow, reflecting late summer conditions; mean depths decreased from 32 m in the subarctic to a local minimum (26 m) in the transition zone, and then increased to 51 m in the subtropics (Figure 2). kO2 was fairly low (3–5 m d−1), and was highest in the subarctic and northern transition and lowest in the southern transition zone (Figure 3). Due to variation in mixed layer depth and kO2, the O2 residence time in the surface layer increased southward from a mean of 8 days in the subarctic and transition to 14 days in the subtropics.
 In the subarctic, we observed high biological O2 saturation (%O2bio) and chl-a and low ΔpCO2, with high meridional and zonal variability (Figure 3). %O2bio and chl-a were highly correlated. pCO2 was highly negatively correlated with %O2bio and chl-a, and only weakly correlated with SST (Table 2). At the southern edge of the subarctic along 145°W, strong peaks in %O2bio and chl-a were coupled with troughs in ΔpCO2 (Figure 3). Surface nutrient concentrations at OSP (6 μM nitrate and 4 μM silicate, respectively) were lower than typical late summer concentrations of 9 μM nitrate and 14 μM silicate [Harrison et al., 2004] (Figure 3).
Table 2. Mean Values and Correlation Statistics for Underway Measurements
Values for NCP and CO2flux are mean ± error of 1 σ (regional s.d. is in brackets). Negative CO2flux represents oceanic CO2 uptake. Correlation coefficient (r) is shown between two parameters listed at top of each column. All correlations were significant (P < 0.001 for all cells except cell marked with an asterisk, P < 0.02). Micro, microphytoplankton; nano, nanophytoplankton; pico, picophytoplankton; fmicro, fnano and fpico are the respective fractions of total cell count made up by these phytoplankton classes.
Climatological CO2flux is presented for comparison, aligned with climatological boxes that most closely correspond to it: subarctic, 46°N–50°N; transition, 30°N–46°N; northern transition, 38°N–46°N; southern transition, 30°N–38°N; subtropics, 22°N–30°N.
25.8 ± 4.6 [9.3]
−4.1 ± 0.9 [3.1]
−1.7 ± 0.3
11.8 ± 3.2 [11.8]
0.2 ± 0.0 [4.0]
1.5 ± 1.3
17.1 ± 4.4 [13.7]
−2.1 ± 0.5 [4.0]
1.1 ± 1.0
5.4 ± 1.8 [2.3]
3.0 ± 0.5 [0.7]
2.6 ± 1.8
8.1 ± 2.1 [2.6]
0.1 ± 0.0 [1.3]
0.8 ± 0.1
Bloom region (30°N–31.2°N)
NCP and CO2 influx had similar spatial trends to %O2bio and ΔpCO2, and their mean values were highest in the subarctic among cruise regions (25.8 mmol C m−2 d−1 and 4.1 mmol C m−2 d−1, respectively; Figure 4). NCP was highly correlated with CO2 influx (r = 0.89), with a slope of 2.6 (Table 2). Flow cytometry results (Figure 4) show high concentrations of all three phytoplankton classes in the subarctic. NCP was most highly correlated with concentrations of microphytoplankton (r = 0.50) in this region.
4.3. Transition Zone
 The transition zone represented a shift from the high-chl-a, high-%O2bio, low-pCO2, high-nutrient conditions of the subarctic to opposite conditions in the subtropics. At the boundary between the northern transition zone and subarctic, in the vicinity of the TZCF, we observed coincident peaks in %O2bio and chl-a and troughs in ΔpCO2, with high zonal variability (Figure 3). In the southern transition zone (32°N–40°N), %O2bioand chl-a decreased. At ∼42°N,pCO2 shifted from undersaturated values to the north to supersaturated values to the south, and remained supersaturated in the southern transition zone.
 Throughout the transition zone, chl-a and %O2bio were positively correlated (r = 0.85), and pCO2 and %O2bio were negatively correlated (r = −0.75), similarly to the subarctic. But in contrast with the subarctic, pCO2 was positively correlated with SST (r = 0.87; Table 2). Surface nitrate and silicate declined from the subarctic to northern transition zone; nitrate approached the detection limit at the TZCF and remained <1 μM in the southern transition zone.
