Estimates of Micro-, Nano-, and Picoplankton Contributions to Particle Export in the Northeast Pacific

The contributions of micro-, nano-, and to particle export were estimated from measurements of size-fractionated particulate 234 Th, organic carbon, and phytoplankton indicator pigments obtained during ﬁve cruises between 2010 and 2012 along Line P in the subarctic northeast Paciﬁc Ocean. Sinking ﬂuxes of particulate organic 5 carbon (POC) and indicator pigments were calculated from 234 Th– 238 U disequilibria and, during two cruises, measured by sediment trap at Ocean Station Papa. POC ﬂuxes at 100 m ranged from 0.65–7.95 mmol m − 2 d − 1 , similar in magnitude to previous results at Line P. Microplankton pigments dominate indicator pigment ﬂuxes (averaging 69 ± 19 % of total pigment ﬂux), while nanoplankton pigments comprised the majority of 10 pigment standing stocks (averaging 64 ± 23 % of total pigment standing stock). Indicator pigment loss rates (the ratio of pigment export ﬂux to pigment standing stock) point to preferential export of larger microplankton relative to smaller nano- and picoplankton. However, indicator pigments do not quantitatively trace particle export resulting from zooplankton grazing, which may be an important pathway for the export of small 15 phytoplankton. These results have important implications for understanding the magnitude and mechanisms controlling the biological pump at Line P in particular, and more generally in oligotrophic gyres and high-nutrient, low-chlorophyll regions where small phytoplankton a component of the autotrophic


Introduction
Phytoplankton community structure exerts an important influence on the strength and efficiency of the biological pump (Michaels and Silver, 1988;Boyd and Newton, 1999;Thibault et al., 1999;Brew et al., 2009;Lomas and Moran, 2011). Small nano-and picoplankton dominate the phytoplankton community in the oligotrophic gyres and high-nutrient, low-chlorophyll (HNLC) oceanographic regions. It has traditionally been 25 thought that small phytoplankton represent a relatively small fraction of the downward and Jackson, 2007; Stukel and Landry, 2010;Lomas and Moran, 2011). A better understanding of the controls on the relative importance of small phytoplankton in POC export is needed to refine our understanding of the magnitude and mechanisms controlling the biological pump, particularly as recent climate models predict an expansion of the oligotrophic gyres where small cells dominate (Irwin et al., 2006;Polovina et al., 10 2008; Morán et al., 2010).
Ocean Station Papa (OSP, 50 • N, 145 • W), the site of one of the longest-running ocean time-series, is located in the northeast Pacific Ocean in one of three major HNLC regions. Previous attempts to resolve the apparent paradox of low phytoplankton biomass and high nitrate concentrations at OSP concluded that a bottom-up control 15 related to iron limitation is most important for large phytoplankton (Muggli et al., 1996;Harrison, 2006;Marchetti et al., 2006), while microzooplankton grazing exerts a strong top-down control on pico-and nanoplankton (Landry et al., 1993;Harrison et al., 1999;Rivkin et al., 1999). Primary production at the stations proximal to the coast on Line P (P4 & P12) is not iron-limited and diatom blooms are typically observed in spring 20 and late summer (Boyd and Harrison, 1999;Thibault et al., 1999). At the offshore stations (including OSP) the phytoplankton community is dominated by cells < 5 µm and the seasonal variability of primary production is relatively low (∼ 25 mmol C m −2 d −1 in winter and ∼ 67 mmol C m −2 d −1 in summer) (Boyd and Harrison, 1999;Thibault et al., 1999;Choi et al., 2014). In contrast to the low variability in primary produc- nual net community production (ANCP) at OSP (6.3 ± 1.6 mmol C m −2 d −1 ), suggesting that the majority of organic carbon export is due to active transport by zooplankton and/or DOC export (Timothy et al., 2013;Emerson, 2014). This study builds upon prior investigations of phytoplankton community composition and export production along Line P by examining the distributions of organic carbon, 5 phytoplankton indicator pigments, and 234 Th in three particle size-fractions. Sinking fluxes of POC and indicator pigments from the upper waters (∼ 100 m) were calculated from the 234 Th-238 U disequilibrium and, during two cruises, measured at OSP using free-floating sediment traps. A comparison of indicator pigment fluxes with the respective standing stocks suggests that microplankton (20-200 µm) make up a higher 10 percentage of POC export than biomass, whereas pico-and nano plankton (0.2-2 µm and 2-20 µm) make up a lower percentage of POC export than biomass.

