The material was collected during cruise # 69 of the RV “Akademik Mstislav Keldysh” to the Laptev Sea from 31 August to 20 September 2017. In the western part of the Laptev Sea, a transect of 11 stations from the inner Khatanga Gulf northward to the continental slope was performed ( Figure 1 ). The northernmost station 5635 was located at the ice edge. Two of the stations (5591_2 and 5590_2) were sampled at the same locations where the pilot stations 5591 and 5590 have been sampled two weeks earlier to follow a short-term variability in zooplankton composition and abundance. To estimate if there were diel changes in the vertical distribution of zooplankton, day and night sampling was done at the pilot station 5591.

In the eastern part of the Laptev Sea, three stations were carried out in the shelf adjacent to the Lena Delta on 2-3 September. Two of the stations (5596 and 5592) were repeated on 15-16 September, and one more station (5623), furthest from the Lena delta, was sampled on 15 September ( Figure 1 ).

Simultaneously, data on temperature, salinity, and chlorophyll fluorescence were obtained from vertical CTD sounding with the SeaBird 19+ probe at all the stations.

Mesozooplankton was sampled using a Juday closing net (0.1 m² mouth area, 180 μm mesh size) towed vertically from 2-3 m above the bottom to the surface at all shallow stations. At deeper shelf stations, the layers below and above the pycnocline, which was determined according to the CTD profiles, were usually sampled. Five separate layers were sampled with a MultiNet multiple plankton sampler (mouth area 0.125 m², mesh size 180 μm) at the deepest station 5635 ( Table 1 ). The volume of filtered water was calculated by multiplying mouth area value by the sampled depth layer. For determination of zooplankton abundance and biomass, samples were preserved in 4% borax-buffered formalin. Zooplankton were identified, staged, measured and counted in the laboratory under a stereomicroscope under x40 magnification. Not numerous large specimens (the older stages of Calanus spp., Oikopleura vanhoeffeni, Limacina helicina, Parasagitta elegans) were counted in the whole samples, while more numerous forms were counted in subsamples so as not less than 50 specimens were recorded. The copepodites of the closely related copepod species Calanus finmarchicus and C. glacialis were distinguished according to morphology and prosome lengths (Kwasniewski et al., 2003). For Calanus spp., all copepodite stages were distinguished, for Pseudocalanus spp. and Drepanopus bungei, copepodite stages CI to CIV were pooled. The wet weight (WW) of each species was calculated using nomograms by Chislenko (1968). These tables (1968; www.twirpx.com/file/1588162/) allow the calculation of biovolume (wet weight, WW) of aquatic organisms on the basis of the body shape and length. Dry weight (DW) of crustacean plankton was estimated as 0.16 WW (Vinogradov and Shushkina, 1987), DW of chaetognaths was calculated according to Matthews and Hestad (1977), larvaceans – to Hopcroft et al. (1998). To compare seasonal development of C.glacialis population at different stations, a mean developmental stage index (DSI) was calculated as abundance weighted average:

where i is copepodite stage number from 1 to 6, n_i is abundance of stage i, and summations are for i from 1 to 6 (Skjoldal et al., 2021).

Feeding rates of dominant herbivorous zooplankton (C. glacialis, C. finmarchicus, C. hyperboreus, Limnocalanus macrurus, Senecella siberica, Pseudocalanus spp., Drepanopus bungei, Jaschnovia tolli, Oithona similis, Oikopleura vanhoeffeni, Limacina helicina) were assessed with the gut fluorescence method (Mackas and Bohrer, 1976). This approach was subject to discussion because of the possibility of pigment destruction during the passage through the gut (e.g., Båmstedt et al., 2000). There is no general agreement as to extent of pigment degradation, with values ranging from 0 to 95% (Pasternak, 1994). On the other hand, when different methods to estimate feeding were directly compared, reliable results were obtained with the gut fluorescence (Baars and Franz, 1984; Tiselius, 1988; Peterson et al., 1990; see also Pasternak, 1994). This suggests that high degradation rates are not the rule. Therefore, the gut fluorescence method continues to be a highly valuable tool for studies of zooplankton feeding (e.g., Saiz and Calbet, 2011; Valdés et al., 2017; D’souza and Gauns, 2018), and is still the only method that provides information on the in situ feeding rates and feeding impact of herbivorous zooplankton on phytoplankton assemblage.