 From the northern to the southern transition zone, NCP declined (mean 17.1 and 5.4 mmol C m−2 d−1, respectively), and CO2flux varied from an influx to an efflux (Figure 4 and Table 2). Meridional and zonal (145°W–152°W) variability in fluxes were high in the northern transition zone. Throughout the transition zone, NCP and CO2 influx were correlated with a similar r (0.85) and slope (2.4) to that in the subarctic (Table 2).
 Throughout the transition zone, NCP was highly correlated with microphytoplankton and nanoplankton (r = 0.89 and 0.83, respectively), and less strongly correlated with picophytoplankton (r = 0.58). NCP was also highly correlated with the fraction of total cell count made up by microphytoplankton (r = 0.72). In the northern transition zone, high abundance of micro- and nanophytoplankton coincided with highNCP and CO2 uptake (Figure 4).
 In the subtropics, %O2bio, chl-a, and nutrients were low, similar to conditions in the southern transition zone, with the exception of a small peak in %O2bio at 30.5°N. ΔpCO2generally decreased southward from a maximum in the southern transition zone. Throughout the subtropics, chl-a was highly correlated with %O2bio (r = 0.86). pCO2 was weakly negatively correlated with both %O2bio and SST (Table 2). Surface nitrate was below detection limit throughout most of the subtropics, and surface silicate averaged 3 μM.
 Mean NCP in the subtropics (8.1 mmol C m−2 d−1; Figure 4) was slightly higher than that in the southern transition zone. Mean CO2flux was close to zero, and CO2 influx was weakly correlated with NCP (Figure 4 and Table 2). NCP was weakly correlated with picophytoplankton (Prochlorococcus) throughout the subtropics (Table 2). A peak in NCP at ∼30.5°N of 22 mmol C m−2 d−1 corresponded to a small CO2 influx. In the region of this peak (30°N–31.2°N), NCP was highly correlated with chl-a (r = 0.97) and CO2 influx (r = 0.69), and the NCP:CO2 influx slope (3.3) was similar to that observed in the subarctic and northern transition zone (Table 2).
5.1. Rates of NCP and CO2 Uptake
 In the subarctic, NCP (25.8 ± 4.6 mmol C m−2 d−1) and CO2 influx (4.1 ± 0.9 mmol C m2 m−2 d−1) reached their highest mean values along the cruise track. These values were ∼2× higher than the mean of previous estimates of NCP and climatological CO2 influx for this region (Table 1 and Takahashi et al. [2009, Table 2]). Mean chl-a (0.8 mg m−3) was almost twice as high as usual for September based on the SeaWiFS 9-kmchl-a time series for 1997–2010 (Figure 5), reflecting a phytoplankton bloom in this region. This bloom was most likely stimulated by iron deposited with volcanic ash in the early August 2008 eruption of the Aleutian island volcano Kasatochi [Hamme et al., 2010; Langmann et al., 2010].
 The high NCP and CO2 influx in the subarctic estimated during our cruise were consistent with bloom conditions, although there is evidence that we sampled during a declining stage of this bloom. SeaWiFS chl-a in our cruise region declined from August to September and onward (Figure 5). At OSP, our O2/Ar-basedNCP on 30 August (30 mmol C m−2 d−1) was lower than O2/Ar-basedNCP (43 mmol C m−2 d−1) estimated on 21 August [Hamme et al., 2010].
 In the subtropics, mean NCP (8.1 ± 2.1 mmol C m−2 d−1) was very similar to that previously observed at ALOHA (8 ± 4 mmol C m−2 d−1; Table 1), indicating that observations at this station are generally representative of the NE Pacific subtropical region. CO2flux showed a weak efflux (0.1 ± 0.0 mmol C m−2 d−1), ∼8× lower than climatological September CO2 efflux (Table 2). Chl-a in the subtropics (0.1 mg m−3) was similar to typical September values from the SeaWiFS time series (Figure 5).
 Although NCP in the subtropics was ∼3× lower than in the subarctic during our cruise, we observed a peak in NCP at 30°N–31°N, 152°W (Figure 4) associated with the late summer bloom that has been previously observed in this region (∼30°N, 130°W–160°W) and that can last up to 3–4 months [Wilson et al., 2008]. This recurrent bloom has been attributed both to N2 fixation by unicellular cyanobacteria and endosymbionts, and to Rhizosolenia diatom mats that can vertically migrate into the nitracline through ballasting [Wilson, 2003; Wilson et al., 2008]. NCP estimates at 30°N–31°N peaked at 19 mmol C m−2 d−1, approximately twice the mean subtropical rate.