Study location
Sample collection was conducted at five stations along Line P (P4, P12, P16, P20, 15 and P26 (OSP)) during cruises aboard the CCGS John P. Tully in August 2010, February 2011, June 2011, February 2012, and June 2012 ( Fig. 1, Table 1). Line P is located at the southern edge of the Alaskan Gyre, and the prevailing winds and surface currents are west-east (Bograd et al., 1999). Because precipitation and continental run-off exceed evaporation, a permanent halocline exists at ∼ 100 m impeding deep winter 20 mixing. In addition, a seasonal thermocline forms at ∼ 50 m in spring and shoals to ∼ 20 m in summer (Freeland et al., 1997;Thibault et al., 1999;Freeland, 2013;Timothy et al., 2013

Net primary production by 14 C incubation
Rates of net primary production (NPP) were determined following the protocols outlined in Lomas et al. (2012). Samples were collected with Niskin bottles from seven depths in the euphotic zone corresponding to 1, 5,9,17,33,55, and 100 % of surface irradiance. Three "light" bottles, a single "dark" bottle, and a single initial (T) bottle were 5 each spiked with ∼ 10 µCi NaH 14 CO 3 . A sub-sample to confirm total added activity was removed from the T bottle at each light depth and immediately added to an equal volume of β-phenylethylamine. Bottles were incubated under simulated in situ conditions, using neutral density screening to mimic light levels at the depth of sample collection, in an on-deck incubator for ∼ 24 h. After incubation, 125 mL sub-samples from each light 10 and dark bottle were filtered through an Ahlstrom 151 (0.7 µm nominal pore size) and a Whatman Track Etch 5 µm filter and rinsed with 10 % HCl. Samples were counted on a Perkin Elmer TriCarb 2900LR ∼ 48 h after the addition of 5 mL of Ultima Gold (Perkin Elmer, USA) scintillation cocktail.

15
Total 234 Th (dissolved + particulate) analysis followed the procedures outlined in Bauman et al. (2013). Briefly, samples (4 L) were collected by Niskin bottle at 12 depths (surface to ∼ 500 m) and spiked with 230 Th to monitor Th recovery. Samples were then treated with 7-8 drops of concentrated NH 4 OH solution, followed by 25 µL of 0.2 M KMnO 4 , and finally with 11.5 µL of 1.0 M MnCl 2 to form a MnO 2 precipitate that 20 quantitatively scavenges Th Buesseler et al., 2001;van der Loeff et al., 2006 Study Laboratory (Knap et al., 1997). Fucoxanthin (FUCO), peridinin (PER), 19'hexanoyloxyfucoxanthin (HEX), 19'-butanoyloxyfucoxanthin (BUT), alloxanthin (ALLO), total chlorophyll b (TChl b), and zeaxanthin (ZEA) were analyzed as indicator pigments based on their correspondence to particular phytoplankton taxonomic groups. Indicator proportion factors (PFs) were calculated to further analyze the size-distribution of the 25 phytoplankton community (Hooker et al., 2005;Lomas and Moran, 2011 nanoplankton proportion factor (nPF), and ZEA was used to determine the picoplankton proportion factor (pPF) (Hooker et al., 2005;Moran, 2011). Hooker et al. (2005) included TChl b in pPF, but because Prochlorococcus is not found in the study region, it was assumed in this study that any Chl b would be found in cells (e.g., chlorophytes and euglenophytes) in the nanoplankton size-class.   (Knap et al., 1997). Filters for anal- 15 ysis of POC and 234 Th were frozen at −2 • C until analysis. A sub-sample (∼ 30 % by weight) was cut with acetone-cleaned stainless steel scissors from each 234 Th filter for POC analysis, and these sub-samples were dried and fumed with concentrated HCl as described above. POC was then measured using a CE 440 CHN Elemental Analyzer (Exeter Analytical, Inc., Chelmsford, MA). The 234 Th filter subsample was dried at 60 • C 20 in a drying oven and counted on a RISØ beta detector as noted above. iment traps at any other stations sampled as part of this study. The trap design and sampling procedure is described in Baumann et al. (2012). Four tubes (72 mm diameter, 450 mm length) were used at each depth, and tubes were filled with non-poisoned, 0.4 µm filtered brine (S =∼ 85 ‰) prior to deployment. Upon recovery trap brines were combined, particles were re-suspended and filtered onto pre-combusted GF/F filters, 5 and swimmers were removed. Filters were stored frozen and later analyzed for POC, 234 Th, and indicator pigments as described above.

