Distribution of Arctic and Pacific copepods and their habitat in the northern Bering and Chukchi seas

The advection of warm Pacific water and the reduction in sea ice in the western Arctic Ocean may influence the abundance and distribution of copepods, a key component of food webs. To quantify the factors affecting the abundance of copepods in the northern Bering and Chukchi seas, we constructed habitat models explaining the spatial patterns of large and small Arctic and Pacific copepods separately. Copepods were sampled using NORPAC (North Pacific Standard) nets. The structures of water masses indexed by principle component analysis scores, satellite-derived timing of sea ice retreat, bottom depth and chlorophyll a concentration were integrated into generalized additive models as explanatory variables. The adequate models for all copepods exhibited clear continuous relationships between the abundance of copepods and the indexed water masses. Large Arctic copepods were abundant at stations where the bottom layer was saline; however they were scarce at stations where warm fresh water formed the upper layer. Small Arctic copepods were abundant at stations where the upper layer was warm and saline and the bottom layer was cold and highly saline. In contrast, Pacific copepods were abundant at stations where the Pacific-origin water mass was predominant (i.e. a warm, saline upper layer and saline and a highly saline bottom layer). All copepod groups showed a positive relationship with early sea ice retreat. Early sea ice retreat has been reported to initiate spring blooms in open water, allowing copepods to utilize more food while maintaining their high activity in warm water without sea ice and cold water. This finding indicates that early sea ice retreat has positive effects on the abundance of all copepod groups in the northern Bering and Chukchi seas, suggesting a change from a pelagic–benthic-type ecosystem to a pelagic–pelagic type.


Introduction
The Arctic sea-ice reduction is remarkable during these decades and has been a public concern as this may cause the changes in marine ecosystem in Arctic Ocean. Arctic marine food webs are supported by primary production in the seasonal sea-ice zone and profoundly influenced by the timing of blooming of the ice algae and the timing 5 of stabilization of water column by formation of sea-ice (Grebmeier, 2012). The recent sea-ice reduction progresses the timing of phytoplankton bloom (Hunt et al., 2002(Hunt et al., , 2011Kahru et al., 2011) and increases the annual primary production (Arrigo et al., 2008). However, Clement et al. (2004) suggested that the earlier sea-ice retreat leads the stratification, the trapping of the nutrient in the surface and lower primary produc- 10 tion with insufficient sunlight. In Arctic, food webs are short and efficient, even small changes in production pathways can affect on higher trophic organisms (Grebmeier et al., 2006). The change of the timing and location of primary production and associated grazing by zooplankton have a direct influence on the energy and material transfer to benthic community (Grebmeier et al., 2010). Thus recent reduction of sea-ice 15 extent leads shifts in the distribution of benthos, fishes and marine mammals (Mueter and Litzow, 2008;Doney et al., 2012). Further these shifts in the species distributions may change the interaction of species including predation and competition (Moore and Huntington, 2008).
The northern Bering Sea and Chukchi Sea may be amongst regions where the sea-20 sonal sea-ice coverage has been changed drastically in this decade (Comiso et al., 2008;Parkinson and Comiso, 2013), possibly because of the increase of the inflow of the Pacific water from the Bering Sea through the Bering Strait (Shimada et al., 2006). In these seas, the water masses have been identified based on salinity and temperature ( Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 32.3-33.3 • C) originated from the Gulf of Anadyr at the depth along the continental shelf of the Bering Sea; BSW and AW merged into Bering Sea Anadyr water (BSAW) (Coachman et al., 1975;Springer et al., 1989). Further, cold/lower salinity ice melt water (IMW; < 2.0 and < 30.0 • C) originates from sea-ice and colder/high salinity dense water (DW; < −1.0 and 32.0-33.0 • C) formed in the previous winter during freezing 5 over the both Bering Sea and Chukchi Sea  are identified. These water masses often show vertical settings geographically and seasonally (Iken et al., 2010;Eisner et al., 2013;Weingartner et al., 2013).
