Phaeodaria: An Important Carrier of Particulate Organic Carbon in the Mesopelagic Twilight Zone of the North Pacific Ocean

Phaeodaria, which comprise one group of large, single‐celled eukaryotic zooplankton, have been largely ignored by past marine biological studies because Phaeodaria and their delicate skeletons are liable to collapse. As a result, collection and quantification of specimens are difficult, and seasonal changes of phaeodarian abundance have not been thoroughly studied. The transport of biogenic elements by sinking phaeodarians has been estimated for only a few representative species. Sinking particles >1 mm in size and swimmers have traditionally been excluded when estimating sinking particle fluxes. The focus of this study is the large number of phaeodarians among the >1‐mm sinking particles collected in the western North Pacific from June 2014 to July 2015. Careful sorting by microscopic examination and chemical analyses revealed that phaeodarians accounted for up to about 10% of the organic carbon in all sinking particles and accounted for a mean of 33% of the organic carbon in the >1‐mm sinking particles. The high‐standing stocks of phaeodarians at depths of 150–1,000 m in the mesopelagic twilight zone suggested that particles sinking from the euphotic zone as aggregates and fecal pellets can be efficiently exported to the deep sea by the ballasting effect of large phaeodarian particles rich in organic carbon.


Introduction
The biological carbon pump (BCP), which transports particulate organic carbon (POC) as sinking particles from the surface ocean to the ocean's interior, plays an important role in regulating atmospheric carbon dioxide (CO 2 ) concentrations (e.g., Henson et al., 2011). Shipboard and satellite observations have been used to study biogeochemical cycling by the BCP with respect to the standing stocks and production of zooplankton and phytoplankton in water shallower than 200 m. At depths greater than 1,000 m, where hydrodynamic disturbances are small, the flux and chemical composition of sinking particles have been studied by employing time series sediment traps. In contrast, there is limited information about the processes of POC transport in the mesopelagic "twilight zone" at depths of 200-1,000 m (Buesseler & Boyd, 2009). The major processes affecting the efficiency of POC transport in the twilight zone have been discussed with respect to the proportion of fecal pellets versus phytoplankton aggregates, the fraction of export associated with ballast minerals, and the sinking rates of particles (Armstrong et al., 2002;Buesseler et al., 2007Buesseler et al., , 2008Francois et al., 2002;Honda & Watanabe, 2010;Klaas & Archer, 2002;Passow et al., 2003). However, the role of mesopelagic zooplankton communities, with the exception of abundant crustaceans, in POC transport has received relatively little attention.
In recent years, a worldwide in situ imaging survey of plankton larger than 600 μm has revealed that phaeodarians make a substantial contribution to zooplankton biomass at depths of 100-500 m in the global ocean (Biard et al., 2016), and their importance to biogeochemical cycles has become recognized Guidi et al., 2016;Stukel et al., 2018). Phaeodaria represent a group of holoplanktonic marine protozoa with mostly fragile and delicate siliceous skeletons that belongs to the phylum Cercozoa of the infrakingdom Rhizaria (Nakamura & Suzuki, 2015). Phaeodarians have been largely ignored in marine biology up to now because their delicate skeletons and hollow structure make them liable to collapse, and it has been difficult to collect and quantify specimens (Nakamura & Suzuki, 2015). Previous phaeodarian studies mainly focused on the <1-mm fraction because this is where important phytoplankton components can be found (e.g., Ikenoue et al., 2012Ikenoue et al., , 2015Okazaki et al., 2005;Takahashi & Honjo, 1981;Takahashi, 1987Takahashi, , 1991aTakahashi, , 1991b, and only a handful of studies have estimated the contribution to carbon and silica cycles of selected phaeodarian species Gowing, 1986;Michaels et al., 1995;Stukel et al., 2018;Takahashi, 1991aTakahashi, , 1991bTakahashi et al., 1983). There has still been no quantitative estimate of the contribution of all phaeodarian species to vertical POC export in the mesopelagic zone.