Hydrography and NPP
Depth sections of temperature and density anomaly (sigma-t) were generated using re-10 sults from all CTD casts for a given cruise to improve horizontal data resolution (Fig. 2). The seasonal change in water temperature is largely confined to the upper ∼ 100 m. Surface temperatures in August 2010 were ∼ 14 • C, while during the February cruises, surface temperatures were slightly cooler offshore (∼ 6 • C) than inshore (∼ 8 • C). During the June cruises, inshore temperatures were warmer (∼ 10-12 • C) while offshore 15 temperatures remained relatively cool (∼ 8 • C). Density anomaly did not vary greatly between cruises below ∼ 100 m. During the winter, a pool of less dense water (density of 1023-1025 kg m −3 ) was observed toward the coast (east of ∼ 126 • W). During the June cruises, this pool was observed extending west to ∼ 130 • W and during August 2010, it extended out to OSP (145 • W). These data follow the expected seasonal 20 pattern of a well-mixed water column in winter and increasing stratification moving from spring to summer. Total NPP and > 5 µm size-fractionated NPP values were trapezoidally integrated over the euphotic zone ( a factor of two with the seasonal averages reported by Boyd and Harrison (1999). A maximum > 5 µm NPP of 39.6 mmol m −2 d −1 was at station P4 during June 2012 and a minimum of 2.2 mmol m −2 d −1 was measured at station P12 in February 2012.

Small-and large-volume POC concentrations
Suspended POC concentrations from Niskin bottle samples collected in the photic 5 zone range from 1.1-7.1 µmol L −1 . POC concentrations were generally lowest at the base of the photic zone, though decreasing concentrations with depth were not observed at all stations (Table S1). The highest suspended POC concentrations were measured at station P4 during all cruises. POC concentrations were also measured in three size-fractions of particles collected with large-volume in situ pumps (Table S2).

10
Concentrations of each size-fraction tended to decrease with depth and were typically less than 0.5 µmol L −1 at all depths. One exception was at station P26 during February 2011 when POC concentrations at 30 m were between 1.8 and 2.9 µmol L −1 for all size-fractions. The concentrations of POC collected using small-volume and large-volume methods 15 often do not agree for samples collected at the same location and depth (Gardner, 1977;Moran et al., 1999;Liu et al., 2005Liu et al., , 2009. As reported in these previous studies, POC concentrations measured by large-volume in situ pumps (summed for all size-fractions) are significantly (ANOVA, p < 0.05) less than small-volume POC measurements from the same station and similar depth (Fig. 3a). Explanations put forth 20 to account for this discrepancy include DOC adsorption to filters, pressure effects on particle retention in pump samples, the collection of zooplankton by Niskin bottles but not pumps, and particle washout from pump filters Liu et al., 2005Liu et al., , 2009. In this study, the smallest pump size-fraction was collected using a 1 µm Nitex screen, not a GF/F, resulting in the pumps missing the portion of the POC on parti-25 cles between 0.7 and 1 µm, which may further contribute to the difference observed between the two methods. Lomas and Moran (2011) reported that sonication of in situ BGD 11,2014 Estimates of micro-, nano-, and picoplankton contributions to particle export B. L. Mackinson et al. pump samples to resuspend particles from the Nitex screens had no significant effect on measured POC concentrations.

