Summertime Amino Acid and Carbohydrate Patterns in Particulate and Dissolved Organic Carbon Across Fram Strait

Amino acids (AA) and carbohydrates (CHO) are important components of the marine organic carbon cycle. Produced mainly by phytoplankton as part of the particulate organic carbon (POC) fraction, these compounds can be released into the outer medium where they become part of the dissolved organic carbon (DOC) pool and are rapidly taken up by heterotrophs (e.g., bacteria). We investigated the quantity and quality of POC and DOC, AA and CHO composition in both pools in three different water masses in the Fram Strait (Arctic Ocean) in summer 2017. Polar Waters and Atlantic Waters showed similar concentrations of particulate and dissolved AA and CHO, despite Polar Waters showing the highest DOC concentrations. In Mixed Waters, where the two water masses mix with each other and with melting sea ice, the concentrations of particulate and dissolved AA and CHO were highest. AA and CHO composition differed substantially between the particulate and dissolved fractions. The particulate fraction (>0.7 μm) was enriched in essential AA and the CHO galactose, xylose/mannose, and muramic acid. In the dissolved fraction non-essential AA, several neutral CHO, and acidic and amino CHO were enriched. We further investigated different size fractions of the particulate matter using a separate size fractionation approach (0.2–0.7 μm, 0.7–10 μm and >10 μm). The chemical composition of the 0.2–0.7 μm size-fraction had a higher contribution of non-essential AA and acidic and amino sugars, setting them apart from the 0.7–10 μm and >10 μm fractions, which showed the same composition. We suggest that the relative differences observed between different size fractions and DOC with regards to AA and CHO composition can be used to evaluate the state of organic matter processing and evaluate the contribution of autotrophic phytoplankton or more heterotrophic biomass. In the future, changing conditions in the Central Arctic Ocean (Atlantification, warming, decreasing ice concentrations) may increase primary production and consequently degradation. The AA and CHO signatures left behind after production and/or degradation processes occurred, could be used as tracers after the fact to infer changes in microbial loop processes and food web interactions.


INTRODUCTION
Through photosynthesis phytoplankton cells (and also seaice algae in higher latitudes) take up inorganic carbon and incorporate it into numerous organic molecules. The four major biomolecules, amino acids (AA), carbohydrates (CHO), fatty acids, and DNA/RNA make up the bulk part of cell biomass (Geider and LaRoche, 2002). Their synthesis depends on availability of light, temperature and the overall physiological state of phytoplankton cells, however, intra-cellular proportions are also massively regulated by N and P availability, consequently affecting biomolecule synthesis rates, cellular C:N:P ratios and overall growth rates (Arrigo et al., 1999;Falkowski, 2000;Geider and LaRoche, 2002). Under nutrient limiting conditions the partitioning of C into C-rich storage compounds such as carbohydrates or triglycerides is increased and the partitioning between essential AA (EAA) and non-essential AA (NEAA) is affected (Grosse et al., 2017(Grosse et al., , 2018. Additionally, the contribution of individual AA and CHO to total AA and CHO concentrations in phytoplankton can depend on community composition (Markou et al., 2012 and references therein).
Phytoplankton cells can actively excrete organic matter (OM) (Baines and Pace, 1991;Thornton, 2014) hence AA and CHO become part of the dissolved organic matter pool (DOM) that small heterotrophic consumers (e.g., bacteria) depend on. Other pathways by which OM enters the dissolved pool include passive release through the cell membrane (Marañón et al., 2004), through cell destruction via viral lysis or grazing (Brussaard, 2004;Saba et al., 2011), or enzymatic degradation of particulate organic matter (Wetz et al., 2008). Overall, the released AA and CHO can contribute up to 20% to open ocean DOC pool and even more in shelf areas (Pakulski and Benner, 1992;Davis et al., 2009;Kaiser and Benner, 2009;Panagiotopoulos et al., 2014).
