The recent increasing of atmospheric turbulence has had considerable impact on the oceanic environment and ecosystems of the Arctic. To understand its effect on phytoplankton community structure, a Eulerian fixed-point observation (FPO) was conducted on the Chukchi shelf in fall 2013. Temporal and vertical distributions of the phytoplankton community were inferred from algal pigment signatures. A strong wind event (SWE) occurred during the observation term, and significant convection supplied nutrients from the bottom layer to the surface. Before the SWE, pigment composition in the warmer, less saline, and nutrient-poor surface waters was diverse with low concentration of chlorophyll-a (chla). Vertical mixing induced by the SWE weakened the stratification and brought sufficient nutrients to enhance diatom-derived pigment concentrations (e.g., fucoxanthin and chlc3), suggesting increases in diatoms. We also developed a model to predict the distribution of major phytoplankton pigment/chla ratios using a profiling multi-wavelength fluorometer ( Multi-Exciter) with higher spatio-temporal resolution. The Multi-Exciter also captured changes in pigment composition with environmental changes at the FPO site and at four observation sites 16 km from the location of the FPO. Furthermore, we investigated the change in grazing rates of the major Arctic copepod Calanus glacialis copepodid stage five to assess the interaction between primary and secondary producers during the fall bloom. Increased diatom biomass caused a significant increase in the grazing rate on microphy - toplankton (> 20 µm) and a decrease on nanophytoplankton (2–20 µm), indicative of a strong cascade effect because of the reduction of microzooplankton due to the grazing from C. glacialis. We conclude that SWEs during fall might affect food webs via the alternation of seasonal succession of phytoplankton community structure. Keywords Fall bloom · Wind-induced mixing · Phytoplankton community structure · Zooplankton grazing Introduction Recent warming of the Arctic Ocean has caused reduc- * Amane Fujiwara tion in the area of seasonal sea ice cover; its early break-up email@example.com in spring and delayed freezing in fall (e.g., Stroeve et al. 2007; Comiso et al. 2008; Markus et al. 2009). Thus, Arctic Japan Agency for Maine-Earth Science and Technology, Ocean ecosystems are facing drastic modification because 2-15 Natsushima-cho, Yokosuka, Kanagawa 237-0061, Japan of these changes. A number of studies have reported shifts in species composition, biomass, and distribution of many Faculty of Fisheries Sciences, Hokkaido University, 3-1-1 Minato-cho, Hakodate, Hokkaido 046-8611, Japan trophic level organisms because of the decline of sea ice and related environmental changes (e.g., Grebmeier 2012 Atmosphere and Ocean Research Institute, The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8564, and references therein). For example, in phytoplankton com- Japan munities, a taxonomic shift from large to small has been Faculty of Environmental Earth Science, Hokkaido found in the upper layer of the Canada Basin owing to the University/JST-CREST, N10 W5, Kita-ku, Sapporo, nutricline deepening (Li et al. 2009). Earlier sea ice retreat Hokkaido 060-0810, Japan has been found to cause change in the seasonal succession of National Institute of Polar Research, 10-3 Midori-cho, surface phytoplankton communities in the northern Chukchi Tachikawa, Tokyo 190-8518, Japan Vol.:(0123456789) 1 3 1280 Polar Biology (2018) 41:1279–1295 Sea (Fujiwara et al. 2014). Furthermore, enhanced ocean cir- increase in the number of pennate diatoms after the SWE. culation has been reported to stimulate primary productivity Mesozooplankton also responded to the increase in the in the Eurasian Basin but to reduce it within the Beaufort large-sized phytoplankton (> 10 µm) that they feed on (Mat- gyre (Nishino et al. 2011). Changes in secondary produc- suno et al. 2015). Not only planktonic organisms, but also ers have also been reported for the Chukchi and adjacent bacterial abundance and production responded and increased seas. Higher abundance and biomass of mesozooplankton significantly after the SWE (Uchimiya et al. 2016 ). How- during summers in 2000s than 1990s in the Chukchi Sea is ever, the response of the phytoplankton community structure considered to be less ice and warmer temperature (Matsuno including small assemblages (e.g., prasinophytes and hap- et al. 2011). Warmer temperature of less ice condition in the tophytes), generally dominate in the surface during strongly northern Bering Sea is expected to enhance zooplankton stratified season (e.g., Hill et al. 2005; Fujiwara et al. 2014), grazing activity on phytoplankton (Coyle et al. 2011). is still unclear. Since identification of these small assem- Changes in atmospheric conditions have been reported blages by microscopy requires a high level of taxonomic as well as sea ice conditions. For example, an increase in skills, we used pigment signature as a biomarker to infer storm frequency and a northward shift of storm tracks in the phytoplankton taxonomic composition at the class level the Arctic Ocean due to the sea ice loss have been reported (e.g., Jeffrey and Vesk 1997; Wright and Jeffrey 2006). To (e.g., Serreze et al. 2000; Zhang et al. 2004; Sepp and Jaa- understand the roles of the fall bloom on biogeochemical gus 2011). It is becoming important to take into account cycles and food web, it is important to comprehend the tax- the ocean–atmosphere interactions in the Arctic, because onomic response of phytoplankton community structure, recent sea ice loss may lead an increase of wind energy into because different phytoplankton assemblages play different the surface ocean (Rainville and Woodgate 2009; Martini roles in biogeochemical cycles and ecosystems (e.g., Cush- et al. 2014). In temperate seas, enhanced surface mixing ing 1989; Lochte et al. 1993; Sunda et al. 2002; Bopp et al. caused by storms could trigger sudden development of phy- 2003; Ardyna et al. 2011; Leu et al. 2011). In this study, toplankton blooms because of nutrient enrichment from we also focused on the ecological impact of the fall bloom below the nutricline (Lin 2012; Zhao et al. 2015). Ardyna assessing how increased phytoplankton can transport to the et al. (2014) documented that the occurrence of fall bloom secondary producer, the Arctic copepod Calanus glacia- in the Arctic, rarely seen previously (a single spring bloom lis, the key species in the Chukchi Sea, by measuring their is the predominant blooming feature), has increased in pan- grazing rate on phytoplankton. Integrating the knowledge of Arctic seas because of the growth in the number of stormy predator–prey interactions is essential to comprehend recent days. Because of the short-chained food web in the Arctic, Arctic changes in ecosystem. Hence we aimed to evaluate a small change in primary producers can have considerable the response of phytoplankton community structure and zoo- effect on higher trophic level organisms (Grebmeier et al. plankton grazing activity on the increased phytoplankton 2010). Therefore, it is crucial to comprehend the detailed to short-term environmental changes during the fall bloom. response of phytoplankton communities to novel blooming characteristics in fall. To document the physical, chemical, and biological Materials and methods oceanic effects of strong wind, we conducted a fixed-point observation (FPO) on the Chukchi shelf for two weeks Cruise summary and water sampling (Nishino 2013). Fortunately, a strong wind event (SWE) −1 defined as wind speed exceeds 10 m s occurred during The FPO site was located at 72.75°N, 168.25°W (Fig. 1a). the FPO. It persisted for a few days because of the slow We remained on station at the FPO site from September movement of an anticyclone over the Siberian Sea (Inoue 10–25, 2013 and revisited on September 30, during the et al. 2015). Several studies have reported about physical cruise of R/V Mirai (JAMSTEC). CTD observations and and biogeochemical impacts of the SWE during the FPO. routine water sampling were conducted every 6 h at the Briefly, the SWE magnitude was sufficiently large to induce FPO site (00:00, 06:00, 12:00, and 18:00 UTC). In addi- internal waves and weaken the vertical stratification (Kawa- tion, four observation sites (sub-FPO sites; stations A, B, C, guchi et al. 2015; Nishino et al. 2015). A significant increase and D) were set 16 km from the FPO site and CTD profiles in nutrients from deeper water was found at the surface due were conducted between the FPO site samplings. On most to the weakening of the pycnocline. Subsequently, both the CTD casts, a multi-spectral excitation/emission fluorom- chlorophyll-a (chla) fraction attributed to large-celled phyto- eter (Multi-Exciter, JFE-Advantech Co. Ltd.) was attached plankton, and the depth-integrated primary production were to the CTD frame to measure the vertical profile of chla also found to have increased. Yokoi et al. (2016) examined fluorescence. the taxonomic changes in large phytoplankton (> 20 µm) and Water samples were collected using a clean plastic bucket microzooplankton communities, and reported a remarkable for surface samples and 12-L Niskin-X bottles for each 1 3 Polar Biology (2018) 41:1279–1295 1281 Fig. 1 a