 The transition zone (32°N–45°N) had mean NCP intermediate between the subarctic and subtropics (11.8 ± 3.2 mmol C m−2 d−1; Table 2), and within the range of previous NCP estimates in both the subarctic and transition zone (Table 1). Notably, NCP in the transition zone had much higher spatial variability than in the subarctic or subtropics, with regional s.d. of 11.8, equivalent to the mean (Table 2). It had mean CO2 efflux (0.2 ± 0.0 mmol C m−2 d−1) that was ∼7× lower than September climatological efflux (Table 2).
 In the northern transition zone, mean NCP (17.1 ± 4.4 mmol C m−2 d−1) was similar to a previous estimate in this region (Table 1) and on the high end of previous summertime NCP estimates in the subarctic. Both meridional and zonal (145°W–152°W) NCP variability were high in this region (regional s.d. = 13.7, Table 2). CO2 influx (2.1 ± 0.5 mmol C m−2 d−1) here was greater than climatological September estimates, which show a mean efflux of 1.1 ± 1.0 mmol C m−2 d−1 (Table 2); this discrepancy could be attributable to the anomalous bloom conditions in the subarctic, or to interannual variability in the frontal dynamics of the transition region. Based on the SeaWiFS time series (Figure 5) chl-a was not anomalously high during our cruise in either of these regions, as it was in the subarctic. In the southern transition zone, NCP at 5.4 ± 1.8 mmol C m−2 d−1 was similar to previous estimates at ALOHA, and mean CO2 efflux was 3.0 ± 0.5 mmol C m−2 d−1, similar to the September climatological value in this region (Table 2).
 Compared with a wealth of NCP estimates at OSP in the subarctic and ALOHA in the subtropics, we have only a few snapshots of NCP in the transition zone. Based on our results and previous studies, NCP in the northern transition zone resembles the subarctic, with comparably high summertime daily NCP rates (Table 1) and a strong link between NCP and CO2 uptake. We observed the highest spatial variability in NCP and CO2flux in this region, and it has high interannual variability in satellite chl-a (Figure 5). Frontal dynamics in this region likely play a role in this temporal and spatial biological patchiness, as we discuss further below. In contrast, the southern transition zone resembles the subtropical gyre, with lower daily NCP rates that peak in spring, lower spatial variability and a summertime CO2 efflux.
5.2. Impact of NCP on CO2 Uptake
 Through the high spatial resolution of our measurements, we observed several regions in which NCP and CO2 influx were highly coupled, demonstrating a link between the biological pump and atmospheric CO2uptake on a ∼2-week timescale. In the subarctic, we observed this high correlation during the decline of the anomalous bloom (including micro- and nanophytoplankton) described above, most likely stimulated by volcanic deposition of iron. This could be characteristic of the “event-driven, mass sedimentation carbon pump” described byKarl et al. in which perturbations such as pulsed delivery of iron lead to blooms and aggregation of large phytoplankton like diatoms, resulting in rapid, efficient export events. These results indicate that intermittent iron supply in this region (through volcanic events, eddy transport from the continental shelf or atmospheric deposition) would lead to a stronger carbon sink on short-term (seasonal) timescales.
 Throughout the transition zone, we observed similarly high correlation between NCP and CO2 influx. NCPwas highly correlated with concentrations of micro- and nanoplankton, which were abundant in the northern transition zone, and was highly correlated with the fraction of total cell count comprising microphytoplankton. The SAFZ, STFZ and TZCF have been found to be associated with highPP, NCP and abundance of microphytoplankton by other researchers [Leonard et al., 2001; Juranek, 2007; Howard et al., 2010; Juranek et al., 2012].
 In the subtropics, phytoplankton abundance (chl-a) and %O2bio were highly correlated as in the subarctic (Figure 3), but there was a much lower correlation between NCP and CO2 influx (Table 2; r = 0.20 versus r = 0.89 in the subarctic). This probably results from the strong influence of temperature on the summertime CO2flux in this region, yielding a weaker link between NCP and CO2uptake. The dominance of picophytoplankton over nano- and microphytoplankton in the subtropics (Figure 3) may also lead to less efficient export.