The accessibility of total dissolved AA and dissolved combined CHO (TDAA and DCCHO) for bacteria depends on the molecular complexity. For example, monosaccharides and single (free) AA are labile, taken up quickly by bacteria, and have short turnover times of minutes to days. In their study Luo et al. (2010) determined that labile compounds alone cannot support bacterial biomass in the ocean but that up to 40% of bacterial C-demand is satisfied by semi-labile OM. The semi-labile fraction has two major contributors: Oligo-and polysaccharides (combined CHO) and peptides (combined AA) (Søndergaard et al., 2000;Engel and Haendel, 2011). Because of their higher complexity, turnover times of the semi-labile fraction range from months to years .
Individual AA and CHO are taken up preferentially based on their molecular structure and can be reworked into more degraded compounds (Dauwe and Middelburg, 1998;Veuger et al., 2012). For example, loss rates of the individual amino acids isoleucine, threonine, and valine were high during a longterm degradation experiment, while glycine showed low rates, and no loss was detected for lysine and proline (Veuger et al., 2012). As some molecules are precursors or breakdown products of others, the synthesis/degradation of one will lead to a relative accumulation of another. Using the molecular composition of the AA pool, a degradation index was developed to infer the relative age or "freshness" of a sample (Dauwe and Middelburg, 1998;Kaiser and Benner, 2009). Similarly, dissolved carbohydrates degradation alters neutral CHO composition, for example by a relative decrease in arabinose and galactose and a relative increase in glucose (Amon and Benner, 2003;Goldberg et al., 2009;Engel et al., 2012).
So far, studies have focused on either the dissolved or the particulate pool or on either AA or CHO (Borchard and Engel, 2015). But a concurrent assessment of the AA and CHO composition in different size fractions (e.g., phytoplankton, bacteria, DOC) would be necessary to characterize sourcesink interactions.
Our study aims to determine the quantity and quality of individual biomolecules within the particulate and dissolved OM pools. To evaluate if the biomolecule composition of producers is mirrored by heterotrophic bacteria, we also investigated biomolecule composition of three different POC size fractions. The three studied fractions targeted larger primary producers such as Phaeocystis sp. colony stages and larger or chain-forming diatoms (>10µm), smaller primary producers and protozoa, i.e., Phaeocystis sp. single flagellates or other nano-and dinoflagellates (0.7-10 µm) and small auto-and heterotrophs including bacteria and archaea (consumers, 0.2-0.7 µm).
Our study took place in the Fram Strait, the only deep gateway to the central Arctic Ocean, an ocean heavily impacted by climate change. The eastern Fram Strait is influenced by warm and saline Atlantic Water (AW), which is transported north via the West Spitsbergen Current (WSC, e.g., Cottier et al., 2005;Beszczynska-Möller et al., 2012;Appen et al., 2016). In the west, the East Greenland Current (EGC) transports cold and fresher Polar Water (PW) along the Greenland continental shelf break toward the South (e.g., de Steur et al., 2009). Compared to the AW, the PW contains higher DOC concentrations that originate from terrigenous sources supplied by Arctic Rivers via the Transpolar Drift (Hansell et al., 2004). The East Greenland Polar Front System separates these two water masses (Paquette et al., 1985). However, substantial horizontal mixing and exchange by eddies can occur (Wekerle et al., 2017) and together with meltwater from drifting ice floes produce an Mixed Water (MW) body.
We hypothesize that water masses differ in the quantity and quality of their biomolecule composition. To address this hypothesis, we focus on two main questions: (1) to what extent do water masses affect the biomolecule composition of POC and DOC in terms of concentration and composition, (2) does the biomolecule composition differ between size fractions of particulate matter?

Study Area and Sample Collection
During the summer season of 2017 (23 July-19 August), a total of 15 stations were sampled within the LTER observatory HAUSGARTEN with the research vessel RV Polarstern (Figure 1). The sampling was carried out using 12L Niskin bottles attached to a CTD rosette (Sea-Bird Electronics Inc. SBE 991 plus probe), equipped with temperature and conductivity FIGURE 1 | Map of stations sampled for dissolved and particulate compounds (black squares and gray triangles). Size fractionation was performed on stations EG3, HG1, and N5, depicted as gray triangles. The dotted black line indicates the sea ice monthly median edge retrieved from https://www.meereisportal.de (Spreen et al., 2008). sensors, as well as a pressure sensor, altimeter, chlorophyll a (Chl a) fluorometer, and transmissometer. Hydrographic data, including temperature and salinity, was retrieved from PANGAEA (von Appen and Rohardt, 2018). Water samples were taken from 5 depths: Above Chl a maximum (10 m), at the Chl a maximum (according to the fluorescence sensor in 15-35 m), below Chl a maximum (20-45 m), at 50 and 100 m. Subsamples were taken as described below and stored for later analysis in the home laboratory.