Zooplankton for the gut pigment analysis was sampled from the layers above and below the pycnocline similarly to collection for the identification. The content of the cod-end was diluted in a 1L plastic bucket and the zooplankters were immediately anaesthetized with carbonated seawater. Undamaged animals were then sorted under a dissecting microscope and 1 to 50 animals per replicate, depending on size/stage, were picked with forceps and placed in test tubes with 3 ml of 90% acetone. Two to five replicates, if possible, for each species/stage were analyzed. Pigments were extracted for 24 h at 5 °C in the dark. Chl a and phaeopigments were measured by a standard fluorometric procedure (Holm-Hansen et al., 1965) with a Trilogy fluorometer (Turner Designs) before and after acidification with 1 N HCl. Gut content of the animals (G) in units of Chl a equiv. ind^{- 1} was calculated as G = (Chl a +1.51 x Phaeopigment) (Dagg and Wyman, 1983). The amount of pigment ingested daily (I, ng Chl a ind^-1 day^-1) was estimated as: I =G*24/GPT, where GPT is the gut passage time (hours). We used the gut passage values obtained in our previous studies and the published data ( Table 2 ). We could not find published data on the gut passage time in Clausocalanidae D. bungei, so the value of 0.9 h obtained for the related Clausocalanus laticeps (Atkinson et al., 1996) was used. All the data on GPT were adjusted to the temperature of the habitat where the animals were collected with the Q₁₀ = 2.2 (Irigoien, 1998). To assess ingestion rates of zooplankton in carbon units, we calculated the phytoplankton carbon (C, mg m^-3) to Chl a (mg m^-3) ratio of 15 ± 7 (n=11) in the Khatange region and 18 ± 8 (n=6) in the Lena region. Phytoplankton biomass and Chl a concentration were assessed from the same samples. The data on phytoplankton biomass in carbon units calculated according to Menden-Deuer and Lessard (2000) were taken from Sukhanova et al. (2019) for the Khatanga and I.N. Sukhanova (personal communication) for the Lena region. Chl a was assessed fluorometrically by Demidov et al. (2019).

Grazing impact of each of the dominant species on the integrated (0-bottom) Chl a content (E_Chl , mg Chl a m^-2) was estimated using the individual ingestion rates (I_i , ng Chl a ind^-1 d^-1) and the abundance of the given species (N_i , ind m^-2):

where i is the layer number, n is the total number of layers. Grazing impact on primary production was assessed with the use of the obtained C: Chl a ratio. We used the data on primary production measured in this cruise using on-deck ¹⁴C incubations (Demidov et al., 2020 for the Khatanga region and A.B. Demidov, personal communication, for the Lena region).

Detailed description of the hydrography of the studied areas is given in (Flint et al., 2018; Osadchiev, 2020). Here, we present the features of physical environment which are most important for zooplankton.

Khatanga Transect. Hydrography of the studied area was strongly influenced by the river run-off. Variations of surface salinity and temperature along the transect are shown in Figures 2A, B . Salinity and temperature of the upper mixed layer along the Khatanga transect varied from 3.4 and +3.6 ^оС at st. 5628 (Khatanga Gulf) to 32.2 and -1.3 °C at st. 5635 (slope area). According to the changes in salinity, three zones could be distinguished: inner Khatanga Gulf (st. 5627-5629, surface salinity from 3.4 to 11), shelf zone influenced by the Khatanga plume (st. 5630-5591_2, surface salinity from 17 to 21), shelf-and-slope zone outside the plume (st. 5633-5635, salinity > 27). Chlorophyll a concentration varied from 1-1.9 mg m^-3 (inner Gulf) to 0.1-0.9 mg m^-3 (plume zone) and 0.2-0.4 (shelf-and-slope zone outside the plume) ( Table 1 ). At the shallow stations in the Khatanga Gulf, there was no pronounced vertical stratification. It was recorded at all stations outside the Gulf ( Figures 2A, B ). Salinity in the upper mixed layer at stations 5590_2 and 5591_2 was similar to that measured at these locations at the corresponding pilot stations two weeks earlier (32.1 and 31.5, 21.5 and 22.3, respectively).