In the northern Bering Sea and Chukchi Sea, copepods are primary consumers of phytoplankton and are main prey of forage fish (e.g., Polar cod Boreogadus saida, 10 Nakano et al., 2015), seabirds (e.g., Phalaropes, Shearwaters and Crested Auklets Aethia cristatella, Piatt and Springer, 2003;Hunt et al., 2013), and baleen whales (e.g., Bowhead whale Balaena mysticetus, Lowry et al., 2004). Therefore, copepods could be a key component in the Arctic marine food webs (Lowry et al., 2004). The communities of copepods are associated with the structure of water masses (e.g., Springer 15 et al., 1989;Eisner et al., 2013); Pseudocalanus species are abundant in the ACW and Pacific species are abundant in the AW. To quantify the response of copepods to the recent reduction of sea-ice extent, environmental factors affecting their distributions have to be explored more systematically and quantitatively. In this region, large sized Arctic copepods (Calanus glacialis) and small sized Arc-20 tic copepod (e.g., Acartia hudsonica, Centropages abdominalis, Eurytemora herdmani and Pseudocalanus acuspes) are distributed (Springer et al., 1996). Also Pacific copepods (Calanus marshallae, Eucalanus bungii, Meridia pacifica, Neocalunus cristatus, N. flemingeri and N. plumchrus) are often transported from the Bering Sea (Lane et al., 2008;. Therefore, for communities of copepods in this region, the 25 inflow of warm Pacific water as well as the reduction of sea-ice extent may be important factor. In 2007, for an example, sea-ice disappeared at earliest and its coverage was minimum and water temperature was highest and annual transport of the Pacific water was greatest during these 17 years (Woodgate et al., 2010(Woodgate et al., , 2012 2007(Matsuno et al., 2011. To predict these responses of zooplankton to the environmental changes can be happened in future, it is important to understand the tendency of the spatial patterns of the abundance of zooplankton and the key environmental factors for them by using statistical model, quantitatively.

5
The objective of this study is to examine the factors affecting the spatial pattern of the abundance of copepods based on the data collected by T/S Oshoro-maru in the summer of 2007, 2008 and 2013. We categorized copepods into three groups; large and small Arctic, and Pacific copepod. Life cycles of large Arctic copepods are one or less generation per year, while that of small Arctic copepods are multiple genera-10 tions in Arctic (e.g., Dvoretsky and Dvoretsky, 2009;Falk-Peterson et al., 2009). Pacific copepods are only advected from the Pacific Ocean through the Bering Strait and not established in Arctic Ocean (Springer et al., 1989;Matsuno et al., 2015). To quantify the factors affecting the spatial pattern of abundance of each copepods group, we used the Generalized Additive Models (GAMs). The relationships between the abun-15 dance of copepods and traditionally defined water masses are reported Eisner et al., 2013) where the surface and bottom water masses were characterized based on the temperature and salinity. However, it is difficult to evaluate the effects of complicated water properties quantitatively on the abundance of copepods. To apply GAMs, explanatory variables that were correlated with other vari-20 able have to be removed to avoid the problem of multi co-linearity. This procedure may fail to denote the important oceanographic features such as the combination of water masses in upper and bottom layers since temperature and salinity of waters in both layers are often correlated strongly. In this study, to denote the combination of water masses in upper and bottom, we summarized the water mass properties in upper and bottom layers into the scores using principal component analyses (PCA). These scores can be used as continuous explanatory variables in the habitat models. Introduction  Dvoretsky and Dvoretsky (2009), and we summarized the copepods species into three groups: large Arctic (Cop arc -L, generation length is more than one 15 year and reproduction occur at one time), small Arctic (Cop arc -S, generation length is less than one year and multi-times reproduction occur in a year) and Pacific copepods (Cop pac , generation length is more than one year and reproduction occur at one time) (  (Suzuki and Ishimaru, 1990), and chlorophyll a concentrations were determined by fluorometric method using a Turner Designs 10-AU fluorometer (Welschmeyer, 1994). In order to investigate the relationships between the abundance of copepods and sea-ice condition, we derived SSM/I Daily Polar Gridded Introduction Sea Ice Concentration (SIC) data, calculated by using NASA Team, from the National Snow and Ice Data Center (http://nsidc.org/).

