Dietary records in cosmopolitan copepod Calanus spp. revealed large contribution of dinoflagellates in subarctic Pacific Authors: Hyuntae Choi, Bohyung Choi, Jee-Hoon Kim, Nayeon Park, Haemin Won, Jieun Kim, Jinyoung Jung, Youngju Lee, Eun Jin Yang, Min-Seob Kim, Wonseok Yang, Yeonjung Lee, Dong Han Choi, Jae Hoon Noh, Brian N. Popp, Kyung-Hoon Shin Published: 2026 ABSTRACT Global warming and the resulting surface seawater stratification have led to dominance of small phytoplankton, driving substantial changes in planktonic ecosystem structure. This study investigates the composition and nitrogen sources of phytoplankton and other basal resources by analyzing the cosmopolitan copepod Calanus spp. in western North Pacific. Carbon isotope ratios of amino acids (δ13CAA) in Calanus spp. tissue revealed geographically distinct basal diets: N2-fixing organisms (19.6 ± 5.8%) and heterotrophic bacteria (30.6 ± 5.3%) in the tropical and subtropical Pacific, and dinoflagellates (54.4 ± 10.5%) in the Bering Strait and southern Chukchi Sea. The proportion of dinoflagellates in Calanus spp. diets was strongly correlated with the dietary protist contribution estimated using nitrogen isotope ratios of amino acids (δ15NAA). In addition, nitrogen sources assimilated into total eukaryotic microalgae varied regionally: biologically fixed N2 in the tropical Pacific (29.2 ± 7.9%), while nitrate was the primary source in the Pacific Arctic Ocean (91.4 ± 2.3%). By linking isotopic compositions to ecological function, this study provides insight into how regional nutrient shifts are altering planktonic food webs, with cascading consequences for marine biodiversity, biogeochemical cycling, and ecosystem resilience in a rapidly changing ocean. INTRODUCTION Climate change has significantly altered the composition of the phytoplankton community in marine environments (1). Intensified surface seawater stratification has reduced the dominant cell size of phytoplankton (2, 3), likely due to diminished concentrations of inorganic nutrients in the euphotic zone. These nutrient limitations favor smaller phytoplankton species that are better adapted to low-nutrient conditions (4, 5). Such shifts in phytoplankton size and composition cascade through the food web, influencing the abundance and diversity of protozoa, zooplankton, and higher trophic levels (6). Analysis of 18S rDNA from zooplankton gut contents (7, 8) and ecological modeling studies (9, 10) have demonstrated that spatiotemporal changes in phytoplankton communities directly affect zooplankton dietary composition. Consequently, the structure of basal producer composition and the physicochemical characteristics of oceanic habitats are tightly linked to trophic dynamics and energy transfer efficiency. Understanding these relationships is important for predicting ecosystem responses to ongoing climate-driven changes in marine biogeochemistry and food web structure. The western North Pacific is a region of strong climate feedbacks and rapid oceanographic change (11–14). The northward-flowing Kuroshio Current, which originates east of the Philippines, and forms the Kuroshio Extension after interacting with the Oyashio Current (15), has intensified over the last two decades, partly due to strengthened tropical cyclones (16). Concurrently, increased northward transport of warm Pacific water through the Bering Strait during summer has accelerated sea ice retreat in the western Arctic Ocean (17, 18). These hydrographic shifts strongly influence planktonic ecosystem structure (2, 19, 20), including enhanced heterotrophy in dinoflagellates (21) and the expanding ecological prominence of mixotrophic plankton (22, 23). In addition, contributions from N2-fixing cyanobacteria in the western Bering Sea (24) and Arctic Ocean (25, 26) may be more substantial than previously recognized. Yet, the trophic dynamics of these autotrophic and mixotrophic plankton, characterized by high phylogenetic and functional diversity, remain difficult to resolve, particularly the regional pathways by which inorganic nutrients are transformed and transferred to consumers. Compound-specific stable isotope analysis of amino acids (AA) provides a powerful tool for disentangling these pathways (27, 28). The δ13C values of essential AA in consumer tissues remain largely unchanged through food webs because internal biosynthetic pathways of essential AA are absent in heterotrophs (27, 29). This biochemical consistency enables stable isotope fingerprinting to identify basal dietary sources such as bacteria, microalgae, and terrestrial plants based on the distinct δ13C patterns in essential AA (27, 30). These isotopic patterns can further resolve marine phytoplankton taxa, including diatoms, dinoflagellates, cyanobacteria, and prasinophytes (31, 32). In parallel, δ15N values of amino acids (δ15NAA) distinguish between source and trophic AA, allowing reconstruction of inorganic nitrogen pathways and precise trophic position estimates (33–35). Because δ15N values differ between nitrogen gas (N2, ∼0‰) and nitrate (NO3−, ∼5 to 8‰) (14, 36), and because primary producers fractionate these sources to varying degrees, δ15NAA values can reveal the dominant nitrogen source supporting production. For example, enhanced stratification in the Baltic Sea increased N2 fixation, producing lower δ15NAA values in zooplankton relative to nitrate-dominated eutrophic sites (36). Moreover, zooplankton trophic position (TP) is sensitive to the size structure of basal producers, with nano- and pico-phytoplankton supporting longer food chains than microphytoplankton (36–38). Despite these advances, most studies have applied δ13CAA