Temperature and phytoplankton cell size regulate carbon uptake and carbo overconsumption in the ocean

Introduction Conclusions References


Introduction
Accounting for approximately half of annual global primary production (Falkowski et al., 1998), phytoplankton exert a fundamental control on the uptake of carbon dioxide (CO 2 ) by the ocean and thus, atmospheric CO 2 concentration.The growth of phytoplankton is largely controlled by water column structure that regulates the availability of nutrients through mixing and stratification and light via the turbulent mixing of cells throughout the euphotic zone.Recent investigations have pointed to the effects of increasing water column stratification on oceanic phytoplankton communities, caused by an ensemble of processes associated with global climate change (Cermeño et al., 2008;Li et al., 2009;Riebesell et al., 2009;Morán et al., 2010;Taucher and Oschlies, 2011).These studies show a change in the phytoplankton community size structure towards smaller cells apparently in response to the impoverished nutrient conditions resulting from stronger stratification which limits mixing of nutrients from deeper waters Introduction

Conclusions References
Tables Figures

Back Close
Full into the euphotic zone.An alteration of phytoplankton size structure has profound implications for energy transfer to higher trophic levels.Longer food chains, and therefore decreased energy transfer efficiency to higher trophic levels, are associated with assemblages dominated by small cells (Barnes et al., 2011), and a relationship between fish production and nutrient-controlled cell size has been suggested (Sommer et al., 2002).Export of carbon from the surface ocean has also been shown to be strongly dependent on phytoplankton cell size; larger cells tend to be dominant in higher nutrient and turbulence regimes (Margalef, 1978) and are associated with greater sinking rates, higher cellular carbon values and, consequently, greater carbon export to the deep ocean (Laws et al., 2000;Finkel et al., 2010).Yet, how such a climate-driven alteration of phytoplankton size structure might affect the biologically mediated uptake of atmospheric CO 2 by the ocean has rarely been explored.In this report we utilize observations from a "natural laboratory" provided by a long-term study site on the Scotian Shelf in the western North Atlantic.Over an annual cycle, observations at this site encompass a dramatic warming of surface waters from sub-zero to ∼ 20 • C and the concomitant response of the phytoplankton community, thus providing an analogue of the physical and biological responses of the upper ocean likely to result from climateinduced warming.
2 Materials and methods

In situ data
Phytoplankton, chemical, hydrographic and partial pressure of CO 2 (pCO 2 ; µatm) data were collected from station HL2 (44.4 • N-63.viously (Li and Dickie, 2001;Li and Harrison, 2001;Shadwick et al., 2011).Briefly, high temporal resolution measurements of pCO 2 at approximately 2 m were obtained from a CARIOCA buoy moored at station HL2.Ship-based measurements were also collected bi-weekly from station HL2 and included CTD casts, microscopic enumeration of microphytoplankton (20-200 µm) and water sample analyses for chlorophyll a concentration (Chl a; mg m −3 ) and nutrients using the methodologies detailed by Mitchell et al. (2002b).Flow cytometric counts of pico-(0.2-2µm) and nanophytoplankton (2-20 µm) measured weekly at a nearby coastal monitoring site in Bedford Basin (44.69 • N, −63.64 • W) were used as proxies for the abundance of these size fractions at HL2.The use of these counts as proxies for HL2 was considered robust as previous studies have demonstrated that they are temporally coherent with the less frequent measurements of the same size fractions made at HL2 (Li et al., 2006;Li and Harrison, 2008).Monthly climatologies were then constructed for each of these parameters.
The spectral phytoplankton absorption coefficient (a ph (λ); m −1 ) was also measured at HL2 and the wider Scotian Shelf region approximately 2-3 times per year using the quantitative filter technique (Mitchell et al., 2002a) and a pathlength amplification factor derived by Hoepffner andSathyendranath (1992, 1993) and modified by Kyewalyanga et al. (1998).erage of this box.Additionally, Pathfinder 5 AVHRR 4 km sea surface temperature (SST;

