Dissolved organic matter represents an important component of oceanic carbon cycle (Stein et al., 2004; Vetrov and Romankevich, 2010) that influences on ecosystem functioning controlling microbial food webs and through many physical and biogeochemical processes, such as sunlight absorption, photo- and biodegradation, flocculation, formation of metal-organic complexes, etc. (Jaffé et al., 2008). In the context of serious concerns about global climate change, DOM is considered as a key component of the biological pump - a series of processes through which CO₂ is fixed as organic matter by photosynthesis and then transferred to the ocean interior by means of particulate organic matter (POM) sedimentation and transport of DOM to deep waters. This mechanism includes the transformation of the labile DOM of euphotic zone into recalcitrant DOM through the microbial carbon pump (Jiao et al., 2010). Since the production, transformation, and consumption of organic matter in aquatic ecosystems is strongly dependent on DOM composition (D’Andrilli et al., 2019), DOM quality represent a perspective indicator of biogeochemical trends related to climate change and anthropogenic stress (Wohlers et al., 2009; Savun-Hekimoğlu and Gazioğlu, 2021).

Recently, optical techniques have become widespread in the study of CDOM. It is related to multiple factors, such as i) fast and simple preparation of seawater for laboratory analysis, ii) developments of in situ fluorescence sensors (Carstea et al., 2020), iii) development of algorithms for CDOM and DOC determination with the use of satellite remote sensing (Griffin et al., 2018), iv) incubation experiments aimed to investigate production of FDOM by marine biota, i.e. (Castillo et al., 2010; Romera-Castillo et al., 2011), v) development PARAFAC decomposition of EEMs (Bro, 1997) and OpenFluor spectral library (Murphy et al., 2014), vi) investigation of the influence anthropogenic factors, such as oil pollution (Gonnelli et al., 2016) and microplastics (Chen et al., 2018) on the fluorescence spectra of natural waters. Consequently, in the study of Jaffé et al. (2008) the importance of absorbance and fluorescence measurements as a part of long-term ecological research and monitoring programs were emphasized, since they may provide an advanced understanding of organic matter dynamics in aquatic ecosystems.

In the recent decades, the Arctic is undergoing significant transformations induced by climate change. Carbon fluxes and sources of DOM are changing due to increase in the discharge of major rivers terminating in the Arctic, reduce the ice cover, and melting of permafrost (Box et al., 2019 and references therein). Sub-sea permafrost thaw and formation of taliks result in release of organic matter of glacial origin or modern production from ancient organic matter (Anderson et al., 2011; Vonk et al., 2012). The resulting microbial decomposition of frozen carbon pool is a potential feedback process affected global climate dynamics accelerating global warming (Schuur et al., 2008). The Siberian shelf seas, in particular the East Siberian Sea, are considered to be the most affected by the above changes.

Although a lot of studies were devoted to the CDOM optical properties in the Arctic Ocean, see, for example (Stedmon et al., 2011; Gonçalves-Araujo et al., 2015; Mann et al., 2016), СDOM and especially FDOM of the East Siberian Sea shelf, remains one of the least investigated. In the paper of Mann et al. (2012) DOC concentration and the composition of CDOM across the hydrograph in Siberia’s Kolyma River is discussed. Water sampling was performed in the vicinity of the Northeast Science Station near Cherskiy, Russia. The composition of DOM within the Kolyma basin was characterized using absorbance-derived measurements and fluorescence spectroscopy, including EEMs + PARAFAC analysis. The B, T, M, A, and C fluorophores according to Coble (Coble et al., 1998) were identified, and substantial losses of DOC occurring via biological and photochemical pathways were demonstrated. The study of Guéguen et al. (Guéguen et al., 2012) examined seawater FDOM outside the East Siberian Sea shelf north of 75^○N at depths >100 m. Two protein-like (B, T) and three humic-like (A/C, M) components were identified by PARAFAC. High protein-like and terrestrial humic-like FDOM signals were most often associated with surface waters. Microbial/marine humic-like component was associated with the middle halocline at depth 50-200 m. It was suggested to be of Pacific origin. Another study of East Siberian Sea FDOM was performed recently by Chen et al. (2021) as a part of investigation of organic carbon release from the sediments of East Siberian subsea permafrost. The shelf (north of 73^○N) and slope sites were examined. Six FDOM components, namely four humic-like and two protein-like, were identified from PARAFAC modeling. Strong signals of terrestrial humic-like FDOM in seawater and pore waters allowed suggesting the predominance of permafrost-derived DOM over the products of marine sediment mineralization. Marine/microbial humic-like component was found to be more abundant in the seawaters on the slope sites (28-47%) compared to East Siberian shelf waters (8-16%).