and δ15NAA separately, limiting our ability to resolve how primary producer diversity and inorganic nitrogen sources jointly shape planktonic trophodynamics. A dual-isotope approach can identify dominant phytoplankton subgroups and simultaneously trace the flow of inorganic nitrogen (N2 versus nitrate) into consumers, whether through direct grazing or protistan intermediaries. To capture these dynamics across contrasting oceanographic regimes, we analyzed the cosmopolitan calanoid copepod Calanus spp., a filter-feeding zooplankton whose diet integrates regional phytoplankton and protist communities (7, 8). Calanus exhibits substantial trophic plasticity (37, 39) and plays a key role in transferring energy to zooplanktivorous fishes and sea birds (40), making it an ideal indicator of basal resource composition and nitrogen pathways. We sampled across a broad latitudinal gradient in the western Pacific during summer and autumn, spanning oligotrophic tropical and subtropical waters, high-nutrient low-chlorophyll (HNLC) regions of the subarctic Pacific and Bering Sea basin, and highly productive shelf regions of the Bering Sea, Bering Strait, and southern Chukchi Sea. Our objective was to quantify how nutrient sources and primary-producer communities shape zooplankton diets and nitrogen transfer pathways across these dynamic regions. We used δ13CAA values to estimate contributions of cyanobacteria, eukaryotic microalgae, and bacteria to Calanus diets, and δ15NAA values to distinguish protistan-mediated versus direct grazing pathways and to infer the δ15N values of inorganic nitrogen sources assimilated by N2 fixers and eukaryotic microalgae. By integrating these dual isotopic tracers, we test the hypothesis that regional variation in nutrient regimes drives distinct pathways of carbon and nitrogen transfer from inorganic sources to zooplankton AA across the western North Pacific and Pacific Arctic (Fig. 1). This dual-isotope framework provides an integrated reconstruction of both basal producer identity and inorganic nitrogen provenance in zooplankton AA, offering a mechanistic lens for resolving shifts in planktonic trophodynamics. Study sampling sites. Detailed information on cruises and corresponding sampling sites is described in table S1. Dashed lines indicate boundary between regions based on previous literature (4, 19, 52, 63). Symbol color indicates the year of sampling. The map was created by QGIS Desktop 4.0.2 (106). RESULTS Biogeochemical settings of sampling sites The tropical and subtropical Pacific Ocean exhibited oligotrophic conditions, while the Kuroshio Extension showed occasional nutrient enrichment around 35°N (figs. S1 and S2, A and B). Nutrient concentrations peaked from the subarctic Pacific to Bering Sea Basin, then declined northward from the Bering Strait (figs. S1 and S2, A and B). Surface chlorophyll-a (Chl-a) concentrations derived from MODIS (Moderate Resolution Imaging Spectroradiometer) remote sensing data (fig. S2C) were regionally variable (Kruskal-Wallis test, P < 0.05, table S10) and reflected these gradients: low values (0.03 to 0.13 mg/m3) in the tropical and subtropical Pacific Ocean, intermediate levels (0.23 to 1.31 mg/m3) in subarctic Pacific and Bering Sea Basin, and high concentrations (0.35 to 7.86 mg/m3) in the Bering Strait and southern Chukchi Sea (Spearman’s r = 0.841, P < 0.05, table S13). Regional variations in phosphate inventory, nitrate inventory, and N* were significant across all sampling sites [one-way analysis of variance (ANOVA) or Kruskal-Wallis test, P < 0.05 for all three parameters, table S10]. Among these, only N* exhibited a clear latitudinal trend, showing a strong negative correlation with latitude (Spearman’s r = −0.744, P < 0.05 for the 0 to 100 m dataset, table S13). Inorganic nutrient inventories from the subarctic Pacific to southern Chukchi Sea, measured across multiple cruises (ARA06A/06B and ARA08A/08B), showed no detectable interannual differences but were distinctly regional in character: The subarctic Pacific and Bering Sea were clearly differentiated from the Bering Strait and southern Chukchi Sea (two-way ANOVA, table S12). Heterotrophic bacterial abundances in surface seawater (fig. S3) were generally below 50 × 104 cells/ml in the tropical and subtropical Pacific, but exceeded this threshold in most high-latitude regions (Kruskal-Wallis test, P < 0.05, and table S10). Notably, Synechococcus was detected not only in the tropical and subtropical waters but also at several sites in the southern Chukchi Sea (fig. S3), indicating its broader ecological range (one-way ANOVA, P = 0.127, table S10). Diatoms dominated the eukaryotic microalgae community from the subarctic Pacific to the southern Chukchi Sea with insignificant distribution among regions (one-way ANOVA, P = 0.654, table S10). However, several sampling sites had comparable or even greater proportions of dinoflagellates (autotrophic and mixotrophic) relative to diatoms (fig. S3). The regional variation was not significant in autotrophic dinoflagellates (Kruskal-Wallis test, P = 0.137, table S10) but significant in heterotrophic dinoflagellates (Kruskal-Wallis test, P < 0.05, table S10). Ciliate abundances (fig. S3) remained uniform across these northern regions (Kruskal-Wallis test, P = 0.168 and table S10). Stable isotope fingerprinting using carbon isotope ratios of AA Stable isotope fingerprinting based on δ13C values of essential AA in Calanus spp. was compared with six distinct endmembers representing basal primary producers: diatoms, dinoflagellates, other eukaryotic microalgae, heterotrophic bacteria, N2-fixing cyanobacteria, and non–N2-fixing