 We observed a peak in NCP at ∼30°N, 130°W–160°W, where recurrent late summer phytoplankton blooms have been observed. In this region (30°N–31.2°N), NCP was highly correlated with chl-a and CO2 influx, and the slope of this correlation was similar to that observed in the subarctic and transition zone (Table 2). At this location, microphytoplankton and picophytoplankton (Prochlorococcus) dominated, and their abundances were highly correlated with NCP (r = 0.70 and 0.45, respectively).
5.3. Impact of NCP on CO2 Uptake Over Annual Cycle
 Our snapshot observations demonstrate the strong link between NCP and CO2 uptake in the subarctic and transition zone. However, our larger goal is to put these estimates in the context of determining the impact of NCP on annual CO2 uptake in the North Pacific. To do this, we made two box model calculations to examine the seasonal cycle of pCO2 and DIC in our cruise region using a climatological surface pCO2 data set in boxes of 4° latitude × 5° longitude [Takahashi et al., 2009, section 3.5]. We first examined the seasonal cycle of pCO2 in a simplified “abiotic” system in which only temperature and gas exchange influence pCO2 and CO2flux. Second, we used a simplified DIC mass balance approach to compare the annual climatological CO2flux with the contribution of NCP and physical input of DIC.
5.3.1. Abiotic pCO2 Model
 A simplified mass balance of DIC in the mixed layer can be written as:
where FCO2(in)is the sea-air flux of CO2 (kCO2*kH(pCO2atm − pCO2SW)), with positive FCO2(in) defined as into the ocean (note this is the opposite of the sign convention in equation (3) and our results), h is the mixed layer depth, Jphys is the physical input of DIC due to advection, diffusion and entrainment (with supply of DIC defined as positive and removal as negative), and Jncp is the export of carbon due to NCP (with respiratory source of DIC defined as positive and biological uptake of DIC as negative). The annual cycle of FCO2(in) is mainly controlled by seasonal changes in surface ocean pCO2 with a secondary effect of variation in kCO2 due to seasonal wind speed variation.
 To separate the influence of temperature on pCO2 from the biological and physical effects, we described seasonal changes in pCO2in an “abiotic” ocean in which only warming or cooling and air-sea gas exchange influenced seasonal changes inDIC and pCO2. In this case, the DIC mass balance simplifies to:
 We initialized this abiotic model with the 1 March climatological pCO2SW and DIC. We calculated the initial FCO2(in) based on the 1 March climatological pCO2SW,pCO2atm, kH and kCO2. We distributed this FCO2(in) over a daily time step and recalculated dDIC/dt and pCO2SW based on thermodynamic equilibrium (using climatological daily interpolated values of h, alkalinity, pCO2atm, temperature, salinity, kH and kCO2; see Methods for details), and repeated the process for each subsequent daily time step. In the region of our study, the mixed layer generally begins to shoal in March, and therefore we assume that any influence of entrainment on DIC is negligible until late summer or early fall, when the mixed layer begins to deepen again.
 The observed (climatological) pCO2 peaks in summer in all boxes in our cruise region along 152°W, yielding a summertime CO2 efflux in all regions (Figure 6). In the abiotic model, however, the estimated summertime increase in pCO2 (and thus CO2 efflux) is greater than the observed pCO2 in all boxes. Biological (Jncp) and physical (Jphys; equation (4)) input or removal of DIC must be responsible for the difference in annual cycles of the predicted abiotic and observed pCO2. In spring and summer in the subarctic and transition zone, when the mixed layer is shoaling or stable and wind speeds are low, removal of DIC by Jncp is much greater than net removal by Jphys [Ayers and Lozier, 2012], and is the dominant cause of the observed pCO2 (pCO2SW) being less than the abiotic pCO2 (pCO2abio). In the subtropics in spring and summer, based on studies at ALOHA, Jncp is the only term removing DIC, because Jphys is a net DIC source [Keeling et al., 2004; Quay and Stutsman, 2003]. In contrast, in fall and winter, when the mixed layer is deepening, addition of DIC through positive Jphys outweighs removal of DIC by Jncp in the subarctic, transition zone and subtropics [Ayers and Lozier, 2012; Keeling et al., 2004], and causes pCO2SW to be greater than pCO2abio.