Phytoplankton size class seasonal patterns
The annual spring bloom in the western North Atlantic typically occurs during late March to early April at the temperature minimum and mixed layer depth maximum (Fig. 2), and, over the period of just a few weeks, draws down approximately one third of the total carbon fixed over the annual cycle in this region (Fournier et al., 1977).
During the intense carbon uptake of the spring bloom, diatoms in the microphytoplankton (20-200 µm) size range dominate the phytoplankton assemblage (Fig. 3a,  b).These cells are adapted to rapid growth in the nutrient replete conditions (Fogg, 1991) that result from intense winter mixing, and reach their climatological maximum of 2.4 × 10 8 cells m −3 in April.The spring bloom collapses precipitously when nitrate and silicate are largely depleted (Fig. 2c, d), and beginning ∼ May, a new assemblage of cells dominated by much smaller but numerically more abundant pico-(0.2-2µm) and nanophytoplankton (2-20 µm) flourish in the warming, relatively nutrient poor conditions (Figs. 2 and 3).Pico-and nanophytoplankton cell abundance increases steadily, reaching its maximum of 10×10 10 cells m −3 in September.This seasonal pattern is mirrored in the dinoflagellate population, although their abundance is approximately four orders of magnitude less than the pico-and nanophytoplankton fraction.Dinoflagellates are generally categorized in the microphytoplankton size class, and, in the context of size scaling with temperature (Peters, 1983), their co-occurrence with the smaller cell sizes seems incongruous.However, dinoflagellates possess a collection of unique features (e.g.mixotrophy and the ability to vertically migrate (Smayda, 1997)), which Introduction

Conclusions References
Tables Figures

Back Close
Full collectively may reduce their dependence on nutrients delivered from aphotic depths increasing the likelihood of peak abundance during the summer months.Following the spring bloom, the diatom contribution to total cell abundance remains low at < 0.5 × 10 8 cells m −3 , although occasionally, a modest autumnal bloom occurs caused by wind-driven mixing (Greenan et al., 2004), and this is reflected in the small elevation in the diatom climatology in November (Fig. 3a, b).It should be noted that, despite being numerically more abundant than diatoms by several orders of magnitude in the summer months, pico-and nanophytoplankton biomass is not correlated with chlorophyll a standing stock (Fig. 3c), an important observation also made by Claustre (1994) in several oceanic provinces.Chl a is persistently less than 1 mg m −3 following the collapse of the spring bloom, yet the cell counts reveal substantial biomass in the pico-and nanophytoplankton fractions, emphasizing the fact that Chl a primarily mirrors patterns in the diatom fraction of the assemblage (Fig. 3d), but not the dinoflagellate fraction (Fig. 3e).For this reason, the assumption that Chl a is a robust proxy for biomass should be applied with care (Cullen, 1982).

Characterisation of community composition from satellite ocean colour
The shape of a ph (λ) is strongly related to phytoplankton size (Bricaud et al., 2004), which in turn, can be related to trophic state (Chisholm, 1992), making it a powerful means to detect the transition from diatom dominated assemblages in the spring to pico-and nanophytoplankton dominated assemblages in the summer.To investigate the potential of satellite measurements to reveal phytoplankton size class information, a ph (λ) was first derived from MODIS Aqua level 2 LAC 1 km R rs (λ) from the Scotian Shelf region using a regionally trained ocean colour algorithm (Craig et al., 2012) (See Supplement).The model of Ciotti et al. (2002) was then used to estimate a dimensionless size factor (S f ) from the satellite-derived a ph (λ) from the Scotian Shelf.S f is essentially derived from a spectral mixing model parameterised by two absorption spectra -one from very small picophytoplankton and the second from very large microphytoplankton.The model additively combines these two spectra and S f specifies Introduction