Data analysis
We divided the water column into two layers; i.e., the layers above and under the pycnocline and defined them as the upper and bottom layers, respectively. The pycnocline 5 was defined as the depth of maximum density gradient. The density (ρ) was calculated from temperature and salinity measured by CTD profiles with a vertical data resolution of 1 m. Then, we calculated vertical density gradient ( dρ dD ) at each depth (D). We defined the depth of the maximum dρ dD (hereafter dρ dD max ) as the depth of maximum density gradient (D dρ dD max ). In this approach, to examine the water mass properties at 10 the upper and bottom layers, environments (temperature, salinity and log-transformed chlorophyll a) were vertically averaged within the upper and bottom layers (i.e., above and under D dρ dD max ) and defined them as T UPP , T BOT , S UPP , S BOT , Chl a UPP and Chl a BOT , respectively (see Table 3). Principal component analysis (PCA) was applied to determine water mass structure using dρ dD max , T UPP , T BOT , S UPP and S BOT at all 88 stations 15 together. Because principal water masses in the Bering Sea and Chukchi Sea were characterized by temperature and salinity of water column (Coachman et al., 1975), Chl.a UPP , Chl.a BOT and SIC were not used in PCA for determining structure of water masses. Given these five parameters in PCA were standardized prior to analysis to reduce problems with unequal variation and expected nonlinearity. Several principal 20 components and its factor loadings (correlations of factors to the derived principal components) are presented. The scores of PCAs were used as covariates of water mass structures in the habitat models. In addition, we used the anomaly of timing of sea-ice retreat (aTSR) at each sampling station as index of sea-ice condition. aTSR was calculated using satellite derived sea-ice image in 1991-2013. Although sea-ice concentration images had been projected to polar stereographic coordinates with a 25 km spatial resolution, we interpolated them using the nearest neighbour method and resampled Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | them into 9 km spatial resolution. Considering the missing value and land contaminations, we defined SIC < 50 % as non-ice covered pixels, and aTSR was defined as the anomalous last date when the SIC fell below 50 % prior to the date of annual sea-ice minimum in the Arctic Ocean. 5 Prior to conducting the habitat models, we examined the multi co-linearity between the explanatory variables by correlation analysis. To examine the relationships between copepods abundance (large and small Arctic, and Pacific copepods) and environments, we constructed the habitat models using Generalized Additive Models (GAMs). GAMs are a non-parametric extension of Generalized Linear Models (GLMs) such as multiple 10 regression models (Eq. 1); with the only underlying assumption that functions are additive and that the components are smooth (Eq. 2). The basic concept is replacement of the parametric GLMs structure:

Statistical analysis
with the additive smoothing function structure: where the α and the ε are the intercepts, and the β i and the s i are coefficients and smooth functions of covariates, respectively. To select the most adequate model in our approach, we used Akaike's Information Criteria. Then model validation was applied on the optimal models to verifying assumptions and reproducibility. Specifically, we plotted 20 original value vs. fitted value, and judged the adequacy of our optimal models based on R 2 . Deviance explained indicated how many percent can explain the variance of the most adequate model. All statistical analyses were undertaken using R (version.2.15.0 http://www.r-project.org).

Principal component analysis and water mass
The first principal component explained 47.1 % of total variability. In the score of first principal component (PC1), the coefficient of loading was positive for dρ dD max , indicating that the magnitude of stratification increased with increasing PC1, while strongly neg-5 ative for T UPP and T BOT , indicating that the temperature in the whole of water mass was lower with higher PC1 (Table 4). And also, it was negative for S UPP indicating that there was low salinity water mass in the surface layer with higher PC1, while weakly positive for S BOT . According to Fig. 2a showing the T -S diagram coloured by the score of PC1, the higher PC1 (> 1) value indicated the combination of the cold/lower salinity ice-melt 10 water (IMW) in upper layer and the colder/high salinity dense water (DW) in bottom layer. On the contrary, low PC1 indicated the warm water mass in both layers and/or low salinity water in the surface (Table 4). According to Fig. 2a, lower PC1 ( < −1.5) value indicated the combination with warmer/low salinity Alaskan coastal water (ACW) in the upper layer and warm/saline Bering shelf water (BSW) or cold/higher salinity 15 Anadyr Water (AW) or merged water (BSAW) in bottom layer. When the score of PC1 showed low-medium (−1.5-0.5), it indicated the combination of water mass with BSW and AW or BSAW (Fig. 2a) (Table 4). These indicated that there was the highly saline water at both layers which leads to de- 25 crease the magnitude of stratification and become single layered structure with higher PC2. Shown in Fig. 2b, the medium-high PC2 (> 0.5), indicated waters with singlelayered structure, warm/saline BSW or colder/higher salinity AW or BSAW, and the 18669 Introduction  (Table 4), especially positive correlated especially with T UPP and S BOT . According to T -S diagram, 10 coloured by the value of PC3 (Fig. 2c), relatively high PC3 (> 0.5) with warmer T UPP (> 4.0 • C) and/or high S BOT (> 32.0) suggested that the water columns were composed by warmer ACW at the surface and/or high salinity BSW/AW at the bottom. PC3 was higher in 2007 than in 2008 and 2013, especially at the stations in the north of Bering Strait (Fig. 3), indicating that warm BSW and warmer ACW in upper layer and/or higher 15 salinity AW or BSAW or DW in bottom layer.