Location of the sampling domain colored with climatologic mean chla derived by Aqua-MODIS (2003–2015). The central station (star) is the main FPO site where observa- tions were conducted every 6 h. Sub-FPO stations 16 km from the FPO site are denoted A, B, C, and D (circles). b Time series of satellite-derived chla at the FPO site from 2003 to 2015. Color scale indicates year of the data plots standard collection depth (5, 10, 20, 30, 40, 45, and 50 m) of PAR using a PRR-800 spectroradioimeter with a freefall and optical depth (38, 14, 7, 4, 1, 0.6% relative to surface profiler (Biospherical Inc.). photosynthetically available radiation, PAR), which were Total chla concentration was measured at each target attached to the CTD/Carousel sampler. The temperature depth by filtering 300 mL of seawater onto Whatman GF/F and salinity (defined by the dimensionless practical salinity glass fiber filter (25 mm diameter) immediately after the scale) were measured using a thermometer and a Guildline sampling. The filters were soaked in DMF for 24–48 h AUTOSAL salinometer, respectively, or by a CTD system (Suzuki and Ishimaru 1990), and the chla concentrations (SeaBird Electronics Inc., SBE 9plus). Nutrient concentra- were measured using a fluorometer (10-AU, Turner Design). tions (nitrate, nitrite, ammonia, phosphate, and silicate) Samples for analysis of phytoplankton pigments (chlo- were determined onboard using auto-analyzers (QuAAtro, rophylls and carotenoids) were collected from six or seven SEAL Analytical) within 24 h after sampling according to depths every FPO observation day at 18:00 UTC (09:00 the GO-SHIP Repeat Hydrography Manual (Hydes et al. local time). Sample water (2.2 L) was filtered onto a GF/F 2010) using the Reference Materials of Nutrients in Sea- filter (25 mm diameter) and stored in liquid nitrogen for water (Aoyama and Hydes 2010; Sato et al. 2010). Total 3 months until analysis in the laboratory. The filter samples alkalinity of the water was measured using a spectrophoto- were soaked and sonicated in 3 mL of N, N-dimethylfor- metric system (NIPPON ANS, Inc.) and the scheme of Yao mamide (DMF) (Suzuki et al. 2002). Then, the extracted and Byrne (1998). We calculated the fraction of freshwater algal pigments were separated by HPLC following the contents (i.e., the sea ice meltwater, f , and other fresh- method of van Heukelem and Thomas (2001). The pig- SIM water, f , carried by the water of Pacific origin) from the ments used for the analysis were chlorophyll-c3 (chlc3), OF values of total alkalinity and salinity following the method chlorophyll-c1 + c2 + magnesium divinyl pheoporphyrin of Yamamoto-Kawai et al. (2005), and the endmembers of a5 monomethyl ester (chlc1 + c2 + MgDVP), peridinin salinity and total alkalinity were obtained from Nishino et al. (peri), 19′-hexanoyloxyfucoxanthin (hex), fucoxanthin (2016). We conducted the measurement of vertical profile (fuco), 19′-butanoyloxyfucoxanthin (but), diadinoxanthin 1 3 1282 Polar Biology (2018) 41:1279–1295 (diadino), alloxanthin (allo), zeaxanthin (zea), prasinox- or phytoplankton community/size composition using EOF anthin (prasi), lutein (lut), chlorophyll-b (chlb), and chla. analysis of spectral optical signatures, such as remote-sens- In this study, we simply interpreted pigment/chla ratios as ing reflectance and absorption (Craig et al. 2012; Bracher proxy of taxonomic composition in combination with cluster et al. 2015; Wang et al. 2015). Similarly, we extracted the analysis (see “statistical analysis” section for details), and dominant modes of excitation spectra from the EOF analysis inferred the temporal and vertical distribution of taxonomic using the MATLAB statistical toolbox (MathWorks Inc.). composition (Hill et al. 2005; Fujiwara et al. 2014). To minimize the effect of chla concentration, which directly affects the fluorescence spectral shape and magnitude, the Measurement of high‑resolution pigment spectra were standardized by subtracting the mean and distribution using Multi‑Exciter dividing by the standard deviation. Then, the variability of the spectral shapes became comparable with the pigment/ High-performance liquid chromatography (HPLC) pigment chla ratios. Chlc3, peri, but, fuco, hex, pras, allo, and chlb signature is used widely to infer phytoplankton taxonomic were chosen for the modeling of their ratios against chla. composition at the class level (reviewed in Jeffrey and Vesk The partial regression coefficients were selected by stepwise 1997; Wright and Jeffrey 2006). Several studies have used method such that all coefficients were statistically significant HPLC pigment signatures to assess the algal taxonomic (t test, p < 0.05). composition in Arctic seas (e.g., Hill et al. 2005; Coupel et al. 2012, 2015; Fujiwara et al. 2014; Alou-Font et al. Chla from satellite ocean color data 2013; Vidussi et al. 2004), including the Chukchi Sea. How- ever, in the absence of direct measurements of phytoplank- Aqua-MODIS level-3 standard mapped images of the ton pigment, taxonomic composition has also been derived spectral remote-sensing reflectance (R ) data (daily and 9 rs based on excitation/emission fluorescence spectra (Yentsch km resolution) were acquired from Goddard Space Flight and Yentsch 1979; Yentsch and Phinney 1985). Both labo- Center/Distributed Active Archive Center, NASA. Using the ratory and in situ observations have successfully revealed continuous time series R data (2003–2015), we derived rs its practical use in identifying or inferring algal community chla concentration from the Arctic OC4L algorithm (Cota structures (e.g., MacIntyre et al. 2010; Houliez et al. 2012; et al. 2004), which was optimized for the optical properties Kuwahara and Leong 2015; Wang et al. 2016). An in situ of phytoplankton in the Arctic Ocean. September climato- multi-excitation fluorometer is a powerful tool for inferring logic mean chla were also computed for the Chukchi Sea vertical distribution of phytoplankton community structure (Fig. 1a). with high resolution. However, the spectral fluorescence method still has some limitations when applied to in situ Statistical analysis observations. For example, the instrument must be calibrated using isolated algal cultures from the study area either before To group samples of similar pigment composition, cluster or after observation (Wang et al. 2016). Fortunately, as men- analysis was conducted for accessory pigment/chla ratios. tioned above, several studies have referred to the relationship This method has been used widely for dividing large vol- between phytoplankton pigment and taxonomic composi- umes of pigment data into several groups (e.g., Hill et al. tion in the Chukchi Sea. Thus, the derivation of pigment 2005; Fujiwara et al. 2014; Goés et al. 2014; Isada et al. signatures is considered sufficient to infer the distribution 2015). For clustering, the unweighted pair group method of phytoplankton communities in the western Arctic Ocean. using arithmetic averages (UPGMA) algorithm with Euclid- The Multi-Exciter (JFE-Advantech Inc.), which measures ian distance was chosen following Isada et al. (2015). Clus- fluorescence response (emission) between 630 and 1000 nm ter analysis was also performed for the pigment/chla ratios excited at nine wavelengths (375, 400, 420, 435, 470, 505, predicted by the Multi-Exciter, but the k-means clustering 525, 570, and 590 nm), is designed to derive temporally method was applied to divide the sample into the same num- or vertically continuous distribution of phytoplankton taxa. ber of cluster as the in situ clusters. The optimum number Instead of calibration with pure culture samples (Chaetoc- of cluster was determined using Calinski–Harabasz index eros sp., Nannochloropsis sp., and Microcystis sp. are used (Calinski and Harabasz 1974). as default reference groups), we developed a model to derive To illustrate the similarity of environmental variables accessory pigment/chla ratios by quantifying the relation- coincident with pigment samples, we conducted a principal ship between fluorescence spectra and HPLC pigments/ component analysis (PCA). Temperature, salinity, total inor- chla ratios obtained at the same depth. The development of ganic nitrogen (TIN = nitrate + nitrite + ammonia), phos- the model was based on an empirical orthogonal function phate, silicate, f , f , and PAR were used as the inputs OF SIM (EOF) analysis of the fluorescence spectra. Several previous of the PCA after standardization. Temporal and vertical studies have successfully modeled water constituents, chla changes of environmental properties were assessed using 1 3 Polar Biology (2018) 41:1279–1295 1283 the PCA scores. These statistical analyses were performed (2015) for the FPO considering the large fresh water fraction using the MATLAB statistical toolbox. attributed to sea ice meltwater (f ). The surface water is SIM categorized as sea ice meltwater (SIMW) (Fig. 2j). The low Grazing rate of Calanus glacialis C5 salinity was coincident with relatively large amounts of f SIM (> 0.02) (Fig. 2 g). In contrast, the lower layer temperature Incubation experiments to assess temporal changes in the was near freezing (< − 1 °C) and its salinity was ~ 32.8. grazing activity of C. glacialis copepodid stage five (C5), The water mass around pycnocline was