cyanobacteria (fig. S4 and table S3). The essential AA δ13C fingerprints of the potential source groups using linear discriminant analysis (LDA) exhibited 93.88% of total reclassification success (table S5). The correct classification percentage was the lowest in N2-fixing cyanobacteria (50.00%), which showed a large overlapping area of 95% confidence ellipse with other eukaryotic microalgae (fig. S4). LD1 primary separated heterotrophic bacteria, diatoms, dinoflagellates, and other eukaryotic microalgae and accounted for 51.64% of the variance, whereas LD2 distinguished non–N2-fixing cyanobacteria and N2-fixing cyanobacteria and explained an additional 36.17% of the variance. Calanus spp. centered on the plot and located among these six endmembers. Using a Bayesian stable isotope mixing model (MixSIAR), we quantified the relative contributions of these basal sources to Calanus spp. diets across the study region (Fig. 2 and table S6). Credible intervals (Q1 to Q3) for dietary contributions were relatively broad for endmembers that overlap in the LDA plot (fig. S4): 4.0 to 20.4% for diatom, 4.7 to 13.9% for dinoflagellates, 3.4 to 15.9% for other eukaryotic microalgae, and 3.4 to 17.2% for N2-fixing cyanobacteria (fig. S5). These intervals are comparable to those in previous marine food web applications of MixSIAR (41, 42). Although the correct classification rate for N2-fixing cyanobacteria was relatively low, this uncertainty is unlikely to affect confidence in the broad regional patterns we report. The dietary contribution of N2-fixing cyanobacteria was also similar to the MixSIAR result using a conventional endmember dataset (fig. S6), suggesting that the regional pattern of N2-fixing cyanobacteria is reliable. Result of Bayesian mixing model (MixSIAR) using AA δ13C fingerprinting. Sampling sites were aligned with latitude. Endmember information is described in table S3 and fig. S4. The total contribution of eukaryotic microalgae (diatoms, dinoflagellates, and other eukaryotic microalgae) to Calanus spp. gradually increased with latitude from 47.2 ± 5.7% in the tropical and subtropical Pacific to 74.3 ± 3.8% in the southern Chukchi Sea (Fig. 2 and tables S6 and S13). This trend was mirrored by heterotrophic bacteria and N2-fixing cyanobacteria, which accounted for 30.6 ± 5.3% and 19.6 ± 5.8%, respectively, in low-latitude regions, but declined to 16.3 ± 4.0% and 7.1 ± 1.9% in the southern Chukchi Sea (Fig. 2 and tables S6 and S13). Dinoflagellates, N2-fixing cyanobacteria, and heterotrophic bacteria exhibited significant regional variation (one-way ANOVA, P < 0.05 for all three parameters, table S10), with their contributions at low-latitude sites clearly distinguishable in Tukey’s post hoc comparisons (table S11). In contrast, diatoms showed no regional differences (one-way ANOVA, P = 0.779), nor did other eukaryotic microalgae (Kruskal-Wallis test, P = 0.175) or non–N2-fixing cyanobacteria (Kruskal-Wallis test, P < 0.109) (table S10). None of the endmembers displayed interannual variation between 2015 and 2017 across the subarctic Pacific to the southern Chukchi Sea (two-way ANOVA, P > 0.05, table S12). In the tropical and subtropical Pacific, dinoflagellates (19.2 ± 3.7%) and diatoms (17.5 ± 4.1%) contributed similarly to Calanus spp. diets (Fig. 2 and table S6). However, in the southern Chukchi Sea, dinoflagellates dominated dietary composition, contributing 52.0 ± 5.4%, while diatom contributions remained relatively stable at 15.1 ± 5.3%. Notably, dinoflagellates contributions were positively correlated with surface Chl-a concentrations (R2 = 0.648, P < 0.001), whereas other basal resources showed no such relationship (fig. S7). TP estimated by nitrogen isotope ratios of AA TP values of Calanus spp. were estimated using two complementary approaches: TPGlu, which reflects metazoan trophic transfer, and TPAla, which incorporates both metazoan and protistan pathways (43–45). TPGlu values ranged widely from 1.9 to 3.2 (Fig. 3A), while TPAla values were consistently higher, ranging from 2.9 to 3.8 (Fig. 3B). Although neither TPGlu and TPAla exhibited clear spatial variation (Kruskal-Wallis, P = 0.108 and one-way ANOVA, P = 0.092, respectively, table S10), the difference between them (ΔTP) and the corresponding microbial percentages, which indicate the number of protistan trophic steps (45), showed a pronounced increase toward the Arctic Ocean (Spearman’s r = 0.497 and 0.548, respectively, P < 0.05 for both, table S13). Both ΔTP and microbial percentages were positively correlated with the contribution of dietary dinoflagellates (fig. S7). Among all trophic indices (TPGlu, TPAla, ΔTP, and microbial percentages), interannual variations were detected only in microbial percentages between 2015 and 2017 from the subarctic Pacific to the southern Chukchi Sea (two-way ANOVA, P < 0.05, table S12). Trophic indices of Calanus spp. TPGlu (A), TPAla (B), ΔTP (difference between TPAla and TPGlu) (C), and microbial percentages (D) of sampling sites. Datasets were plotted as whole data (gray circle), mean (white diamond), and boxplot components including median (thick line), box (Q1 to Q3), and whisker (5 to 95%). Symbol color indicates the year of sampling. Details of the Kruskal-Wallis test and one-way analysis of variance (ANOVA) are described in tables S11 and S12. Nitrogen sources assimilated by total eukaryotic microalgae Among individual AA, phenylalanine in consumers serves as an effective proxy for the δ15N value of primary producers, owing to its minimal trophic enrichment (0.4 ± 0.5‰) during transfer through the food web (46). To improve accuracy, the nitrogen isotope ratio of the basal diet