 We calculated the integrated CO2 flux resulting from the difference between pCO2abio and pCO2SW by summing daily CO2 flux = kCO2kH (pCO2abio − pCO2SW) for the period in each box when pCO2abio is greater than pCO2SW, typically March–September. The trends in this integrated flux along the cruise track largely correspond to the trends in NCP and FCO2(in) we observed on our cruise (Figure 6). The integrated flux is greatest in the subarctic (46°N–50°N) and northern transition zone (38°N–46°N), implying high summer drawdown of pCO2 by NCP (assuming Jphys is small), in agreement with our cruise observations of high NCP and strong coupling of NCP and CO2 uptake in these regions. The integrated flux is lower in the southern transition zone (30°N–38°N) and subtropics (22°N–30°N), in agreement with our cruise observations in these regions of lower NCP, CO2 efflux, and little coupling between NCP and FCO2(in). pCO2SW is lower than pCO2abio for the longest interval (>240 days) in the subtropics, as a result of a long growing season, and in the subarctic and northern transition zone (which has both a spring and a fall bloom [Longhurst, 2007]).
5.3.2. DIC Mass Balance
 We next used a simplified version of the DIC mass balance approach of Lee  to compare the annual climatological FCO2(in) with the contribution of NCP. Rearranging the DIC mass balance (equation (4)), the magnitude of DIC drawdown plus CO2 invasion must be the net result of NCP plus physical supply:
 We again initialized the model with the 1 March value of pCO2 and climatological DIC. We calculated daily FCO2(in)and salinity-normalized dDIC/dt using daily interpolated data described above and solved for Jncp + Jphys. Initially, we assumed Jphys = 0 during the well-stratified spring-summer season (when observedpCO2 is less than abiotic pCO2, typically March–September) to obtain an estimate of Jncp. This would result in an upper-boundary estimate for NCP in the transition zone (whereJphys removes a small amount of DICin spring-summer;Ayers and Lozier, 2012) and a minimum estimate of NCP in the subtropics (where Jphysis positive in spring-summer [Keeling et al., 2004]). We integrated this estimate of summer NCP by integrating the values of Jncp + Jphys for the period 1 March–15 September (Figure 7).
 To obtain the observed summertime pCO2 decrease (if Jphys = 0) requires mean daily NCP in the subarctic and northern transition zone of 10 and 13 mmol C m−2 d−1, respectively (Table 3). These values are slightly lower but within the range of uncertainty of previous field estimates for these regions (Table 1). Our cruise-basedNCP estimate was more than 2× than the model estimate in the subarctic (26 mmol C m−2 d−1), likely reflecting the anomalous conditions there during our cruise, and 31% greater in the northern transition zone.
Table 3. Results of Model of Climatological pCO2 and DIC Seasonal Cycle
Summer-integrated modeledNCP and FCO2(in) are integrated for the period 1 March–15 September. In model, positive FCO2(in) represents influx into ocean. Regions were defined as described in Table 2.
 In the southern transition zone and subtropics, modeled summer NCP was about half that in the subarctic and northern transition zone, at 5–6 mmol C m−2 d−1, similar to our cruise estimates of 5–8 mmol C m−2 d−1 and within the uncertainty of previous field estimates of NCP in these regions (Tables 1 and 3). The ∼2× lower summer NCP in the subtropics compared to the northern transition zone is similar to the meridional trends in NCP estimated from our cruise observations. This suggests that our O2/Ar-basedNCP in the transition zone and subtropics during September 2008 may be representative of the climatological NCPduring the spring-summer growing season. ModeledNCP in the subtropics (5 mmol C m−2 d−1versus our cruise-based estimate of 8 mmol C m−2 d−1) is probably underestimated because physical supply of DIC in the subtropics is positive during the summer, as has been shown at ALOHA [Keeling et al., 2004; Quay and Stutsman, 2003]. In support of this, the stability (Δσ/Δz) in the upper water column in the subtropics and southern transition zone was much less than that in the subarctic during our late summer cruise (Figure 2).