A size fractionation of PAA and PCHO was performed at stations HG1, EG3, and N5 (Figure 1). First, 600 ml of sample was filtered over a 10 µm gauze. Visible zooplankton individuals (e.g., copepods) were carefully removed using tweezers. Contents from the gauze were washed onto a pre-combusted GF/F filter (4 h, 500 • C) using filtered seawater (0.2 µm) resulting in one replicate for >10 µm. The filtrate was again filtered over pre-combusted GF/F filters in duplicates of 300 ml and a cut-off at 0.7 µm. The filtrate was then filtered over Supor Filters (0.2 µm, Pall) to collect the size fraction 0.2-0.7 µm. All samples were stored in pre-combusted aluminum-foil packs (4 h, 500 • C) at-20 • C and analyzed as described below.

Dissolved and Particulate Organic Carbon (DOC, POC)
Samples for DOC were taken in duplicates by filtering 20 ml of seawater through GMF filters (0.45 µm) and collected in combusted glass ampoules (4 h, 500 • C). Samples were acidified with 20 µl of 30% hydrochloric acid, flame sealed, and stored at 4 • C in the dark until analysis. In the lab, DOC samples were analyzed by high-temperature catalytic oxidation (TOC -VCSH, Shimadzu, Sugimura and Suzuki, 1988) using the modified protocol after Engel and Galgani (2016). POC samples were taken by filtering aliquots of 1-6 L of seawater onto combusted GF/F filters (4 h at 500 • C). Filters were stored frozen (-20 • C) until analysis. In the lab, filters were soaked in 0.1N HCl for removal of inorganic carbon and dried at 60 • C. POC concentrations were determined with a Carlo Erba CHN elemental analyzer (Sharp, 1974).

Nutrients, Chlorophyll a, TEP
Samples for the analysis of inorganic nutrients were collected directly from Niskin into sterile 50 ml Falcon tubes. The tubes were rinsed with sample water three times before the sample was drawn. Samples were then stored at -20 • C for later analysis on land. Inorganic nutrients (nitrate + nitrite, nitrite, phosphate, silicate, and ammonium) were analyzed with a QuAAtro Seal Analytical continuous flow autoanalyzer following the manufacturers methods, which are based on common colorimetric techniques after Grasshoff et al. (2009). All measurements were calibrated using a set of five nutrient standards prepared with commercially available Merck solutions (traceable to SRM from NIST). These standards were prepared in an artificial seawater (ASW) matrix, which was also used as wash-water between the samples. Chlorophyll a (Chl a) samples were taken by filtering 0.5-2 L seawater onto glass fiber filters (Whatman GF/F) under low vacuum (<200 mbar). All filters were stored at −20 • C until pigments were extracted with 5-10 ml of 90% acetone. First, filters were sonicated in an ice bath for <1 min, and subsequently extracted at 4 • C for 2 h. Before the measurement, samples were centrifuged for 10 min at 5,000 rpm at 0 • C. Chl-a concentration was determined fluorometrically (Turner Designs), together with total pheophytin concentration after acidification (HCl, 1.0 N) slightly modified to the methods described in Edler (1979) and Evans et al. (1987), respectively. The standard deviation of replicate test samples was (<10%).
The acidic polysaccharide-containing transparent exopolymer particles (TEP) were analyzed by photometric analysis. Therefore, 60-150 ml samples were filtered onto 0.4 µm polycarbonate filters, stained with a calibrated Alcian Blue solution and rinsed with several milliliter of ultrapure water (Passow and Alldredge, 1995). The filters were stored at-20 • C until spectrophotometric analysis in the home laboratory. The adsorption of Alcian Blue to the sample was detected colorimetrically. In order to dissolve particles, each filter was incubated for 3 h in 6 ml of H 2 SO 4 (80%). Subsequently, the solution was measured at 787 nm with an UV-Vis spectrophotometer (Shimadzu UV-1700 PharmaSpec). Xanthan gum was used for calibration, and total TEP concentrations are given in mg xanthan gum equivalent (Xeq) L −1 .