Particulate 234 Th and POC/ 234 Th ratios
Size-fractionated particulate 234 Th activities in samples collected by in situ pump generally decrease with depth, and are typically less than 0.1 dpm L −1 (Table S2). As with 5 in situ pump POC concentrations, station P26 during February 2011 is an exception, with values exceeding 0.1 dpm L −1 for all size fractions at 30 m and throughout most of the water column for the 1-10 µm fraction. Size-fractionated POC/ 234 Th ratios (Fig. 4, Table S2) are less than ∼ 6 µmol dpm −1 for all size-classes at most stations, with higher values measured at stations P4 and P12 in February 2012 and P4 in June 2012.

10
POC/ 234 Th ratios tend to decrease or remain constant with depth, with one exception at station P12 during February 2012 where the maximum POC/ 234 Th was at 100 m for all size fractions. Also, the POC/ 234 Th ratio does not vary greatly between sizefractions ( Fig. 4) as was observed in Speicher et al. (2006) and Brew et al. (2009). The accuracy of 234 Th as a tracer of POC export depends on the assumption that 15 234 Th and POC are sinking on the same particles, and therefore sinking at the same rate (Moran et al., 2003;Smith et al., 2006;Speicher et al., 2006;Burd et al., 2007;Brew et al., 2009). A high degree of correlation between the size-fractionated distributions of 234 Th and POC ( Fig. 4) along Line P provides evidence in support of this assumption. All correlations were statistically significant (p < 0.05) and imply a strong BGD 11,2014 Estimates of micro-, nano-, and picoplankton contributions to particle export B. L. Mackinson et al.

Small-volume Chl a and indicator pigments
Concentrations of total Chl a and > 5 µm Chl a measured by fluorometer (Table S1) were trapezoidally integrated over the photic zone to determine respective standing stocks. During August 2010, the > 5 µm fraction accounted for > 30 % of the Chl a at all stations, with a maximum of 50 % at station P26. During the other four cruises, the 5 > 5 µm size-fraction generally accounted for < 30 % of the total Chl a, except at station P26 in February 2012 and station P4 in June 2012. Previous studies have reported that larger cells are more abundant at stations closer to the coast (Boyd and Harrison, 1999), though this was not always apparent. The highest > 5 µm percentage of Chl a was measured at station P26 during August 201, June 2011, and February 2012. Phy-10 toplankton indicator pigments and Chl a concentrations in samples from the euphotic zone samples were also measured by HPLC (Table S1). HPLC and fluorescence Chl a concentrations generally agreed to within a factor of two, and the correlation between the two measurements was statistically significant (p < 0.05) (Fig. S1). The correlation between the sum of the indicator pigment concentrations and the Chl a concentration 15 was statistically significant (p < 0.05) and roughly 1 : 1, suggesting that the indicator pigments examined in this analysis accounted for most of the phytoplankton biomass (Fig. S2). Furthermore, the correlation between the > 5 µm fraction of Chl a and mPF is statistically significant (p < 0.05), suggesting that this PF is a reasonable representation of that size-fraction of the phytoplankton community. Profiles of indicator pigment 20 concentrations were trapezoidally integrated over the photic zone to quantify standing stocks (Table 3). FUCO was the most abundant microplankton pigment, and HEX was the most abundant nanoplankton pigment at most stations. Indicator pigment PFs (Fig. 5, Table S3) reveal that the phytoplankton community was typically dominated by nanoplankton, although at P4, and to a lesser extent at P20 in June 2012, microplank-BGD 11,2014 Estimates of micro-, nano-, and picoplankton contributions to particle export B. L. Mackinson et al.