 Integrated annual climatological FCO2(in)was similar in magnitude to modeled summer-integratedNCP along the cruise track, peaking in the northern transition zone at 2.0 mol C m−2 yr−1, slightly lower in the subarctic and southern transition zone (1.2–1.4 mol C m−2 yr−1), and lowest in the subtropics at 0.8 mol C m−2 yr−1 (Figure 7 and Table 3). On average, the magnitude of summer-integratedNCPbalances annual-integratedFCO2(in) across the cruise track, with ratios of summer NCP to annual FCO2(in) of 0.7–1.4 across all regions (Table 3). Some carbon exported through NCP during summer could be remineralized and returned to the mixed layer in winter, as has been observed in the North Atlantic [Körtzinger et al., 2008]; this would lower O2/Ar-basedNCP and CO2 uptake. However, winter mixed layer depths in the NE Pacific are much lower than in the North Atlantic, so the effect may not be as pronounced there.
 We conclude that NCP exerts a significant control on the high CO2 uptake in the NE Pacific, with the remainder of annual CO2 uptake impacted by surface cooling and physical processes. It is difficult to directly estimate the contribution of NCP to annual CO2 uptake, because the response time of CO2 gas exchange with respect to changes in temperature and DIC is relatively sluggish (∼1 year) [Broecker and Peng, 1982]. Therefore, NCP that takes place in summer likely drives uptake of CO2 in fall or winter.
 Mean NCP and CO2 influx along our cruise track were greatest in the subarctic (25.8 ± 4.6 and 4.1 ± 0.9 mmol C m−2 d−1, respectively) and the northern transition zone (17.1 ± 4.4 and 2.1 ± 0.5 mmol C m−2 d−1, respectively). In the subarctic, our observations of NCP were ∼2× the mean of previous summer observations, reflecting the influence of an anomalous phytoplankton bloom. In the transition zone, we detected high zonal and meridional variability in NCP and CO2fluxat <5-km scales, suggesting the impact of frontal dynamics on biogeochemical fluxes in this region, and the need for observations throughout the year to determine its mean annual state. In the southern transition zone and subtropics,NCP (5.4 ± 1.8 and 8.1 ± 2.1 mmol C m−2 d−1, respectively) was about 2–3× lower than in the subarctic and northern transition zone, and CO2 had mean efflux (3.0 ± 0.5 and 0.1 ± 0.0 mmol C m−2 d−1, respectively).
NCP and CO2 influx were highly correlated in the subarctic and transition zone with a slope of 2 to 3, indicating the strong coupling between biological uptake and CO2 influx during summertime in these regions. NCP in these regions was also highly correlated with concentrations of microphytoplankton. In contrast, NCP and CO2 influx were not strongly coupled in the subtropics, suggesting that the temperature effect on solubility dominates CO2 flux during summertime there. In the subtropics, NCPwas uncorrelated with microphytoplankton and weakly correlated with picophytoplankton, potentially indicating the lower efficiency of a picophytoplankton-dominated biological pump.
 Our analysis and modeling of the climatological DIC and pCO2 cycles in the transition zone and subtropics yielded a similar meridional pattern in integrated summertime NCP and CO2flux as we found during our cruise in September 2008. NCP in the northern transition zone was ∼2–3× that in the southern transition zone and subtropics, suggesting that our snapshot estimates in these regions are representative of the seasonal influence of NCP on CO2 influx. Based on our model of the climatological seasonal DIC and pCO2 cycle in the Northeast Pacific, we conclude that summertime NCP is similar in magnitude to the annual atmospheric CO2 uptake in the NE Pacific, indicating that NCP has a significant impact on CO2 uptake in this region. More continuous observations of the temporal and spatial variability of the biological pump in this region and more sophisticated physical modeling would improve our ability to distinguish the magnitude of the influences of temperature, biological uptake, and physical dynamics in this region of high CO2 uptake.
 This work was supported by a grant from NSF to P.D.Q., S. Emerson, R.F. and R. Letelier, and a NSF Graduate Research Fellowship (D.L.). We thank Mark Haught, Johnny Stutsman, Fran Janny and David Munro for field and laboratory assistance, and the captain and crew of the R/V Thomas G. Thompson. We are grateful to Steven Emerson, David Nicholson and an anonymous reviewer for thoughtful comments that helped us to improve the paper.