Statistical Analysis
Statistical comparison of water mass composition of OM constituents and nutrients was conducted using the Kruskal-Wallis ANOVA because datasets showed no normal distribution after Kolmogorov-Smirnov-test. Post hoc comparison of means (Wilcox test) was used to assess significant differences among water masses.
Principle component analysis (PCA) was performed to explore differences in individual AA and CHO composition between dissolved and particulate pools. Data for the relative contribution (%) of individual AA and CHO concentrations to total AA and CHO concentrations (nmol L −1 ) was used. The package CRAN:factoMineR in the open-source software R was used for the PCA analysis using a correlation matrix (Lê et al., 2008;R Development Core Team, 2014).

RESULTS
Three different water masses were identified using temperature, salinity, and density data according to Cokelet et al. (2008, Figure 2). The main water mass in the eastern part of the Fram Strait was Atlantic Water (AW) defined by temperatures > 2 • C and salinities > 34.9. Polar Water (PW) was encountered with temperatures < 0 • C and salinity < 34.7, in the western Fram Strait. Stations, where the two water masses mix and the sea surface layer was impacted by sea-ice melt, were referred to as Mixed Water (MW). MW temperatures ranged between 0 and 7.2 • C and MW salinities ranged between 29.6 and 34.9. Water mass specific differences in nutrient distribution occurred as well (Supplementary Figure 1). In general, PW was characterized by higher concentrations of PO 4 3− and Si(OH) 4 . Significant differences in NO 3 − only occurred between AW and PW in sampling depths below the Chl a maximum. The upper two water depths (Chl a maximum and above) showed no significant differences. Average concentrations and concentration ranges for Chl a and TEP were similar for AW and PW and highest in MW ( Table 2 and Supplementary Figure 2).

Distribution of Dissolved and Particulate Constituents
Highest POC, PAA, and PCHO concentrations were found in MW. Their respective concentrations ranged between 1.1 and 21 µmol L −1 for POC, 0.27-6.5 µmol C L −1 for PAA and 0.02-0.86 µmol C L −1 for PCHO (Figure 3 and Supplementary  Figure 3). Contribution of PAA to POC was 28 ± 16% (n = 62), PCHO contributed 4.8 ± 2.7% to POC (n = 63,  Similarly, 67.6% of the variance in CHO can be explained by PC1 and PC2 (Figure 4B). PCHO showed increased contributions of Gal, Xyl/Man, and MurA; Rha was not present in any of the samples. DCCHO had increased contributions of Ara, Rha, Fuc, of amino sugars (GlcN, GalN) and acidic sugars (GlcA, GalA). Furthermore, MurA was not detected in any DCCHO sample, and no GlcUA was detected in DCCHO and PCHO samples, with the exception of 2 sampling depths during size fractionation (see below). Glc was the main driver of PC2 and showed increased contributions in both DCCHO and PCHO.
Additional PCA analyses were performed separately for the dissolved and particulate pools to determine the effect of water masses and depth on biomolecule composition (Figure 5). An effect of water mass was observed in the TDAA composition (Figure 5A), where PW showed an association with GABA (green symbols). Additionally, a separation by sampling depth in the AW and MW was detected where samples from the Chl a maximum and above (blue and black triangles) were enriched in EAA (e.g., Tyr, Leu, Phe), and deeper samples were rather associated with higher contributions of NEAA Ala, Thr, Gly (circles). A less clear trend emerged in the PAA (Figure 5B), where PW samples from below the nutricline (green circles) associated with GABA and Arg and were slightly separated from many of the AW and MW surface samples (blue and black triangles), which associated closer to the EAA.
CHO did not show water mass separation in either fraction, but some patterns emerged as well. Within the DCCHO ( Figure 5C) Glc contribution seems to increase in deeper samples (circles and squares) while surface samples of AW and MW (black and blue triangles) are associated with all other investigated sugars, including amino and acidic sugars. Similarly, in the PCHO (Figure 5D), many deep-water samples (circles) were associated with Xyl/Man and Ara, and many surface samples were associated with Glc.