Lena Transect. Two zones could be distinguished along the Lena transect, shelf zone influenced by the Lena plume and shelf zone outside the plume. At the first sampling, two of the stations (5597 and 5596) were located inside the Lena plume where salinity and temperature in the upper layer varied from 13.4 and 5.3 °C at the closest to the Lena Delta st. 5597 to 21.8 and 3.9 °C at st. 5596 ( Figures 2C, D ). Station 5592 with temperature of 2.1 °C and salinity of 25.7 was outside the plume. Two weeks later, salinity of the upper layer at st. 5596 decreased from 21.8 to 6.6 psu, while temperature increased from 3.9 to 4.6 °C. At another repeated station (st. 5592), changes in hydrographical parameters were less pronounced: salinity decreased from 25.8 to 19, while temperature remained almost unchanged (about 2.1 °C, Figures 2E, F ). Chlorophyll a concentration increased by almost an order of magnitude at st. 5596 and did not change at st. 5592 ( Table 1 ).

Two peaks of total zooplankton biomass (DW, mg m^-3) were recorded along the Khatanga transect: the first, in the inner Khatanga Gulf (st. 5627-5629), and second, at the shelf edge (st. 5590_2). Biomass values were more than 50 mg DW m^-3 in both areas. In the plume zone, biomass was much lower, constituting, on average, 17 mg DW m^-3 ( Figure 3A ). Minimum of 2.2 mg DW m^-3 was recorded at the most oceanic station 5635.

Species composition of zooplankton along the Khatanga transect changed with the distance from the river mouth. Limnocalanus macrurus which was responsible for the first maximum of zooplankton biomass dominated inside the Khatanga Gulf. It was followed by Drepanopus bungei. At the two stations closest to the river, Senecella siberica played a significant role ( Figure 3 ). When moving seaward, Calanus spp. accounted for a larger share of zooplankton biomass (15-60%). In this area, copepods of the genus Pseudocalanus and chaetognaths Parasagitta elegans contributed substantially to zooplankton biomass. Metridia longa made up 27% of zooplankton biomass at the slope stations. Comparison of zooplankton composition and biomass at the repeated stations revealed that composition of the dominant species did not change in two weeks, while biomass increased by a factor of 1.8 to 2.6. Increase of biomass was mostly due to C. glacialis at st. 5590_2 and C. glacialis together with P. elegans at st. 5591_2 ( Figure 3 ).

Zooplankton biomass and composition at the three stations in the Lena area performed on 2-3 September were pretty similar. Calanus glacialis, Pseudocalanus spp. and P. elegans played the most pronounced roles at these stations ( Figure 3 ). Beside them, O. vanhoeffeni were important at the nearest to the delta st. 5597. When sampling was repeated two weeks later, both composition and biomass of zooplankton demonstrated considerable changes. The most striking changes considered st. 5596_2: total biomass there turned out to be three times that at st. 5596. High abundance of L. macrurus at st. 5596_2, which was not found at all at the first sampling, was mostly responsible for this difference. Drepanopus bungei also contributed to biomass increase at this station, although to a lesser extent than L. macrurus. The latter was also found at another repeated station (5592_2), but its biomass was much less than at st. 5596_2.

Vertical distribution of zooplankton biomass at the pilot stations 5591 and 5590 was similar irrespective of the time of the day ( Figure 4 ). This suggests that zooplankton did not ascend into the upper layer at night, at least, at the shelf shallow water stations. Relatively high biomass of C. glacialis in the upper layer at these stations at all times of the day is associated with high abundance of the young development stages (CI-CIII) of this copepod.

Vertical distribution of zooplankton biomass on the shallow shelf depended on the vertical distribution of salinity and temperature. In both of the areas, the bulk of the biomass was concentrated in the layer below the pycnocline regardless of Chl a vertical distribution and time of the day. In the relatively warm freshened upper mixed layer, biomass of zooplankton was lower ( Figures 5 , 6 ). The biomass maximum was determined mostly by the vertical distribution of Calanus glacialis, Pseudocalanus spp. and Parasagitta elegans. The only exception was st. 5596_2 in the Lena influenced area where biomass in the upper layer was twice that in the lower layer. Limnocalanus macrurus, which strongly dominated zooplankton biomass at this station, occupied the upper 10-0 m layer inhabited also by the second important species, D. bungei ( Figure 6 ).

In the shelf edge and slope region, the biomass in the upper mixed layer was low ( Figure 5 ). The higher values were recorded in the 40-20 m layer below pycnocline. Calanus glacialis, C. hyperboreus and M. longa were responsible for this biomass increase. An increase of zooplankton biomass related to these species was recorded in the near-bottom layer at these stations ( Figure 5 ).

The demographic structure of the population of C. glacialis changed from the south to the north, with similar trends in the both studied regions. At the southernmost stations where C. glacialis was first found (the species did not inhabit strongly freshened regions), the population consisted of only older stages (mainly, CV) ( Figures 7A–C ). To the north, the share of younger stages gradually increased. The youngest population was observed at the ice edge (st. 5635) where CI and CII comprised more than a half of the population abundance and the role of CV decreased to less than 5% ( Figure 7A ). Note that at this station a high abundance of Calanus spp. nauplii was recorded, up to 600 ind m^-3 in the 0-20 m layer. Distinct demographic changes were also recognized when comparing the population structure at the repeated stations: in two weeks, the share of the younger stages decreased, and the older stages increased. Thus, at st. 5591 the share of CI-CIII decreased from 11 to 3%, at st. 5590 – from 62 to 18%, at st. 5596 – from 37 to 0%, at st. 5592 – from 46 to 10% ( Figures 7B–D ). To follow the ageing of the population, we calculated the mean developmental stage index (DSI), which varied from 2.2 at the ice edge to 4.5-5.2 at the southernmost stations ( Table 3 ). At the repeated stations 5590, 5592, 5596, DSI in two weeks increased from 3.1-3.8 to 4.4-5.0 and almost did not change (4.7 to 4.9) at st. 5591. At the latter station, abundance of C. glacialis increased by a factor of 2 (from 710 to 1500 ind m^-2), while no significant changes in abundance were found at the other three repeated stations (st. 5590: 9500 and 9700 ind m^-2, st. 5592: 1000 and 1200 ind m^-2, st. 5596: 670 and 440 ind m^-2, Table S2 ) which suggests that the observed demographic changes were related rather to the ageing of the population than to advection. We noted clear positive effect of temperature on the DSI (DSI = 0.344 t + 0.346, R² = 0.38; n = 14; p = 0.017).

Overall, gut pigment content of dominant zooplankters in both regions was not high, usually, not exceeding 3 ng Chl ind^-1 ( Table S1 ). Higher gut pigment content (> 10 ng Chl ind^-1) was recorded in large copepods C. hyperboreus and in large (> 2 mm) larvaceans O. vanhoeffeni. The maximum values (> 1000 ng Chl ind^-1) were measured in large (4-6 mm diameter) specimens of pteropods L. helicina.

Specific daily rations of small-sized copepods (including the young stages of C. glacialis) through herbivory varied from less than 1% to 4% ( Table 4 ). Specific rations of large copepods made up around 1%. Contrary to copepods, both large pteropods and larvaceans demonstrated high values of specific rations: from 24% to 45% and 11% to 20%, respectively ( Table 4 ).

