Pigmented microbial eukaryotes fuel the deep sea carbon pool in the tropical Western Pacific Ocean

This study provides first details of the abundance, vertical profile, and size structure of pigmented nano-sized eukaryotes in the water columns of the tropical Western Pacific Ocean down to abyssal depths. We provide a thorough description of the community structure of phototrophic microbial eukaryotes in the deep Western Pacific Ocean by sequencing both 18S rRNA and psbA gene transcripts. The phototrophic microbial eukaryote signals caught in the present study may come from surface organisms that descend through the water column due to various fast-sinking mechanisms, thus contributing to the biological pump and fueling the deep-sea communities by supplying fresh organic carbon. We further demonstrate the potential presence of mixotrophic microbial eukaryotes in the deep sea that may graze prokaryotes and mediate nutrient cycling in the microbial loop of deep-sea ecosystems


Originality-Significance Statement
This study provides first details of the abundance, vertical profile, and size structure of pigmented nano-sized eukaryotes in the water columns of the tropical Western Pacific Ocean down to abyssal depths.We provide a thorough description of the community structure of phototrophic microbial eukaryotes in the deep Western Pacific Ocean by sequencing both 18S rRNA and psbA gene transcripts.The phototrophic microbial eukaryote signals caught in the present study may come from surface organisms that descend through the water column due to various fast-sinking mechanisms, thus contributing to the biological pump and fueling the deep-sea communities by supplying fresh organic carbon.We further demonstrate the potential presence of mixotrophic microbial eukaryotes in the deep sea that may graze prokaryotes and mediate nutrient cycling in the microbial loop of deep-sea ecosystems.

Summary
Phototrophic microbial eukaryotes dominate primary production over large oceanic regions.
Due to their small sizes and slow sinking rates, it is assumed they contribute relatively little to the downward export of organic carbon via the biological pump.Therefore, the community structure of phototrophic cells in the deep ocean has long been overlooked and remains largely unknown.In this study, we used an integrative approach, including epifluorescence microscopy, sequencing of 18S rRNA and photosystem-II psbA gene transcripts, to investigate phototrophic microbial eukaryotes in samples collected from the tropical Western Pacific Ocean.It was found that: 1) pigmented nano-sized eukaryotes (PNEs) are ubiquitous in the deep Western Pacific Ocean down to 5,000 m depth; 2) the PNE community is dominated by cells 2-5 µm in size; 3) their abundance is significant, averaging 4 ± 1 (± s.e.) cells mL -1 in waters below 1,000 m which is comparable to that of heterotrophic nanoflagellates; 4) the active pigmented microbial eukaryotes in the deep waters are highly