Phytoplankton most likely dominated in the size fractions >0.7 µm which contained 4.3 ± 2.4 times more C in PAA than in PCHO (n = 30). The size fraction < 0.7 µm was most likely dominated by bacteria and archaea and contained only 0.8 ± 0.4 times of C in PAA than PCHO (Figure 6 and Supplementary Table 1).

Where Water and Melting Ice Meet
The Fram Strait is the gateway to the Arctic Ocean where northward-flowing, warm and more saline AW in the east meets southward flowing, cold PW in the west. The two water masses meet and mix with each other and melting sea-ice in the marginal ice zone, forming MW. These mixing zones extend over several km and are characterized by higher biological productivity (Thomsen et al., 2016;Baltar and Arístegui, 2017;Fadeev et al., 2020). Our research area encompassed all three water masses and the study supports earlier findings that MW is indeed more productive than either PW or AW (based on Chl a and POC concentrations). A detailed study of such a mixing event showed that upwelling on the outside of the filament brings nutrients to the surface, which enhances productivity and can eventually export POC in the filament center on timescales of days to weeks (von . In the same filament, Fadeev et al. (2020) investigated the composition of bacteria and phytoplankton and found elevated densities of both diatoms and Phaeocystis sp. compared to outside-filament waters, and a bacterial community that is usually associated with a summer aging phytoplankton bloom. Blooming of fast-growing phytoplankton species, such as diatoms, were also observed in other low-latitude, submesoscale-frontal systems (Basterretxea Oyarzabal and Aristegui, 2000;Clayton et al., 2014;Allen et al., 2015). A similar mixing process and the concurrent injection of new nutrients was potentially responsible for increased biomass in MW in our study. The subsequent DOM release from new phytoplankton biomass and its degradation after nutrient depletion fuels the existing bacterioplankton community, emphasizing that this narrow streak of MW between AW and PW is important for primary and secondary production in the area. According to Nöthig et al. (2020) AW is more productive than PW in terms of both POC and Chl a concentration, opposite to us finding no statistical differences in POC concentration between the two. The reason for this could be two fold; (1) the sampling in 2017 took place in post-bloom conditions (Engel et al., 2019) and low overall biomass resulted in relatively similar concentrations between water masses. Secondly, Nöthig et al. (2020) only distinguish between PW and AW, using the 0 • Meridian as a boundary, rather than physical settings, and therefore included MW as part of AW. However, they also emphasize that productive zones were found at the dynamic ice edge of the Fram Strait where warm and nutrient rich AW partly recirculates and comes into contact with the sea ice cover of the western Fram Strait allowing favorable growth conditions for phytoplankton. MW were also characterized by increased contribution of PAA to POC in our study, indicating sufficient N availability for phytoplankton growth. Available N supports the synthesis of PAA (structural proteins and enzymes required for phytoplankton cells functioning and growth), while increased PCHO can indicate either an increased contribution of storage PCHO under nutrient limitation (Granum et al., 2002;Borsheim et al., 2005) or high amounts of excreted heteropolysaccharides such as transparent exopolymeric particles (TEP) (Passow, 2002;Engel et al., 2017). Based on the similarity seen in Chl a, POC and TEP concentrations, probably both mechanisms contributed. Nutrient distribution in the upper two water depths was patchy and ranged from concentrations below the detection limit to sufficient concentrations across all three WM, hence we probably encountered plankton communities at different physiological stages. While POC concentrations can be used to determine biomass distribution, DOC concentrations across Fram Strait are not appropriate to deduce the availability of labile and semi-labile organic substances. DOC concentrations in PW were much higher than in AW, but due to increased terrigenous contributions from large rivers such as Lena, Mackenzie, Ob, or Yukon (Kattner et al., 1999;Anderson and Amon, 2015). The terrigenous DOC contains numerous complex biomolecules (e.g., humic substances) that are resistant to bacterial degradation (Engel et al., 2019). Thus, evaluation of TDAA and DCCHO concentrations is a more appropriate approach. The freshness of DOC increased from PW < MW < AW with highest contributions per water mass of 4.3% < 6.1% < 8.2% of DOC, respectively. These values are low but in a similar range to other studies. In the Beaufort Sea the contribution ranged between 4 and 16% (Panagiotopoulos et al., 2014). Amon and Benner (2003) reported neutral sugar contributions to DOC ranging between 1.3 and 9.6% throughout the entire Arctic Ocean, and up to 22% in ice flow melt water.