Given the low feeding activity demonstrated by the main groups of zooplankton grazers and their not very high biomass, it is no surprise that grazing pressure was not high. It was less than 1% of chlorophyll in the inner Khatanga Gulf and at the ice edge, while consistuted about 3% in the plume-affected shelf area. At the pilot st. 5590, grazing impact reached maximum of 8%. In the Lena area, grazing imact was 1.5-2% with one exception of 6% at st. 5623 close to the Lena plume periphery ( Figure 8 ). Grazing impact at the repeated stations decreased by a factor of 2. Grazing pressure on primary production varied from 1.4 to 13% at most of the stations with the maximum of 29% at st. 5590_2 (outer shelf off Khatanga) and 72% at st. 5623. The grazing pressure within the distinguished zones was determined by different consumers. Limnocalanus macrurus and D. bungei were the main grazers in the inner Khatanga Gulf and at st 5596_2 in the Lena plume area. Larvaceans O. vanhoeffeni contributed 99% to total grazing at the southernmost station in the Lena plume. At all of other stations, C. glacialis and Pseudocalanus spp. were the main grazers.

Several previous studies estimated zooplankton biomass and composition in the Lena influenced area (Jashnov, 1940; Kosobokova et al., 1998; Abramova, 1999; Abramova and Sokolova, 1999; Lischka et al., 2001; Abramova and Tuschling, 2005; Arashkevich et al., 2018) but the data in the Khatanga plume area are scarce (Jashnov, 1940; Kosobokova et al., 1998) and no data, to our knowledge, were so far obtained for the Khatanga Gulf. Our values for zooplankton biomass in the Khatanga plume area (11-28 mg DW m^-3) are close to the values (18-36 mg DW m^-3) reported by Kosobokova et al. (1998) for the same season. Lower biomass values of 0.3-6 mg DW m^-3 reported by Lischka et al. (2001) were probably connected with the later sampling in this study (October). Our estimates of the biomass in the Lena plume area (12-42 mg DW m^-3) are similar to those reported for this region by Lischka et al. (2001) in August-September and Arashkevich et al. (2018) in September: 15-60 and 10-30 mg DW m^-3, respectively. Similar values of around 23 mg DW m^-3 were recalculated from the data by (Sorokin and Sorokin, 1996). Lower values (2-13 mg DW m^-3) were obtained for the same season by Kosobokova et al. (1998) which was explained by the scarcity of sampling in the shallow water region (Lischka et al., 2001). In the outer shelf/slope zone outside the Khatanga plume, our estimates (2-52 mg DW m^-3) are close to the values of 4-25 mg DW m^-3 presented by Kosobokova et al. (1998). Thus, the comparison of the data obtained in 1991, 1993, 1995 and in 2015, 2017 did not reveal considerable trends in changes of the biomass of zooplankton in the studied regions of the Laptev Sea, despite the fast warming of the Arctic and ice cover decrease (e.g., Perovich et al., 2019; Overland et al., 2019; Timofeeva and Sharatunova, 2021).

In general, our results on the species composition and zooplankton assemblages in the distinguished zones are in accordance with the previous studies in the Siberian Arctic seas (e.g., Vinogradov et al., 1995; Lischka et al., 2001; Fetzer et al., 2002; Abramova and Tuschling, 2005; Hirche et al., 2006; Arashkevich et al., 2010; Drits et al., 2020). Brackish species Limnocalanus macrurus, Senecella siberica and Drepanopus bungei made up the bulk of biomass in the most freshened areas closer to the river mouth. Their role decreased in the plume-affected areas where Pseudocalanus spp. and Calanus glacialis contributed significantly to total biomass. In the outer shelf/slope outside the plume, C. glacialis was by far the most important species, followed by M. longa and P. elegans.