Introduction
Marine phytoplankton encompasses prokaryotic and eukaryotic organisms that are capable of oxygenic photosynthesis and inhabit the photic zone of all oceans (Falkowski et al., 2004;Worden et al., 2004).They account for almost half of all photosynthetic activity on Earth and thus play a fundamental role in marine ecosystems and global biogeochemical processes.In vast areas of the oceans, especially the oligotrophic regions, the majority of primary production is performed by small-sized cells, i.e. the picoplankton (less than 2 -3 m in diameter) and nanoplankton (2 -20 m in diameter) (Marañón et al., 2001).Among these minute cells, the abundance of phototrophic microbial eukaryotes is two to three orders of magnitude lower than their prokaryote counterparts, the cyanobacteria.However, because of their equivalent or higher growth rates and relatively larger cell size, they can dominate carbon biomass and net production both in coastal waters (Massana et al., 2004;Worden et al., 2004) and in the open ocean (Li, 1995;Not et al., 2008;Jardillier et al., 2010).In addition, compared with cyanobacteria that are dominated by just two genera (Prochlorococcus and Synechococcus), photosynthetic microbial eukaryotes are much more diverse and include representatives of almost all major clades of the eukaryotic phylogenetic tree (Massana, 2011;Decelle et al., 2015).
The biological pump involves a series of processes through which CO 2 is fixed as organic matter by photosynthesis and then transferred to the deep ocean, resulting in the temporary or permanent storage of carbon (Sigma and Haug, 2006).Traditionally, large and "heavy" phytoplankton cells such as diatoms are thought to be the main contributors to the downward transportation of organic carbon through the biological pump.The pico/nano-sized phototrophic eukaryotes are generally believed to contribute relatively little to carbon export due to their smaller sizes, slow sinking rates, and rapid utilization in the microbial loop in the photic zone.It has been argued, however, that the contribution of picoplankton to carbon export from the photic zone is proportional to their total net primary production, which is significant considering their high abundances and production rates (Richardson and Jackson, 2007).
Early studies reported the occasional presence of well-preserved phytoplankton cells in the deep-sea samples, although these observations were considered anecdotal and have long been overlooked (Kimball et al., 1963;Smayda, 1971;Wiebe et al., 1974).More recently, however, intact, viable diatom cells have been reported from deep-sea sediment traps and downward export pulse has been observed at Station ALOHA (Scharek et al., 1999;Karl et al., 2012).Furthermore, a global survey employing DNA sequencing has also suggested the presence of phototrophic signals in the deep ocean (Pernice et al., 2015b).The ubiquitous presence of healthy photosynthetic cells (mainly diatoms) and 16S rRNA gene sequences of photosynthetic cells (cyanobacteria and algal chloroplasts) at depths of up to 4,000 m in the dark ocean on a global scale has been reported (Agusti et al., 2015).Finally, based on the sequencing both of the 18S rRNA and its gene, active phototrophic microbial eukaryotes were confirmed to be present in the deep waters of the South China Sea (Xu et al., 2017b).
Although these pioneering studies have revealed the presence of phototrophic microbial eukaryotes in deep oceans, knowledge of their source, community structure, biogeography, and function remain largely unknown.
In the present study, an integrative approach including fluorescence microscopy, highthroughput sequencing (HTS) of the hyper-variable V9 region of 18S rRNA, and cloning and sequencing of photosystem-II psbA gene transcripts, was used to explore the abundance (nano-sized only), size fractions (2 -5 µm, 5 -10 µm, and 10 -20 µm), and phylogenetic identity (evidence from both 18S rRNA and photosystem-II psbA gene transcripts), of phototrophic microbial eukaryotes in the deep tropical Western Pacific Ocean.In so doing, the community structure of phototrophic microbial eukaryotes in deep oceanic waters is revealed for the first time, confirming their contributions to the downward transportation of organic carbon through the biological pump.The potential presence of mixotrophic microbial eukaryotes in the deep tropical Western Pacific Ocean is also revealed.

Distribution of environmental variables
In total, 110 water layers from 11 stations in the tropical Western Pacific Ocean were sampled (Fig. 1).The water temperature, salinity, and chlorophyll a (Chl a) at each depth interval are summarized in Supporting Information Fig. S1.A well-defined pycnocline was found with the steepest gradients in salinity (27.0 to 35.3) and temperature (30.5 to 1.6 °C) between 20 and 100 m (Supporting Information Fig. S1A and B).Surface (5 m) Chl a concentration was low (0.02 -0.16 µg L -1 ) throughout the study area reflecting the oligotrophic conditions in the tropical Western Pacific Ocean.The deep chlorophyll a maximum (DCM) layer was found at 50 m at P14, 75 m at P12, and 100 m at the rest of sampling stations (Supporting Information Fig. S1C).Phytoplankton pigments measured at 5 m, DCM, and 200 m of all sampling stations revealed that the contribution of eukaryotes to total Chl a was 9 -19% (on average 13%), 9 -50% (on average 32%), and 10 -100% (on average 43%), respectively (data not shown).Using CHEMTAX, the relative contributions of six major eukaryotic chemotaxonomic groups, i.e., haptophytes, prasinophytes, cryptophytes, chlorophytes, diatoms, and dinoflagellates, to total eukaryotic Chl a were determined.At all depths of all sampling stations, haptophytes were clearly the dominant pigmented group among eukaryotes, their relative contributions to total eukaryotic Chl a being 57 -74 % (on average 64 %) at 5 m, 58 -94% (on average 77%) at DCM, and 0 -92 % (on average 67 %) at 200 m.Excluding P14, where no haptophyte pigment signal was recovered at 200 m, the contribution of haptophytes to total eukaryotic Chl a reached 42 -92% (on average 75%) at This article is protected by copyright.All rights reserved.

Accepted Article
200 m (Supporting Information Fig. S2A, B, and C).Generally, haptophytes were the highest contributor to the total eukaryotic Chl a at DCM, with a lower relative contribution at 200 m and lower still at 5 m.The absolute Chl a concentration of haptophytes ranged 5.9 -32.2 ng L -1 (on average 14.1 ng L -1 ) at 5 m, 20.6 -170.9 ng L -1 (on average 118.1 ng L -1 ) at DCM, and 0 -16.2 ng L -1 (on average 2.7 ng L -1 ) at 200 m, respectively (Supporting Information Fig. S2D).