Quality of AA and CHO Pools
Organic carbon production and degradation determine the quantity and quality of AA and CHO pools. PCA analysis of AA and CHO from all water masses showed clear separation of particulate (producers) and dissolved (degraded) pools, demonstrating fundamental differences in their compound composition. PAA were rich in the EAA supporting the hypothesis that they contain actively growing phytoplankton communities that synthesize all investigated AA. EAA are important to sustain growth in higher trophic levels as these have no capabilities to synthesize EAA. Several groups of bacteria have also lost synthesis pathways for individual AA and rely on scavenging them from the environment (auxotrophy, Tripp et al., 2009). The TDAA pool is enriched in AA, such as Gly, AsX, and Ala, emphasizing continuous degradation by heterotophic bacteria (Kaiser et al., 2017).
Hence, PC1 ( Figure 4A) is a reflection of the degradation index established by Kaiser and Benner (2009). Interestingly, our data showed a separation between of GlX and Ser along PC2. A similar separation was found between Atlantic and Pacific Water ) and could potentially be an indicator for differences in bacterial communities composition. For example, it is known that peptidoglycan, peptides in gramnegative bacteria, exclusively contain GlX, Ala and MurA (see below), while other peptides can additionally include AsX, Ser, and ornithine (Schleifer and Kandler, 1972). GlX is used by some Gram-positive Bacilli to form a capsular polypeptide (Hanby and Rydon, 1946;Troy, 1973) and Ser frequently appears in iron scavenging siderophores (Demange et al., 1990;Bernardini et al., 1996;Martinez et al., 2000) and lipo-peptides (Morikawa et al., 1993;Vater et al., 2002).
Distinguishing further between water masses and depths (Figures 5A,B) revealed a separation of TDAA and PAA associated with the degradation index (along PC1). Essential TDAA are associated with productive surface waters (Chl a maximum and above) especially in AW and MW. Deeper waters are associated with higher contributions of Ala, Thr and Gly. PW show the most degraded TDAA pool associated with Gly, Thr and especially GABA, indicating highly degraded TDAA pools in the entire PW water column. GABA is a non-protein AA produced by marine bacteria but also by fungi (Morse et al., 1980;Masuda et al., 2008). Differences in GABA, GlX and Ser distribution could again indicate compositional differences in microbial communities between ice-free AW and ice-covered PW as demonstrated by Fadeev et al. (2018) and others (Pernthaler et al., 1998;Crump et al., 1999).
Similarly, PCHO and DCCHO pools show differences in composition. Gal has increased contributions to the PCHO pool, which could be associated with diatoms (Markou et al., 2012). MurA was also found in PCHO. It is however only found in the unique cell wall biopolymer peptidoglycan, which is synthesized by Gram-negative and Gram-positive bacteria (Schleifer and Kandler, 1972). It consists of glycan strands of the disaccharide N-acetyl-glucosamine and N-acetyl muramic acid which are cross-linked by small peptides consisting of specific amino acids (Schleifer and Kandler, 1972). Muramic acid has been identified in DOM (Benner and Kaiser, 2003) and might be associated with bacteria colonized particles in our study. Therefore, the MurA signal in the PCHO fraction could signal the presence of gellike particles, which can be colonized by bacteria. Transparent polymer particles (TEP) are one form of gel-like particles and contain large amounts CHO. Acidic sugars (GlcA, GalA, GalUA) in the DCCHO can be an indicator for freshly excreted DOM (Borchard and Engel, 2015) while in the PCHO fraction they are an additional indicator for the presence of TEP (Alldredge et al., 1993). The occurrence of MurA in the particulate fraction, e.g., as indicator for gel-like particles should be investigated further.