Conclusions References
Tables Figures

Back Close
Full  (Ciotti et al., 2002;Ciotti and Bricaud, 2006).The satellite-derived S f was compared with S f derived from matching in situ filterpad a ph (λ) to ensure that no bias or errors were introduced by using satellitederived a ph (λ) (Fig. 4a).The two values were found to agree well (R 2 = 0.763, N = 54, RMSE = 0.065), and satellite-derived S f was considered to be a reliable means of describing the dominant size class of the phytoplankton assemblage.The S f model was then applied to a ph (λ) spectra derived from monthly climatological R rs (λ) (MODIS Aqua level 3 mapped 4 km monthly R rs (λ) composites) from the HL2 box.S f is observed to mirror clearly the patterns revealed by the in situ data (Fig. 3b) -larger cells dominate the phytoplankton assemblage in the spring (S f < 0.1), smaller cells in the summer months (S f ≈ 0.3) and an intermediate size around the autumnal bloom (S f ≈ 0.26).
The Ciotti et al. (2002) approach is just one of many optically-based approaches that have been successfully implemented to delineate phytoplankton community composition (Alvain et al., 2005;Uitz et al., 2006;Brewin et al., 2011;Devred et al., 2011;Hirata et al., 2011), and the success at this study site highlights the potential utility of remote sensing in revealing synoptic seasonal patterns in phytoplankton succession.

Cell size and carbon uptake
Phytoplankton cell counts (cells m −3 ) were converted into total cellular carbon concentration (mol C m −3 ) using average literature values for carbon content per cell for pico-, nano-and microphytoplankton (Mullin et al., 1966;Verity et al., 1992;Li et al., 1993;Menden-Deuer and Lessard, 2000;Li and Harrison, 2001).These values were then used to examine the relationship between water temperature and phytoplankton carbon concentration (Fig. 5a).We present temperature on the abscissa rather than date to account for the fact that the same temperature may occur in multiple seasons.Max-

BGD Introduction Conclusions References
Tables Figures

Back Close
Full in concert with pico-and nanophytoplankton abundance, steadily increases by ∼ 85 % attaining values comparable to the those in the spring (Fig. 5a).Throughout this period, Chl a is persistently < 1 mg m −3 (Fig. 3b), reinforcing the fact that this important fraction of the phytoplankton assemblage is almost completely decoupled from the chlorophyll standing stock (Fig. 3c; c.f. Claustre, 1994) -significant given the fact that Chl a is used ubiquitously as the biomass term in many different types of global and regional ocean models (e.g.Behrenfeld et al., 2006;Fennel et al., 2008;Boyce et al., 2010).
The inability to accurately estimate carbon uptake from Chl a during this summer period was also identified by Shadwick and colleagues (Shadwick et al., 2010(Shadwick et al., , 2011)), and reinforces the concept that Chl a closely mirrors the patterns of assemblages with high intracellular Chl a, e.g.large diatoms (Li et al., 2006) (Fig. 3b, d), but does not accurately represent the more numerically abundant smaller size fractions with lower intracellular Chl a that dominate in the summer months (Fig. 3c).
To different phytoplankton communities is clearly evident in the bulk CO 2 system parameters.Springtime maximum phytoplankton carbon concentration, which is associated with large diatoms and high NCP p values (Figs. 5a,b,3a,b) and that occurs at the temperature minimum, is reflected in the rapid drop in pCO 2, norm (Fig. 5a, c).The collapse of the bloom and the onset of surface warming in May results in rising pCO 2, norm concentrations, and at a water temperature of ∼ 5-6 • C in May/June, the diatom community is succeeded by smaller pico-and nanophytoplankton and dinoflagellates associated with lower NCP p values (Figs. 3a,b,5b).The increase in biomass and concomitant uptake of carbon by these communities (Fig. 5a) consistently lowers pCO 2, norm (Fig. 5c) throughout the summer months until the temperature reaches its maximum.
At the end of the summer period, respiration resulting from decay of phytoplankton biomass (Fig. 5a) and wind-induced or convective entrainment of CO 2 from deeper waters (Greenan et al., 2004) raises pCO 2, norm back to pre-bloom winter conditions.