Enumeration of pigmented nano-sized eukaryotes (PNEs) by microscopy
Fluorescence microscopy examinations of PNEs (2 -20 µm) were performed for all 110 samples collected (Supporting Information Fig. S4).This showed not only the vertical distribution pattern of abundance but also the cell-size structure of PNEs in water columns of the Western Pacific Ocean.Interestingly, PNEs were found not only in all 66 euphotic zone samples (5 -200 m) where photosynthesis could be achieved but also in all 11 mesopelagic zone samples (200 -500 m) and 61% of the 33 deep-sea (≥ 1,000 m) samples including the greatest depth sampled (5,000 m at P12).The abundance of PNEs varied among the three depth zones with the highest abundance generally occurring in the DCM (Fig. 2).In the waters at 5 -50 m, the average (± s.e.) abundance of PNEs was 768 ± 59 cells mL -1 .At 75 -This article is protected by copyright.All rights reserved.
Accepted Article 100 m, their abundance averaged 672 ± 64 cells mL -1 .At 200 -500 m, their abundance dropped to 31 ± 10 cells mL -1 and at 1,000 -3,000 m their abundance averaged 4 ± 1 cells mL -1 (Table 1).Generally, the abundance and biomass of PNEs did not change much in waters ≤ 100 m but decreased sharply from 100 to 200 m and maintained low abundance in waters > 200 m (Fig. 3A and B).In general, the abundance of PNEs decreased with depth with log-log regression slopes (Fig. 3C) and declined more steeply than that of bacteria and archaea (Fig. 3D).Additionally, epifluorescence microscopy observation allowed an approximate estimate of the size structure of the PNE community.The PNE communities (both abundance and biomass) were dominated by cells in the size range 2 -5 µm at all depths followed by cells in the size range 5 -10 µm.In waters below 200 m, no PNEs larger than 5 µm were found.PNEs with size range 10 -20 µm were only found in certain samples at 5 m (Fig. 3A and B).

Active microbial eukaryote communities revealed by high-throughput sequencing of V9 regions of 18S rRNA
Following high-throughput sequencing of V9 regions of the reverse-transcribed 18S rRNA and the removal of low quality reads, potential chimeras, reads that were not assigned as eukaryotes, and singletons, there were 371,834 reads representing both protistan and metazoan taxa, ranging 25,388 to 48,813 reads per sample.Reads affiliated with Metazoa were removed from the dataset resulting in 7,708 to 40,497 reads per sample (Supporting Information Table S1).Notably, a large portion (81.8%) of the reads from the deepest sample (5,000 m) collected at station P12 (P12.5K) was affiliated with Cnidaria (data not shown).To normalize sampling effort, OTU counts were rarefied at a uniform sequencing depth based on the lowest sequence count affiliated with microbial eukaryotes (7,708 reads) of sample

P12.5K.
This article is protected by copyright.All rights reserved.

Accepted Article
Nonmetric multidimensional scaling ordination (nMDS) analysis showed that all samples were clustered into two groups according to the depth where samples were collected, one containing samples from the euphotic zone (25 and 75 m) and the other containing samples from deep waters (≥ 500 m) (Fig. 4A).Statistical analyses showed the composition of the two groups differed significantly from one another (ANOSIM, R = 0.7255, p = 0.0032).This grouping pattern was also supported by principle component analysis (PCoA) of community taxonomic relatedness quantified by the Unweighted Unifrac metric which compares samples based on the phylogenetic relatedness of OTUs in a community without taking into account relative OTU abundance (Fig. 4B).Therefore, the community composition is described below according to this vertical distribution pattern.At the supergroup level, the active microbial eukaryote communities in the euphotic zone were characterized by high relative sequence abundances of Stramenopiles (on average 48 %), followed by Alveolata (on average 25%), and Hacrobia (on average 16%) (Supporting Information Fig. S5A).The other supergroups, i.e., Amoebozoa, Apusozoa, Archaeplastida, Excavata, Opisthokonta, Picozoa, and Rhizaria, were only minor contributors to the microbial eukaryotic communities.By contrast, the most dominant supergroup in the deep-sea samples (500 -5,000 m) was Alveolata (on average 51%), followed by Stramenopiles (on average 19%), Hacrobia (on average 6%), Rhizaria (on average 8%), and Opisthokonta (on average 8%) (Supporting Information Fig. S5A).The third most dominant supergroup was not always consistent, however, but shifted between Rhizaria and Hacrobia in different samples (Supporting Information Fig. S6).In terms of OTU richness at the supergroup level, the contributions of Stramenopiles and Alveolata were almost equal both in the euphotic zone (on average 31.4% and 30.5%, respectively) and in the deep-sea (on average 27.6% and 28.5%, respectively), while that of Hacrobia (on average 13.5% and 10.5%, respectively) and Rhizaria (on average This article is protected by copyright.All rights reserved.