Large-volume size-fractionated Chl a and indicator pigments
Size-fractionated Chl a and indicator pigment concentrations were also measured by in situ pump (Table S4). Chl a was once again strongly correlated in a roughly 1 : 1 ratio with the sum of the indicator pigments (p < 0.05) (Fig. S3). The highest Chl a concentrations were measured in the 10-53 µm fraction during all cruises. In February 2012, 5 the > 53 µm fraction generally had the lowest concentrations, while in June 2012 and June 2011 the lowest concentrations were generally in the 1-10 µm fraction. Ideally, small-volume and large-volume concentrations of Chl a and indicator pigments should agree for samples collected at the same station and depth, but this was not observed in this study (Fig. 3). Although differences between small-and large-10 volume measurements of POC have been reported (Gardner, 1977;Moran et al., 1999;Liu et al., 2005Liu et al., , 2009, few studies have compared Niskin bottle and in situ pump measurements of indicator pigments (Lomas and Moran, 2011). Relative to bottle samples, the pump samples indicate higher concentrations of microplankton pigments FUCO and PER and lower concentrations of ZEA and TChl b, which are pigments associated 15 with pico-and nanoplankton ( Fig. 3b-d). Large-volume pump and small-volume bottle measurements of the nanoplankton indicator pigments HEX, BUT, and ALLO generally agree within a factor of two ( Fig. 3b-d). Given the small size of ZEA-containing Synechococcus and TChl b-containing chlorophytes and prasinophytes, it is likely that many of these cells pass through the 1 µm Nitex screen which would lead to under-20 sampling by the pumps (Liu et al., 2005). Bottles may undersample large, rare cells because the small volume might not be a statistically representative sample (Lomas and Moran, 2011). Furthermore, larger cells may settle below the spigot of the Niskin bottles, leading to a further bias against the collection of large cells (Gardner, 1977;Gundersen et al., 2001). Pumps sample higher concentrations of Chl a than bottles 25 (Fig. 3a) at stations with high concentrations of Chl a, but when Chl a concentrations are low (< 200 ng L −1 ), the pumps tend to undersample relative to the bottles. Given these sampling differences, it is important to note that although the total concentrations (summed for all size-fractions) measured by the in situ pumps may be inaccurate, it is still possible that the > 53 µm fraction accurately represents the composition of sinking particles. The disruption of loosely-bound aggregates during collection by the pumps could cause an error in the > 53 µm fraction, but this is considered un-5 likely due to the presence of nanoplankton (and in some cases picoplankton) pigments in this fraction. Furthermore, a recent study in the Sargasso Sea employed a similar methodology and also found picoplankton pigments in three particle size-classes, each > 10 µm (Lomas and Moran, 2011).

BGD
Indicator pigment PFs calculated for the size-fractionated particles (Table S3) and 10 plotted against depth (Figs. 6-8) reveal that while the overall indicator pigment concentrations vary with depth and across size-fractions, the PFs do not exhibit a systematic pattern of variation across size classes, depths, or seasons. The picoplankton pigment ZEA typically represents < 10 % of the total indicator pigments for all size classes. Microplankton pigments dominated samples at station P4 in February 2012 15 and June 2012, with mPFs typically exceeding 0.5 and 0.8, respectively, for each cruise. In addition, mPFs were high at station P26 during these times, with values generally exceeding 0.5 ( Figs. 7 and 8). Nanoplankton pigments dominated at station P12 in February 2012 cruise with nPFs exceeding 0.5 for most samples. As with the small volume samples, FUCO was usually the most abundant microplankton pigment 20 while HEX was usually the most abundant nanoplankton pigment (Table S4).

Total
where A U is the activity of 238 U, λ Th is the 234 Th decay constant, A Th is the activity of 234 Th, P Th is the vertical flux of 234 Th on sinking particles, K h is the eddy diffusion 5 coefficient, and U h is the current velocity (Coale and Bruland, 1985;Charette et al., 1999). Assuming a steady-state (∂A Th /∂t = 0) over several weeks to months, and that the diffusive flux of 234 Th is small relative to advection and can therefore be ignored, the vertical flux of 234 Th is defined by, where z is the depth of the water column over which the flux is measured. In this study, the gradient of thorium (∂A Th /∂x) was only measured in the east-west direction (along Line P). Therefore, x is the east-west distance across which the gradient will be measured and U h is the east-west current velocity. Given that the currents in the 15 region generally flow west-east, and with no data at stations north and south of Line P, the north-south transport of 234 Th by advection had to be assumed to be negligible. At stations P12, P16, and P20, the 234 Th gradient was measured between the adjacent stations. For stations P4 and P26 (at either end of Line P), the gradient of 234 Th was determined from the adjacent station assuming a linear change extended beyond the 20 measured transect. 234 Th fluxes (P Th ) calculated using the 2-D model are within 5 % of fluxes determined using a steady-state 1-D model that ignores advection (Fig. S4) higher during the August and June cruises than during the February cruises (Fig. 9a) . Also, 234 Th fluxes did not exhibit a consistent trend along Line P.