Satellite estimates of NCP p
Having shown the strong effect of seasonal phytoplankton succession on the drawdown of pCO 2 on in situ single point measurements, an obvious question that arises is whether remotely sensed satellite data can be used to predict NCP p .It was clearly shown in the preceding sections that Chl a is a poor proxy of small cell biomass (Fig. 3c) making it an unsuitable remote sensing parameter for estimating biologically mediated carbon drawdown -especially in the summer.However, studies performed in the North Atlantic have shown that water temperature can be used as a holistic simplifier of mechanistically complex processes (e.g. the response of phytoplankton to seasonal variability in nutrients and mixed layer depth) and can predict the abundance of pico-and nanophytoplankton (Li et al., 2006;Morán et al., 2010).We use the same approach and examine the relationship between log 10 [picophytoplankton + nanophytoplankton] abundance and water temperature at HL2 during 1999-2011 (Fig. 6a) to reveal a strong correlation (R 2 = 0.739, N = 156, p 0.01), comparable to the findings of previous studies at other North Atlantic sites (Li et al., 2006;Morán et al., 2010).Temperature Introduction

Conclusions References
Tables Figures

Back Close
Full was also found to predict the log 10 abundance of dinoflagellates, but this relationship was revealed only when the data were organised into climatological months.The resulting regression equation for estimating dinoflagellate abundance ( y dino ; cells m −3 ) was y dino = 10 6.300+0.0378T(R2 = 0.819, N = 12, p 0.01), where T is water temperature ( • C).In the case of diatoms, there is essentially no relationship between temperature and log 10 abundance as diatom peak abundance is during the spring bloom that occurs prior to appreciable increase in water temperature (Fig. 3b).Instead, diatom abundance ( y dia ; cells m −3 ) was estimated using the following simple relationship: where dia clim and (pico + nano) clim are the climatological cell concentrations for diatoms and (picophytoplankton + nanophytoplankton) respectively and y pico+nano is estimated using the regression in Fig. 6a.Cell abundances were estimated from Pathfinder 5 AVHRR 4 km satellite measurements of SST from station HL2 using the relationships described above, converted to cellular carbon using literature carbon cell −1 values, then used to calculate satellite NCP p (NCP psat ; mol C m −3 month −1 ) for the period 1999-2010 (Fig. 6b).We note here that a strong relationship (R 2 = 0.961, N = 12, p 0.01) was found for monthly climatological values of log 10 [picophytoplankton + nanophytoplankton] abundance, but it was decided to use the fully time resolved monthly data (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011) as this produced the best estimates of NCP psat (see below), likely due to the fact that more of the variance was accounted for.It is evident that the performance of this approach is variable, with satisfactory estimates (R 2 > 0.5) of NCP psat obtained for only some of the years or portions of those years.In order to estimate the error associated with each NCP psat calculation step, statistics are presented in Table 1 to describe the agreement between NCP p and NCP psat for the total phytoplankton assemblage (tot) and for each of the assemblage components (p + n -(picophytoplankton + nanophytoplankton); dia -diatoms; dino -dinoflagellates).R Introduction