source (δ15NBase) can be estimated by correcting for trophic enrichment in the consumers, offering a more reliable indicator than the direct use of δ15N values of phenylalanine alone (47). In Calanus spp., δ15NBase values significantly varied among regions (one-way ANOVA, P < 0.05, tables S10 and S11) and were strongly correlated with the relative dietary proportions of total eukaryotic microalgae and N2-fixing cyanobacteria, as determined by MixSIAR (Fig. 4, A and B). These values also showed a significant correlation with N* values (Fig. 4C). Regression results using δ15NBase values with other variables. Total eukaryotic microalgae (A) and N2-fixing cyanobacteria (B) estimated using MixSIAR results. Average excessive DIN parameter (N*) from surface to 100-m water depth (C). Symbol color indicates sampling site latitude. Gray-shaded areas indicate 95% confidence intervals of the regression lines. To identify the source of dissolved inorganic nitrogen (DIN) assimilated into Calanus spp. tissues, we estimated the δ15NDIN_assimilated values using isotope mass balance informed by the relative dietary proportions of total eukaryotic microalgae (Fig. 5A) and N2-fixing cyanobacteria (Fig. 5C). The approach aimed to clarify nitrogen assimilation pathways at each study site, including tropical and subtropical regions where nitrate is scarce in the surface mixed layer. Regional averages of MixSIAR results and nitrogen baseline estimated. Proportions of total eukaryotic microalgae (diatoms, dinoflagellates, and other eukaryotic microalgae) (A) and N2-fixing cyanobacteria (C) estimated using MixSIAR. δ15NBase (B) and δ15NDIN_estimated values (D) estimated in this study. Orange dashed lines indicate δ15N values of nitrate obtained in previous studies (tropical and subtropical Pacific, Kuroshio Extension, Subarctic Pacific, and Bering Sea) (14, 53) and this study (Bering Strait and southern Chukchi Sea). Datasets were plotted as whole data (gray circle), mean (white diamond), and boxplot components including median (thick line), box (Q1 to Q3), and whisker (5 to 95%). Symbol color indicates the year of sampling. Details of the Kruskal-Wallis test and one-way ANOVA are described in tables S11 and S12. In the tropical and subtropical Pacific, δ15NDIN_assimilated values were slightly elevated (by 0.3 to 0.6‰) relative to δ15NBase values, whereas in the Bering Strait and the southern Chukchi Sea, the two metrics were nearly identical (Fig. 5, B and D). The latitudinal trend in δ15NDIN_assimilated values was significant (Spearman’s r = 0.793, P < 0.05, table S13) and slightly stronger than that observed for δ15NBase values (Spearman’s r = 0.743, P < 0.05, table S13). When compared with the previously reported nitrate isotope values (δ15NNO3) measured in summer (14), δ15NDIN_assimilated values in low-latitude regions diverged from subsurface δ15NNO3 values (4.9 ± 0.8‰ at 100 to 200 m) and instead aligned more closely with the δ15N values of atmospheric N2 (∼0‰) (Fig. 5D). This pattern was consistent across the subarctic Pacific (0.1 ± 1.1‰) and Bering Sea (1.6 ± 2.1‰), despite samples for δ15NNO3 measurements taken from shallower depths (<100 m). In contrast, δ15NDIN_assimilated values in the Bering Strait (3.4 ± 1.9‰) and southern Chukchi Sea (4.3 ± 1.2‰) more closely matched δ15NNO3 values measured at surface to 50 m (Fig. 5D). This regional variation was significant, distinguishing productive regions (Bering Strait and southern Chukchi Sea) from oligotrophic and HNLC regions (one-way ANOVA, P < 0.05, tables S10 and S11). However, neither δ15NBase nor δ15NDIN_assimilated showed interannual variation between 2015 and 2017 (ARA06A/06B and ARA08A/08B, respectively) across the subarctic Pacific to the southern Chukchi Sea (two-way ANOVA, P > 0.05, table S12). DISCUSSION This study applied compound-specific stable isotope analysis to resolve regional variations in planktonic ecosystem structure. Using MixSIAR, we extracted basal producer information at the phytoplankton subgroup level from δ13CAA values of Calanus spp., leveraging their filter-feeding ecology. In parallel, TPs estimated from δ15NAA values of Calanus spp. captured the dietary contribution of protists. By integrating δ13CAA and δ15NAA values, we introduce a new metric, δ15NDIN_assimilated, to trace nitrogen assimilation and transfer pathway from inorganic sources to consumers. We then examined geographical patterns emerging from these isotopic indicators, with particular attention to the elevated dinoflagellate contributions to Calanus spp. diets at high latitudes and the ecological mechanisms underlying this pattern. Geographical variations in basal diet sources The relative contributions of the basal producers to Calanus spp. diets were shaped by distinct marine biogeochemical conditions across the study region. The tropical and subtropical Pacific, characterized by strong surface stratification and oligotrophic conditions, exhibited low phytoplankton biomass and a dominance of small-sized phytoplankton (11, 48, 49), as reflected in the inorganic nutrient profiles (figs. S1 and S2, A and B). N2-fixing cyanobacteria play a key role in such oligotrophic systems by assimilating atmospheric N2 as “new” nitrogen (50), which can be mineralized by heterotrophic bacteria into DIN and subsequently used by phytoplankton (51). In these regions, total eukaryotic microalgae (diatoms, dinoflagellates, and other eukaryotic microalgae) and N2-fixing cyanobacteria contributed 47.2 ± 5.7% and 19.6 ± 5.8% to the basal diet of Calanus spp., respectively (Fig. 2). This coexistence of multiple primary producer groups in the tropical Pacific is consistent with previous