Accepted Article
10.9% and 13.8%, respectively) varied in the two groups relative to each other (Supporting Information Fig. S5B).
In common with the results from epifluorescence microscopy examination, HTS also recovered possible phototrophic signals in all deep-sea samples.For the comparison of phototrophic signals from shallow and deep waters, only major phototrophic groups, i.e.
Chrysophyceae-Synurophyceae, Pelagophyceae, Dictyochophyceae, Bolidophyceae, Bacillariophyta, Haptophyta (mainly Prymnesiophyceae), Cryptophyceae, and Prasinophyceae, were considered in the present analysis.The contributions of major phototrophic groups to total microbial eukaryote communities in the shallow waters ranged 28% -55%, which is higher than those in the deep-sea which ranged from 18% -28% (Fig. 4C).Even in the sample from the greatest depth (5,000 m: P12.5K), the contribution of phototrophic signal to the total eukaryotic community still reached 16.3%.In the shallow waters, the three most dominant phototrophic groups were Prymnesiophyceae, Pelagophyceae, and Bacillariophyta, the contributions of which differed in different samples (Fig. 4C).In deep-sea samples, phototrophic groups were dominated by Bacillariophyta, Cryptophyceae, and Prasinophyceae with their contributions to total microbial eukaryote communities differing among samples (Fig. 4C).In terms of the entire microbial eukaryote communities, phototrophic groups ranked among the top 10 dominant groups in most deepsea samples (Supporting Information Fig. S7).

Phototrophic microbial eukaryotes revealed by sequencing of psbA gene transcripts
To avoid the masking of heterotrophic microorganisms resulting from rRNA sequencing, the functional protein-coding gene for D1 of photosystem-II reaction center (psbA) was cloned and sequenced to reveal the active phototrophic microbial eukaryotes in the water column.
After removing potential chimeras and sequences of bacteria/archaea and virus/phage origin, This article is protected by copyright.All rights reserved.

Accepted Article
617 sequences (~ 750 bp) affiliated with microbial eukaryotes (Dinophyceae, Cryptophyta, Euglenozoa, Haptophyta, Cercozoa, Bacillariophyta, Chrysophyceae, Dictyochophyceae, Pelagophyceae, Pinguiophyceae, and Chlorophyta) remained.Because of the lack of phenotypic and systematic data on phototrophic microbial eukaryotes in deep oceanic waters, sequences were further assigned taxonomy by blasting against GenBank database to reveal both of their nearest neighbor (NN) and nearest named neighbor (NNN).Generally, as expected, the similarities of the representative sequence to their NN were higher than to their NNN.In some groups, similarities to their NNN were rather low, e.g.Chrysophyceae (89.0%-91.6%, on average 90.3%) and Chlorophyta (89.5% -99.6%, on average 91.4%) (Table 2).
Similar to the result of 18S rRNA HTS, sequencing of psbA gene transcripts also revealed the presence of active phototrophic microbial eukaryotes in all deep-sea samples, with their relative sequence abundances differing among sites and depths (Supporting This article is protected by copyright.All rights reserved.

Accepted Article
Information Fig. S8).For phototrophic microbial eukaryotes in shallow and deep waters, Haptophyta was the main contributor (56% and 60%, respectively) to both communities.
Bacillariophyta and Chlorophyta contributed 15% and 12% respectively to the group of shallow water communities and 8% and 16% respectively to the group of deep-sea communities (Fig. 5A and B).
To determine the possible origin of deep-sea phototrophic microbial eukaryotes, all psbA OTUs were clustered into three groups, one containing OTUs occurring in shallow waters only (the "shallow water only" group), one containing OTUs occurring in deep waters only (the "deep water only" group), and one containing OTUs occurring in both shallow and deep waters (the "both shallow and deep waters" group).Fifty-one percent of all phototrophic microbial eukaryote OTUs were found in the "deep water only" group and 34% were present in the "shallow water only" group, the remaining 15% occurring in the "both shallow and deep waters" group (Fig. 5C).In terms of the compositions of these three groups, Haptophyta contributed the most to the three groups (77% of the "both shallow and deep waters" group, 57% of the "deep water only" group, and 37% of the "shallow water only" group) (Fig. 8D).