This paper focuses on characterization of FDOM in East Siberian Sea waters influenced by Indigirka and Kolyma rivers in the end of the productive period when the finality of phytoplankton development was detected (Drits et al., 2019). It continues the investigation of East Siberian Sea DOM (Drozdova et al., 2021), that revealed high variability of seawater absorbance and DOC that cannot be described by conservative mixing often reported for the river-influenced Arctic coastal regions. Other areas under investigation are the western and central parts of the Laptev Sea influenced by the Khatanga and Lena river runoff. With respect to the East Siberian Sea, the Laptev Sea represents a contrasting example of a river-influenced shelf region where DOC was distributed conservatively. The study is based on excitation-emission matrices of seawater, sampled during the 69^th cruise of R/V Akademik Mstislav Keldysh in September 2017. We analyze the distribution of PARAFAC components along the 4 cross-shelf transects to get a better understanding of DOM quality in the region.

Component C1 matched a lot of EEM - PARAFAC models describing CDOM from various regions, including Arctica (Brogi et al., 2018; Painter et al., 2018) and Antarctica (Kida et al., 2019). It was usually assigned as terrestrial humic material and was abundant mostly in rivers, lakes, estuarine, costal and shelf regions of the oceans. The spectral characteristics of C1 were reported to be similar to the ones of syringaldehyde which, along with vanillin, is a product of oxidative degradation of lignin (Murphy et al., 2011). We conclude that C1 in the Laptev and East Siberian seas is supplied by river runoff. Component C2 exhibited protein-like fluorescence ( Table 2 ) typical for proteins and/or aromatic amino acids and often linked to aquatic productivity (Castillo et al., 2010). In the study of DOM composition and bioavailability during the sea ice formation (Jørgensen et al., 2015) the similar component with excitation and emission maxima of 273 nm and 327 nm, respectively, was reported to represent a combined signal of tryptophan and tyrosine amino acids referred to the T and B peaks (Coble, 2007; Stedmon and Nelson, 2015). The PARAFAC components with spectral characteristics similar to C3 were found in various studies. For example, C3 absorption and emission maxima ( Table 2 ) are close to C3 (λ_ex = 270 (360) nm, λ_em = 478 nm) in the study of DOM in the catchment and estuary of Horsens Fjord (Denmark) (Stedmon et al., 2003), and C2 (λ_ex = 240 (370) nm, λ_em = 480 nm) in the PARAFAC model of EEMs of Atlantic Ocean FDOM (Kowalczuk et al., 2013). Although the fluorescence maximum is blue-shifted by ~30 nm compared to conventional peak C (Coble, 1996), in the above studies the components were identified as peak C and assigned to humic material exported from terrestrial sources (Stedmon et al., 2003; Kowalczuk et al., 2013). Peak C was also reported to represent a product of microbial transformation of phytoplankton exudates (Romera-Castillo et al., 2011). The C4 component resembles fluorescence typical for marine humic substances peak M (Coble, 2007). In the experiments, provided Romera-Castillo et al., peak M fluorophore consumed by bacteria had been produced by phytoplankton (Castillo et al., 2010). A component BERC6 very similar to C4 was identified by a PARAFAC modeling of Canadian Arctic surface water FDOM fluorescence (Walker et al., 2009). Low fluorescence of this component was interpreted as minimal microbial productivity it the studied water area.

The relationship between the PARAFAC component values and various metadata of our samples was explored with the RDA analysis. Since the full set of parameters are available for 78 samples, RDA analysis was performed for them only. Using a permutation test, we validated 2 canonical axes as significant at α=0.01 level ( Figure 7 ). The angle between the variable arrows indicates the correlation between them, while the length of the variable arrows reflects the degree of its influence on the fluorescent components. Based on the obtained results, fewer parameters have been chosen to study their correlation with PARAFAC component fluorescence intensity in more detail, namely, salinity, DOC, chlorophyll-a, a _CDOM(350), S, HIX, BIX, SUVA, and FIMcK. Pearson’s correlation coefficients were calculated individually for the Khatanga, Lena, Indigirka, and Kolyma transects ( Table 3 ) with the Origin7.0 software (Edwards, 2002).