observations, showing the dominance of cyanobacteria, diatom, and dinoflagellates in this region (11). In contrast, the Kuroshio Extension and subarctic Pacific showed increased contributions from eukaryotic microalgae (>70%), while N2-fixing bacteria and heterotrophic bacteria decreased to <20% (Fig. 2). Nutrient mixing and vertical nutrient supply from the Kuroshio and Oyashio currents occasionally enhance Chl-a concentrations above 35°N (52), yet nondiatom phytoplankton dominate even during spring blooms, likely due to silicate and iron limitations (4). Offshore subarctic Pacific and Bering Sea Basin, classified as HNLC regions, are similarly constrained by iron availability (13, 53), suppressing diatom productivity and favoring other eukaryotic microalgae, which are reflected in Calanus spp. diets. The Bering Sea shelf, Bering Strait, and southern Chukchi Sea are biological hotspots with high biomass and diversity driven by nutrient-rich waters (54), supporting a high abundance of phytoplankton in ice-edge zones (19). During our cruises, however, open water conditions and low inorganic nutrient concentrations (fig. S2) indicate that sampling occurred in the postbloom period. Under these nutrient-deficient conditions, dinoflagellates and other small phytoplankton typically increase in relative abundance, even though diatoms still dominate total biomass (19). Thus, dinoflagellates would emerge as the dominant dietary contributor to Calanus spp. in these northern regions, with total eukaryotic microalgae comprising ∼80% of the Calanus spp. diet (Fig. 2). The relatively low dietary contribution of diatoms may reflect such factors as their slower digestibility (55), diel shifts in feeding behavior, or selective grazing by zooplankton (56). The contribution of N2 fixers to the Calanus spp. diet ranged from 3.8 to 23.2%, even in the Arctic Ocean, consistent with prior observations (24, 25). Symbiotic relationships between N2 fixers and protozoa may facilitate nitrogen transfer to mesozooplankton in polar regions (57), bypassing direct grazing limitations due to cyanobacterial size. Trophic dynamics of Calanus spp. In oligotrophic systems dominated by small phytoplankton, energy transfer to mesozooplankton likely requires intermediate protistan steps (6, 9), resulting in regional variation in microbial trophic pathways (fig. S7). The observed range of TPGlu values (1.9 to 3.2) for Calanus spp. reflects their omnivorous particle feeding, consistent with previous studies (37, 38, 58). The differences between TPGlu and TPAla values (ΔTP) along with microbial percentages can quantify the contribution of protistan trophic transfer to the food chain (44, 45, 59). Among all endmembers, dietary contributions of dinoflagellates were strongly and positively correlated with surface Chl-a concentrations (fig. S7) and ΔTP and microbial percentage (fig. S8). Such a result suggests that mixotrophic and heterotrophic protists play a significant role in mediating energy flow in high-latitude ecosystems. In contrast, these indices were negatively correlated with the contributions of N2-fixing cyanobacteria and heterotrophic bacteria (fig. S8), likely indicating a shift away from microbial loop dominance in regions where dinoflagellates prevail. The wide variation in hetero/mixotrophic dinoflagellate contributions is supported by laboratory findings, where copepods fed heterotrophic dinoflagellates showed substantial increases (∼1.0) in TPAla values but negligible changes in TPGlu (44, 60). ΔTP values were <1.0 in tropical and subtropical regions but exceeded 1.0 in subarctic and Arctic sites (Fig. 3C), indicating greater protist consumption at higher latitudes during summer. Elevated Chl-a concentrations (fig. S2C) and bacterial cell abundance (fig. S3) in these regions suggest a more complex microbial food web (61). This pattern mirrors findings from the Southeastern Pacific, where coastal upwelling zones exhibited high ΔTP values (>1.0), while oligotrophic offshore areas showed lower values (0.5 to 1.0) (62). In this study, the strong correlation between dietary dinoflagellates and protistan trophic steps in Calanus spp. stresses the role of dinoflagellates in microbial food web dynamics and provides field verification for previous laboratory culture study (44). Microbial percentages varied widely in Kuroshio Extension and HNLC regions, even at similar latitudes, reflecting differences in trophic link complexity. Primary production in these regions is episodically enhanced by nutrients inputs from regional eddy formation (52), leading to patchy phytoplankton distributions (4). Moreover, anthropogenic nitrogen (12) and iron (63) deposition in the western North Pacific Ocean influences primary producer communities, with cascading effects on grazers. Future research should aim to resolve planktonic food web structure at finer temporal and spatial scales, particularly in HNLC regions, using multidisciplinary approaches that integrate isotopic, molecular, and ecological data. Nitrogen source identification of basal diets recorded in Calanus spp. The increase in δ15NBase values toward the Arctic Ocean (Fig. 5B) suggests a latitudinal shift in the dominant nitrogen source from biologically fixed gaseous N2 in oligotrophic regions to nitrate in more productive northern waters. This transition aligns with patterns in N* values, which are calculated from nitrate and phosphate concentrations (64). Positive N* values indicate nitrogen addition via fixation or atmospheric deposition, while negative values reflect nitrate removal through processes such as denitrification (65). Negative N* values, indicative of nitrogen-deficient