Habitats of copepods
We constructed the habitat models using the anomaly of the timing of sea-ice retreat (aTSR), quantitative index of water masses (PC1, PC2 and PC3), bottom depth (Bdepth) and averaged log-transformed chlorophyll a in upper layer (Chl.a UPP ) and bottom layer (Chl.a BOT ) as potential explanatory variables. Averaged physical factors in 5 upper layer and bottom layers were excluded from potential explanatory variables, as these were already included in the quantitative index of water masses. The most adequate model explaining the abundance of Cop arc -L included all explanatory variables (Table 5). Cop arc -L were abundant at the station with the lower aTSR ( < 0 days), with deeper Bdepth, especially in the area where the bottom depth 10 was deeper than 45 m (Fig. 5). Cop arc -L appeared to be abundant at stations with medium-higher PC1 (> −0.5), low-high PC2 (−1-1), low-medium PC3 (−1-0). The abundance of Cop arc -L was slightly high in the water of low ( < −0.5) and high (0.2-0.5) Chl.a UPP , however, the effects of Chl.a UPP and Chl.a BOT on Cop arc -L were not clear. 15 The most adequate model explaining the abundance of Cop arc -S included all explanatory variables except PC2 (Table 5). Cop arc -S were abundant at the station with the lower aTSR ( < 5 days), with deeper Bdepth, especially in the area where the sea depth was deeper than 40 m (Fig. 5). The abundance of Cop arc -S was high in low-high PC1 ranged from −1.5 to 2 and medium PC3 (0-1.2), and in medium-high Chl.a UPP 20 (> 0) (Fig. 5). The effect of Chl.a BOT was not clear on the abundance of Cop arc -S. The most adequate model explaining the abundance of Cop pac included all explanatory variables except Chl.a UPP (Table 5). Cop pac were abundant at stations with low aTSR ( < 0 days), deeper Bdepth with clear positive effects in waters deeper than 35 m, low-medium PC1 (−2-0.5) and PC3 (−0.5-1) and PC2 ( < −0.5), and less abundant at 25 stations with medium-high PC2 (> −0.5) and high PC1 (> 0.5) (Fig. 5). The abundance of Cop pac was high in the water of low ( < −0.2) and high (> 0.5) Chl.a BOT , however, the effects of Chl.a BOT on Cop pac was not clear.

The application of principal component analysis
In the northern Bering Sea and Chukchi Sea, the major six water mass (Alaskan coastal water (ACW), ice melt water (IMW) and dense water (DW), Bering Shelf water (BSW), Anadyr water (AW) and merged Bering Shelf Anadyr water (BSAW)) are dom-5 inated (e.g., Coachman et al., 1975;Springer et al., 1989). These water masses and their combinations have mostly been described by clustered analysis using temperature and salinity (e.g., Norcross et al., 2010;Eisner et al., 2013;Ershova et al., 2015). This study quantitatively characterized these water masses using PCA; the combination of water masses, the number of composing layers (single or double layered) and  (Fig. 3). In the summer of 2007, however, the water off Point Hope (southern part of the Chukchi Sea) was characterized by warmer (2.0-10 • C)/low salinity ( < 31.8) ACW as upper layer and AW as bottom layer (Figs. 2 and 3), giving low PC1 ( < −1.5), medium-high PC2 (> 0.5) and low PC3 ( < −1) (Fig. 3). Combination and distributions of water masses are known to be affected by Pacific inflow (Wein-25 gartner et al., 2005) and related to sea-ice retreats (Coachman et al., 1975;Day et al., 2010). This indicates that the inflow of warmer Pacific ACW was dominated in 2007 (Woodgate et al., 2010), and this strong inflow was believed to trigger of sea-ice retreat 18672 Introduction in the western Arctic Ocean (Woodgate et al., 2012). Thus, variability of water masses and its combination indexed by PCs were agreed with the conventional description of the dynamics of water masses. Our index can be used to evaluate the effects of water mass combination with multiple components of water properties quantitatively and so be useful for predicting the distribution of copepods under climate changes.