Fluorescent component distribution along the salinity gradient for salinity ranges of 3.5 – 33.8 in the Laptev Sea and 15.2 – 32.7 in the East Siberian Sea is shown in Figure 8 . The C1 appeared to be linearly anti-correlated with salinity for all the transect with correlation coefficient varying between -0.99 and -0.92. Conservative mixing of C1 confirms our assumption about its supply by river runoff. Intercept values of the equations describing linear correlations are very close for the Khatanga and Lena, as well as for the Indigirka and Kolyma transects. At the same time, they differ for the Laptev (average value is 8.5×10^-2) and East Siberian (6.4×10^-2) seas (see Table 4 ). Such a difference in optical patterns between the rivers can be related to their individual watershed characteristics. So, we assume that the Khatanga and Lena rivers are characterized by higher content of terrigenous humic substances with a fluorescence maximum at 430 nm. This is consistent with the study of variation of FDOM of large Arctic rivers, in which lower a _CDOM(350) values in the Kolyma River, compared to the Lena River, were explained by lower vascular plant inputs during freshet and its more extensive microbial degradation in the Kolyma watershed (Walker et al., 2013).

We also suggest that C1 component makes a dominant contribution to the CDOM absorption, since a strong correlation was revealed also between C1 and CDOM absorption coefficients at wavelengths from 300 nm to 400 nm, in particular, with a _CDOM(350), often used as a quantitative measure for lignin concentration and input of terrestrial DOM to the Arctic Ocean (Spencer et al., 2009; Stedmon et al., 2011; Mann et al., 2016) ( Figure 9 ). Component C1 correlates with DOC concentration in the river plume influenced areas in the Laptev Sea ( Figure 10A ), where conservative DOC behavior along the salinity gradient was demonstrated recently (Drozdova et al., 2021). Chl-a correlates with C1 to a lesser extent (0.62 < r < 0.82), showing higher values at the southern stations of the transects most affected by river runoff ( Figure 10B ). A positive linear correlation between C1 and humification index HIX – an indicator of humic substance content or extent of humification – was demonstrated for the Khatanga, Lena and Kolyma transects, while BIX - an indicators of autochthonous DOM input (Stedmon and Nelson, 2015) – was lower in the river-influenced areas ( Figures 10C, D ). A weak positive correlation was observed also between C3 and HIX and negative – between C3 and BIX ( Figure 11 ). The HIX is based on the idea that the emission spectra of fluorescing molecules shift toward longer wavelengths as humification of DOM proceeds. Higher values indicate an increasing degree of humification. So, the increase of the content of C3 with fluorescence maxima at 478 nm, necessarily leads to an increase of HIX ( Table 1 ).

The C2 and C3 component fluorescence intensities were generally higher at low salinities (and high CDOM absorption a _CDOM(350)). A moderate negative correlation (-0.85 < r < -0.53) was observed for the Khatanga, Lena, and Indigirka transects ( Figure 8 and Table 4 ). Higher concentration of C2 and C3 in the river-influenced regions is in a good agreement with reports of their allochthonous source, see, for example, the studies (Yamashita and Jaffé, 2008; Sun et al., 2014; Lee and Kim, 2018). Non-conservative behavior of protein substances (C2) in many estuaries is explained by relatively rapid degradation labile DOM (Yamashita and Tanoue, 2003; Vignudelli et al., 2004; Yamashita and Tanoue, 2008). Another possible reason for non-conservative distribution of C2 and C3 FDOM components is their production in marine environment. Thus, the study of Amerasian Basin seawaters revealed high abundance of protein-like fluorescence suggesting potential influence of fall phytoplankton blooms in the Chukchi and East Siberian seas in August – September (Chen et al., 2017) and its production from sediments (Chen et al., 2016). Taking into account weak correlations of C2 and C3 with salinity caused mostly by the local maxima of C2 and C3 fluorescence in the continental slope regions and in bottom waters ( Figure 6 ), we suggest the mixed autochthonous-allochthonous origin of C2 and C3 components with the predominance of allochthonous (mostly riverine runoff) sources in the river plumes - waters with salinity below 27 (Kang et al., 2013). The local maxima of the C2-C4 fluorescence intensity outside the plumes, in turn, are associated with autochthonous FDOM production, in particular with the fall phytoplankton blooms and, possibly, with a processes of organic matter reworking in the bottom water layer. Indeed, the areas of the continental slope of the Arctic shelf are known as potential depots for the accumulation of large masses of organic matter, and the mineralization on the bottom surface was estimated as about 89% of the total input (Romankevich et al., 2009).