As seen in Figure 4B, PC1 shows qualitative differences between PCHO and DCCHO, while PC2 is driven by Glc. Glc contributions to PCHO are usually explained by nutrient limitation. When photosynthesis is limited by either N or P, the further synthesis of Glc into AA, structural lipids or DNA/RNA is limited, reducing community growth rates (Falkowski, 2000). Instead, Glc is then incorporated into storage products such as Glc-rich chrysolaminarin or processed further into storage lipids (Janse et al., 1996;Kroth et al., 2008). Alternatively, increased contributions of TEP could be reflected in the shifting relative composition of PCHO.
Looking closer into water mass and depth-related distribution (Figures 5C,D) confirm the special role of Glc, especially in the DCCHO fraction where it is associated foremost with waters below the Chl a maximum and the nutricline. Glc is increased in the PCHO pool in the surface (Chl a max and above), where when excreted Glc is taken up rapidly by bacteria and hardly accumulates (Kirchman et al., 2001). The relative increase of DCCHO-Glc at depth could be associated with the diagenesis of CHO and the uptake of more labile sugars from sinking OM (Amon and Benner, 2003). The melting ice might also release degradation products from ice algae together with ballast material, which causes particle sinking rates to increase 2-3-fold (Iversen and Robert, 2015), and would relocate degradation processes away from the surface and deeper into the water column.

Differences in Size Fractions
The three size fractions targeted larger primary producers such as Phaeocystis sp. colony stages and larger or chain-forming diatoms (>10 µm), smaller primary producers, i.e., Phaeocystis sp. single flagellates or other nano-and dinoflagellates (0.7-10 µm) and bacteria and archaea (0.2-0.7 µm). Comparing PAA and PCHO between these fractions shows the importance of smaller sized phytoplankton in the Arctic Ocean during post-bloom summer month as they contribute higher concentrations to both pools at all the investigated stations. Larger phytoplankton (>10 µm) had lowest contribution in PW. As to be expected, concentrations are highest at the surface and in the Chl a maximum (≤20 m) and based on PAA concentrations, smaller phytoplankton also contribute relatively more at greater depths. The contribution of PCHO (>0.7 µm) to POC in this study was low in comparison to the literature. Finkel et al. (2016) concluded from a meta-data analysis that CHO constitute approximately 15% to cell biomolecules under nutrient-sufficient, exponential growth conditions. They found the lowest values for Ochrophyta and Bacillariophyta (<10%), comparable to values in this study (4.8 ± 2.7%). These values are at a similar range as values found for our samples, suggesting that nutrients were still available throughout the study area. To increase accuracy in follow-up studies, volumes for PAA and PCHO filtration should be similar or higher than volume filtered for POC analysis, especially in low biomass areas. During our study we encountered postbloom conditions with communities comprising of diatoms and Phaeocystis spp. as shown by Fadeev et al. (2020). While diatoms contain less PCHO because they have a silica shell instead of a cell wall, Phaeocystis is mucus-, hence PCHO-rich, especially when encountered as colonies (Alderkamp et al., 2007). Moreover, high concentration of TEP associate with blooms of Phaeocystis sp. have been reported for the Fram Strait (Engel et al., 2017). Exponentially growing as well as senescent cells, both have been shown to induce TEP formation (Passow and Wassmann, 1994;Reigstad and Wassmann, 2007). Lipids have to be considered as an alternative storage compound in phytoplankton and partitioning of storage C between CHO and lipids depends on the community composition and the experienced limitation (N, P, or light) (Granum et al., 2002;Borsheim et al., 2005;Grosse et al., 2017).
POC:PON ratio in the >0.7 µm size-fractions were close to the Redfield ratio (C:N = 6.6, Table 3, Redfield et al., 1963) in surface waters and increased with depth indicating degradation The molar ratios were calculated by (C AA + C CHO )/N AA for size fractions and the dissolved pool.