Bering shelf water which is a key zooplankton species and one of the most (BSW), and bottom water was Pacific winter water (PWW) abundant species within the region, were conducted twice (Fig. 2j). Both the thermocline and the halocline were before the SWE (September 12 and 16) and twice after the located between depths of 20–30 m, where PAR relative to SWE (September 24 and 25) using an on-deck incubator. the surface was 3–10%. Vertical profiles of TIN (Fig. 2d) Prior to the water sampling for the incubation, we deter- and phosphate (Fig. 2e) followed those of temperature and mined the maximum chla fluorescence depths using a CTD salinity with depletion in the upper layer and 2–3 magni- equipped with a fluorometer. Then, seawater was collected tudes higher in the lower layer. Silicate generally exhibited from the layer of maximum chla fluorescence and dispensed a profile similar to the other two nutrients but a subsurface into an acid-rinsed 20-L polyethylene bottle. The sample maximum occasionally appeared around the pycnocline with water was divided into 2.3-L acid-cleaned and Milli-Q relatively higher f (> 0.03) (Fig. 2f, h). OF rinsed polycarbonate bottles, screened with 330 µm mesh to During the FPO term, we observed two occasions when −1 remove mesozooplankton. Calanus glacialis C5 individuals the wind exceeded 10 m s : September 14 (257, Julian were collected using ring net (mouth diameter 80 cm, mesh day) and September 19–21 (262–264, Julian day) (Fig. 2a). size 335 µm) with a 2-L cod-end. Collected zooplankton was Since the major biological responses were found after the retained in ~ 1000 mL of cool filtered seawater prior to being stronger and longer second wind event, we hereafter call placed into an incubation bottle. Active, freshly collected the second event as SWE in this paper. The strong wind was individuals of C. glacialis C5 were chosen and placed in an also recorded at the other four sites around the FPO. We incubation bottle, one individual per bottle. Each incubation found the apparent responses of the marine environmental bottle was then sealed with Parafilm, after confirming no variables to the strong wind. The pycnocline weakened after bubbles were inside the bottle, and tightly capped. We set the first strong wind (Fig. 2b, c) as a result of the enhanced two treatments and two control bottles (without copepods) internal gravity waves (Nishino et al. 2015; Kawaguchi et al. and incubated them for 24 h. During the experiment, the 2015). Furthermore, the temperature and salinity in the temperatures of the bottles were controlled by running sur- upper layer showed a slight decrease and increase, respec- face water over them, and the light condition was adjusted tively. The temperature decrease and salinity increase were to 7% relative to surface PAR to be close to the optical depth accelerated by the strong mixing in the upper layer on Sep- of chla maximum. Chla concentration was measured before tember 19. As the salinity increased, f in the upper layer SIM and after the incubation for total, > 20, 2–20, and < 2 µm decreased (Fig. 2g), and SIMW changed to BSW (Fig. 2j). It particles by filtering 500 mL of water through a 20, 2 µm was remarkable that nutrients in the upper layer showed dif- polycarbonate filters and 0.7 µm GF/F filter (47 mm diam- ferent responses; TIN did not change significantly (Fig. 2d, eter, Whatman), and concentrations were measured using j) but phosphate (Fig. 2e) revealed a significant increase and the fluorometer as described in “Cruise summary and water silicate showed a signic fi ant decrease (Fig. 2f). The response sampling” section. Then, the grazing rates for each size of of chla was similar to that of phosphate, which showed a −3 chla were calculated following the method of Dagg et al. gradual increase from ~ 0.3 to ~ 1.0 mg m in the upper (2006). layer after the SWE (details are described in “Short-term changes in phytoplankton groups” section). Both pre-bloom −1 −1 chla (< 0.5 µg L ) and fall bloom chla (~ 1.0 µg L ) were Results similar magnitude of satellite-derived chla (Fig. 1a, b). Times series of environmental variables Short‑term changes in phytoplankton groups At the beginning of the FPO, the water column was strongly Inference of taxonomic compositions stratified by both temperature and salinity, and it showed a clear two-layered structure (surface mixed layer was ~ 25 m) Cluster analysis was applied to the pigment/chla ratios, and (Fig. 2b, c). The upper layer temperature was > 2.5 °C and we divided the phytoplankton community into three groups the salinity was < 31.5. The water mass classification dur - with dissimilarity criteria = 0.114 (Fig. 3a). The clustered ing the FPO is shown in Fig. 2j modified from Itoh et al. groups showed clear distribution differences with date and 1 3 1284 Polar Biology (2018) 41:1279–1295 Fig. 2 Time series of a wind speed, b temperature, c salinity, d TIN, (j). TIN is indicated by the color scale. Black lines denote the bound- e phosphate, f silicate, g f , h f , and i chla during the FPO term. ary of the different water masses. The water masses are: ACW Alas- SIM OF −1 Red solid line in (a) indicates 10 m s . Black triangles denote onset kan coastal water, BSW Bering shelf water, SIMW Sea ice melt water, of the SWE. Temperature/salinity diagram is also provided in panel PWW Pacific winter water Fig. 3 a Dendrogram of cluster analysis based on the UPGMA and each cluster. Boxplots of chla concentration for the clusters are also Euclidian distance. Colors represent cluster groups divided with dis- provided (right axis) indicating values of median (horizontal bars), 25 similarity criteria = 0.114 (dashed line) and b median fractional con- and 75% quartiles (box ranges), confidence intervals (whiskers), and tribution of accessory pigments to total pigment concentration for outliers (crosses) depth (Fig. 4). Clusters 1 and 2 appeared around the pyc- distribution to the upper layer and replaced cluster 2. Cluster nocline layer (around the chla maximum layer) and surface 3 was found in the lower layer throughout the FPO term. layer before the SWE, respectively. Cluster 1 expanded its We inferred major phytoplankton taxonomic groups for each 1 3 Polar Biology (2018) 41:1279–1295 1285 Fig. 4 Time series of vertical distribution of ratios of a chlc3, b peri, (HPLC cluster) and j pigment/chla ratios predicted by the Multi- c but, d fuco, e pras, f hex, g allo, and h chlb against chla. Vertical Exciter instrument (ME cluster) are also shown. Black triangles distributions of cluster group divided by i in situ pigment/chla ratios denote onset of the SWE cluster using pigment composition and pigment/chla ratios and it spread into the upper layer after the SWE. However, fully referring to previous related studies (Hill et al. 2005; the other pigment/chla ratios are significantly smaller than Fujiwara et al. 2014). The average algal pigment compo- cluster 2, suggesting a small contribution from non-diatom sition is indicated in Fig. 3b and the mean pigment ratios cultures. Conversely, both the hex/fuco (0.059 ± 0.039, are listed in Table 1. Fuco dominated all the clusters (fuco/ n = 45) and but/fuco (0.038 ± 0.031, n = 45) ratios show chla > 0.3), which is a typical characteristic of the Chukchi much smaller values than cluster 2. Such low values of the shelf during fall that has been reported in earlier studies hex/fuco and but/fuco ratios indicate that the fuco of cluster (e.g., Fujiwara et al. 2014; Coupel et al. 2015). However, 1 was derived mainly from diatoms (Hill et al. 2005). We there is large diversity in the secondary pigment/chla ratio determined diatoms with highest biomass group that occu- among the cluster groups. The taxonomic interpretations for pied cluster 1. each cluster are documented as below. Cluster 2 in considered influenced by northern sur- Cluster 1 has the highest chl a concentration face water communities because of the high presence (mean ± standard deviation: 0.667 ± 0.226, n = 45) and of accessory pigments such as but, hex, peri, allo, and the second highest fuco/chla ratio (0.391 ± 0.024, n = 45) fuco. These pigment/chla ratios are the highest among (Fig. 3b and Table 1). The cluster was found in the chla the clusters. Although the fuco/chla ratio (0.308 ± 0.023, maximum layer (around the pycnocline) before the SWE n = 34) is close to that of diatom communities found in Table 1 Means and standard Pigment Cluster 1 Cluster 2 Cluster 3 deviations of pigment/chla and pigment/pigment ratios chlc3/chla 0.101 ± 0.024 0.081 ± 0.019 0.163 ± 0.037 used in this study to infer the [chlc1 + c2 + MgDVP]/chla 0.129 ± 0.019 0.134 ± 0.012 0.177 ± 0.027 phytoplankton community peri/chla 0.020 ± 0.014 0.033 ± 0.023 0.000 ± 0.000 composition of each cluster but/chla 0.015 ± 0.013 0.020 ± 0.009 0.019 ± 0.016 fuco/chla 0.391 ± 0.024 0.308 ± 0.023 0.518 ± 0.049 pras/chla 0.018 ± 0.007 0.029 ± 0.019 0.002 ± 0.003 hex/chla 0.022 ± 0.015 0.049 ± 0.019 0.002 ± 0.004 allo/chla 0.006 ± 0.006 0.012 ± 0.011 0.003 ± 0.006 zea/chla 0.000 ± 0.000 0.000 ± 0.001 0.000 ± 0.001 lut/chla 0.000 ± 0.000 0.000 ± 0.001 0.000 ± 0.001 chlb/chla 0.051 ± 0.032 0.045 ± 0.031 0.003 ± 0.006 hex/fuco 0.059 ± 0.039 0.160 ± 0.062 0.004 ± 0.009 but/fuco 0.038 ± 0.031 0.064 ± 0.027 0.036 ± 0.030 1 3 1286 Polar Biology (2018) 