High-standing stocks of phaeodarians in the mesopelagic zone have been reported in the Western Subarctic Gyre (WSG) of the North Pacific . The WSG is well known as the terminal area of deepwater circulation. The WSG is therefore rich in nutrients and is associated with high BCP efficiency (Honda et al., 2002;. Phaeodarians may therefore be an important player in the WSG mesopelagic food web and contribute to the high efficiency of its BCP. Sinking particles larger than 1 mm have been recognized as swimmers (swimming zooplankton that are thought to actively enter into sediment trap collection cups and augment the trap contents) and have been traditionally separated from the sinking particles in most sediment trap studies (Steinberg et al., 1998). In this study, we often found phaeodarians among sinking particles larger than 1 mm in the WSG. Here we present the seasonal variability of the POC export fluxes derived from phaeodarians >1 mm in size collected in a sediment trap moored at a depth of 1,000 m in the WSG from June 2014 to July 2015. To determine the original depths of the habitats of the phaeodarians observed among the sinking particles, we determined the vertical distribution of the standing stocks of each of the phaeodarian taxonomic groups in the WSG. Our study is the first attempt to measure the organic carbon contents in phaeodaria in >1-mm fraction employing an elemental analyzer. Biard et al. (2016) and Stukel et al. (2018) quantified phaeodarian biovolumes and estimated phaeodarian carbon contents from the biovolumes and existing conversion factor. However, there have been actually no direct measurements of carbon contents in phaeodarians, other than our present study. Our study revealed that >1-mm sinking particles exported substantial amount of particulate organic carbon in the western North Pacific and that one of the major carriers in the twilight zone was Phaeodaria.

Sediment Trap Samples
Sinking particles were collected with a sediment trap (McLane Mark 7G-21, aperture area: 0.5 m 2 ; Honjo & Doherty, 1988) at Station K2 in the western North Pacific (47°N, 160°E; seafloor depth of 5,280 m) for 400 days from 1 June 2014 to 6 July 2015. The trap was deployed at a fixed depth (1,000 m) and collected sinking particles with 20 bottles that rotated at consecutive intervals of 20 days. To preserve the trapped particles, each sample bottle was filled with prefiltered deep-sea water with 5% formalin (pH neutralized) before the sediment trap deployment. The mooring deployment and recoveries were operated from the research vessel Hakuho Maru of the Japan Agency for Marine-Earth Science and Technology during Cruises KH14-02 and KH15-J01. The samples were water-sieved (deep-sea water) through a 1-mm-mesh Nytex screen and were divided into the >1-and <1-mm fraction at a shore-based laboratory. Swimmers (Figure 1a), zooplankton carcasses (ZC samples; Figure 1b), phaeodarians, and particles adhered to phaeodarian skeletons were major components of the >1-mm fraction. Swimming zooplankton with no physical defects were defined as swimmers, and other zooplankton fragments were defined as ZC. The >1-mm fractions were split into two equal aliquots with a Motoda box splitter (Motoda, 1959) and then they were examined under a Nikon SMZ18 stereomicroscope to remove swimmers and to sort the remainder of the samples. Identification of swimmers followed Chihara and Murano (1997). The >1-mm fractions, from which swimmers had been removed, were defined as >1-mm sinking particles. The >1-mm sinking particles were sorted into taxon groups in Figures 1b-1e, and the organic carbon in each group was measured. Organic carbon content was also measured for the <1-mm sinking particles. Details are described below, and the process is also summarized in the flowchart in Figure 2.