Habitats of copepods
In the northern Bering Sea and Chukchi Sea, it has been well documented that the community structure and the abundance of zooplankton species were varied in water masses (e.g., Lane et al., 2008;Matsuno et al., 2011). The abundance of all copepod groups in this study was also related to water mass (Fig. 5).
For example, large Arctic copepods (Cop arc -L) were slightly abundant in the water with cold/lower salinity IMW at upper layer and the colder/high salinity DW in bottom layer corresponding to higher PC1 and low-medium PC2 and PC3, or cold/high-higher salinity BSAW and AW in both layer corresponding to medium PC1, medium-high PC2 and low-medium PC3. These observations support the previous findings that C. glacialis 15 in the Arctic population is distributed in Winter Water (Ershova et al., 2015) and concentrated further offshore in higher salinity BSAW at the bottom (Eisner et al., 2013). However, Cop arc -L in this study were less abundant in ACW at upper layer and BSAW at the bottom layer corresponding to low-medium PC1 and high-medium PC3 (Fig. 5). This contradicted with Eisner et al. (2013) that found that C. glacialis was more abun-20 dant in the water with BSAW at the bottom and ACW at the upper layer. In contrast to Cop arc -L, small Arctic copepods (Cop arc -S) were common through the study area, they were abundant in waters of medium PC1 medium PC3, indicating that this group is distributed in waters having wide range of temperature and saline, i.e., warm/saline BSW. However, Cop arc -S were less abundant in the waters of higher PC1, i.e., colder/low salinity IMW at upper layer and cold/high salinity DW. These support the previous findings that small Arctic copepods (e.g., Pseudocalunus sp., A.  (Eisner et al., 2013;Ershova et al., 2015). Thus, abundance of Cop arc -L could be associated with cold water mass where Cop arc -S were less abundant. The difference in water masses between Cop arc -L and Cop arc -S might be related to body size and life cycle. Cop arc -S are smaller and have 5 multiple generations per year (e.g., McLaren et al., 1989), while Cop arc -L have 1-3 year life cycle (Melle et al., 1998). Larger Cop arc -L can accumulate more lipid than Cop arc -S and larger energy storage enables Cop arc -L to diapause under sea-ice in winter (Seuthe et al., 2007;Falk-Peterson et al., 2009) and to be active and graze icealgae under the sea-ice in cold IMW and DW in pring. Although few studies about lipid 10 of smaller-sized copepods (e.g., Kattner and Hagen, 2009), Cop arc -S might not be able to diapause so long time in cold water because of their small body size, so abundant in the warmer ACW in upper and higher nutrient BSW in bottom layers.
Pacific zooplanktons are advected into the western Arctic Ocean through the Bering Strait (Springer et al., 1989). Previous studies showed that Pacific zooplankton com-15 munities were observed in the high salinity water (BSW/AW) in the northern Bering Sea and Chukchi Sea (Springer et al., 1989;Lane et al., 2008;Matsuno et al., 2011;Eisner et al., 2013). In this study, Pacific copepods (Cop pac ) were abundant in the Bering Strait and Chukchi Sea south of Point Hope, giving lowmedium PC1 and PC2; associated with warmer/low salinity ACW in upper layer and 20 cold/higher salinity AW and warm/saline BSW or BSAW in the bottom layer, or single layered AW, BSW and BSAW, supporting these previous observations. Our study further confirmed the effects of the interannual variation of the water masses on the abundance of copepods. During the summer of 2007, Pacific water masses (ACW, BSW and BSAW) extended to the north of 69 • N (Fig. 3) and transported Cop pac into 25 the Chukchi Sea (Matsuno et al., 2011). But in the summer of 2008 and 2013 when IMW and colder/high salinity DW were dominant, and little Cop pac were collected in the northern part of the Chukchi Sea (Fig. 4).