41:1279–1295 the earlier CHEMTAX (CHEMical TAXonomy) studies in the Arctic seas (Vidussi et al. 2004; Coupel et al. 2015), the highest hex/fuco (0.160 ± 0.062, n = 34) and but/fuco (0.064 ± 0.027, n = 34) ratios reveals that fuco could some- how be attributed to haptophytes. The highest peri/chla (0.033 ± 0.023, n = 34), pras/chla (0.029 ± 0.019, n = 34), and hex/chla (0.049 ± 0.019, n = 34) ratios indicate some fractions of dinoflagellates, prasinophytes, and hex-contain- ing haptophytes next to the diatoms in this cluster, indicative of the highest taxonomic diversity of the clusters. Cluster 3 shows the highest fuco/chl a ratio (0.518 ± 0.049, n = 31) but the lowest chla (0.411 ± 0.533, n = 31) of all the clusters. Pigment/chla ratios are absent or extremely low, except diatom-related pigments (chlc3, Fig. 5 Result of principal component analysis (PCA) for the envi- chlc1 + c2 + MgDVP, fuco) and the but/fuco ratio. This ronmental variables where HPLC data were collected. The variance in the data was majorly explained by principal component 1 (PC1) cluster showed very small temporal and vertical change (64.3%) and component 2 (PC2) (17.2%). Plots are distinguished in its distribution and it was located below the pycnocline according to the clustered group and sampled timing; crosses and tri- throughout the FPO term. Because the cluster is adapted angles denote samples taken before and after the SWE, respectively. to low chla and high nutrient concentration, we assumed The vectors of loadings are also shown in the discrete panel; tmp tem- perature, sal salinity, TIN total inorganic nitrogen concentration, phos that the growth of phytoplankton assemblages in cluster 3 phosphate concentration, sil silicate concentration, %PAR percent was strongly light limited. Therefore, we suggest that less PAR relative to surface PAR, f fraction of sea ice melt water, f SIM OF productive senescent diatoms mostly contribute to this clus- fraction of other fresh water ter. Unlike the general pigment composition of this cluster (i.e., high fuco/chla and low chla), we should also note that Cluster 1 first appeared at intermediate depth with moder - some of the highest values of chla (> 1.0) observed during ate nutrient conditions where a subsurface f maximum the FPO belonged to this group (Fig. 3b). A few samples OF occurred (PC1 = 0–2 and PC2 = 0–3). However, its dis- attributed to this group appeared sporadically around the tribution expanded to the surface along with the nutrient pycnocline with the water column chla maximum depth supply after the occurrence of the SWE (PC1 = − 3 to − 1 (261–262, Julian day, Fig. 4i), i.e., the diatoms with high and PC2 = − 2 to 1). Cluster 2 was suited largely to the chla have similar pigment composition to the less productive less saline, high temperature, and low nutrient conditions communities of the lower layer. of the upper layer (PC1 < − 2, PC2 < 0), but it disappeared after the SWE. Cluster 3 was distributed mainly in the cold, Relationship between the cluster group saline, high nutrient, and low irradiance conditions of the and environmental variables bottom layer (PC1 > 2 and PC2 = − 2 to 2). However, as noted in “Inference of taxonomic composition” section, clus- PCA was applied to the environmental variables (tempera- ter 3 also appeared with high chla (> 1.0) with similar pig- ture, salinity, %PAR relative to surface value, TIN, phos- ment composition to the lower community, which is plotted phate, silicate, f , and f ) to enable the visualization of SIM OF in the range of PC1 < 2 and PC2 > 0. the similarities and differences of the environmental condi- tions. PC1 and PC2 explained 64 and 17% of the vertical and temporal environmental variability, respectively. A scatter Interpretation of pigment/chla ratio using plot of PC1 and PC2 colored by the cluster groups of pig- the Multi‑Exciter ment composition is shown in Fig. 5. It is helpful to under- stand the relationship between the suitable habitat for the The EOF analysis of the standardized excitation spectra indi- clustered phytoplankton communities and the environmental cated that the first five modes explained 88.7, 8.29, 1.71, conditions, as well as the vertical and temporal variations 0.58, and 0.34% of the spectral variance, respectively, i.e., of the environmental variables. PC1 was determined prin- their cumulative contribution accounted for 99.62%. Then, cipally by temperature and salinity, which almost explains we selected the target pigments (chlc3, peri, but, fuco, pras, the vertical features of the environmental variables. The hex, allo, and chlb) that were used for the cluster analysis other environmental characteristics modified by the SWE (see “Inference of taxonomic compositions” section) for the are represented by PC2. Warm, fresh, and oligotrophic sur- prediction. Because the accessory pigment/chla ratios had face water at the beginning of the FPO term was occupied a lower limit of 0 and would never exceed 1, we chose a by cluster 2 (PC1 = − 4 to − 2 and PC2 = − 3 to − 1). 1 3 Polar Biology (2018) 41:1279–1295 1287 sigmoidal model to represent the variation of the pigment/ S (n = 1–5) for unknown samples were reconstructed as chla ratios: below: Pigment/chla S = (z() − ())∕l (), (2) n n = 1∕ 1 + exp − + S + S + S + S + S , 0 1 1 2 2 3 3 4 4 5 5 where z(λ), µ(λ), and l (λ) are the standardized fluorescence value, estimated mean value of z(λ), and loadings of EOF (1) where S are the EOF scores and β are the partial regres- mode n for each wavelength λ, respectively. The values of 1–5 0–5 µ(λ) and l (λ) are listed in Table 3. The time series of the sion coefficients listed in Table 2. Thus, the pigment/chla ratios could be derived from Eq. 1. The comparisons and clustered phytoplankton groups predicted by the Multi- Exciter for the FPO site is shown in Fig. 4j. The vertical and statistics of the modeled and in situ pigment/chla ratios are shown in Fig. 6. The root mean square errors (RMSEs) and temporal distributions of the clustered groups at the FPO site match the groups predicted from in situ pigment ratios determination coefficients between the modeled and in situ pigment/chla ratios are also listed (Table 2). The RMSEs well (Fig. 4i, j). To evaluate the spatial generality of the response of range between 0.007 and 0.020, and the values of r exceed 0.5 in all focused pigment/chla ratios except allo (r = 0.05). phytoplankton community structure to the SWE, we investigated the time series of the environmental vari- The Multi-Exciter was also used to assess the short-term changes in phytoplankton community composition with ables (Fig. 7) and pigment/chla ratios at the four sub- FPO stations (A, B, C, and D) (Fig. 8). Atmospheric and high temporal and vertical resolutions. The EOF scores CTD observations without water sampling captured the Table 2 Regression coefficients chlc3 peri but fuco pras hex allo chlb for Eq. (1) β − 2.121 − 5.296 − 4.040 − 0.422 − 4.596 − 4.039 − 5.041 − 4.225 β − 0.318 2.700 − 0.317 − 0.233 1.414 1.241 0.546 2.503 β − 0.565 2.044 0.714 − 0.676 1.781 1.500 n.s. n.s. β − 0.875 − 1.001 − 1.499 n.s. − 1.304 n.s. − 3.278 − 3.095 β − 1.733 6.717 n.s. − 1.255 1.854 1.731 n.s. 4.015 β − 0.979 8.119 − 5.739 n.s. − 2.601 − 5.154 n.s. n.s. r 0.74 0.72 0.63 0.67 0.71 0.74 0.05 0.54 RMSE 0.023 0.010 0.008 0.051 0.008 0.012 0.007 0.018 Coefficients that satisfied p < 0.05 (t test) were selected. n.s. denotes no significant coefficient was obtained. r and RMSE between in situ and predicted pigment/chla ratio are also provided Fig. 6 Comparison of HPLC measured and Multi-Exciter predicted ratios of chlc3, peri, but, fuco, pras, hex, allo, and chlb against chla. Dashed lines indicate 1:1 line. Statistics between HPLC and reconstructed pigment/chla ratios are also shown in Table 2 1 3 1288 Polar Biology (2018) 41:1279–1295 Table 3 Loadings of the 1st λ (nm) l l l l l µ 1 2 3 4 5 to 5th EOF mode (l ) and 1–5 estimated mean of standardized 375 − 0.371 − 0.190 − 0.258 − 0.106 − 0.430 − 0.106 fluorescence value (µ) for 400 − 0.531 − 0.159 − 0.285 − 0.155 − 0.032 0.319 the Multi-Exciter measured 420 − 0.190 0.029 − 0.183 0.294 0.632 0.887 wavelengths (λ) 435 0.289 0.083 − 0.236 0.587 0.082 1.205 470 0.656 − 0.086 − 0.423 − 0.303 − 0.213 1.153 505 0.117 − 0.138 0.235 − 0.078 0.141 − 0.725 525 0.116 − 0.498 0.534 − 0.252 0.230 − 1.164 570 − 0.046 0.793 0.144 − 0.426 0.116 − 0.448 590 − 0.040 0.167 0.472 0.438 − 0.526 − 1.120 SWE, and water convection was found to have occurred the temperature decrease and salinity increase were found homogenously at all the stations; however, the initial con- at all stations (Fig. 7). The fuco/chla ratio also increased ditions were slightly different. The vertical water mass at all stations but conversely, the hex/chla and pras/chla structure at the near-shelf break sites (FPO, stations A and ratios decreased significantly (Fig. 8). It is notable that low B) (Figs. 2b, c, 7b, c, e, f) was similar, but rather weaker salinity water (< 31.5), probably warmed SIMW (Fig. 2j), stratification and a shallower surface mixed layer depth appeared suddenly in the upper layer at station C during were found at shelf side sites (stations C and D) (Fig. 7 h, 263–265 (Julian day) with a slightly lower temperature i, k, l). The predicted pigment/chla ratios followed the (~ 1 °C) (Fig. 7h, i). A slightly higher hex/chla and pras/ physical structure; a smaller fraction of fuco/chla ratio chla ratios co-occurred with this peculiar water mass was found at the FPO site (Fig. 2d) and at stations A and (Fig. 8h, i). B (Fig. 8a, d) before the SWE compared with stations C and D (Fig. 8g, j). Since the observation sites were set at Change in grazing rate of Calanus glacialis C5 the edge of the shelf region, initial physical and biologi- cal condition showed a weak gradient from shelf to shelf The grazing rates of C. glacialis on different size phyto - break area. However, the SWE seems to have mixed the plankton before and after the SWE are compared in Fig. 9. upper layer homogenously throughout the region and thus, The grazing rates were nearly zero on all size classes before Fig. 7 Time series of wind speed (left panels, a, d, g, j), temperature (center panels, b, e, h, k), and salinity (right panels, c, f, i, l) for the sub- −1 FPO stations (A, B, C, and D). Red solid lines in the panels of wind speed indicate 10 m s and black triangles represent the onset of the SWE 1 3 Polar Biology (2018) 41:1279–1295 1289 Fig. 8 Fuco/chla (left panels, a, d, g, j), hex/chla (center panels, b, e, h, k), and pras/chla (right panels, c, f, i, l) for the sub-FPO stations (A, B, C, and D). Black triangles represent the onset of the SWE (2015) using the gut pigment method is also illustrated in Fig. 9. It shows a very similar response in grazing rate on microphytoplankton, with a significant increase after −1 −1 the SWE (from 1.16 to 2.27 ng pigment ind h , U test, p = 0.010). Discussion Fig. 9 Comparison of grazing rate of Calanus glacialis C5 before Interpretation of taxonomic composition using (BS) and after the SWE (AS). Grazing rates were calculated for chla HPLC pigments attributed to microphytoplankton (GR ), nanophytoplankton > 20 µm (GR ), picophytoplankton (GR ), and total phytoplank- 2–20 µm < 2 µm ton (GR ). Grazing rate calculated by gut pigment (Matsuno et al. Pigment signatures have been used widely to infer the verti- total 2015) is also provided for comparison (GR ) with right axis. gut pig. cal and horizontal distributions of phytoplankton taxonomic The boxplots indicate values of median (horizontal bars), 25 and groups at the class level in the western Arctic Ocean (Hill 75% quartiles (box ranges), confident intervals (whiskers), and outli- et al. 2005; Coupel et al. 2012, 2015; Fujiwara et al. 2014). ers (crosses). Statistical differences between grazing rates before and after the SWE are shown (p-values, U test) Several methods exist for the interpretation of phytoplank- ton groups from accessory pigment concentrations, e.g., multiple regression analysis and the CHEMTAX method the SWE. Remarkable but different responses were found (Wright and Jeffrey 2006). The multiple regression method among the various sizes after the SWE. The most signifi- quantifies the contribution of accessory pigments toward cant change in grazing rate was found on nanophytoplankton chla using several accessory pigments. It is suitable for data (chla ). The median grazing rate on nanophytoplank- where the taxonomic composition is unknown (Wright and 2–20 µm −1 −1 ton decreased from − 0.22 to − 1.92 ng chla ind h (U Jeffrey 2006). However, we were unable to obtain the proper test, p = 0.044). The second largest change was found in equation (not shown). Only few pigments showed statisti- chla , which showed a gradual increase in median value cally significant coefficients, which could be attributed to > 20 µm −1 −1 from 0.33 to 2.99 ng chla ind h (U test, p = 0.065). How- the narrow range of chla variation during the FPO term. ever, there were no significant changes in grazing rates on Thus, a larger range of chla variation is required for proper picophytoplankton (chla ) and total chla (chla ). For chla quantification using accessory pigments. On the other < 2 µm total comparison, the grazing rate measured by Matsuno et al. hand, the CHEMTAX method is not restricted by chla range 1 3 1290 Polar Biology (2018) 41:1279–1295 and it is suitable for the detection of the chla contributions of the northern basin area during late summer to fall (e.g., of even minor pigments (Mackey et al. 1996). However, Yamamoto-Kawai et al. 2005) where non-diatom communi- CHEMTAX is sensitive to the initial specific pigment/chla ties such as prasinophytes, green-algae, and haptophytes are ratios for the target taxa, which vary largely with region and generally predominant (Hill et al. 2005; Sukhanova et al. season, and they have not yet been established for our study 2009; Joo et al. 2012; Coupel et al. 2012; Fujiwara et al. area. In addition, because we unfortunately did not conduct 2014). In such waters, nutrients are generally depleted and microscopic analysis for the detailed identification of phy - small-celled phytoplankton and regenerated production are toplankton species, target taxa are difficult to determine and common (e.g., Sherr et al. 2003; Hill et al. 2005; Matsuno validate. For these reasons, the taxonomic composition was et al. 2014). Community structures for the subsurface chla inferred from the patterns of pigment composition fully maximum were also reported and generally pennate and cen- referring to previous studies of Arctic waters (e.g., Booth tric diatoms were predominant in terms of cell abundance or and Horner 1997; Vidussi et al. 2004; Hill et al. 2005; Sukh- chla concentration (Sukhanova et al. 2009; Joo et al. 2012; anova et al. 2009; Joo et al. 2012; Coupel et al. 2012, 2015; Coupel et al. 2012). Consistent with these past studies, a Fujiwara et al. 2014). higher fuco/chla ratio was found at the depth of chla maxi- mum during the FPO term, which increased after the SWE. Short‑term changes of phytoplankton community An increase in the chla and fuco/chla ratio in the upper structure layer was probably attributed to the remarkable increase in the pennate diatom Cylindrotheca closterium (Yokoi et al. Before the SWE, the water column structure of the FPO sites 2016). The replacement of the algal community of cluster exhibited typical stratified two-layered condition, i.e., a large 2 by cluster 1 in the upper layer could be the consequence fraction of warmed sea ice meltwater in the upper layer and of changes in pigment composition with pulsed production saline Pacific winter water in the lower layer. The strong of pennate diatoms. Such a dramatic increase in pennate stratification of the water column meant that nutrients were diatoms is likely to have contributed to the doubling of the almost depleted in the top layer consequence of primary water column primary production (Nishino et al. 2015). production after the sea ice melt. The SWE enhanced inter- These results also suggest that the phytoplankton community nal waves and wind-induced mixing, which weakened the in the subsurface chla maximum plays an important role for stratification (Kawaguchi et al. 2015; Nishino et al. 2015). the development of the fall bloom phytoplankton owing to The upper layer mixing and subsequent nutrient supply from increased light and nutrient availability via the wind-induced −1 lower layer enhanced surface chla from 0.33 to 0.84 µg L mixing. in average. Such magnitude of surface chla is not normally The relationships between the environmental variables seen around the FPO site from the satellite observation and phytoplankton community structures are visualized well (Fig. 1a). Although we acquired 35 scenes of satellite chla by the PCA plot (Fig. 5). Suitable environmental conditions during Septembers of 2003–2015, we found that only the for each clustered group can be explained approximately −1 4 scenes exceeded 0.8 µg L (Fig. 1b). Therefore, it can by PC1; clusters 1, 2, and 3 are adapted to small, moder- be said that the fall bloom in the shelf region of the central ate, and large PC1, respectively. In contrast, PC2 can be Chukchi Sea is an episodic event. treated as the proxy of pycnocline strength, where the f OF During the FPO term, fuco/chla ratios were always maximum occurs (Nishino et al. 2015). The homogenization higher than other pigment/chla ratios. This characteristic of water properties in the upper layer by the SWE reduced is widely observed in the surface layer of the shelf region PC2, i.e., the stratification weakened and cluster 2 disap- during fall (Fujiwara et al. 2014). Coupel et al. (2012) also peared after the SWE. Such replacement of the upper layer reported that the contribution of fuco to total accessory community of cluster 2 by cluster 1 can be explained by the pigments exceeded 70% on the Chukchi shelf during sum- positive and negative contributions of phosphate and f . SIM mer. Although fuco is commonly used as a proxy of diatom However, TIN and silicate apparently showed very weak pigment (e.g., Jeffrey and Vesk 1997), microscopic analy - contributions to the taxonomic changes in the upper layer. sis suggests that fuco in the surface layer of the Chukchi Because of the immediate use of TIN for primary produc- shelf and Canada Basin originates from nanophytoplank- tion (Nishino et al. 2015), TIN supply from the lower layer ton. During the FPO term, Nishino et al. (2015) reported was difficult to detect. This is why TIN and silicate showed that small phytoplankton (< 20 