Classification and Chemical Analysis of the >1-mm Sinking Particles
The >1-mm sinking particles consisted of zooplankton carcasses (ZC samples; Figures 1b and 2), phaeodarians, and particles adhered to phaeodarian skeletons. The particles adhered to most phaeodarian skeletons consisted of detritus, microplankton such as diatoms, silicoflagellates, radiolarians, foraminifers, and phaeodarians. Zooplankton carcasses were sorted with tweezers (Hammacher, catalog number HWC 118-10) and placed into the preservative solution, separately from other particles. In most cases the phaeodarian taxa were identified at the family level based on Boltovskoy (1999), Takahashi and Anderson (2002), and Nakamura and Suzuki (2015) (Table 1 and Figure 3). In this study, any phaeodarian or colony of phaeodarians with a size >1 mm was defined as Phaeodaria. That is, phaeodarians <1 mm in size, which constituted the adherent particles, were not treated as Phaeodaria in our results. Adherent particles were manually separated from the phaeodarians with tweezers if the particles were larger than 500 μm. The phaeodarians were then carefully rinsed several times through a 500-μm sieve. A 500-μm sieve was used to wash phaeodarian skeletons in order to prevent undersieving of broken and deformed phaeodarians as well as individual phaeodarians <1 mm in size that had formed colonies. Adherent particles <500 μm in size that washed through the sieve were gathered and placed into a preservative with the manually collected particles >500 μm in size. This mixture of adherent particles >500 and <500 μm in size was defined as adherent particles (AP samples; Figures 1c and 2). Most of the microplankton in adherent particles mentioned above was included in the AP samples. The rinsed phaeodarians without adherent particles were also placed into the preservative solution separately. However, substantial amounts of adherent particles still remained on the surface of some phaeodarians after rinsing. Therefore, for accurate determination of POC derived from phaeodarians, the phaeodarians with remaining adherent particles were dissected under a stereomicroscope and separated into skeletons with adherent particles and the internal soft parts, which consisted mainly of the central capsule and phaeodium. The phaeodarian skeletons with adherent particles were defined as SAP samples (Figures 1d and 2). Table 1 shows details of the dissection method for each taxon. Figure S1 shows examples of dissections of phaeodarians into skeleton and internal soft parts. The phaeodarian internal soft parts and the rinsed phaeodarians without adherent particles were combined, and the mixture was defined as phaeodarians and their internal soft parts (PHA samples; Figures 1e and 2). The phaeodarian cell body is composed mainly of a siliceous skeleton (scleracoma) and internal soft parts (malacoma). The phaeodarian siliceous skeleton contains no organic matter within itself (Takahashi et al., 1983;Takahashi & Hurd, 2007). The malacoma contains the phaeodium and protoplasm such as the central capsule and ectoplasm (Nakamura & Suzuki, 2015). The organic carbon is presumed to be concentrated within the central capsule and phaeodium . Therefore, POC derived from whole phaeodarian communities was determined by measuring the organic carbon contents of PHA samples.
The above taxonomic groups are arranged as follows: Each sample for AP, SAP, and PHA analysis was filtered onto a weighed 25-mm Whatman GF/F filter with 0.7-μm pores. The filters had been burned for 2 hr at 450°C in advance to remove organic matter and were desalted with Milli-Q water. Each swimmer sample and ZC sample was filtered onto a weighed 47-mm Whatman polycarbonate membrane filter with 0.4-μm pores and was desalted with Milli-Q water. The filters with samples were dried in a constant temperature oven at 50°C for 24 hr. The dried filters with samples were weighed on an electronic balance (A&D Company, BM-252). The ZC samples were ground into a homogeneous powder with an agate pestle and mortar. Each powdered ZC sample was placed in a silver capsule and moistened with Milli-Q water. All of the samples were placed in a glass desiccator and fumigated with HCl mist for approximately 24 hr to eliminate CaCO 3 (decalcification). The samples were then dried and neutralized with P 2 O 5 and NaOH in a desiccator for 72 hr. After decalcification, each sample in its silver capsule was wrapped within a tin capsule. The wrapped samples were placed in the autosampler of an elemental analyzer (CHN Analyzer Flash EA-1112; Thermo Finnigan, CA) to measure organic carbon concentrations. The combination of GF/F filter, silver capsule, and tin capsule wrapping each sample was treated as a blank. The average value of three or more blanks was subtracted from the carbon signal of the sample. The detection limit of carbon content was 1 μg or less, and the measured values were at least 70 times higher than the lower limit of detection. Organic carbon concentrations were determined by comparison with a calibration curve, which was prepared with an official certified standard material (acetanilide). Relative standard deviation of carbon contents by repeated measurements on acetanilide were ±0.7% (n = 13).

Chemical Analysis of the <1-mm Sinking Particles
The fine particle fraction (<1 mm) was split into 10 equal aliquots for chemical analysis using a rotary splitter. The samples on the Nuclepore filters were dried at around 50°C for 24 hr, and the dry weights were measured to estimate mass fluxes. The organic carbon contents were determined for decalcified samples. Particulate organic carbon and nitrogen in the samples were measured with a CHN Analyzer NCS2500, Thermo Quest. Details of all procedures have been previously reported by Honda et al. (2016).

Plankton Tow Samples (>1 mm)
Plankton samples in water column (hereinafter plankton tow samples) were collected with a verticalmultiple-plankton sampler to investigate the depth distribution of phaeodarians observed in sinking particles. The instrument (net mesh size: 62 μm, aperture area: 0.25 m 2 ) was towed within nine discrete water layers (0-30, 30-50, 50-100, 100-150, 150-200, 200-500, 500-1,000, 1,000-1,500, and 1,500-2,000 m) at Station K2 on 8 July 2015 (Table S1). The volume of seawater filtered through the net was estimated by using a flowmeter mounted in the mouth of the vertical-multiple-plankton sampler. The samples collected by vertical-multiple-plankton sampler were split equally into two or four aliquots with a Motoda box splitter (Motoda, 1959). The split samples were water-sieved through a 1-mm-mesh stainless steel screen, and the material retained on the screen was fixed with 99.5% ethanol for taxonomic counts and chemical analyses. Tuscaroridae, Castanellidae, and Circoporidae usually occurred as colony in sediment trap samples but occurred as individual cells in plankton net samples. Since individual cells of Tuscaroridae and Circoporida were usually larger than 1 mm, they remained on the 1-mm sieve. Individual cells of Castanellidae were sometimes smaller than 1 mm and passed through a sieve although they usually occurred as colonies in the sediment trap samples. Therefore, we handpicked up Castanellidae and categorized them as >1-mm fraction. Phaeodarians and other zooplankton taxa were sorted under a stereomicroscope and preserved separately in the preservative solutions. In order to measure the dry weight and organic carbon content of phaeodarians, phaeodarians were dissected and pretreated in the same way as the >1-mm sinking particles. Other zooplankton were pretreated in the same way as the ZC sample. For each sample, the amount of organic carbon was measured with a CHN Analyzer (NCS2500, ThermoQuest). Abundance and standing stocks per unit volume of seawater were calculated based on the counts, dry weight, and volume of water filtered through the plankton net.
Because the SAP samples consists mainly of phaeodarian siliceous skeletons and organic matter (OM) in adherent particles, dry weight of whole phaeodarian communities were estimated as follows:  Redfield et al. (1963) and Richards (1965) (Honda et al., 2002).
In addition to the plankton samples, hydrographical data (temperature, salinity, and dissolved oxygen) were obtained with a CTD (Conductivity-Temperature-Depth profiler) system (SBE9Plus, Sea-Bird Electronics) at Station K2 at midnight between 7 and 8 July 2015. These environmental data were collected during Cruise KH15-J01 of the research vessel Hakuho Maru and are available from the Japan Agency for Marine-Earth Science and Technology data search portal (http://www.godac.jamstec.go.jp/darwin/cruise/hakuho_ maru/kh-15-j01/e). Figure 4 and Table S2 show the temporal variation of apparent mass fluxes in >1-mm fractions at Station K2. The contribution of >1-mm sinking particles to the apparent mass fluxes in the >1-mm fractions averaged 42% (range 18-73%) during the sampling period. Swimmers accounted for more than half of the >1-mm mass fluxes. In the swimmer samples, Copepoda were observed most frequently, followed by Gastropoda ( Figure S2 and Table S3).

Contribution of >1-mm Sinking Particles to Total Mass Flux and Their Composition
We defined the sum of the mass fluxes of the >1-mm sinking particles and <1-mm sinking particles as the total mass flux. The total mass fluxes at Station K2 averaged 158 mg-dry m −2 day −1 (range 42-403 mg-dry m −2  day −1 ). The maximum flux was observed during October 2014 ( Figure 5). The mass fluxes associated with the >1-mm sinking particles at Station K2 averaged 4.2 mg m −2 day −1 (range 2.1-8.2 mg-dry m −2 day −1 ). The maximum flux was observed during July 2014. The fluxes were relatively low (2.1-2.5 mg-dry m −2 day −1 ) during January-February 2015 ( Figure 6 and Table S2). The mass fluxes in the <1-mm sinking particles at Station K2 averaged 154 mg-dry m −2 day −1 (range 40-399 mg-dry m −2 day −1 ; Figure 5 and Table S4). The mass flux in the <1-mm sinking particles increased during August-November 2014, reached a maximum during October 2014, and decreased thereafter; it tended to be relatively low after December 2014, but increased slightly during March-May 2015 ( Figure 5). The <1-mm sinking particles accounted for >90% of the total mass fluxes throughout the sampling period. The corresponding contribution to the mass fluxes of the >1-mm sinking particles averaged 3.6% (range 0.9-7.2%) of the total.