The freshwater discharge into the Laptev Sea forms a salinity gradient over the shelf and, to a large extent, determines the structure and functioning of zooplankton community (Jashnov, 1940; Kosobokova et al., 1998; Lischka et al., 2001; Abramova and Tuschling, 2005; Arashkevich et al., 2018, and others). Most of the data were obtained close to the Lena Delta and in the shelf plume-affected area. Our results make it possible to track if the influence of the Khatanga discharge on zooplankton community differs from that of the Lena. Despite large spatial extension of the Khatanga plume, the impact of Khatanga discharge on zooplankton species composition was restricted mainly to the inner Gulf where the brackish species (S. siberica, L. macrurus, D. bungei) shaped the community. These species were almost absent in the shelf plume area. The specific feature of the Khatanga estuary is a strong tidal forcing (Osadchiev et al., 2020) which could constitute the base of the mechanism of zooplankton retention (Gagnon and Lacroix, 1983; Morgan et al., 1997; Menéndez et al., 2012) within the Gulf. As an indirect support of this suggestion, results on zooplankton distribution in the not-so-far located Yenisei gulf in the Kara Sea which is similar in size, geomorphology, and climatic conditions to the Khatanga Gulf are considered. Contrary to the latter, the Yenisei Gulf is less affected by tidal mixing due to low tidal velocities (Osadchiev et al., 2020), and the distribution of the brackish zooplankton is not limited to the gulf. Instead, they inhabit the extended area of the shelf affected by the river plume (Vinogradov et al., 1995; Fetzer et al., 2002; Hirche et al., 2006). Similarly, the brackish species are distributed far to the north within the Lena plume area (Lischka et al., 2001; Abramova and Tuschling, 2005; Arashkevich et al., 2018). Our results showed highly variable pattern of spatial distribution of the brackish zooplankton assemblage: at the first sampling (1-2 September), when northern boundary of the plume was located at 75° N, the distribution resembled that in the Khatanga plume area. Neither L. macrurus nor D. bungei were found even at the closest to the Lena Delta station 5597, instead, C. glacialis, Pseudocalanus spp. and O. vanhoeffeni formed the bulk of the biomass. Two weeks later, when the Lena plume extended northward up to 77° N, L. macrurus constituted more than 50% of zooplankton biomass at 100 km distance from the Lena Delta and was found as far as 75.5° N. The wide distribution of brackish zooplankton in the plume area indicates that there are no mechanisms of zooplankton retention in the vicinity of the Lean Delta. Thus, contrary to the Khatanga input, the Lena freshwater discharge governs the structure of zooplankton community over a vast shelf area. Position and area occupied by the Lena plume strongly depend on the local wind forcing conditions (Osadchiev et al., 2021). In particular, meridional extension of the plume at the repeated sampling was related to the prevailing eastern and south-eastern winds in this area at the time of the study (Flint et al., 2018).

Distribution of total zooplankton biomass along the Khatanga transect is similar to that reported in the Lena area in autumn (Arashkevich et al., 2018). In both cases, biomass over the shelf influenced by the river plume was twice lower than in the vicinity of the freshwater inflow. Increased biomass of zooplankton close to the Lena Delta was also noted by Lischka et al. (2001) who explained this by better feeding conditions, increased water temperature and local hydrography. In our study, chlorophyll concentration and temperature were higher in the inner Khatanga Gulf which is in agreement with this explanation. In the Khatanga plume zone, besides lower chlorophyll concentration, its maximum was in the upper layer above pycnocline, the layer avoided by the marine species. This further worsened feeding conditions of herbivores in the plume-affected shelf area.

The vertical distribution pattern of zooplankton biomass in most of the cases resembled the pattern found in the Laptev Sea by Jaschnov (1940) and Kosobokova et al. (1998). They reported that biomass was confined mainly to the layer below pycnocline while the surface layer in the Lena and Khatanga plumes was poor. However, our results show that under increased spreading of freshened upper layer in the Lena plume, the vertical distribution pattern changed: maximum of zooplankton biomass made up by the brackish species, L. macrurus and D. bungei, was found in the layer above pycnocline.