µm) was predominant in the much weaker contributions to the switching of the surface upper layer before the SWE. Our data revealed hex, pras, and phytoplankton community. Overall, the PCA revealed that peri ratios against chla were larger before the SWE in the temporal changes in the phytoplankton community structure upper layer where a relatively large value of f was found. were concomitant with environmental variations. SIM Such a large value of f is typical of the surface water SIM 1 3 Polar Biology (2018) 41:1279–1295 1291 Multi-Exciter with in situ optimization is applicable for the Interpretation of phytoplankton groups using Multi‑Exciter high-frequency observations and for inferring phytoplank- ton community structure. Recently published analysis using Measurement of multi-spectral excitation/emission fluo- the Multi-Exciter has also demonstrated the effectiveness of in situ optimization using CHEMTAX-derived taxa in the rescence is a rapid and costless method for the determina- tion of phytoplankton taxonomic composition (MacIntyre East China Sea (Wang et al. 2016). It is expected to advance our knowledge of the variability of the temporal and spatial et al. 2010). However, it requires suitable calibration of pigment–taxonomy relationships using pure cultures of the distributions of phytoplankton assemblages by attachment to CTD profilers, moorings, or by application to continuous target species. The Multi-Exciter also requires pre- or post- calibration using the spectral fluorescence features of pure monitoring of surface water. culture samples of the target species (Yoshida et al. 2011). During field sampling, it is difficult to predict those types Response of grazing rate of Calanus glacialis C5 of species that will appear at the observation site. In the case of the western Arctic Ocean, several past studies have Calanoid copepods comprised 60% of the total zooplank- ton abundance during the FPO term. Pseudocalanus spp. already reported the relationship between taxonomy and pigment composition (Hill et al. 2005; Coupel et al. 2012, and C. glacialis were the dominant species of zooplankton (Matsuno et al. 2015), accounting for 60 and 35% of the total 2015). Therefore, the derivation of pigment/chla ratio using the Multi-Exciter could provide important information for copepod abundance, respectively. Considering its large body size and biomass, C. glacialis is considered a key species in inferring and monitoring phytoplankton communities in the western Arctic Ocean. the waters of the Arctic shelves (Conover and Huntley 1991; Lane et al. 2008). It is important to assess their response of The relationships between in situ and predicted pigment/ chla ratios showed good agreement except for allo and chlb feeding activity to the eventual fall bloom. Matsuno et al. (2015) has reported a temporal change in grazing rate and (Fig. 6, Table 2). We should note that the pigments that appear constantly, such as fuco, but, and pras, are better pre- grazing impact of C. glacialis C5 during the FPO using the gut pigment approach. They showed that the grazing dicted than those that appear sporadically, such as allo and chlb. Despite the poorer predictions of chlb/chla and allo/ rate increased significantly with chla after the SWE (from −1 −1 0.11 to 0.18 ng pigment ind day ). In addition to their chla ratios, the cluster analysis applied to the Multi-Exciter- predicted pigment/chla ratios also showed good agreement study, we measured the grazing rate of C. glacialis C5 for three size classes of phytoplankton busing the incubation with the in situ clustered groups, in both their vertical and their temporal patterns (Fig. 4i, j); the timing of the switch- approach. Our results revealed an increase of grazing rate on microphytoplankton (> 20 µm) after the SWE that was ing of surface communities and vertical locations of the clusters were well matched. This is because allo is one of the consistent with the gut pigment approach. Another notable point is that the grazing rate for nanophytoplankton showed major accessory pigments that contribute to chla variability. However, the allo/chla ratio was quite low (~ 0.03 maxi- negative values and a significant decrease after the SWE. A similar phenomenon for nanophytoplankton was reported in mum) compared with other pigments and thus, it did not have much effect on the clustering result. Our results reveal another region under grazing experiment conditions of cope- pods with incubation bottles. For example, Liu and Dagg that interpolative use of the Multi-Exciter for the monitoring of phytoplankton pigment is practical for high-resolution (2003) reported chla of sizes < 5 µm or 5–20 µm increased after adding mesozooplankton, suggesting that mesozoo- observation. Unfortunately, we did not obtain samples inde- pendent of the model development data for the validation plankton grazing caused the removal of microzooplankton, which reduced the grazing pressure on small phytoplankton. of the model; however, such matching of the distribution patterns in cluster groups supports proper use of the model. Similarly, selective grazing of Neocalanus spp. on larger particles in the Pacific Ocean has been found to enhance Moreover, as shown in Fig. 8, every 24 h observation using the Multi-Exciter at the stations 16 km from the FPO site smaller phytoplankton growth in the incubation environment through the cascade effect (Liu et al. 2005; Dagg et al. 2006, also showed significant responses in pigment signatures to the SWE and subsequent environmental changes. That is, 2009). In the study region, Campbell et al. (2009) reported strong food preference of C. glacialis for microzooplankton the initial oceanic conditions at the FPO and sub-FPO sites were different, though changes in phytoplankton community rather than diatoms because of the decline of food quality of the diatoms in post-bloom conditions. Although we did not structure occurred at the four other stations as well as at the FPO site. It suggests that such extrapolative use to confirm measure the grazing rate of microzooplankton, a decrease in the grazing rate on nanophytoplankton after the SWE is the spatial generality of phytoplankton response to environ- mental forcing is also practical. Therefore, we would like to believed to result from reduced microzooplankton grazing pressure. We also would like to note that grazing rates on highlight that the prediction of pigment/chla ratios using the 1 3 1292 Polar Biology (2018) 41:1279–1295 all size classes of phytoplankton were nearly zero before the the post-bloom conditions. We developed a model to enable SWE, indicating the minimal cascade effect and minimal high-frequency measurements of the pigment signature from feeding activity of C. glacialis on both microzooplankton multi-wavelength excitation/emission fluorescence spectra and microphytoplankton. This might be because C. glacialis by quantifying the relationship between in situ pigment prepared for diapause (Matsuno et al. 2015). However, the concentrations and excitation spectra. The changes in phy- fall bloom apparently enhanced the feeding activity of C. toplankton pigment signature were also observed by Multi- glacialis not only on microphytoplankton but also on micro- Exciter. Moreover, the grazing experiment of C. glacialis C5 zooplankton. Estimated food requirements of C. glacialis C5 revealed a significant increase of feeding activity, supporting during the FPO term suggest that the enhanced phytoplank- potential increased consumption of diatoms together with ton biomass is still insufficient to maintain their population microzooplankton, which could affect the success of over - and thus, another food source was suggested by Matsuno wintering and reproduction in the following spring. Thus, et al. (2015) The significant cascade effect found in this the SWEs on the Chukchi shelf during fall have remarkable study supported their suggestion that microzooplankton is impact on both primary and secondary producers (Matsuno likely consumed by C. glacialis C5 together with increased et al. 2015; Yokoi et al. 2016). The occurrence of the fall quantities of microphytoplankton. bloom, or changes in its magnitude and timing, is notewor- We captured the enhanced diatom biomass that is clearly thy to comprehend the recent drastic ecosystem changes in transferred to C. glacialis, which is the key species linking the Arctic Ocean. the primary producers and higher trophic level organisms Acknowledgements We thank the captain, officers, and crews of the in the waters of the Arctic shelves (Søreide et al. 2010). R/V Mirai. We also appreciate the staff of Marine Works Japan Ltd. However, it is still unknown whether the fall bloom posi- and Global Ocean Development, Inc. for their skillful samplings and tively ae ff cts their life cycle. It seems that the fall bloom can analysis of the data. This research was funded by the Japan Society for the Promotion of Science (JSPS) (KAKENHI, 7112771), GRENE provide additional energy for secondary producers before Arctic Climate Change Research Project, and Arctic Challenge for Sus- overwintering. Seasonal succession of phytoplankton spe- tainability (ArCS) Project. cies is reported to have large impact on secondary producers. For example, Leu et al. (2011) reported that yearly changes