At present, it seems to be complicated to relate unambiguously the spectral composition of FDOM to specific biogeochemical processes and DOM sources. Nevertheless, we assume that the increase of C2 and C3 content at the continental slope region of the Khatanga transect (station 5634) is most likely related to the recent near-ice phytoplankton bloom. It is supported by a deep maximum of chl-a (about 1 µg l^-1 at the station 5635 and 0.6 µg l^-1 at the station 5634). The bottom waters of the northmost station of the Kolyma transect (5612), exhibited high C2 – C4 fluorescence intensity ( Figure 6 ). The study of Sukhanova et al. (2021) revealed the maximum abundance and biomass of algae of the genus Chaetoceros with high content of spores and dead cells (up to 40%) in the lower layers of the water column. This region was also characterized by the maximal nitrate concentration along the transect of 4.1 µM (Sukhanova et al., 2021), high concentration of detrital and mineral particles, (Kopylov et al., 2021), as well as the maximal abundance and biomass of heterotrophic nanoflagellates (Kopylov et al., 2021). The complex composition of DOM in this area may be caused by several factors, for example, destruction of dead algae cells, production of heterotrophic nanoflagellates and bacterioplankton, and DOM released as a result of viral lysis of bacteria (Middelboe and Jørgensen, 2006; Jürgens and Massana, 2008).

Suspended matter at the station 5606 of the Indigirka transect was analyzed as a part of the study of the role of plankton in the vertical suspended matter (SM) flux in the East Siberian Sea (Drits et al., 2019). The dramatic increase of the total vertical SM flux at 35 m depth at the station 5606 was suggested to be due to re-suspension in the bottom nepheloid layer. At the same time, the analysis of physical and chemical data (CTD data, nutrients, oxygen, dissolved inorganic carbon, fCO₂, CH₄) measured in September 2008 in the East Siberian Sea revealed microbial decay of organic matter occurring through much of the water column with its predominance at the sediment interface (Anderson et al., 2011). We assume that at the stations 5605 and 5606 of the Indigirka transect high concentrations of FDOM in mid-waters and bottom water layer can be related to the erosion of the nepheloid layer containing organic matter decay products. It should be taken into account, that the increase in the concentration of SM in bottom waters does not necessarily indicate the resuspension of nepheloid layer. Ungrazed phytoplankton blooms, combined with aggregation and coagulation can lead to rapid sinking of the newly produced organic material. Mesozooplankton can also compact slowly sinking smaller phytoplankton cells into fecal pellets (Wassmann et al., 2004).

The study of FDOM of the Arctic shelf waters showed the difference in the distribution of PARAFAC components in the East Siberian and Laptev seas. In the Laptev Sea, components C1 and C3 dominate, with the maximal fluorescence intensities observed mostly in the Lena and Khatanga river-influenced shelf areas. Protein-like (C2) and marine humic-like (C4) components are poorly represented. In the East Siberian Sea, a similar distribution was observed only for the terrestrial humic-like component C1, while C2 and C4 components are more abundant and distributed irregularly. Component C3, usually assigned as terrestrial-derived FDOM, exhibited local maxima throughout the water column which pointed on autochthonous production of FDOM with emission maximum at 478 nm. The obtained results make the optical studies of FDOM a powerful tool for identifying the most interesting areas in terms of understanding the carbon cycling in the Arctic shelf seas. It is necessary to take into account the high variability of FDOM, and when studying processes in the bottom water layer, much attention should be paid to natural barriers, where organic matter can be accumulated.

FDOM of Siberian shelf seas is described by the set of four PARAFAC components which correspond to conventional fluorophores A, C, T/B and M (Coble, 1996; Coble et al., 1998; Coble, 2007) identified in natural waters. The composition of FDOM was found to be typical for the Siberian shelf seas (Guéguen et al., 2012; Mann et al., 2012; Chen et al., 2017; Chen et al., 2021), but this study showed the difference in FDOM distribution in the Laptev Sea (Khatanga and Lena transects) and in the East Siberian Sea (Indigirka and Kolyma transects). The consideration of fluorescence intensities of individual PARAFAC components along the shelf-crossing transects made it possible to draw the following conclusions:

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AD, IK, and TL contributed to conception and design of the study. TL performed spectral measurements. IK performed PARAFAC decomposition. Analysis of CTD and chlorophyll-a data was performed by AN and EA. AD, and IK wrote the first draft of the manuscript. All the authors revised the draft and approved the submitted version of the manuscript.

This research was funded by the Russian Science Foundation, project 21-77-10064.

We are grateful to Nikolay V. Lobus for water sampling, Mikhail V. Flint for the opportunity to take part in the impressive complex expeditions to the Arctic shelf seas and the crew of the R/V Akademik Mstislav Keldysh for their contribution during field studies. We would like to thank the reviewers for their constructive comments and the editor for his review of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.