Timing of ice retreat is among the main factors controlling seasonal development of the dominant copepod populations in the Arctic seas (Kosobokova and Hirche, 2001; Leu et al., 2011; Drits et al., 2020; Rogachev et al., 2021). According to Kosobokova and Hirche (2001), spawning of C. glacialis in the Laptev Sea occurred at the ice edge, and ageing of the population followed ice retreat. These authors noted that reproduction and development of C. glacialis differed between the eastern and western parts of the Laptev Sea in September 1993. In the east, where a large area extending northward from the mouth of Lena was open by the beginning of August 1993, the young stages CI-CIII constituted up to 40-60% of the population. In the western sea, almost completely covered with ice in early September, the older stages CIV- CVI dominated (Kosobokova and Hirche, 2001). The authors assumed that under the ice conditions in the western sea, reproduction and recruitment of C. glacialis has a sporadic character. Delays in C. glacialis reproduction and population developments due to inter-annual/geographical differences in sea ice conditions were recorded both in the Canadian and Eurasian Arctic (Kwasniewski et al., 2003; Søreide et al., 2008; Leu et al., 2011).

In our study, both eastern and western areas of the Laptev Sea were open from the second week of August in 2017 (http://www.aari.ru/). Demographic structure of C. glacialis did not reveal pronounced differences between the Khatanga and Lena areas at the beginning of September. At the southernmost stations, the population consisted totally of the overwintering stages (CIV-CVI). Northward, the contribution of younger stages gradually increased. Our results on demographic structure of C. glacialis along the Khatanga transect enabled us to follow population advancement in relation to timing of ice retreat. At the southernmost stations, the ice breakup occurred at the beginning of August, while at the northernmost ice-edge station, about 30-35 days later (http://www.aari.ru/). The development stage index of the population at these sites was 5.2 and 2.2, respectively. The positive correlation between DSI and temperature of the upper mixed layer agrees with delayed population development under late ice retreat and low temperatures. The question arises whether the population which we observed in the vicinity of the ice at the second half of September would be able to reach an overwintering stage (CV or CIV) before the formation of solid ice. The calculated duration of a copepodite stage (assuming equal development time of each stage) was around 12 days at the repeated stations. Our estimate is in accordance with other studies stating that C. glacialis needs approximately 2 months to develop from CI to their overwintering stages CIV and CV at 3° C (McLaren and Corkett, 1981; Corkett et al., 1986). This means that development from CII to CIV (the overwintering stage) will take about 24 days at surface temperature about 2-4 °C. Since the ontogenetic development of Calanus spp. is strongly temperature dependent (McLaren and Corkett, 1981; Skjoldal et al., 2021), the development of C. glacialis to the overwintering stage in the marginal ice zone (with the surface temperature below -1°C) is likely too slow to succed before the solid ice cover was formed by the midle of October. Suboptimal conditions for C. glacialis in this zone suggest that only a minor portion of the population survives through the winter season.

The herbivorous feeding rates of the dominant copepods did not reveal any influence of the river discharge. Specific daily rations on the algal food in the overwhelming majority of cases were less than 2% and did not exceed 4% body carbon. Assuming assimilation of phytoplankton of 0.6 (Mauchline, 1998), energy input through herbivory did not cover the metabolic requirements of these copepods (2-10% body carbon) calculated according to Ikeda (2001) with regard to water temperature and body weight. This pattern is similar to what was described by Arashkevich et al. (2018) in the Lena plume area in September 2015: the lack of pronounced river influence on feeding rates, and low daily rations. The estimated budgets showed that during the post-bloom period, ingestion through herbivory could not provide the necessary carbon to sustain the demands of growth and metabolism. Interestingly, that despite low herbivorous feeding, the population of C. glacialis grew and developed. This means that their energy input was supplemented by other sources, such as ciliates (e.g., Levinsen et al., 2000) and detritus (e.g., van der Jagt et al., 2020). For rough estimation of daily rations of copepods on all these sources (phytoplankton, ciliates and detritus), Arashkevich et al. (2018) calculated C. glacialis ingestion rates using suspended POC/Chl (besides C_phyt/Chl) ratio at the stations along the Lena transect. This resulted in an almost an order of magnitude higher daily energy input which considerably exceeded metabolic requirements. Thus, given the observed development of the C. glacialis population in our study, we suggest that the copepods successfully exploited additional food sources beside phytoplankton.