Conclusions References
Tables Figures

Back Close
Full p+n is greater than 0.5 for almost the same subset of years (with the exception of 2001), illustrating the importance of robust p + n estimates for accurate estimation of NCP psat .R 2 dino is never greater than 0.5, even in the years for which reasonably accurate NCP psat estimates (i.e.R 2 tot > 0.5) were obtained.Compared to the contribution in abundance terms of diatoms in the spring and pico-and nanophytoplankton in the summer, the contribution to NCP p from dinoflagellates is small (∼ 4 % maximum) and could likely be omitted from these calculations with little or no impact on NCP psat accuracy (see statistics for p + n + dia in Table 1).Interestingly, y dia is rather accurately estimated (R 2 dia ≈ 0.5-0.9) in most years -surprising given the simple climatological ratio method used to estimate diatom abundance and suggestive of a fairly stable ratio of diatoms to p + n from year to year.It is presently unclear why the approach works reasonably well for some years but not for others.Temperature is used here as a means to integrate many complex mechanisms and their resulting effects on small cell abundance, and it is clear that the relationship breaks down under certain conditions that are not resolved by our analyses.However, as a preliminary exercise, the approach does show promise, is comparable in terms of accuracy of NCP p estimates to other approaches (e.g.Serret et al., 2009) and points to the potential utility of satellite measurements to provide synoptic scale insight into the biological mediation of carbon.This approach could be particularly useful during the summer season when Chl a, typically < 1 mg m −3 at this site and the surrounding shelf area, is not representative of the small cell biomass and its associated and significant carbon uptake.

Carbon overconsumption and temperature effects
The increase in phytoplankton biomass, and thus NCP p , during the summer months is in spite of apparent depletion of surface mixed layer nitrate (Fig. 2c), a finding also reported at this site by Shadwick and colleagues (Shadwick and Thomas, 2011;Shadwick et al., 2011).They observed supersaturation of O 2 with respect to atmospheric Introduction

Conclusions References
Tables Figures

Back Close
Full levels and a decrease in DIC in surface waters at HL2 throughout the post bloom summer months, which they attributed to phytoplankton primary production.The phenomenon of elevated carbon consumption relative to nitrogen that exceeds the classical Redfield ratio (Redfield et al., 1963) of 6.6 -so called "carbon overconsumption" -has been reported extensively in the literature (Sambrotto et al., 1993;Toggweiler, 1993;Thomas et al., 1999;Osterroht and Thomas, 2000;Körtzinger et al., 2001;Koeve, 2004;Taucher et al., 2012;Jiang et al., 2013), and the authors of these studies observed that overconsumption appears to be associated with summertime nutrient poor conditions.Here, we place overconsumption of carbon in the context of the seasonal succession of phytoplankton size classes, that can be related to trophic state (Chisholm, 1992), and that is governed by the paradigm described by Margalef of organisms of differing sizes inhabiting specific regions of turbulence-nutrient space (Margalef, 1978).We postulate that, given the persistent near zero nitrate concentrations present in the summer months (Fig. 2c), the pico-and nanophytoplankton dominated assemblage is responsible for carbon overconsumption and achieves this by intensive recycling of biomass degradation products (e.g.NH + 4 ) and, perhaps, also by bacterivory (Zubkov and Tarran, 2008).
The effect of temperature on NCP for pico-and nanophytoplankton (NCP p+n ), and thus on overconsumption, was examined by plotting the annual summer maxima during 1999-2010 for NCP p+n and for water temperature (Fig. 7a), which revealed a positive correlation (R 2 = 0.412, N = 11, p < 0.05).This effect was further explored through a simple exercise in which 0.5 • C, 1 • C and 2 • C were added to the climatological seasonal temperature cycle, and found to increase NCP p+n by 12.5 %, 26.6 % and 60.2 % respectively (Fig. 7b), highlighting the potential sensitivity of the system to temperature.This is a much-simplified estimate that ignores the effects of thermal stratification and other complex physical phenomena that may result from increasing water temperature.However, Umoh and Thompson (1994) showed that the vertical eddy diffusivity on the Scotian Shelf reaches its annual minimum during these months, thereby restricting nutrients to the upper water column.It seems reasonable to postulate that increas-

Conclusions References
Tables Figures

Back Close
Full ing temperatures would further reduce nutrient diffusion and so enhance the role of smaller phytoplankton size classes associated with nutrient recycling (Claustre, 1994) and possible carbon overconsumption.A further amplification of the effect of temperature on the summertime phytoplankton assemblage is suggested by recent studies by Taucher and colleagues (Taucher and Oschlies, 2011;Taucher et al., 2012) who postulate that nutrient recycling and phytoplankton metabolism may be elevated at higher temperatures.