Effects of sea-ice, phytoplankton and bottom depth
The influences of the changes in sea-ice on Arctic and subarctic ecosystem have been implicated in several previous studies about phytoplankton, benthic and higher trophic level organisms (Arrigo et al., 2008;Moore and Huntington, 2008;Grebmeier et al., 2012). In this study, the positive effects of earlier sea-ice retreat on the abundance of all copepod groups have shown in the result of GAM (Fig. 5). There has been little study that captures the influence of the timing of sea-ice retreat on the abundance of copepods, whereas the relationship between sea-ice retreat timing and the phytoplankton bloom has been well discussed in previous studies (e.g., Hunt et al., 2002Hunt et al., , 2010Kahru et al., 2010;Brown et al., 2011). The timing of phytoplankton bloom which 10 is affected by the timing of sea-ice retreat plays an important role in the recruitment of copepods in the southeastern Bering Sea (Hunt et al., 2002(Hunt et al., , 2011Overland and Stabeno, 2004;Stabeno et al., 2007). Brown and Arrigo (2013) showed the spring bloom inevitably formed at the ice edge and its timing was controlled by the timing of sea-ice retreat in the northern Bering Sea. In addition, Sigler et al. (2014) suggested 15 that no ice-associated bloom occurs under the earlier sea-ice retreats (before mid-March), because sunlight is not sufficient to an ice-associated bloom throughout the eastern Bering Sea. In these years with earlier sea-ice retreat, the spring bloom occurs in late spring (May-early June) (Sigler et al., 2014). In the late spring bloom with warmer temperature, grazing by copepods can increase. Thus, earlier sea-ice retreat 20 might have positive effects on reproduction of copepods in the northern Bering Sea and Chukchi Sea. Cop arc -S (e.g., Pseudocalanus spp.) graze phytoplankton and reproduce in the surface layer during day and night in summer (Norrbin et al., 1996;Plourde et al., 2002;Harvey et al., 2009). We, therefore, expected positive effects of the surface chloro- high-grazing and reproduction season when they require large amount of food intake. Cop arc -L reproduce during the spring phytoplankton bloom (e.g., Falk-Peterson et al., 2009), so our sampling period was not the time of their reproduction. The phytoplankton cells sinking into the bottom layers might be important food for copepods (Sameoto et al., 1986). Thus we also expected positive effects of the bottom chlorophyll a con-5 centration (Chl.a BOT ) on all copepod groups. However, clear positive effects were not observed again (Fig. 5). It is difficult to link chlorophyll a concentration to copepod abundance using time lag between the bloom of phytoplankton and copepods. There were a few previous studies reporting the associations between the abundance of copepods and the bottom depth in shelf of the northern Bering Sea and 10 Chukchi Sea (e.g., Ashjian et al., 2003). The reason why all copepod groups were less abundant in waters shallower than 32 m bottom depth was unclear. These sampling stations were near the land with relative low salinity (ρ = 0.53, spearman's rank correlation test in S UPP vs. Bdepth p < 0.01). Although oceanic copepods survive in waters 0-25 m bottom depth with low salinity in the Arctic Canada basin , the sallower area in this study was considered to be affected by fresh waters from land and unsuitable for all copepods.
The association between environmental factors and the abundance of copepods and its communities have been well documented (e.g., Springer et al., 1989;Lane et al., 2008;Matsuno et al., 2011). Recently these relationships have been analyzed using 20 clustered water masses (Eisner et al., 2013;Ershova et al., 2015). This study indexed the water mass and then modelled the relationship between the water mass characteristics and spatial patterns of copepods abundance quantitatively. Our evaluation of the changes in the timing of sea-ice retreat on the abundance of copepods insisted the suitable environment for copepods are forming by the early sea-ice retreat. The 25 influence of the changes in sea-ice in Arctic ecosystem has been well documented. However, to the best of our understanding, this is the first quantitative study to show the relationships between the early sea-ice retreat and the abundance of copepods. Quantitative analyses using the habitat models are useful for understanding various phenomena and risks faced by organisms (e.g., sea-ice loss, increased water temperature and fresh water content). Furthermore, it can be adapted to predict the changes on ecosystem in future by incorporating climate and predicted environmental data. It can also be used to understand the responses of organisms to environmental change in the northern Bering Sea and Chukchi Sea.   Matsuno, K., Yamaguchi, A., Hirawake, T., Nishino, S., Inoue, J., and Kikuchi,  Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Seuthe, L., Darnis, G., Riser, C. W., Wassmann, P., and Fortier, L.: Winter-spring feeding and metabolism of Arctic copepods: insights from faecal pellet production and respiration