234 Th-derived POC fluxes
The POC/ 234 Th ratio in the > 53 µm size-class and P Th for a given depth horizon were 5 used to calculate POC fluxes (P POC ) (Fig. 9). In most cases, P POC decreases with depth, although in some cases, the maximum P POC in a given profile occurs at 50 or 100 m. P POC fluxes at 100 m range from 0.65-7.95 mmol m −2 d −1 ; they are generally higher in summer than winter, and highest at station P4, consistent with previous studies at Line P Wong et al., 1999;Timothy et al., 2013). 10 The ratio of P POC flux to NPP, referred to as the ThE-ratio, is an estimate of efficiency of the biological pump (Buesseler, 1998). ThE-ratios determined using P POC fluxes at the base of the photic zone (Table 2, Fig. 10) are similar to those reported by Charette et al. (1999), and are also in line with an annual average e ratio determined using average sediment trap POC fluxes (Wong et al., 1999) and annual average NPP (Harrison,15 2002) (Fig. 10).

Sediment trap 234 Th and POC fluxes
The particle fluxes of both 234 Th and POC fluxes determined by the PITS traps (F Th and F POC respectively) generally decreasee with depth (Table 4). F Th was higher in June 2012 than in June 2011, though there was no clear difference between the two 20 cruises for F POC . A comparison of the F Th with the P Th from corresponding stations and depths indicates that the F Th is consistently higher than the P Th , though usually not by more than a factor of two. F POC is also consistently higher than P POC , though again not by more than a factor of two (Fig. 11a). The POC/ 234 Th ratios of particles caught in sediment traps (Table 9) tend to be slightly higher (generally within a factor of 2) than 25 the ratio of particles sampled by pumps at the corresponding station and depth.

234 Th-derived and sediment trap pigment fluxes
Sinking fluxes of Chl a (P Chla ) and indicator pigments (P Pigment ) were calculated from P Th and the Pigment/ 234 Th ratio measured on > 53 µm particles. Chl a and indicator pigment fluxes (Table 3, Fig. 11a-c) are generally highest at station P4 and decrease moving offshore. The highest indicator pigment fluxes were typically observed for mi-5 croplankton pigments (FUCO and PER) whereas the lowest were observed for the picoplankton pigment ZEA (Table 3, Fig. 12a-c). Sediment trap pigment fluxes (F Pigment ) were typically lower than P Pigment (Table 3, Fig. 11b). The maximum sediment trap fluxes of Chl a and most indicator pigments were determined at 50 m in June 2011 and at 30 m in June 2012 (Table 3). For both 10 deployments the deepest fluxes were generally the lowest, presumably due to the progressive degradation of sinking phytoplankton and resulting loss of pigments. Chl a and indicator pigment fluxes were generally higher in June 2011 than in June 2012, which is the opposite of the trend observed for F Th .
Pigment PFs determined for material captured by the PITS traps do not vary greatly 15 with depth, suggesting that the quality of material sinking to depth is similar to that in the surface water, despite the general decrease of material (Figs. 6 and 8). Microplankton PFs are higher for trap samples than for bottle samples but not as high as for pump samples, while nPFs and pPFs are higher for trap samples than for pump samples but lower than for bottle samples.

Discussion
The results presented in this study build on previous investigations of export production in the northeast Pacific by providing estimates of the relative contributions of different phytoplankton size-classes to particle export.  Fig. 12). Sediment trap pigment fluxes indicate a lower, but still substantial, relative contribution of microplankton to export, with microplankton pigments making up 47 % and 33 % of the total sediment trap indicator pigment flux in June 2011 and June 2012 respectively, as compared to 81 % and 85 % of total P Pigment fluxes.