waters, are commonly observed in the subtropical North Pacific, subarctic regions (14), and the Chukchi Sea (66). The observed correlation between δ15NBase and N* values (Fig. 4C) can indicate the tight coupling between DIN dynamics and δ15N values of basal food web sources. By integrating AA δ13C fingerprinting, we further resolved the relative contributions of gaseous N2 versus nitrate to the δ15NBase signal. The δ15NDIN_assimilated values—calculated by removing the dietary influence of N2 fixers—provide a refined estimate of the DIN source assimilated into Calanus tissues. In summer of the tropical and subtropical Pacific, δ15NDIN_assimilated values (−0.9 ± 1.2‰) were lower than δ15NNO3 values (4.9 ± 0.8‰, measured at a seawater depth of 100 to 200 m), likely indicating limited nitrate utilization due to strong surface stratification (Fig. 5D). In parallel, HNLC regions such as the Subarctic Pacific and Bering Sea Basin retain surface nutrients due to micronutrient limitations (e.g., dissolved iron; fig. S1), likely allowing partial nitrate assimilation by primary producers in summer (14, 53) resulting in δ15NDIN_assimilated values lower than ambient δ15NNO3 values in the surface seawater. In the Kuroshio Extension, direct comparison of auttumn δ15NDIN_assimilated values with previously reported summer δ15NNO3 values (7.1 ± 1.2‰, within 100 m seawater depth) is difficult (14). However, the consistently low δ15NBase values (∼0‰) estimated from zooplankton in this region (58) and the patchy phytoplankton distributions generated by small-scale vertical nutrient injections in autumn (67), suggest that δ15NDIN_assimilated likely reflects partial nitrate utilization in this region, similar to patterns observed in other HNLC systems. The difference in δ15NDIN_assimilated values between Eddy1 (0.6‰) and Eddy3 (2.7‰) may reflect contrasting hydrographic conditions that alter the relative availability of nitrogen sources (68). In the Bering Strait and southern Chukchi Sea, δ15NDIN_assimilated values (3.4 ± 1.9‰ and 4.3 ± 1.2‰, respectively) closely matched δ15NNO3 values (4.0 ± 0.9‰ and 4.9 ± 0.3‰, respectively, measured at a seawater depth of 0 to 50 m) in summer (Fig. 5D), suggesting active nitrate assimilation by primary producers and direct transfer to Calanus spp. This isotopic pattern mirrors previous findings, where baseline δ15N values for copepods were near 0‰ in the Kuroshio Extension and the North Pacific Ocean, but approached 10‰ in the Bering Sea shelf (58). Regional differences in nitrate assimilation rate further support this interpretation, with low rates (<5 nmol N liter−1 day−1) in subtropical and subarctic Pacific and considerably higher rates (>200 nmol N liter−1 day−1) in the Bering Strait and the southern Chukchi Sea (69). On the basis of these findings, we propose three nitrogen transfer pathways operating during summer (Fig. 6). In oligotrophic regions, organic nitrogen produced by N2 fixers is remineralized by heterotrophic bacteria and subsequently assimilated by eukaryotic microalgae. Calanus spp. therefore obtain energy either indirectly through the microbial loop or directly from eukaryotic microalgae, consistent with the small cell sizes and low dinoflagellate abundance characteristic of these waters (Fig. 6A). Although N2-fixing cyanobacteria and heterotrophic bacteria are central to nitrogen cycling, their overall contribution to the microbial loop is lowest here (18.7 ± 4.6%), likely reflecting low primary production. Conceptual diagram of nitrogen transfer mechanisms in planktonic ecosystems of sampling site groups in summer. Oligotrophic regions including tropical and subtropical Pacific (A), HNLC regions including Subarctic Pacific and Bering Sea (B), and relatively productive regions including Bering Strait and southern Chukchi Sea (C). Color bars represent basal source proportions of each endmember estimated from MixSIAR and material or biological pools and are connected by arrows indicating major transfer pathways. Basal source proportion and ΔTP values are averages of each grouping. In HNLC regions, nitrate supports only partial primary production due to iron limitation, while episodic atmospheric deposition can supply both nitrogen and iron (Fig. 6B). These intermittent inputs enhance the role of eukaryotic microalgae, particularly dinoflagellates in our dataset, and strengthen protistan trophic transfer to Calanus spp. The sporadic nature of nutrient and trace-metal supply likely contributes to the large variability observed in dietary proportions, trophic indices, and δ15N parameters (Figs. 3 and 5). In relatively more productive regions, sustained nitrate assimilation would support high eukaryotic microalgae biomass (Fig. 6C). The high contribution of dietary eukaryotic microalgae rather raises TPAla of Calanus spp., likely because of increased dinoflagellate abundance. Thus, dinoflagellates may contribute not only to the primary production but also to the microbial food chain as mixotrophs in summer oceans. Our results show that combining δ13CAA and δ15NAA values in Calanus spp. provides a robust framework for reconstructing dietary records, including basal producer composition and assimilated nitrogen sources. The cosmopolitan distribution of Calanus spp. make them ideal indicators of spatial variation in nitrogen assimilation and transfer pathways in planktonic food webs. The ecological function of Calanus spp., which can cope with oligotrophic or HNLC condition, should be highlighted by their wide range of available diets and energy transfer to predators, as the δ15NDIN_assimilated metric offers a promising tool for assessing DIN sources