Ocean
Marine photosynthesis contributes ca.50% of total primary production on Earth (Field et al., 1998) and that of the phototrophic microbial eukaryotes is agued to surpass other phototrophic organisms in both coastal and oceanic waters (Li, 1995;Worden et al., 2004;Li, 2009).Despite their high contributions to the primary production in the surface ocean, the contributions of phototrophic microbial eukaryotes, especially the pico-and nano-sized cells, to the downward transportation of organic carbon through the so-called biological pump has This article is protected by copyright.All rights reserved.

Accepted Article
not been appreciated until relatively recently (Bienfang, 1980).Previous studies on pico-sized phytoplankton showed that these minute cells indeed occurred in both deep oceanic waters and sediments.For example, aggregates of intact Emiliania huxleyi and diatom cells were reported from sediment traps (Cadée, 1985;Lampitt et al., 1993;Scharek et al., 1999) and live Prochlorococcus cells were found in mesopelagic waters (Jiao et al., 2014).Based on the Malaspina circumnavigation expedition, viable diatoms (mainly micro-sized cells, 20 -200 µm) were reported in the deep sea globally (Agusti et al., 2015).Richardson and Jackson (2007) proposed that the contribution of picophytoplankton to carbon export is proportional to their total net primary production, which would be extremely high considering they are the main contributors to primary production in the ocean.Compared with pico-sized phytoplankton, however, the contribution of nano-sized (2 -20 µm) phototrophic eukaryotes to the downward transportation of organic carbon was largely unknown.
In the present study, pigmented nano-sized eukaryotes (PNEs) collected from 110 samples in the Western Pacific Ocean were examined by fluorescence microscopy.We found that PNEs not only occurred in all 66 samples in the euphotic zone and 11 samples in the twilight zone but were also present in 20 of all 33 deep-sea (≥ 1,000 m) samples including one sample from the greatest depth of 5,000 m (Fig. 2).Both the abundance and biomass of PNEs were high in the euphotic zone and relatively low in deep waters with a sharp decrease occurring between 100 to 200 m, suggesting the rapid mineralization of PNEs in the euphotic zone (Fig. 3A, B).Previous studies employing microscopy showed large variability of the abundance of deep-sea flagellates (0 -20 cells mL -1 ) in different oceanic regions although without differentiating phototrophs from heterotrophs (see review by Nagata et al., 2010).A recent study based on a global expedition which surveyed deep-sea heterotrophic protists (HP) showed that the average abundances of HP ranged from 72 ± 19 cells mL -1 in mesopelagic waters down to 11 ± 1 cells mL -1 in bathypelagic waters (Pernice et al., 2015a).
This article is protected by copyright.All rights reserved.

Accepted Article
It seems that the abundance of deep-sea PNEs (0 -19 cells mL -1 ) discovered in the present study is slightly lower than, but at the same order of magnitude as, that of the HPs (Pernice et al., 2015a).In the present study, the size structure of the PNE community was also investigated which revealed that 2 -5 µm cells were the most frequently encountered sizegroup at all depths, suggesting the considerable contributions of these to the PNE community throughout the water column in tropical oligotrophic oceanic waters.

Phylogenetic identity of phototrophic microbial eukaryotes in deep Western Pacific Ocean
In the present study, both 18S rRNA and psbA gene transcript sequencing were used in order to reveal the active microbial eukaryotes in the deep waters of the Western Pacific Ocean.
Signals of phototrophic microbial eukaryotes were found indicating that active phototrophic microbial eukaryotes were widespread in the deep Western Pacific Ocean, the dominant groups being Prasinophyceae, Bacillariophyta, and Cryptophyceae as revealed by 18S rRNA sequencing (Fig. 4C).In a recent study, high throughput sequencing of both 18S rRNA and its gene also revealed the presence of active phototrophic microbial eukaryotes in the deep South China Sea.In the same study it was reported that, using RNA/DNA ratio as a proxy, the relative activities of some phototrophic Stramenopiles (e.g.Chrysophyceae, Synurophyceae, Dictyochophyceae, and Pelagophyceae) in the deep waters were almost equal with those in the shallow waters (Xu et al., 2017b).
In the euphotic zone, 18S rRNA sequencing revealed the dominance of Prymnesiophyceae in P8.25m, Pelagophyceae in P8.75m and P14.75m, and Bacillariophyta in P14.25m (Fig. 4C).This finding was not, however, supported by pigment composition analysis which clearly showed that Haptophyta was the dominant phototrophic microbial eukaryote group in the euphotic zone, DCM, and 200 m waters (Supporting Information Fig.

S2). Previous studies reported that the 18S rRNA gene of haptophytes has a high GC content
This article is protected by copyright.All rights reserved.

Accepted Article
(57%), which can explain why this group was under-represented in the environmental survey by PCR amplification using eukaryote universal 18S rRNA primers (Moon-van der Staay et al., 2001;Massana et al., 2004;Liu et al., 2009).In the present study, direct sequencing of 18S rRNA confirmed the presence of active phototrophic microbial eukaryotes in the deep sea but failed to reflect their relative abundances.
Previous studies have found that sequencing of the rRNA gene results in a bias in favor of heterotrophic over phototrophic microorganisms (Vaulot et al., 2002).To remediate this effect, psbA, which codes the photosystem II core protein D1, has been recommended as a good marker for a variety of phototrophic marine microbes including picoeukaryotes and cyanobacteria (Zeidner et al., 2003;Zeidner and Béjà 2004;Man-Aharonnovich et al., 2010).
In the present study, both pigment analysis and psbA transcript sequencing showed that the phototrophic microbial eukaryote community was dominated by Haptophyta in the euphotic zone (Fig. 5A).Indeed, previous studies have shown the pervasive presence of pigment 19'hexanoylfucoxanthin (an accessory photosynthetic pigment found exclusively in chloroplasts of haptophyte origin) in the photic zone of the world's oceans (Lovejoy et al., 2006;Not et al., 2008;Liu et al., 2009) and the dominance of Haptophyta-affiliated psbA transcripts in surface and DCM of the eastern Mediterranean Sea (Man-Aharonovich et al., 2010).
The active deep-sea phototrophic microbial eukaryotes in the Western Pacific Ocean included members from Haptophyta, Bacillariophyta, Chlorophyta, and other "common" groups normally encountered in oceanic surface waters.This is consistent with observations by epifluorescence microscopy which revealed PNEs with a variety of different cell shapes in deep-sea waters (Supporting Information Fig. S4), confirming their high diversity in the deep Western Pacific Ocean.To the best of our knowledge, this is the first use of a functional gene to infer deep-sea phototrophic microbial community structure.Analyses of selected deepwater samples revealed that Haptophyta was the major component (comprising 28% -80% This article is protected by copyright.All rights reserved.

Accepted Article
of total phototrophic microbial eukaryotes) of deep-sea phototrophic microbial eukaryote communities (Supporting Information Fig. S8).Further analyses showed that the nearest named neighbor (NNN) for most (62.6%)Haptophyta-affiliated sequences was Emiliania huxleyi (GenBank accession number JN022705) with genetic similarities ranging from 91.4% to 99.5% (on average 95.6% and median 95.8%).Owing to the low species number and coverage of psbA sequenced for phototrophic microbial eukaryotes in reference databases, the cutoffs of psbA for differentiating species have not reached a consensus.Consequently, it is premature to conclude whether the Emiliania huxleyi-type sequences found in the deep Western Pacific Ocean were E. huxleyi or its congeners or even other haptophyte genera.
To examine the possible source of the deep-sea phototrophic microbial eukaryotes, all psbA OTUs were clustered into three groups, i.e., "shallow waters only" group, "deep waters only" group, and "both shallow and deep waters" group.Notably, 15% of all psbA OTUs recovered belonged to the "both shallow and deep waters" group, indicating these deep-sea phototrophic microbial eukaryotes may come from the euphotic zone (Fig. 5C).This hypothesis is consistent with observations by epifluorescence microscopy examinations which showed that both the abundance and biomass of PNEs dropped sharply in the depth range 100 to 200 m indicating that most were utilized in the microbial loop at these depths while the rest escaped mineralization through various fast sinking mechanisms such as forming/attaching to phytodetritus, aggregates and fecal pellets (Komar et al., 1981;Simon et al., 2002;Ploug et al., 2008;Iversen and Ploug, 2010).It is known that mesoscale eddies are present in the northwestern subtropical Pacific Ocean (Yuan et al., 2007;Yang et al., 2013).
Both anticyclonic and cyclonic eddies can induce downwelling transportation of particles from the surface ocean.Thus, active physical transportation may also expedite the downward transportation of surface PNEs to the deep-sea in the present study similar to previous observations of Prochlorococcus reported from the deep-sea of a nearby area (Jiao et al., This article is protected by copyright.All rights reserved. Accepted Article 2014).Once the PNEs reached the deep sea, they likely meet one of two fates: either 1) they may sink to the sea floor and become buried in the sediment, thereby contributing to the biological pump, or; 2) they die and their organic remains are released serving as source of labile dissolved organic matter fueling deep-sea communities.Thus, phototrophic microbial eukaryotes are probably not only major players in the primary production of the euphotic zone but also contribute to the export of organic carbon from the surface to the deep waters of the ocean.However, since the measurement of sinking rates of phototrophic microbial eukaryotes was outside the scope of the present study, this process needs further exploration in order to fully appreciate their contribution to the downward transportation of organic matter.
Many protists are well known for their capabilities of mixotrophy, i.e. they can use chloroplasts to perform photosynthesis and feed and feed heterotrophically, for example by ingestion of bacterioplankton through phagotrophy (Zubkov and Tarran, 2008;Stoecker et al., 2017;Terrado et al., 2017).Although occupying only a very small fraction of total phototrophic microbial eukaryote communities, investigations of 18S rRNA and psbA gene transcripts revealed the presence of Chrysophyceae-affiliated sequences in the deep Western Pacific Ocean (Fig. 4C, 5A, Supporting Information Fig. S6, S8).Chrysophytes are reported to be capable of phagotrophy, osmotrophy, and phototrophy (Holen and Boraas 1995).
Chrysophytes have been reported in the deep South China Sea (Xu et al., 2017b) and were identified as the second most abundant class in deep-sea protistan communities (Pernice et al., 2015b).Combined with their prevalence in the deep sea globally, it has been proposed that chrysophytes may be important bacterivores in deep waters (Pernice et al., 2015b).In addition to chrysophytes, haptophytes have been proposed to carry out most of the bacterivory in the surface layer of Atlantic Ocean (Zubkov and Tarran, 2008).In the present study, the NNN of some psbA sequences (i.e.34 in total, 21 of which were recovered from ≥ This article is protected by copyright.All rights reserved.

Accepted Article
1000 m waters) is Chrysochromulina sp.(GenBank accession No. KJ201907, Supporting Information Table S2), a non-calcifying haptophyte belonging to Prymnesiales, some members of which are capable of mixotrophy (Kawachi et al., 1991;Tillmann, 1998;Legrand et al., 2001;Liu et al., 2009).Thus, it is possible that some of the phototrophic eukaryotic signals in the deep sea of the Western Pacific Ocean were from mixotrophs which thrive in the dark ocean by preying on other microbes.The global ubiquity of phototrophic microbial eukaryotes in deep oceanic waters and the contribution of mixotrophs to the total microbial eukaryote community needs further exploration in order to determine: 1) their contributions to the downward transportation of organic carbon and, 2) the role of mixotrophs in microbial food webs through predation in the dark ocean.

Accepted Article
to 250 ml of seawater was fixed with glutaraldehyde (final conc.1%) at 4 °C in the dark for two hours before staining for the enumeration of pigmented nano-sized eukaryotes (PNEs).
Filters were then immersed into RNAlater solution (Ambion, CA, USA) and kept at -20 °C until further processing for sequencing of 18S rRNA and psbA gene transcripts.Four to ten liters of seawater from each of three depths (5 m, DCM, and 200 m) was filtered through a 47 mm GF/F glass fiber filter and immediately frozen in liquid nitrogen prior to phytoplankton pigment analysis in the laboratory.Samples were not available for pigment analysis at stations P8, P10, and P18 so seawater from stations P7 and P11 (not shown in Fig. 1), which were located mid-way between P6 and P8, and between P10 and P12, respectively, were used instead.

Accepted Article
obtain reliable estimates of their abundance, at least 50 cells were counted.PNEs were further classified into three size classes (2 -5, 5 -10, and 10 -20 µm) using an ocular micrometer.

Phytoplankton pigment composition
Phytoplankton composition was inferred using pigment markers as indicators.Chl a, an indicator for total phytoplankton biomass, referred to the sum of monovinyl and divinyl Chl a. Pigment analysis followed the protocol according to Huang et al. (2010).Briefly, phytoplankton pigments were extracted in N,N-dimethylformamide (Furuya and Marumo, 1983) and analyzed using High Performance Liquid Chromatography (Agilent Series 1100, Agilent Technologies) system fitted with a 3.5-mm Eclipse XDB C8 column (100 × 4.6 mm; Agilent Technologies) following a modification of the method of Mantoura and Llewellyn (1983) and Van Heukelem and Thomas (2001).Based on pigment data, the phytoplankton community was determined using CHEMTAX software (Mackey et al., 1996).

RNA extraction, PCR amplification, high-throughput sequencing, and processing of the hyper-variable V9 regions of the 18S rRNA
Total RNA was extracted from each cryopreserved filter membrane using AllPrep DNA/RNA kit (Qiagen, USA) following the protocols of Sun et al. (2017) and Xu et al. (2017b).
Extracted RNA was quantified using a Nanodrop ND-2000 Spectrophotometer (Labtech International, USA) and the integrity was assessed by 1.5% agarose gel electrophoresis.After treatment with DNase (Qiagen, Germany) to remove the carry-over genomic DNA, RNA was reverse-transcribed into cDNA using QuantiTect® Reverse Transcription kit (Qiagen, USA).
Genomic DNA was further removed by gDNA Wipeout Buffer supplied within the kit.The hyper-variable loop V9 regions of the 18S rRNA were amplified by applying the universal This article is protected by copyright.All rights reserved.