5
Though nano-and picoplankton did not form the majority of the algal aggregate flux, their 29±19 % contribution is significant and similar to contributions reported by Lomas and Moran (2011) for cyanobacteria and nano-eukaryotes in the Sargasso Sea. Indicator pigment loss rates determined from both P Pigment fluxes and sediment trap pigment fluxes imply that microplankton are exported more efficiently than nano-or 10 picoplankton (Table 3, Fig. 12d-f). Loss rates of pigments, estimated as the ratio of P Pigment fluxes to pigment standing stock, averaged (for all cruises) 8 ± 12 % for microplankton pigments, 1 ± 2 % for nanoplankton pigments and 0.6 ± 1 % for picoplankton pigments. These results suggest that export of large cells by direct sinking of algal aggregates is more efficient than the export of small cells by the same pathway. 15 Sediment trap loss rates for microplankton were also higher than those for nano-and picoplankton, further indicating preferential export of microplankton. Even though differences between bottle and pump samples may exaggerate the extent to which large cells dominate export, sediment trap loss rates support and confirm the preferential export of large cells by algal aggregation. 20 In contrast to the trends observed for pigment fluxes and loss rates, the low variability of pump indicator pigment PFs with depth (Figs. 6-8) does not appear to indicate preferential export of microplankton. Furthermore, the presence of nano-and picoplankton pigments in the > 53 µm size-fraction and in samples below the mixed layer suggests that nano-and picoplankton are incorporated into aggregates and that some of these 25 aggregates are exported from the surface ocean. If large cells were being preferentially exported, microplankton pigments would be expected to make up a larger percentage of total pigments in samples below the mixed layer than in samples from the mixed layer, but this is not observed in the results of this study. It is possible that some of this Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | discrepancy can be attributed to differences between bottle and pump samples. Because cells < 1 µm in size can pass through the 1 µm Nitex screens used in the pumps, the sum of the pump size-fractions does not accurately reflect the community composition in the euphotic zone, and may miss a change in indicator pigment PFs with depth. In addition, the under-sampling of large cells by Niskin bottles may lead to an underes-5 timate of microplankton standing stocks, and thus and overestimate of microplankton loss rates. These pigment fluxes are likely lower estimates of the total contribution of each phytoplankton group to particle export. The use of indicator pigments as tracers of phytoplankton export only accounts for the direct sinking of healthy, ungrazed cells, because 10 grazing degrades the indicator pigments to an analytically undetectable form (Head and Harris, 1992;Strom et al., 1998;Thibault et al., 1999). Indirect export (via grazing) is thought to be an important pathway for picoplankton export in the HNLC Equatorial Pacific (Richardson et al., 2004;Stukel and Landry, 2010). Given that grazing has been shown to control the biomass of small phytoplankton in the northeast Pacific (Landry 15 et al., 1993;Harrison et al., 1999;Rivkin et al., 1999), indirect export may also be a significant pathway for small cell export in this region. Because this pathway is not accounted for by the methodology employed in this study, the results presented here may underestimate the export of small phytoplankton, which may be less likely to sink directly. 20 Although grazing and fecal pellet export were not directly measured in this study, a comparison of sediment trap and pump measurements of Chl a, indicator pigments, and POC, suggests that zooplankton fecal pellets may be an important component of POC export at OSP, at least in spring (Fig. 11). While F POC fluxes are higher than the corresponding P POC fluxes, F Pigment fluxes are lower than P Pigment fluxes, indicating that 25 the material captured by the sediment traps is enriched in carbon and depleted in Chl a and indicator pigments relative to that sampled by the pumps. Because the trap brine was not poisoned, zooplankton grazing and cell degradation in the trap tube may also have contributed to some loss of pigments over the ∼ 3 day deployment of the PITS traps. However, the collection of carbon-rich and pigment-depleted fecal pellets by the traps but not by the pumps, which do not quantitatively sample fecal pellets (Lomas and Moran, 2011), could also explain these observations. This latter explanation is consistent with the results presented in Thibault et al. (1999), which indicate that fecal pellet export is 3 to 6 times greater than algal aggregate export at Line P. 5