processes. C:N ratios of PCHO and PAA [(C AA + C CHO )/N AA ] were lower (Table 3), suggesting additional C components (e.g., lipids, pigments, detritus) contribute to the larger fraction. In the 0.2-0.7 µm, the (C AA + C CHO )/N AA ratio is increased relative to the larger size-fractions and resembles the C:N ratios typically found in marine bacteria (C:N = 5-8), with coastal areas at the lower side of the range and open ocean bacteria at the higher side (Fagerbakke et al., 1996;Fukuda et al., 1998). The 0.2-0.7 µm fraction was therefore a mixture of bacteria and detritus (see also below). Consumers rework phytoplankton derived DOC to retain precious N resulting in DOC:DON ratios > 9 ( Table 3, Eichinger et al., 2009). Labile compounds can be consumed very rapidly by heterotrophs who rework those compounds and excrete their own DOC characterized by a complex chemical composition (Ogawa et al., 2001). The amino sugars GlcN and GalN are commonly found in bacteria-derived DOC (Amelung, 2003;Benner and Kaiser, 2003). Both amino sugars can accumulate outside the cytoplasm, perhaps in membrane-bound components. They have been found in glycoproteins, extracellular polymers, antibiotics (Messner, 1997;Giroldo and Vieira, 2002;Kudo et al., 2005), and as part of lipopolysaccharides (Trefzer et al., 1999). They also have very low decay coefficients, making them resistant to decomposition (Kawasaki and Benner, 2006).
The concentrations of PAA in the 0.2-0.7 µm fraction can be used to estimate bacterial C concentrations. Overall, AA amount to half of the C in bacteria (Simon and Azam, 1989;Neidhardt, 1996;Kaiser and Benner, 2008) and the percentage of total cellular C as AA seems fairly constant (50-60%) in marine heterotrophic bacteria, independent of cell size (Simon and Azam, 1989;Kaiser and Benner, 2008). However, those numbers were mostly derived from culture studies and natural communities show lower values. Kawasaki et al. (2011) determined bacterial AA-C yields between 15.5 and 42.5% for the open ocean and 19.3-33.8% for the coastal ocean and concluded values are driven by the contribution of degraded particles. Assuming 9.1 fg C bacteria cell −1 (Buitenhuis et al., 2012), bacteria contributed 1.51-8.91 µM C resulting in AA-C yields of 3.2-10.6%. These very low AA-C yields indicate that our sampling stations were probably affected by high amounts of other C-rich material, such highly degraded detritus (Amy et al., 1983;Neidhardt, 1996) or TEP (Engel et al., 2017).
In the PCA, the two phytoplankton size fractions clustered and were enriched in most AA except for AsX and GlX, and in the CHO Gal, Fuc and MurA. Since we only investigated three stations with five depths each, the data is not sufficient to interpret if and where detritus might have played a significant role in any size fraction. Interestingly, Ser seems to be associated with samples taken from greater depths and potentially higher degraded particles, while GABA is associated with the Chl a maximum and above perhaps suggesting a bacterial presence on particles. Gal, Fuc, MurA were mainly associated with MW and AW, while Man/Xyl was mostly associated with PW and greater depths, and therefore probably associated with degradation. This agrees with findings in the PCA of unfractionated PAA.
The relative composition of the biomolecule pools can be used to determine the production and degradation state of the POC and DOC pools and even to distinguish between phytoplankton and bacterial biomass as well as different water masses. With Atlantification AW will carry their respective plankton communities farther North into the Central Arctic Ocean. Increased production in the central Arctic Ocean is expected in the future, and consequently enhance degradation processes in these currently oligotrophic areas. As a consequence, food web interactions and microbial loop processes will be affected. Since sampling campaigns in these areas are infrequent and logistically challenging the AA and CHO signatures left behind by these processes could be used to tracers after the fact.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. This data can be found here: the Pangea data repository https://doi.pangaea.de/10.1594/PANGAEA.932070 and https://doi.pangaea.de/10.1594/PANGAEA.932068.

AUTHOR CONTRIBUTIONS
JG analyzed the data and wrote the manuscript with guidance from E-MN and AE. E-MN and ST-V provided POC and nutrient data. All authors contributed to the final version of the manuscript.

FUNDING
This is a contribution to the MicroARC project (03F0802A) within the Changing Arctic Ocean program, jointly funded by the UKRI Natural Environment Research Council (NERC) and the German Federal Ministry of Education and Research (BMBF). This research has also been supported by the Helmholtz Association. Ship time was provided under grant AWI_PS107_09.