The >1-mm sinking particles consisted of phaeodarians, particles adhered to phaeodarian skeletons, and zooplankton carcasses ( Figure 6). Table S2 shows details of the contents of the >1-mm sinking-particle mass flux at Station K2. The contributions of phaeodarians and adherent particles (sum of the PHA, SAP, and AP samples) to the mass flux of >1-mm sinking particles averaged 76% (range 42-90%) during the sampling period. Figure 7 shows the temporal variation of POC fluxes in >1-mm sinking particles at Station K2. Table S2 provides details about these fluxes. The POC fluxes in the >1-mm sinking particles averaged 0.73 mg C m −2 day −1 (range 0.42-1.54 mg C m −2 day −1 ). The POC fluxes due to phaeodarians averaged 0.23 mg C m −2 day −1 (range 0.07-0.38 mg C m −2 day −1 ) and was relatively high during August-November 2014. The maximum flux was observed during August-September 2014. The POC fluxes due to particles adhering to phaeodarian skeletons (sum of SAP and AP) averaged 0.18 mg C m −2 day −1 (range 0.06-0.30 mg C m −2 day −1 ) and increased from March to May 2015. The maximum flux was observed in May 2015. The POC fluxes due to ZC averaged 0.32 mg C m −2 day −1 (range 0.11-1.06 mg C m −2 day −1 ) and reached a maximum in July 2014. The mean contribution to the POC flux of >1-mm sinking particles in each group during the sampling period was 33% for phaeodarians, 26% for particles adhering to phaeodarian skeletons, and 41% for ZC. When phaeodarians were combined with adherent particles, they were responsible for 59% of the POC flux associated with >1-mm sinking particles.

Contribution of Phaeodarians to POC Flux
We defined the sum of the POC fluxes associated with the >1-and <1-mm sinking particles as the total POC flux. Figure 8 shows the temporal variations of the composition of the total POC flux at Station K2. Table S4 shows the organic carbon fluxes of the <1-mm sinking particles at Station K2. The contributions of >1-mm sinking particles to the total POC flux averaged 11.7% (range 2.7-24.3%). The contributions of phaeodarians to the total POC flux averaged 3.5% (range 1.1-9.7%). The contributions of phaeodarians with adherent particles to the total POC flux averaged 6.4% (range 1.8-12.5%).
The vertical distribution of phaeodarian standing stock and organic carbon standing stock at Station K2 were characterized by relatively high values in the mesopelagic zone (150-1,000 m; Figures 10b and 10c), 0.46-0.87 mg-dry/m 3 and 0.09-0.19 mg C/m 3 , respectively. The phaeodarian standing stock reached a maximum of 0.87 mg-dry/m 3 at depths of 100-200 m. The phaeodarian organic carbon standing stock reached a maximum of 0.19 mg C/m 3 at depths of 200-500 m. The contribution of phaeodarians to the >1-mm zooplankton standing stock averaged 12% (range 5-21%) in depths greater than 30 m (Figure 10d). The contribution of phaeodarians to the organic carbon standing stock of >1-mm zooplankton averaged 6% (range 2-10%) in depths greater than 30 m (Figure 10e). The high-standing stock of phaeodarians in the mesopelagic zone was consistent with the fact that many phaeodarians were collected with the sediment trap at a depth of 1,000 m. However, the species compositions of Phaeodaria in sinking particles and in the plankton tow samples were very different. In the sinking particles, the percentages of taxa with relatively solid skeletons such as Tuscaroridae (average: 20%), Castanellidae (average: 28%), and Circoporidae (average: 49%) were higher than the percentages of Aulacanthidae (average: 2.3%) and Phaeosphaerida (average: 0.1%), both of which have fragile skeletons. In contrast, the percentages of Aulacanthidae (average: 58%) and Phaeosphaerida (average: 30%) were higher in the plankton tow samples in depths greater than 50 m. Figure 10f and Table S8 show profiles of temperature, salinity, and dissolved oxygen (DO) down to a depth of 2,000 m at Station K2. Sea surface temperature was 7.8°C, and there was a clear temperature minimum layer in the temperature profile (dichothermal layer) at a depth of 100 m. The salinity was 32.7 psu at the sea surface, rapidly increased to 33.7 psu at a depth of 200 m, and then kept gradually increased to 34.6 psu at a depth of 2,000 m. The DO concentrations decreased rapidly from the surface to a depth of 200 m. The depth range of the oxygen minimum layer (DO < 25 μmol/kg) spanned from 500 to 1,000 m. Within the mesopelagic layer, high-standing stocks of phaeodarians overlapped with layers of relatively low temperatures (2.7-3.4°C), high salinities (33.5-34.4 psu), and low dissolved oxygen waters (21-155 μmol/kg).