Conclusions
This study has shown that numerically abundant pico-and nanophytoplankton are correlated with persistent uptake of carbon throughout the nutrient poor, post bloom summer months, and appear associated with carbon overconsumption.During the summer, this fraction of the phytoplankton assemblage is uncoupled from the Chl a standing stock, yet accounts for approximately the same amount of net community production as the spring bloom.Carbon uptake is commonly estimated using C : Chl a ratios; however, our findings suggest that using this approach may leave ∼ 20 % of biomass inventory unaccounted for.As the ocean responds to climate change, it is expected that there will be an accompanying shift in the phytoplankton community structure towards smaller cell sizes.However, the effect of this shift on the complex carbon dynamics of the Scotian Shelf, which has been characterized as a net source of carbon to the atmosphere (Shadwick et al., 2010(Shadwick et al., , 2011)), remains uncertain.Recent studies (Taucher and Oschlies, 2011;Taucher et al., 2012) suggested that water temperature increases predicted to occur under global warming scenarios may increase microbial loop nutrient recycling, potentially increasing the effect of the smaller cells on annual carbon uptake.It seems reasonable to postulate that these processes may occur in shelf systems worldwide and could have significant implications for estimates of annual carbon uptake.Our results also point to the potential to characterize many of these effects using remote sensing of ocean colour and sea surface temperature, and future efforts will rely Introduction