Open Access This article is distributed under the terms of the Crea- in the seasonal succession of ice algae and phytoplankton tive Commons Attribution 4.0 International License (http://creat iveco cause yearly differences in C. glacialis recruitment and mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- tion, and reproduction in any medium, provided you give appropriate reproduction in the Rijpfjorden (European Arctic shelf). In credit to the original author(s) and the source, provide a link to the the Beaufort Sea, episodic fall blooms have obvious impact Creative Commons license, and indicate if changes were made. on the recruitment of secondary producers (Tremblay et al. 2011). Similarly, yearly changes in the timing of the spring bloom have significant impact on the production of higher trophic level organisms through the food web in the Bering References Sea (reviewed in Hunt et al. 2002, 2011). The impact on secondary producers as a whole remains unknown because Alou-Font E, Mundy CJ, Roy S et al (2013) Snow cover affects ice algal pigment composition in the coastal Arctic Ocean during the grazing experiments were conducted only for C. glacialis spring. Mar Ecol Prog Ser 474:89–104 C5, even though they comprise one-third of the total meso- Aoyama M, Hydes DJ (2010) How do we improve the comparability of zooplankton abundance. As secondary producers can be sen- nutrient measurements? In: Aoyama M, Dickson AG, Hydes DJ, sitive to changes in primary producers, further research is Murata A, Oh JR, Roose P, Woodward EMS (eds) Comparability of nutrients in the world’s ocean. Mother Tank, Tsukuba, pp 1–10 required to comprehend the roles of fall blooms in the entire Ardyna M, Gosselin M, Michel C et al (2011) Environmental forcing ecosystem with reference to the recent increase in the num- of phytoplankton community structure and function in the Cana- ber of stormy days during fall in the Chukchi Sea (e.g., Ser- dian High Arctic: contrasting oligotrophic and eutrophic regions. reze et al. 2000; Zhang et al. 2004; Sepp and Jaagus 2011). Mar Ecol Prog Ser 442:37–57. https://doi.or g/10.3354/meps09378 Ardyna M, Babin M, Gosselin M et al (2014) Recent Arctic Ocean sea ice loss triggers novel fall phytoplankton blooms. Geophys Res Lett. https ://doi.org/10.1002/2014G L0610 47 Summary and conclusions Booth BC, Horner RA (1997) Microalgae on the arctic ocean section, 1994: species abundance and biomass. Deep-Sea Res II 44:1607– 1622. https ://doi.org/10.1016/S0967 -0645(97)00057 -X HPLC pigment signatures clearly captured the changes in Bopp L, Aumont O, Belviso S, Monfray P (2003) Potential impact of phytoplankton community structure with environmental climate change on marine dimethyl sulfide emissions. Tellus B changes triggered by the SWE that occurred during the 55:11–22 FPO term. The SWE was sufficiently strong to supply nutri- Bracher A, Taylor MH, Taylor B et al (2015) Using empirical orthog- onal functions derived from remote-sensing reflectance for the ents from the lower layer to the upper layer and to enhance diatom biomass, which was severely nitrate-limited during 1 3 Polar Biology (2018) 41:1279–1295 1293 prediction of phytoplankton pigment concentrations. Ocean Sci Phaeocystis globosa. J Plankton Res 34:136–151. h t t p s : / / d o i . 11:139–158. https ://doi.org/10.5194/os-11-139-2015org/10.1093/plank t/fbr09 1 Caliński T, Harabasz J (1974) A dendrite method for cluster analysis. Hunt GL Jr, Stabeno P, Walters G et al (2002) Climate change and Commun Stat 3:1–27. https://doi.or g/10.1080/0361092740 88271 control of the southeastern Bering Sea pelagic ecosystem. 01 Deep-Sea Res II 49:5821–5853. https ://doi.org/10.1016/S0967 Campbell R, Sherr E, Ashjian C et al (2009) Mesozooplankton prey -0645(02)00321 -1 preference and grazing impact in the Western Arctic Ocean. Deep- Hunt GL Jr, Coyle KO, Eisner LB et al (2011) Climate impacts on Sea Res II 56:1274–1289 eastern Bering Sea foodwebs: a synthesis of new data and an Comiso JC, Parkinson CL, Gersten R, Stock L (2008) Accelerated assessment of the oscillating control hypothesis. ICES J Mar Sci decline in the Arctic sea ice cover. Geophys Res Lett 35:L01703. 68:1230–1243. https ://doi.org/10.1093/icesj ms/fsr03 6 https ://doi.org/10.1029/2007G L0319 72 Hydes DJ, Aoyama M, Aminot A, et al (2010) Determination of dis- Conover R, Huntley M (1991) Copepods in ice-covered seas-distribu- solved nutrients (N, P, Si) in seawater with high precision and tion, adaptations to seasonally limited food, metabolism, growth inter-comparability using das-segmented continuous flow analys - patterns and life cycle strategies in polar seas. J Mar Syst 2:1–41. ers. In: Hood EM, Sabine CL, Sloyan BM (eds) The GO-SHIP https ://doi.org/10.1016/0924-7963(91)90011 -I repeat hydrography manual: a collection of expert reports and Cota GF, Wang J, Comiso JC (2004) Transformation of global sat- guidelines, IOCCP report number 14, ICPO publication series ellite chlorophyll retrievals with a regionally tuned algorithm. number 134, UNESCO-IOC, Paris, France. http://www.go-ship. Remote Sens Environ 90:373–377. https ://doi.or g/10.1016/j.org/Hydro Man.html rse.2004.01.005 Inoue J, Yamazaki A, Ono J et al (2015) Additional Arctic observations Coupel P, Jin HY, Joo M et al (2012) Phytoplankton distribution in unu- improve weather and sea-ice forecasts for the Northern Sea Route. sually low sea ice cover over the Pacific Arctic. Biogeosciences Sci Rep 5:16868. https ://doi.org/10.1038/srep1 6868 9:4835–4850. https ://doi.org/10.5194/bg-9-4835-2012 Isada T, Hirawake T, Kobayashi T et al (2015) Hyperspectral opti- Coupel P, Matsuoka A, Ruiz-Pino D et al (2015) Pigment signatures of cal discrimination of phytoplankton community structure in phytoplankton communities in the Beaufort Sea. Biogeosciences Funka Bay and its implications for ocean color remote sensing 12:991–1006. https ://doi.org/10.5194/bg-12-991-2015 of diatoms. Remote Sens Environ 159:134–151. https ://doi. Coyle KO, Eisner LB, Mueter FJ et al (2011) Climate change in the org/10.1016/j.rse.2014.12.006 southeastern Bering Sea: impacts on pollock stocks and impli- Itoh M, Pickart RS, Kikuchi T et al (2015) Water properties, heat and cations for the oscillating control hypothesis. Fish Oceanogr volume fluxes of Pacific water in Barrow Canyon during sum- 20:139–156. https ://doi.org/10.1111/j.1365-2419.2011.00574 .x mer 2010. Deep-Sea Res I 102:43–54. https ://doi.org/10.1016/j. Craig SE, Jones CT, Li WKW et al (2012) Deriving optical metrics of dsr.2015.04.004 coastal phytoplankton biomass from ocean colour. Remote Sens Jeffrey SW, Vesk M (1997) Introduction to marine phytoplankton and Environ 119:72–83. https ://doi.org/10.1016/j.rse.2011.12.007 their pigment signatures. In: Jeffrey SW, Mantoura RFC, Wright Cushing DH (1989) A difference in structure between ecosystems SW (eds) Phytoplankton pigments in oceanography, 1st edn. in strongly stratified waters and in those that are only weakly UNESCO Publishing, Paris, pp 37–84 stratified. J Plankton Res 11:1–13. https ://doi.org/10.1093/plank Joo HM, Lee SH, Jung SW et al (2012) Latitudinal variation of phyto- t/11.1.1 plankton communities in the western Arctic Ocean. Deep-Sea Res Dagg MJ, Liu H, Thomas AC (2006) Effects of mesoscale phytoplank - II 81:3–17. https ://doi.org/10.1016/j.dsr2.2011.06.004 ton variability on the copepods Neocalanus flemingeri and N. Kawaguchi Y, Nishino S, Inoue J (2015) Fixed-point observation of plumchrus in the coastal Gulf of Alaska. Deep-Sea Res I 53:321– mixed layer evolution in the seasonally ice-free Chukchi sea: tur- 332. https ://doi.org/10.1016/j.dsr.2005.09.013 bulent mixing due to gale winds and internal gravity waves. J Phys Dagg M, Strom S, Liu H (2009) High feeding rates on large particles Oceanogr 45:836–853. https: //doi.org/10.1175/JPO-D-14-0149.1 by Neocalanus flemingeri and N. plumchrus, and consequences Kuwahara VS, Leong SCY (2015) Spectral fluorometric characteriza- for phytoplankton community structure in the subarctic Pacific tion of phytoplankton types in the tropical coastal waters of Sin- Ocean. Deep-Sea Res I 56:716–726. https ://doi.org/10.1016/j. gapore. J Exp Mar Biol Ecol 466:1–8. https ://doi.org/10.1016/j. dsr.2008.12.012jembe .2015.01.015 Fujiwara A, Hirawake T, Suzuki K et al (2014) Timing of sea ice retreat Lane PVZ, Llinás L, Smith SL, Pilz D (2008) Zooplankton distribution can alter phytoplankton community structure in the western Arctic in the western Arctic during summer 2002: hydrographic habitats Ocean. Biogeosciences 11:1705–1716. https ://doi.org/10.5194/ and implications for food chain dynamics. J Mar Syst 70:97–133 bg-11-1705-2014 Leu E, Søreide JE, Hessen DO et al (2011) Consequences of chang- Goés JI, Gomes HDR, Haugen EM et al (2014) Fluorescence, pigment ing sea-ice cover for primary and secondary producers in the and microscopic characterization of Bering Sea phytoplankton European Arctic shelf seas: timing, quantity, and quality. Prog community structure and photosynthetic competency in the pres- Oceanogr 90:18–32. https://doi.or g/10.1016/j.pocean.2011.02.004 ence of a Cold Pool during summer. Deep-Sea Res II 109:84–99. Li WKW, McLaughlin FA, Lovejoy C, Carmack EC (2009) Smallest https ://doi.org/10.1016/j.dsr2.2013.12.004 algae thrive as the Arctic Ocean freshens. Science 