in the oligotrophic region, where the concentration and δ15N values of DIN cannot be measured. Comprehensively, these findings establish Calanus spp. as effective integrators of regional trophic and biogeochemical processes, and demonstrate the value of a dual δ13CAA-δ15NAA framework for advancing understanding of how warming-driven changes in stratification and nutrient dynamics reshape marine food webs in changing climate. Implications The continuous increase in surface seawater temperature from 1979 to 2022 is evident in both the subarctic and tropical Pacific (70). The North Pacific Subtropical Gyre and its boundary currents (i.e., Kuroshio Current) have intensified over several decades (71), and warming was accelerated further since 2013, in part due to recurrent marine heatwave events (72). Field observations also show that surface seawater temperatures in the tropical-subtropical Pacific (130°E-135°E, 10°N-30°N) was higher in May 2020 (68) than in August to September 2014 (73). Copepods inhabiting temperate regions generally exhibit high thermal tolerance, but both tropic and polar species remain vulnerable to thermal stress (74). Within our 2015 to 2020 sampling window, the Bering Strait and southern Chukchi Sea experienced elevated summer surface seawater temperature in 2019 and 2020 relative to 2016 to 2018, accompanied by abundances that declined in the Arctic species Calanus glacialis and increased in Pacific species (20). Because C. glacialis is more thermally sensitive than temperate congeners (75), its distribution is expected to shift gradually poleward as seawater temperatures rise (76). Although we did not detect interannual difference in trophic indices for Calanus samples collected between 2015 and 2017 in the subarctic Pacific and Arctic Ocean, such indices in more recent years may differ from those reported here. Expanding the spatial and temporal coverage of δ13CAA and δ15NAA datasets across the North Pacific will therefore be essential for evaluating interannual variability in planktonic ecosystem structure under continued ocean warming. In this study, Calanus spp. includes C. glacialis, C. mashallae (subarctic and Arctic regions), and C. pacificus (temperate and subtropical regions), and were all generally active during our sampling seasons (77–79). Life span varies among species (e.g., 1 to 2 years for C. glacialis and 1 year for C. pacificus), and the diapause timing is strongly influenced by surface seawater temperature (78, 79). In addition, C. glacialis may diapause either once (early-recruited cohorts) or twice (late-recruited cohorts) during their life span (79). Diapause at CIV or CV stages in the mesopelagic layer involves long-term starvation (80), which can influence δ15NAA values and derived trophic indices (81, 82). To minimize these effects, we focused on adult individuals, reducing variability associated with development stage and diapause history. Furthermore, δ15NAA values do not differ significantly among adult males, adult females, and their reproductive eggs of copepods (83, 84), indicating that sex-unidentified adult samples are appropriate for isotopic interpretation. We applied δ13CAA values from phylogenetically distinct groups of phytoplankton (diatoms, dinoflagellates, and cyanobacteria) based on previously reported stable isotope fingerprints (32). These subgroups correlated strongly with eukaryotic microalgae endmembers, which are conventionally treated as a single group (fig. S5, A to C) (27). In addition, N2-fixing cyanobacteria and heterotrophic bacteria proportions were tightly linked (fig. S5, D to F), suggesting that endmembers from multiple studies can be reliably integrated. These findings advance our ability to resolve basal diet sources of Calanus spp. in marine ecosystems using subgroups of phytoplankton. However, the isotopic behavior of mixotrophic plankton remains uncertain. Future studies should consider whether δ13C fingerprinting in mixotrophs resemble those of producers or consumers, as mixotrophic taxa (e.g., dinoflagellate and haptophyte) (22) may blur conventional isotopic boundaries. For example, δ15NAA patterns of microbes grown on inorganic substrates (e.g., ammonia) resemble those of phytoplankton, while those grown on organic substrates (e.g., AA mixtures) resemble consumers (85). Given the minimal 13C enrichment of essential AA during trophic transfer, mixotrophs may reflect their dietary δ13CAA values. Whether these patterns shift within a species or taxonomic groups remain an open question. The low proportion of diatoms in the western North Pacific diet in summer likely reflects nutrient depletion driven by intensified surface stratification. In such environmental conditions, the heterotrophic potential of mixotrophic dinoflagellates increases with rising seawater temperature, allowing them to shift nutritional modes and gain competitive advantages in a warming ocean (21). Investigating microbial food chains associated with mixotrophs will be essential for understanding energy flow and ecosystem structure in future ocean conditions. MATERIALS AND METHODS Sample collection Sampling from the tropical Pacific to Kuroshio Extension was conducted during RV ISABU cruises in October 2017, August 2018, October 2019, and May 2020 (Fig. 1 and table S1). Zooplankton were collected using a multiple opening/closing net and environmental sensing system (MOCNESS, 1 × 1 m2, mesh aperture: 200 μm), which was obliquely towed from 200 m to the surface at each station. Seawater sampling was carried out using a CTD rosette system equipped with 10-liter Niskin bottles at depths from 0 to 500 m. The collected seawater was transferred