Contrary to copepods, the two other groups of dominant zooplankton, large size larvaceans and pteropods, demonstrated high rates of phytoplankton consumption ( Table 4 ). We assume that this is related to their specific feeding mechanism – unselected collection of particles from the large water volume with mucous net (L. helicina) or “house” (O. vanhoeffeni). Even under low phytoplankton concentrations in autumn, these groups not only met the metabolic demands calculated as 6-9% and 3-5% of body carbon (after Lombard et al., 2005; Pasternak et al., 2017), but seemed to provide fast growth through solely herbivorous feeding.

The present study is the first attempt to estimate grazing impact on autotrophic phytoplankton in the Khatanga Gulf and adjacent shelf. Influence of the river runoff is manifested in the lowest grazing pressure (mean impact equaled 0.4% of Chl a) in the inner Gulf despite high zooplankton biomass. This is the result of the two main factors: first, an increased phytoplankton biomass (mainly, the freshwater assemblage) carried by the river runoff (Sukhanova et al., 2019), and second, negligible feeding rates of the absolute zooplankton dominant, L. macrurus. Age composition (solely adults) and low feeding rates point to the final phase of seasonal development of the population found in other estuarine arctic regions in September (Hirche et al., 2003; Drits et al., 2016). As the influence of the river discharge weakened, phytoplankton biomass decreased and grazing in the shelf area constituted, on average, 4% of Chl a. The mean consumption of the newly produced organic carbon by zooplankton over the shelf was also higher than in the Gulf (15 and 6% of primary production, accordingly) but the difference was not significant (p=0.07).

Grazing impact of copepods depends on phytoplankton size structure (e.g., Finkel et al., 2010) as the majority of these grazers consume particles <5 μm less efficiently than larger items (Levinsen et al., 2000). Size structure of phytoplankton communities in the Siberian Arctic seas varies depending on the impact of the river discharge, with larger phytoplankton cells (> 3 μm) prevailing under low salinity and relatively high water temperature in the estuaries, and picophytoplankton (< 3 μm) prevailing under high salinity and low water temperature on the shelf (Belevich et al., 2021; Demidov et al., 2021). Specifically, in the Khatnga Gulf, the share of picophytoplankton was about 10% of total Chl a, while on the shelf, the contribution increased to 35% (Belevich et al., 2021). If we re-calculate the estimated grazing pressure considering Chl a content of only the larger phytoplankton, the values will increase to 7% in the shelf area while will remain almost unchanged in the Gulf.

Zooplankton grazing pressure, assessed with the gut fluorescence approach in the Lena area in September 2015 (Arashkevich et al., 2018), followed the same trend: the lowest values (2% of Chl a and 1% of primary production) were found at low salinity and high temperature at closest to the Lena Delta stations. The values gradually increased up to 5% of Chl a and 6% of primary production with increase of salinity and decrease of temperature with the distance from the Delta. Similarly, our estimates showed highest grazing pressure of 4% of Chl a and 72% of primary production at the Lena plume edge (st. 5623) under higher salinity. Within the plume-influenced area, grazing impact did not exceed 1% of Chl a and 10% of primary production. Note that contrary to the negligible role of grazing by L. macrurus in September 2015 (Arashkevich et al., 2018), the species determined the total zooplankton grazing at st. 5596_2, most influenced by an increased freshwater inflow.