Conclusions References
Tables Figures

Back Close
Full  Full Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 3 • W) on the Scotian Shelf, eastern Canada (Fig. 1), a site of regular monitoring since 1998 by the Department of Fisheries and Oceans (DFO) Canada as part of the Atlantic Zone Monitoring Program (AZMP; http://www.bio.gc.ca/science/monitoring-monitorage/azmp-pmza-eng.php).Sampling methods, experimental procedures and methods have been described in detail pre-Discussion Paper | Discussion Paper | Discussion Paper | Aqua level 2 LAC 1 km remote sensing reflectance spectra (R rs (λ); sr −1 ) of the Scotian Shelf region were downloaded from http://oceancolor.gsfc.nasa.gov/.A regionally tuned version of the Craig et al. (2012) ocean colour model to estimate phytoplankton absorption spectra was derived using the satellite R rs (λ) spectra and in situ a ph (λ) measurements (see Supplement).The resulting model was then used to derive a ph (λ) from MODIS Aqua level 3 mapped 4 km monthly composites of R rs (λ) (http://oceandata.sci.gsfc.nasa.gov/MODISA/Mapped/Monthly/4km/Rrs/) in a box encompassing station HL2 but not beyond the shelf break: 43.5 • N ≤ latitude ≤ 45 • N-65 • W ≤ longitude ≤ −63 • W. All ocean colour derived parameters reported are the av-Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | the complementary contribution of each, where S f values close to 0 indicate an assemblage entirely dominated by microphytoplankton, S f values close 1 an assemblage dominated by picophytoplankton, and intermediate values the mixture of size classes typically found in natural phytoplankton assemblages Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | imum phytoplankton carbon concentration is observed at the temperature minimum during the diatom dominated spring bloom.Minimum phytoplankton carbon concentrations occur at approximately 5 • C, corresponding to both the collapse of spring bloom (∼ May) and early winter (December/January).Over the post spring bloom warming period (∼ 5-20 • C ≈ May-August, Figs. 2 and 3), phytoplankton carbon concentration, resolve the important contribution of seasonal biological signals to total carbon dynamics, phytoplankton net community production (NCP p ; mol C m −3 month −1 ) was calculated from the rate of change of phytoplankton carbon concentration from month to month, i.e.NCP p = C p (t 2 ) − C p (t 1 )/ (t 2 − t 1 ), where C p (t x ) is phytoplankton carbon concentration at time x (t 2 > t 1 ), and where positive values represent an increase in phytoplankton carbon concentration (Fig. 5b).The error bars represent one standard deviation of total NCP p , calculated from the mean of 1999-2011, and indicate the high degree of interannual variability.The NCP p maximum of 0.12 mol C m −3 month −1 occurs in April and is influenced primarily by the rapid increase in diatom cell numbers during the spring bloom.This compares well to pCO 2 -derived springtime values for NCP of 0.08 mol C m −3 month −1 found by both Shadwick et al. (2011) and Thomas Introduction Discussion Paper | Discussion Paper | Discussion Paper | et al. (2012) at this site and represented in Fig. 5b as S2011 and T2012 respectively.The precipitous decrease in diatom abundance in May drives NCP p to a negative value of −0.15 mol C m −3 month −1 , similar in absolute magnitude to the spring value.NCP p is positive throughout June-September reflecting the steady increase in picophytoplankton, nanophytoplankton and dinoflagellate abundance, which, along with temperature, reach their maxima in September (Fig. 2a).An average of the positive NCP p values during June-September gives 0.034 mol C m −3 month −1 , a value comparable to the average of summertime NCP computed by Shadwick et al. (2011) (0.017 mol C m −3 month −1 ).Integrating the positive summertime NCP p values (i.e.September June NCP p (t)dt) yields a phytoplankton carbon concentration of 0.14 mol C m −3 , slightly greater than the 0.12 mol C m −3 value in April associated with the spring bloom.If the entire mixed layer depth, (MLD; m) is considered, i.e. an NCP p inventory, NCP p ( MLD 0 NCP p (z)dz = NCP p ; mol C m −2 month −1 ), then average summer NCP p is 3.81 mol C m 32 % of the spring bloom NCP p (48.32 mol C m −2 month −1 ) respectively.Summing all of the positive NCP p values yields an annual value of 90.47 mol C m −2 .Comparing integrated summer NCP p (= 15.26 mol C m −2 ) to the annual value reveals that the summer phytoplankton assemblage accounts for ∼ 17 % of annual uptake of carbon, and reinforces the fact that summertime productivity is significant.Most importantly, two rates of carbon uptake can be delineated: a spring NCP p of 0.12 mol C m −3 month −1 associated with a microphytoplankton diatom assemblage, and an average summer NCP p of 0.034 mol C m −3 month −1 associated with a pico-and nanophytoplankton dominated assemblage.A composite seasonal cycle of pCO 2 constructed from data spanning 2007-2009 is shown in Fig. 5c.To account for the effect of water temperature, we present pCO 2 corrected to a constant annual mean temperature (pCO 2, norm ; µatm) (c.f.Takahashi et al., 2002; Shadwick et al., 2011).The seasonal evolution and succession of the Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 2000, 2001, 2005 and 2008), although 2001 and 2005 have in situ data points for only ∼ one half of the year.R 2 Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Fig. 3 .Fig. 5 .
Fig. 3. Climatological seasonal patterns in phytoplankton succession.(a) Seasonal cycle of diatoms, nanophytoplankton (nano), picophytoplankton (pico), pico and nano combined (pico + nano) and dinoflagellates (dinos).(b) Seasonal cycle of temperature (red line), Chl a (green line), diatoms (grey circles), dinoflagellates (grey squares) and pico + nano (grey triangles).The grey dotted lines indicate the collapse of the spring bloom in May at ∼ 5 • C, the shoaling of the mixed layer depth and the transition to a phytoplankton assemblage numerically dominated by smaller cells.(c-e) pico + nano, diatom and dino versus Chl a.

Table 1 .
Statistics of NCP p estimates for total phytoplankton assemblage and constituent size classes.