Accepted Article
primers, 1389F and 1510R (Amaral-Zettler et al., 2009).Five individual PCR reactions for each sample were run and pooled to minimize reaction-level PCR bias (Bates et al., 2013) and to collect sufficient amplicons for sequencing.The resulting PCR amplicons (ca.130 bp) were excised from the gel using MiniElute Gel Extraction Kit (Qiagen, USA).All purified amplicons were sent to a commercial company for paired-end sequencing with Illumina MiSeq platform.Sequences generated were deposited in NCBI Sequence Read Archive (accession number ******).
Quality filtering, demultiplexing and assembly of raw data were performed following Li et al. (2018) using Trimmomatic andFlash software (Magoc andSalzberg, 2011;Bolger et al., 2014).Chimeras were identified and removed with UCHIME (Edgar et al., 2011) using both de novo and reference-based chimera searches against the PR2 database (de Vargas et al., 2015).Singletons were further removed and reads were clustered into Operational Taxonomic Units (OTUs) at 95% sequence similarity using UPARSE (Edgar, 2013).
Generation of OTU tables and taxonomic assignment of OTUs were done with QIIME by UCLUST against the PR2 database (Caporaso et al., 2010).OTUs that were affiliated with Bacteria or Archaea, plastidial sequences, those assigned as Metazoa and those "Unassigned" were discarded.To normalize sampling effort, sequences were normalized by randomly resampling 7,078 reads per sample (the lowest number of sequences recovered for all samples).Alpha diversity estimators (Chao1, Shannon, ACE, Inverse of Simpson, and Phylogenetic Diversity) were calculated using QIIME (Caporaso et al., 2010).To assess similarities between communities, pairwise similarities among samples were calculated by Bray-Curtis similarity coefficient (Legendre and Legendre, 1998).Similarity matrices were visualized using Non-Metric Multidimensional Scaling (nMDS) diagrams.The differences among groupings from cluster analysis were tested by ANOSIM.To check the robustness of dissimilarity among communities, beta diversity was calculated using the unweighted Unifrac This article is protected by copyright.All rights reserved.

Accepted Article
metric (Lozupone and Knight, 2005) and community dissimilarity was shown using a twodimensional Principal Coordinate Analysis (PCoA).Amplification, cloning, sequencing, and processing of psbA gene transcripts PsbA gene fragments (ca.750 bp) were amplified from reverse-transcribed cDNA using primers psbA-1F and psbA-2R designed by Wang and Chen (2008).For each sample, PCR products from five separate reactions were pooled and clone libraries were constructed using TA cloning kit (TaKaRa, China) according to the manufacturer's instructions.Between 120-200 positive clones from each sample were commercially sequenced using M13F and M13R primers from both ends using an ABI 3730xl DNA Analyzer.
The obtained psbA sequences were automatically assembled and then manually checked using Sequencher 5.0 (Gene-Codes Corp., MI, USA).Each sequence was blasted against the GenBank database and sequences of prokaryote or virus/phage origin were excluded from the analysis.Potential chimeric sequences were detected using UCHIME (Edgar et al., 2011) using the de novo method and then removed.An OTU table was constructed by clustering sequences to OTUs at a 99% similarity threshold using UPARSE (Edgar, 2013).The most abundant sequence within each OTU was selected as the OTU representative sequence.
OTUs were then taxonomically assigned by BLAST search against nr-NCBI and the nearest neighbor (NN) and the nearest named neighbor (NNN) were identified (Scheckenbach et al., 2010;Xu et al., 2017a).All environmental psbA sequences generated in this study were deposited in GenBank under accession numbers ***** -*****.

Fig. 3 .
Fig. 3. Size-fractionated abundance (A) and biomass (B) distribution of PNEs in the water

Table 1 .
Abundance and biomass of pigmented nano-sized eukaryotes (PNEs) in the tropical Western Pacific Ocean.*Including one sample collected at 5,000 m depth.This article is protected by copyright.All rights reserved.

Table 2 .
Taxonomic identifications of the psbA genes retrieved from the Western Pacific Ocean, with their number of clones, and average similarity (%-S) with the nearest neighbor (NN) and the nearest named neighbor (NNN).