Conclusions
New estimates of phytoplankton indicator pigment loss rates calculated from both 234 Th-derived and sediment trap pigment fluxes suggest that large cells are preferentially exported at Line P. Specifically, microplankton pigments on average made up 69 ± 19 % of the total pigment flux, but only 32 ± 24 % of pigment standing stock, 10 whereas nano-and picoplankton pigments on average formed 31 ± 19 % of pigment flux in spite of representing 68 ± 24 % of the standing stock. These results are consistent with traditional food web models (Michaels and Silver, 1988;Legendre and Le Fèvre, 1995) that suggest nano-and picoplankton are underrepresented in particle flux relative to their contribution to phytoplankton biomass; they also lend support to 15 the conclusions of Choi et al. (2014). However, the methods employed in this study do not quantitatively account for export via zooplankton fecal pellets, which could be significant for small phytoplankton as they are controlled by grazing in this region (Landry et al., 1993;Harrison et al., 1999;Rivkin et al., 1999;Thibault et al., 1999). Furthermore, the determination of pigment loss rates also required a comparison between 20 small-and large-volume samples, and the inherent differences of these sampling techniques likely led to an overestimation of the microplankton contribution to algal aggregate export. Therefore, it is possible that all sizes-classes of phytoplankton contribute to POC export in approximate proportion to their contribution to NPP as predicted by Richardson and Jackson (2007).
This study, conducted in a subarctic HNLC region, contributes to the ongoing discussion of small cell export that has largely focused on tropical and subtropical regions BGD 11,2014 Estimates of micro-, nano-, and picoplankton contributions to particle export B. L. Mackinson et al.  (Richardson et al., 2004(Richardson et al., , 2006Richardson and Jackson, 2007;Stukel and Landry, 2010;Lomas and Moran, 2011). In particular, these results suggest that nano-and picoplankton may contribute significantly to POC export in this subarctic HNLC region, even if they are not as efficiently exported as larger microplankton. If large phytoplankton drive more efficient POC export in the northeast Pacific as suggested by this study, 5 it could have important implications for understanding the biological pump. It has been proposed that decreasing winter mixed layer depths (Freeland et al., 1997;Freeland, 2013) and variations of macronutrient concentrations linked to shifts in climate regime (Pena and Varela, 2007) in the northeast Pacific could lead to shifts in the phytoplankton community composition. This study suggests that such changes in phytoplankton 10 community composition could significantly affect the efficiency of the biological pump, and in turn, the cycling of carbon. While the results indicate that shifts in community composition favoring larger phytoplankton could lead to more efficient particle export, they do not indicate that shifts favoring smaller phytoplankton would lead to a shutdown of POC export as suggested by some previous studies (e.g., Michaels and Silver, 15 1988), but merely that the export of POC could be less efficient.   Table 3. Chl a and indicator pigment standing stocks determined by integrating small volume pigment concentrations (determined by HPLC) across the photic zone, pigment fluxes ( 234 Th and PITS-derived) measured at the base of the photic zone, and pigment loss rates, or the percent of the surface concentration represented by those fluxes. Pigment standing stocks are in mg m −2 and pigment fluxes are in mg m −2 d −1 .  BGD 11,2014 Estimates of micro-, nano-, and picoplankton contributions to particle export  BGD 11,2014 Estimates of micro-, nano-, and picoplankton contributions to particle export    234 Th-derived indicator pigment fluxes determined using the Pigment/ 234 Th ratio on > 53 µm particles plotted for micro-, nano-, and picoplankton pigments. (d-f) Indicator pigment standing stocks plotted against indicator pigment fluxes for micro-, nano-, and picoplankton pigments. The slopes of the dashed lines indicate pigment loss rates. (g-i) The contribution to total pigment standing stock plotted against the contribution to total pigment flux for micro-, nano-, and picoplankton pigments. Data points above the 1 : 1 line indicate preferential export by direct sinking and points below the 1 : 1 line indicate disproportionately low export by direct sinking relative to biomass contributions.