Quantification of Phaeodarian POC Flux With a Dissection Method
The goal of this study is to quantify the magnitude of carbon export due to phaeodarians in the mesopelagic twilight zone. Recent field observations made with an Underwater Vision Profiler 5 (Picheral et al., 2010) and a short-term drifting sediment trap have elucidated the contribution of phaeodarians to the marine carbon cycle (Biard et al., 2016;Stukel et al., 2018). Those estimates of the biomass and POC flux of phaeodarians were limited, however, to depths between the base of the euphotic zone and the   diameter; in this study, we directly measured the organic carbon content of the whole phaeodarian community with a CHN analyzer. We propose a simple dissection method to quantify the export flux of organic carbon due to phaeodarians. Our dissection method allowed direct measurement of POC in phaeodaria. However, our dissection method includes potential caveats regarding measurement and contamination. Since our dissection method has focused on measuring POC derived from phaeodarians without contamination of adherent particles, it may underestimate POC derived from phaeodarians and also overestimate POC in particles adhered to phaeodarian skeletons. Our dissection method sometimes slightly injures the internal soft parts when dissecting the fragile phaeodarian bodies and taking out the internal soft parts. In such a case, very small fragments can be mixed into SAP samples and cause contamination. With regard to the estimation of dry weight of phaeodarians, a bias can also occur because the amount of organic matter in adherent particles must be estimated from organic carbon contents using one single stochiometric ratio (see equation (6)). In light of the above, we conclude that our dissection method is suitable in determining the amount of POC in whole phaeodarian communities.

Comparison of Phaeodarian Taxonomic Compositions Between Sinking Particle and Plankton Tow Samples, and Its Implications for the Oceanic Carbon Cycle
The vertical profile of the phaeodarian abundance at Station K2 was characterized by high percentages of Aulacanthidae and Phaeosphaerida throughout the water column (Figure 10a), but these taxa were not common in the sinking particles collected at a depth of 1,000 m ( Figure 9). The abundance of Aulosphaeridae, which belongs to the order Phaeosphaerida, was also as high in the mesopelagic zone offshore of Point Conception in the California Current ecosystem as the abundance at Station K2 . However, in contrast to Station K2, the fluxes of Aulosphaeridae associated with sinking particles collected at a depth of 150 m  in the shallow mesopelagic layer of the California Current ecosystem were high. Fluxes of Aulosphaeridae at a depth of 150 m in the California Current ecosystem averaged 112 individuals m −2 day −1 (range: 0-231 individuals m −2 day −1 ; Stukel et al., 2018). In other words, the fluxes of Aulosphaeridae alone were frequently an order of magnitude higher within the shallow mesopelagic layer of the California Current ecosystem than the total fluxes of all phaeodarians at a depth of 1,000 m at Station K2. The difference between the taxonomic composition of phaeodarians in sinking particles and in plankton tow samples at Station K2 could suggest that (1) dissolution in the sample bottle occurred during the mooring period, (2) selective dissolution due to taxon-specific vulnerability of phaeodarian skeletons occurred during sinking in the mesopelagic zone, and/or (3) consumption of phaeodarians by other omnivorous animals in the mesopelagic zone decreased the phaeodarian fluxes.
The possibility of in-trap dissolution during the mooring period has been suggested in Biard et al. (2018) with respect to silica fluxes. The one-year mooring of our trap was much longer than the two-to five-day deployment of VERTEX-style particle interceptor traps in Stukel et al. (2018). Therefore, our trap samples might be more prone to in-trap dissolution than the trap samples in Stukel et al. (2018) due to the longer mooring period. However, pH values of preservative solution in the sample bottles ranged from 8.2 to 8.7 at the sediment trap recovery, which were almost same as the initial pH (approximately 8.0) at the sediment trap deployment. Furthermore, insignificant amount of disaggregated phaeodarian skeletons and internal parts were observed in the trap samples, which also suggested that selective dissolution was not progressed significantly in the sample bottles. Therefore, the possibility of significant dissolution in the sample bottles can be ruled out.
The selective dissolution during sinking would be related to the particularly fragile skeletons of Aulacanthidae and Phaeosphaerida in the distance between their habitat depth and the deployment depth of the sediment trap. The maximum abundance of Aulacanthidae and Phaeosphaerida at Station K2 was located at 500-1,000-and 200-500-m depth, respectively. Vertical zonation of Aulacanthidae is very broad (500-1,000 m), but judging from Nakamura et al. (2013), the maximum abundance of Aulographis (Aulacanthidae) would be located between 500-and 750-m depth. Therefore, the distance between their habitat depth and the sediment trap depth at 1,000 m must have been large at Station K2. On the other hand, Stukel et al. (2018) deployed the traps at 100-150-m depth where the maximum abundance of Aulosphaeridae (Phaeosphaerida) was observed. Therefore, we interpret that Aulacanthidae and Phaeosphaerida in our study were more susceptible to dissolution and predation during the sinking process than those in Stukel et al. (2018).

Potential Role of Phaeodarians in the Biological Carbon Pump
The volume-specific carbon content of protists has been estimated by Beers and Stewart (1970) and Menden-Deuer and Lessard (2000) to be 80 and 56 μg C/mm 3 , respectively. Assuming the diameter of a phaeodarian cell to be 2 mm, the above values translate to 335 and 234 μg C/cell, respectively. However, Stukel et al. (2018) have suggested that these values should not be applied to phaeodarians because phaeodarian carbon is likely concentrated within the central capsule and phaeodium of the cell. Our estimate of the average organic carbon content of a phaeodarian cell based on the fluxes of phaeodarian cells and phaeodarian POC at Station K2 averaged 13 μg C/cell (range: 7.2-25 μg C/cell). This average cellular carbon content was several orders of magnitude lower than the estimates based on the carbon:volume ratios in Beers and Stewart (1970) and Menden-Deuer and Lessard (2000). However, the average of 13 μg C/cell is several times the value of 4.8 μg C/cell estimated by Stukel et al. (2018) for Aulosphaeridae cells with diameters of 2 mm. Stukel et al. (2018) assumed the ratio of the diameter of the central capsule to the diameter of the scleracoma to be 1 to 4 in the case of Aulosphaeridae, but that ratio and the number of central capsules varies between species. For example, in the cases of Aulacanthidae, Circoporidae, Tuscaroridae, Castanellidae, and Polypyramidae (Figure 3), the central capsule and phaeodium occupied more than half of the scleracoma. In such cases, the carbon content of the phaeodarians would be more than 8 times the value for Aulosphaeridae in Stukel et al. (2018) if the diameter of the scleracoma was the same. It is therefore reasonable that the amount of organic carbon per cell varies severalfold between species. Thus, we are able to conclude that our results estimated from directly measured phaeodarian organic carbon supported the observation made by Stukel et al. (2018) regarding the likely overestimation of global phaeodarian biomasses depicted in Biard et al. (2016). In fact, at a water depth of 1,000 m at Station K2, the contribution of phaeodarians to the total POC flux of sinking particles was at most about 9.7%, with a mean of 3.5%. However, export of organic carbon contained within phaeodarians is not the only impact of these organisms on the carbon cycle and marine food webs. Little is known about the feeding ecology of phaeodarians, but they are known to be omnivorous zooplankton, and they store their food and waste products in their phaeodium (Gowing, 1986). Microscopic observations of the phaeodium had suggested that phaeodarians feed on sinking detrital particles, which include bacteria, other protozoans, small crustaceans, and organic aggregates (Gowing & Bentham, 1994). Also, phaeodarians sometimes have been reported as a major producer of minipellets (Gonzalez, 1992;Gowing & Silver, 1985;Lampitt et al., 2009;and Riemann, 1989). Stukel et al. (2018) have suggested that a single group of phaeodarians (Aulosphaeridae) could intercept up to >20% of sinking particles between the base of the euphotic zone and a depth of 300 m. The entire phaeodarian community would impact an even greater percentage of the POC. An investigation of the process of dissolution of phaeodarian skeletons and the impact of predation on phaeodarians by other animals should help to clarify the process of decomposition/alteration of sinking POC and the amount of inorganic carbon produced by the decomposition of POC in the mesopelagic zone. Understanding the role of phaeodarians in mesopelagic marine ecosystems is thus a key element to understanding all aspects of the oceanic carbon cycle.
Previous studies (e.g., Honda et al., 2017) have shown that POC from the surface mixed layer is efficiently exported to the deep sea at Station K2. In other words, POC is degraded at lesser rate at Station K2 compared to that in other areas. However, only the ballasting of sinking particles with opal has been invoked as an explanation for the high efficiency (Honda & Watanabe, 2010). We indicate considerable contribution of phaeodarians in POC (33% on average) in >1-mm sinking particles collected at a depth of 1,000 m at Station K2. The high proportion of phaeodarians to >1-mm sinking particles would contribute as ballast of sinking particles and would be one of the additional factors that explain the high efficiency of POC export in the WSG. Furthermore, the substantial amounts of POC that adhered to phaeodarian skeletons (i.e., the POC in the AP) could suggest that the gelatinous body and large skeleton of phaeodarians, which absorb and aggregate other fine particles, contribute as ballast to the high efficiency of the BCP. If a phaeodarian is considered to be a composite particle that includes fine particles attached to its skeleton (i.e., sum of the PHA, AP, and SAP), the contribution of phaeodarians to the total POC export fluxes averaged 6.4% with a maximum of 12.5%. In general, the particles in the >1-mm fraction had previously been customary excluded as swimmers. However, the fact that 42% (mean value) of such coarse particles is in fact sinking particles. Hence, the coarse phaeodarians account for a significant percentage of such sinking particles and this has important implications for estimating export production. The enlargement of particles through the food chain (e.g., formation of zooplankton fecal pellets) and physicochemical aggregation of suspended particles (Passow, 2002) are closely related to the efficiency of the BCP. Therefore, phaeodarians associated with sinking particles would be an additional factor involved in both the enlargement and aggregation of sinking particles in the mesopelagic zone of the western North Pacific. It is expected that in the future phaeodarians will be incorporated into marine ecosystem models such as NEMURO (Kishi et al., 2007) to take into account the high efficiency of the BCP in the WSG.
The >1-mm phaeodaria such as Aulacanthida, Phaeosphaerida, and Phaeodendrida are widely distributed in the world ocean (Nakamura & Suzuki, 2015;Reshetnyak, 1966). Aulacanthida and Phaeosphaerida are even found in the Arctic Basin (Zasko & Kosobokova, 2014). Nevertheless, studies on abundance density and abundance flux of >1 mm phaeodaria have been still scarce, with only a few reported in the North Pacific and Atlantic. The abundance density and flux of >1 mm Phaeodaria reported so far are summarized in Table S9. Since phaeodarians are omnivorous heterotroph, they tend to inhabit the depth where particulate organic matter settles from the upper layer (Nimmergut & Abelmann, 2002). However, high abundances of >1 mm phaeodarians have been reported not only in high BCP areas such as the western North Pacific but also in oligotrophic California Current ecosystem . Therefore, quantification of biomass and POC fluxes of >1 mm phaeodarians in the world ocean would be an essential piece to get all aspects of the global ocean carbon cycle and ecosystem. In addition, in the biomass survey of the >600 μm Rhizaria in the Tara Oceans expedition (Biard et al., 2016), the western North Pacific, Indian Ocean, and Southern Ocean were blank areas for the expedition. The Southern Ocean is an area that stores 40% of anthropogenic carbon dioxide (Frölicher et al., 2015;Khatiwala et al., 2009) and is a high BCP area , so it would be particularly important to know the role of phaeodarians in the area. Finally, we focused on phaeodarians in this study, but locally high abundance of phaeodarians <1 mm in size have also been observed in the western North Pacific (Okazaki et al., 2004), eastern North Pacific (Kling, 1976), and Antarctic Ocean (Morley & Stepien, 1984;Nöthig & Gowing, 1991). The combination of >1 and <1 mm phaeodarians may result in POC export significantly higher than previously thought. Our simple dissection method and direct measurement of the organic carbon content in phaeodarians will be an important step to elucidate the global contribution of phaeodarians to the BCP.