326:539. https Grebmeier JM (2012) Shifting Patterns of Life in the Pacific Arctic ://doi.org/10.1126/scien ce.11797 98 and Sub-Arctic Seas. Annu Rev Mar Sci 4:63–78. https ://doi. Lin I-I (2012) Typhoon‐induced phytoplankton blooms and primary org/10.1146/annur ev-marin e-12071 0-10092 6 productivity increase in the western North Pacific subtropi- Grebmeier J, Moore S, Overland J (2010) Biological response to recent cal ocean. J Geophys Res 117((1978–2012)):3039. https ://doi. Pacific Arctic sea ice retreats. Eos Trans 91:161–162org/10.1029/2011j c0076 26 Hill V, Cota G, Stockwell D (2005) Spring and summer phytoplankton Liu H, Dagg M (2003) Interactions between nutrients, phytoplankton communities in the Chukchi and Eastern Beaufort Seas. Deep-Sea growth, and micro- and mesozooplankton grazing in the plume of Res II 52:3369–3385. https://doi.or g/10.1016/j.dsr2.2005.10.010 the Mississippi River. Mar Ecol Prog Ser 258:31–42 Houliez E, Lizon F, Thyssen M et al (2012) Spectral fluorometric Liu H, Dagg MJ, Strom S (2005) Grazing by the calanoid copepod characterization of Haptophyte dynamics using the FluoroProbe: Neocalanus cristatus on the microbial food web in the coastal an application in the eastern English Channel for monitoring Gulf of Alaska. J Plankton Res 27:647–662 1 3 1294 Polar Biology (2018) 41:1279–1295 Lochte K, Ducklow HW, Fasham MJR, Stienens C (1993) Plankton Søreide JE, Leu E, Berge J et al (2010) Timing of blooms, algal food succession and carbon cycling at 47˚N 20˚W during the JGOFS quality and Calanus glacialis reproduction and growth in a chang- North Atlantic bloom experiment. Deep-Sea Res II 40:91–114. ing Arctic. Glob Chang Biol 16:3154–3163. https://doi.or g/10.11 https ://doi.org/10.1016/0967-0645(93)90008 -B11/j.1365-2486.2010.02175 .x MacIntyre HL, Lawrenz E, Richardson TL (2010) Taxonomic discrimi- Stroeve J, Holland MM, Meier W et al (2007) Arctic sea ice decline: nation of phytoplankton by spectral fluorescence. In: Chlorophyll faster than forecast. Geophys Res Lett 34:L09501. https ://doi. a fluorescence in aquatic sciences: methods and applications. org/10.1029/2007G L0297 03 Springer, Dordrecht, pp 129–169 Sukhanova IN, Flint MV, Pautova LA et al (2009) Phytoplankton of Mackey M, Mackey D, Higgins H (1996) CHEMTAX-A program for the western Arctic in the spring and summer of 2002: structure estimating class abundances from chemical markers: application and seasonal changes. Deep-Sea Res II 56:1223–1236. https://doi. to HPLC measurements of phytoplankton. Mar Ecol Prog Ser org/10.1016/j.dsr2.2008.12.030 144:265–283 Sunda W, Kieber DJ, Kiene RP, Huntsman S (2002) An antioxidant Markus T, Stroeve J, Miller J (2009) Recent changes in Arctic sea function for DMSP and DMS in marine algae. Nature 418:317– ice melt onset, freezeup, and melt season length. J Geophys Res 320. https ://doi.org/10.1038/natur e0085 1 114:C12024 Suzuki R, Ishimaru T (1990) An improved method for the determina- Martini KI, Simmons HL, Stoudt CA, Hutchings JK (2014) Near-iner- tion of phytoplankton chlorophyll using N,N-dimethylformamide. tial internal waves and sea ice in the Beaufort Sea. J Phys Ocean- J Oceanogr 46:190–194. https ://doi.org/10.1007/BF021 25580 ogr 44:2212–2234. https ://doi.org/10.1175/JPO-D-13-0160.1 Suzuki K, Minami C, Liu H, Saino T (2002) Temporal and spatial pat- Matsuno K, Yamaguchi A, Hirawake T, Imai I (2011) Year-to-year terns of chemotaxonomic algal pigments in the subarctic Pacific changes of the mesozooplankton community in the Chukchi and the Bering Sea during the early summer of 1999. Deep-Sea Sea during summers of 1991, 1992 and 2007, 2008. Polar Biol Res II 49:5685–5704 34:1349–1360. https ://doi.org/10.1007/s0030 0-011-0988-z Tremblay JÉ, Bélanger S, Barber DG et al (2011) Climate forcing mul- Matsuno K, Ichinomiya M, Yamaguchi A et al (2014) Horizontal tiplies biological productivity in the coastal Arctic Ocean. Geo- distribution of microprotist community structure in the western phys Res Lett 38:L18604. https: //doi.org/10.1029/2011GL 04882 5 Arctic Ocean during late summer and early fall of 2010. Polar Uchimiya M, Motegi C, Nishino S et al (2016) Coupled response of Biol 37:1185–1195. https ://doi.org/10.1007/s0030 0-014-1512-z bacterial production to a wind-induced fall phytoplankton bloom Matsuno K, Yamaguchi A, Nishino S et al (2015) Short-term changes and sediment resuspension in the Chukchi Sea Shelf, Western in the mesozooplankton community and copepod gut pigment Arctic Ocean. Front Mar Sci 3:533. https://doi.or g/10.3389/fmars in the Chukchi Sea in autumn: reflections of a strong wind .2016.00231 event. Biogeoscienes 12:4005–4015. https ://doi.or g/10.5194/ Van Heukelem L, Thomas CS (2001) Computer-assisted high-perfor- bg-12-4005-2015 mance liquid chromatography method development with applica- Nishino S (2013) R/V Mirai cruise report MR13-06. JAMSTEC, tions to the isolation and analysis of phytoplankton pigments. J Yokosuka Chromatogr A 910:31–49 Nishino S, Kikuchi T, Yamamoto-Kawai M et al (2011) Enhancement/ Vidussi F, Roy S, Lovejoy C et al (2004) Spatial and temporal variabil- reduction of biological pump depends on ocean circulation in the ity of the phytoplankton community structure in the North Water sea-ice reduction regions of the Arctic Ocean. J Oceanogr 67:305– Polynya, investigated using pigment biomarkers. Can J Fish Aquat 314. https ://doi.org/10.1007/s1087 2-011-0030-7 Sci 61:2038–2052. https ://doi.org/10.1139/f04-152 Nishino S, Kawaguchi Y, Inoue J et al (2015) Nutrient supply and bio- Wang S, Ishizaka J, Hirawake T et al (2015) Remote estimation of phy- logical response to wind-induced mixing, inertial motion, internal toplankton size fractions using the spectral shape of light absorp- waves, and currents in the northern Chukchi Sea. J Geophys Res tion. Opt Express 23:10301–10318. https ://doi.or g/10.1364/ 120:1975–1992. https ://doi.org/10.1002/2014J C0104 07OE.23.01030 1 Nishino S, Kikuchi T, Fujiwara A et al (2016) Water mass character- Wang S, Xiao C, Ishizaka J et al (2016) Statistical approach for the istics and their temporal changes in a biological hotspot in the retrieval of phytoplankton community structures from in situ fluo- southern Chukchi Sea. Biogeosciences 13:2563–2578. https://doi. rescence measurements. Opt Express 24:23635–23653. https :// org/10.5194/bg-13-2563-2016doi.org/10.1364/OE.24.02363 5 Rainville L, Woodgate RA (2009) Observations of internal wave gener- Wright SW, Jeffrey SW (2006) Pigment markers for phytoplankton ation in the seasonally ice-free Arctic. Geophys Res Lett 36:1487. production. In: Volkman JK (ed) The handbook of environmental https ://doi.org/10.1029/2009G L0412 91 chemistry. Springer-Verlag, Berlin/Heidelberg, pp 71–104 Sato K, Aoyama M, Becker S (2010) Reference materials for nutrients Yamamoto-Kawai M, Tanaka N, Pivovarov S (2005) Freshwater and in seawater as calibration standard solution to keep comparability brine behaviors in the Arctic Ocean deduced from historical for several cruises in the world ocean in 2000s. In: Aoyama M data of δ18O and alkalinity (1929–2002 A.D.). J Geophys Res et al (eds) Comparability of nutrients in the world’s ocean. Mother 110:C10003. https ://doi.org/10.1029/2004J C0027 93 Tank, Tsukuba, pp 43–56 Yao W, Byrne RH (1998) Simplified seawater alkalinity analysis. Sepp M, Jaagus J (2011) Changes in the activity and tracks of Arc- Deep-Sea Res I 45:1383–1392. ht tp s : // do i. org /1 0. 10 16/ S0 96 7 tic cyclones. Clim Chang 105:577–595. https ://doi.org/10.1007/-0637(98)00018 -1 s1058 4-010-9893-7 Yentsch CS, Phinney DA (1985) Spectral fluorescence: an ataxonomic Serreze MC, Walsh JE, Chapin FS et al (2000) Observational evidence tool for studying the structure of phytoplankton populations. J of recent change in the northern high-latitude environment. Clim Plankton Res 7:617–632. https ://doi.org/10.1093/plank t/7.5.617 Chang 46:159–207. https ://doi.org/10.1023/A:10055 04031 923 Yentsch CS, Yentsch CM (1979) Fluorescence spectral signatures-char- Sherr EB, Sherr BF, Wheeler PA, Thompson K (2003) Temporal acterization of phytoplankton populations by the use of excitation and spatial variation in stocks of autotrophic and heterotrophic and emission spectra. J Mar Res 37:471–483 microbes in the upper water column of the central Arctic Ocean. Yokoi N, Matsuno K, Ichinomiya M et al (2016) Short-term changes in Deep-Sea Res I 50:557–571. https ://doi.or g/10.1016/S0967 a microplankton community in the Chukchi Sea during autumn: -0637(03)00031 -1 1 3 Polar Biology (2018) 41:1279–1295 1295 consequences of a strong wind event. Biogeosciences 13:913–923. Zhang X, Walsh JE, Zhang J et al (2004) Climatology and interan- https ://doi.org/10.5194/bg-13-913-2016 nual variability of arctic cyclone activity: 1948–2002. J Clim Yoshida M, Horiuchi T, Nagasawa Y (2011) In situ multi-excitation 17:2300–2317 chlorophyll fluorometer for phytoplankton measurements: tech- Zhao H, Shao J, Han G et al (2015) Inu fl ence of typhoon matsa on phy - nologies and applications beyond conventional fluorometers. toplankton chlorophyll-a off East China. PLoS ONE 10:e0137863. OCEANS’11 MTS/IEEE KONA, Waikoloa, pp 1–4https ://doi.org/10.1371/journ al.pone.01378 63 1 3
Polar Biology – Springer Journals
Published: Feb 20, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera