Zooplankton samples were obtained at a fixed station in the Chukchi Sea (72∘45′ N, 168∘15′ W; depth 56 m) from 10 to 25 September 2013 (Fig. 1) (Nishino et al., 2015). Zooplankton samples were collected by vertical hauls with a NORPAC net (mouth diameter 45 cm, mesh size 335 µm; Motoda, 1957) and ring net (mouth diameter 80 cm, mesh size 335 µm) from a 49 m depth to the sea surface two to four times per day (total of 47 times), including both day and night. The large mesh size of the NORPAC net (335 µm) may have resulted in underestimation of the smaller zooplankton species and early stages of larger zooplankton species. Zooplankton samples collected with the NORPAC nets were preserved with 5 % buffered formalin immediately after being brought onboard. The ring net samples were used for copepod gut pigment measurements. For the evaluation of the diel vertical migration (DVM) of the copepods, day and night vertical stratified hauls were taken with closing PCP nets (mouth diameter 45 cm, mesh size 62 µm; Kawamura, 1989) from two layers (0–20 and 20–49 m) on 14 and 22 September. The samples from the PCP net were preserved with 5 % buffered formalin. During the sampling period, there was a horizontal advection of the water mass oscillation caused by tidal waves (Kawaguchi et al., 2015). To minimize the effect of the tidal oscillation, day and night sampling times were set at 12 h intervals (day: 10:55–11:34; night: 22:27–22:40). The four-time CTD casts at each sampling date confirmed that the hydrography was similar for the day–night sampling period (the CTD data are presented in the Supplement).

At the fixed station, the temperature, salinity and chlorophyll a (Chl  a) fluorescence were measured using the CTD (Sea-Bird Electronics Inc., SBE911Plus) casts at a frequency of two to four times per day. To evaluate the size-fractionated Chl a, water samples from the sea surface and the maximum fluorescence layer (16.8–27.7 m depth) were collected using a bucket and rosette multi-sampler mounted on the CTD, respectively. The water samples were filtered through 20, 10 and 2 µm pore-size membranes and GF/F filters, and Chl a was extracted with N,N-dimethylformamide and measured with a fluorometer (Turner Designs Inc., 10-AU-005).

For fresh samples collected using ring nets, 10 % v/v soda (saturated CO2 in water) was added to avoid copepod grazing, gut evacuation and the decomposition of gut pigments. Fresh specimens of C. glacialis copepodid stage 5 (C5) were sorted under a stereomicroscope. The sorting of all of the specimens was performed under low temperatures and dim light conditions within 1 h. Batches of 15 specimens were immersed in 6 mL of N,N-dimethylformamide and stored in dark, cold conditions overnight to extract the chlorophyll and phaeopigments. After the extraction of the pigment, the chlorophyll and phaeopigments were measured using a fluorometer (Turner Designs Inc., 10-AU-005). The chlorophyll and phaeopigments were summed and expressed as gut pigments (ng pigment ind.-1) (cf. Mackas and Bohrer, 1976).

The amount gut pigment of C. glacialis C5 was higher at night than during the day. Assuming that grazing primarily occurred at night, the grazing rate (GRind., mg pigment ind.-1 day-1) of C. glacialis C5 was calculated using the following equation: GRind.=GP×k×T/106, where GP is the individual gut pigment at night (ng pigment ind.-1), k is the gut evacuation rate (0.017 min-1; Tande and Båmstedt, 1985), and T is the length of the night (mean 13 h = 780 min during the study period). The grazing impact of C. glacialis C5 on micro-size (> 10 µm) Chl a (GI, % on Chl a biomass day-1) was calculated using the following equation: GI=GRind.×N/Int. Chl a×100, where N is the abundance of C. glacialis C5 (ind. m-2) and Int. Chl a is the biomass of large-sized (> 10 µm) Chl a (mg m-2).

In the laboratory, identification and enumeration by taxa were performed on zooplankton samples collected using NORPAC nets under a stereomicroscope. For the dominant taxa (calanoid copepods), identification was performed at the species and copepodid stage levels. For species identification of calanoid copepods, we referred mostly to Brodsky (1967) and Frost  (1974) for Calanus spp., Miller (1988) for Neocalanus spp. and Frost (1989) for Pseudocalanus spp. For Pseudocalanus spp., species identification was performed only for late copepodid stage C5 females/males (C5F/M) and C6F/M, and their early copepodid stages (C1–C4) were treated as Pseudocalanus spp.

For the evaluation of the DVM of large dominant copepods, we enumerated C. glacialis from PCP net samples. For C. glacialis, the lipid accumulation of C5 was classified into three categories: I (the oil droplet length, ODL, was 0–4 % of the prosome length, PL), II (ODL was 4–40 % of PL) and III (ODL was > 40 % of PL). The gonad maturation of C. glacialis C6F was also classified into three categories: I (immature), II (small oocytes in the ovary or oviduct) and III (large eggs or distended, opaque, filled-in oviducts). For this gonad maturation index, we cited that of C. hyperboreus (Hirche and Niehoff, 1996).

A species diversity index (H′) in each sample was calculated using the following equation: H′=-∑n/Ni×ln⁡n/Ni, where n is the abundance (ind. m-2) of the ith species and Ni is the abundance (ind. m-2) of the total calanoid copepods in the sample (Shannon and Weaver, 1949). Pielou evenness (J′) (J′) was also calculated using the equation: J′=H′/ln(s), where s is the total number of observed species in the community (Pielou, 1966).

From the NORPAC net samples, the mean copepodid stage (MCS) of C. glacialis was calculated using the following equation: MCS=∑i=16i×Ai∑i=16Ai, where i is the number of the copepodid stage (1–6 indicate C1–C6) and Ai (ind. m-2) is the abundance of the ith copepodid stage (cf. Marin, 1987).

During the study period, a SWE was observed on approximately 19 to 22 September (Kawaguchi et al., 2015; Nishino et al., 2015). According to Kawaguchi et al. (2015), there were meteorologically and oceanographically distinct periods between 10 and 18 September and 19 and 26 September, represented as terms I and II, respectively. Term II was characterized by longer, stronger northeasterly winds, which continued for several days between 19 and 22 September, the average intensity of which was greater than 13 m s-1. To evaluate the effect of the SWE, the abundances of each zooplankton taxon and species were compared “before the SWE (10–18 September)” and “after the SWE (19–25 September)” using the U test. This statistical analysis was performed using StatView.

During the sampling period, the temperature ranged from -1.5 to 3.3 ∘C and a thermocline was observed at a depth of approximately 25 m (Fig. 2a). Cold water below 0 ∘C continuously persisted below the thermocline, whereas the temperature above the thermocline decreased from 3.3 to 1.5 ∘C during the study period (Fig. 2a). The salinity ranged from 31.1 to 32.8, and a halocline was observed at approximately 25 m, which ran parallel to the thermocline (Fig. 2b). The salinity below the halocline was constant (ca. > 32), whereas the salinity in the upper layer increased from 31.1 to 31.6 throughout the study period. Chl a ranged from 0.08 to 3.25 mg m-3 and increased after the SWE (Fig. 2c). The T-S diagram showed that the hydrographic conditions in the upper layer changed temporally; thus, the temperature decreased, whereas the salinity increased during the study period (Fig. 2d).

The zooplankton abundance ranged from 23 610 to 56 809 ind. m-2, and the calanoid copepods and barnacle larvae composed 57 and 31 % of the community, respectively (Table 1). For the calanoid copepods, 15 species belonging to 9 genera were identified. Among them, Pseudocalanus spp. and C. glacialis dominated and composed 65 and 30 % of the total copepods, respectively. The Shannon species diversity and Pielou evenness for the copepods community were 1.87 ± 0.12 and 0.81 ± 0.06, respectively. According to a comparison of time periods before and after the SWE, the total zooplankton abundance, the Shannon species diversity and the Pielou evenness did not change, whereas one calanoid copepod, Centropages abdominalis, and the cyclopoid copepods increased after the SWE (Table 1). However, one calanoid copepod, Metridia pacifica; appendicularians; barnacle larvae; euphausiids; and gymnosomes (Clione limacina) decreased after the SWE.

Throughout the study period, the biomass of C. glacialis ranged from 1990 to 14 554 ind. m-2, and no significant changes were detected after the SWE (Fig. 3a, Table 1). For the population structure, all of the copepodid stages (C1 to C6F/M) were present, and C5 was the most dominant stage (36 %) of the population (Fig. 3). The MCS did not vary with the SWE (Fig. 3a). Throughout the study period, the lipid accumulation of C5 was high (Fig. 3b) and the gonad maturation of C6Fs was dominated by immature specimens (Fig. 3c). These parameters exhibited no significant changes with regard to the SWE (U test, p=0.285-0.303).

On both 14 and 22 September, the C1–C4 and C5 populations of C. glacialis were present mostly at lower layers (20–49 m) during the daytime, and they migrated to upper layers (0–20 m) at night (Fig. 4). It should be noted that approximately half of the C5 population remained in the lower layer both during the day and at night. The lipid accumulation was higher for the C5 population residing in the lower layer. C6Fs were present at a lower layer throughout the day.

The gut pigment of C. glacialis C5 ranged from 0.6 to 12.3 ng pigment ind.-1 and showed a significant increase after the SWE (mean values: 2.6 vs. 4.5 ng pigment ind.-1, U test, p<0.01) (Fig. 5a). In a comparison between day and night samplings, most dates, except 18 and 23 September, exhibited higher gut pigment levels at night by a factor of 2 to 5 times greater than those observed during the day. In both day and night samplings, the gut pigments were correlated with the biomass of Chl a (p<0.05) (Fig. 5b).

The population grazing rate of C. glacialis C5 ranged from 0.04 to 0.28 mg pigment m-2 day-1, peaking on 20 September (Fig. 6a) and increasing significantly after the SWE (0.11 vs. 0.18 mg pigment m-2 day-1, U test, p<0.05). During the study period, Chl a peaked on 18 September and the micro-size Chl a (> 10 µm) dominated (54 % of the total Chl a), especially after the SWE (66 %) (Fig. 6b). The grazing impact of C. glacialis C5 on the micro-size Chl a ranged from 0.5 to 4.6 % Chl a biomass day-1 and was high before the SWE from 10 to 15 September (Fig. 6c).

The zooplankton community in the Chukchi Sea is known to have large spatial and temporal changes (Springer et al., 1989; Llinás et al., 2009; Matsuno et al., 2011). The total zooplankton abundance in this study was approximately half (mean: 34 059 ind. m-2) the abundance reported by Matsuno et al. (2012) on the Chukchi shelf (mean: 75 683 ind. m-2), with a low abundance of small copepods (Pseudocalanus spp. and cyclopoids) and a remarkable absence of the Arctic copepod Metridia longa. For the hydrography of this station, Nishino et al. (2015) noted that the upper, warm and less saline water was the Pacific Summer Water, which was transported to the Arctic Ocean during summer, and the cold saline water of the lower layer was the remnant brine of the Pacific Winter Water, which was transported to the Arctic Ocean during winter. Geographically, the present station is located at a primary stream of water from the Pacific Ocean (Weingartner et al., 2005). The high abundance of the Pacific copepods M. pacifica and Neocalanus spp. and the absence of the Arctic M. longa in this study was thought to be a reflection of the water mass covering the station. For these reasons (fewer small copepods and high abundance of Pacific copepods), the Shannon species diversity and Pielou evenness in this study (1.87 and 0.81, respectively) are higher than the reported values for the entire Chukchi Sea (1.79 and 0.62, respectively, calculated from Matsuno et al., 2012).

Seasonal characteristics during summer included the dominance of the meroplankton (barnacle and bivalve larvae), which composed 39 % of the total zooplankton abundance (Hopcroft et al., 2010). The dominance of barnacle larvae also occurred in this study (Table 1). Benthic barnacle adults release their larvae when they meet phytoplankton blooms (Crisp, 1962; Clare and Walker, 1986), and their larvae spend 2 to 3 weeks in the water column and then settle (Herz, 1933). The abundance of barnacle larvae in this study (mean 10 430 ind. m-2) was 13–55 % lower than that in summer (19 114–79 899 ind. m-2; Matsuno et al., 2011). It should also be noted that the abundance of barnacle larvae decreased significantly during the study period (Table 1). These facts suggest that most of the barnacle larvae may have ended in the planktonic phase and settled to the sea bottom during the study period (autumn).

Concerning the effect of the SWE, a few taxa and species showed significant changes in abundance (Table 1). Among the dominant species, cyclopoid copepods increased after the SWE (Table 1). The generation length of cyclopoid copepods was reported to be 2 to 3 months in the Arctic Ocean (Dvoretsky and Dvoretsky, 2009). At ambient temperatures (-1.5 to 3.3 ∘C), the egg hatching of this taxon is estimated to be 11–41 days (Nielsen et al., 2002). These facts suggest that the increase in cyclopoid copepods would not be caused by their reproduction within the study period (16 days). An alternative cause, the horizontal advection of the water mass during the study period, which was reported by Nishino et al. (2015), should be considered. These results suggest that the effect of the SWE on zooplankton abundance was relatively small because of the longer generation length of the mesozooplankton in this region.

Concerning the population structure, Ashjian et al. (2003) reported that C. glacialis around the Northwind Abyssal Plain was dominated by C5 and C6Fs in September. In the present study, the population structure of C. glacialis was dominated by C5 (Fig. 3a) and their MCS (mean ± SD: 3.77 ± 0.20) was similar to the reported value for autumn in this region (3.58; Matsuno et al., 2012). Most of the C6Fs had immature gonads, and no ovigerous C6Fs were observed (Fig. 3c). These results corresponded with the year-round observation around the Northwind Abyssal Plain (Ashjian et al., 2003). Calanus glacialis C6Fs are known to occur at the epipelagic layer in April, immediately before sea-ice melting (Kosobokova, 1999), and reproduces with grazing ice algae and the ice-edge bloom (Campbell et al., 2009). Thus, because this study period (September) greatly varied with regards to their reproduction period (April), most C6Fs were considered to have immature gonads when residing in the lower layer (diapause).

The nocturnal ascent DVM, which is related to nighttime grazing on phytoplankton, was reported for C. glacialis in the Arctic Ocean during spring and autumn (Runge and Ingram, 1988; Conover and Huntley, 1991). In this study, the DVM was observed for C5 (U test, p<0.01) (Fig. 4). At high-latitude seas, the magnitude of the Calanus spp. DVM is known to vary with the season and copepodid stage, and their DVM intensity is greater during spring and autumn, when the diel changes in light penetration are large (Falkenhaug et al., 1997). No DVM of Calanus spp. was reported for the lipids accumulated in C5 (Falk-Petersen et al., 2008). In the present study, approximately half of the C5 population, which was characterized as having a large lipid accumulation, remained in the lower layer throughout the day (Fig. 4). The deep C5 population may have already completed lipid accumulation and ceased DVM in the study period (September), whereas the remaining C5 population with an active DVM may have grazed on phytoplankton in the upper layer during the night and stored lipids in preparation for diapause. These results suggest that the C. glacialis population in this study was at the seasonal phase just before entering diapause, and this interpretation corresponded well with their life cycle in this region (Ashjian et al., 2003).

Calanus glacialis in the Arctic Ocean is known to exhibit higher gut pigment levels at night than during the day (Conover and Huntley, 1991). Higher gut pigment levels at night were also observed in this study (U test, p<0.001) (Fig. 5a). The gut pigments of C. glacialis were correlated with the biomass of Chl a (Fig. 5b) and increased during the high-Chl a period after the SWE (Fig. 5a). These facts suggest that C. glacialis feeding responded to the small phytoplankton bloom, which was enhanced by the nutrient supply and vertical mixing caused by the SWE (Nishino et al., 2015; Yokoi et al., 2015).

Concerning the gut pigment measurement, the underestimation by the decomposition of the phytoplankton pigment through the gut passage has been reported (Conover et al., 1986; Head, 1992). This underestimation is reported to be approximately 0.1–10 % of grazing (Conover et al., 1986) and varies with light conditions, grazing behavior and phytoplankton species (Head, 1992). To estimate the grazing impact, data on the gut evacuation rate (k, min-1) are needed (Mauchline, 1998). The gut evacuation rate is known to have a positive correlation with temperature (Dam and Petersen, 1988). From the equation (k=0.00941+0.002575T; Mauchline, 1998) for k and the temperature (T: ∘C) and ambient temperature in this study (T: -1.5 to 3.3 ∘C), k is estimated to be 0.0055–0.0179 min-1. This range covers the value applied in the present study (k=0.017; Tande and Båmstedt, 1985). The value was also in the range observed by our independent laboratory experiments in September 2010 (0.006–0.041; Matsuno et al., unpublished data). These facts suggest that the value applied in this study (k=0.017) was reasonable for C. glacialis in this region.

Assuming that half of the C5 population performed nocturnal ascent and grazed on phytoplankton at night, using the C : Chl a ratio (29.9; Sherr et al., 2003), the grazing impact (mg C m-2 day-1) of C. glacialis C5 was calculated (Table 2). The grazing impact of this study was estimated to be 4.14 mg C m-2 day-1. We also estimated the food requirement of C. glacialis C5 to support their metabolism under ambient temperatures (Ikeda and Motoda, 1978; Ikeda et al., 2001). The potential contribution of phytoplankton's food-to-food requirements was 12.6 % for C. glacialis C5 (Table 2). This result indicates that C. glacialis C5 could not maintain its population solely on phytoplankton food and that other food sources are important. Regarding food for C. glacialis, Campbell et al. (2009) reported that this species prefers microzooplankton rather than phytoplankton in the Chukchi Sea, and Levinsen et al. (2000) noted that the micro-size (> 10 µm) ciliates and dinoflagellates are important food sources during post-bloom. For the microplankton community during the study period, Yokoi et al. (2015) noted that not only diatoms (1.64–14.11 cells mL-1) but also dinoflagellates (0.54–2.42 cells mL-1) and ciliates (0.14–2.76 cells mL-1) were abundant. From the fatty acid composition, C. glacialis is reported to have a strong connection with the microbial food web (Stevens et al., 2004). However, the low abundance of dinoflagellates and ciliates during the study period makes it difficult to assume that they were an important food source for C. glacialis in this study. As an alternative explanation, because most C. glacialis C5 contained an abundance of lipids in their body (Fig. 3b), they may have entered diapause in the Chukchi Sea during autumn.