to 50-ml conical tubes and immediately stored at 4°C for nutrient analysis. Among inorganic nutrient datasets from IBRV ARAON, three cruises (ARA06B, ARA07B, and ARA08B) were derived from a previous study (86). Sampling across the subarctic Pacific to the Arctic Ocean was carried out during the IBRV ARAON cruises in July to August 2015, August 2016, and July to August 2017 (Fig. 1 and table S1). Zooplankton were collected using a bongo net (500 μm mesh size, 60 cm diameter) at a vertical velocity of 60 m/min from a depth of 200 m to the surface at each station. Seawater samples were collected using 10-liter Niskin bottles mounted on a CTD rosette system at depths from 0 to 200 m. The collected seawater was transferred to 50-ml conical tubes and immediately stored at 4°C for nutrient analysis. Sea ice was mostly melted in sampling sites from the Bering Strait and the southern Chukchi Sea in August, as shown in previous studies (19, 20). During the RV ISABU cruises, adult individuals of Calanus spp. were identified and sorted from zooplankton samples. During the IBRV ARAON cruises, adult individuals of C. glacialis (partially mixed with C. marshallae) were identified and sorted from zooplankton samples. All Calanus samples were kept at −20°C during these cruises and freeze-dried after transportation to the laboratory. Because Calanus samples were not completely identified at the species level, we regarded the samples as Calanus spp. in the present study. Detailed zooplankton sampling protocol for both research vessels are described in previous studies (20, 68). The 8-day composite Chl-a data from MODIS with a 0.1° resolution at the study sites were obtained from the NASA Earth Observation website (http://neo.gsfc.nasa.gov). Abundance of heterotrophic bacteria, cyanobacteria, and plankton For samples of RV ISABU cruises, the abundances of heterotrophic bacteria and Synechococcus were determined using a NovoCyte flow cytometer (Agilent, CA, USA), following previously established methods (87, 88). Seawater samples were prefiltered through a 40-μm nylon strainer before analysis. For bacterial counts, samples were stained with SYBR Green I (Sigma, MO, USA) in the dark for approximately 10 min before measurement. For samples collected during the IBRV ARAON cruises, 2-ml aliquots of seawater were fixed with paraformaldehyde at a final concentration of 1% (v/v) for 15 min at room temperature to analyze the abundances of heterotrophic bacteria. Samples were then stored at −80°C until analysis. Before enumeration, samples were thawed and stained with SYBR green I (Molecular Probes) at a final dilution of 1:10,000 (v/v) for 30 min in the dark. Stained samples were analyzed using an Accuri C6 flow cytometer (Becton Dickinson), equipped with a 488-nm, 15-mW air-cooled argon laser, as described in previous studies (88, 89). The instrument’s fluidic system was calibrated for accurate volume measurements according to the manufacturer’s guidelines. Bacterial populations were identified based on side scatter and green fluorescence signals. Flow cytometry data were processed using FlowJo software (Tree Star, www.flowjo.com). To determine the abundance of cyanobacteria, we preserved 10-ml samples of seawater with glutaraldehyde (1% final concentration), then stored them at 4°C before staining and filtration. Subsample of 5 to 10 ml was filtered onto nuclepore filters (0.2 μm pore size, black) and stained with 4′,6-diamidino-2-phenylindole (DAPI; 5 μg ml−1 final concentration). Filters were mounted onto glass slides with immersion oil and coverslips, and were counted with an epifluorescence microscope (Olympus BX 51) at a magnification of 600× to 1000×. To quantify dinoflagellate abundance, 500-ml seawater samples were preserved with glutaraldehyde (1% final concentration) and stored at 4°C before staining and filtration. A 100- to 300-ml subsample was filtered onto black nuclepore filters (8 μm pore size) and stained with proflavin (0.33%) for 1 hour before filtration. During filtration, the samples were drawn down until 5 ml remained in the filtration tower. Concentrated DAPI (50 μg ml−1 final concentration) was then added and allowed to sit briefly (5 s) before filtering the remaining sample until dry, following the protocol of a previous study (90). Filters were mounted onto glass slides with immersion oil and coverslips. Dinoflagellates were counted with an epifluorescence microscope (Olympus BX 51). To determine the abundance of ciliates and diatoms, 500-ml samples of water were preserved with 4% acid Lugol’s iodine solution and subsequently stored in darkness. Preserved samples were allowed to settle in the mass cylinder for at least 48 hours. The upper water layer was then siphoned out, leaving 50 ml. Subsequently, 5 to 20 ml from each sample was settled in sedimentation chambers before enumeration under an inverted compound microscope (Olympus IX 70). Nutrient analysis Inorganic nutrient measurements were conducted onboard following protocols described previously (86, 91). In brief, nitrate and nitrite (NO3− + NO2−), phosphate (PO43−), and silicic acid [Si(OH)4] were measured using a four-channel continuous Auto-Analyzer (QuAAtro, Seal Analytical, Germany) following the Joint Global Ocean Flux Study (JGOFS) guidelines (92). The analytical precision for nitrate and nitrite, phosphate, and silicic acid was ±0.14, ±0.02, and ±0.28 μmol kg−1, respectively. The parameter N* is a quasi-conservative tracer, which refers to the nitrate deficit (negative values) or excess (positive values) with respect to phosphate (64). The N* value is defined in the following equation (64)N∗=([NO3−]−16×[PO43−]+2.9)×0.87(>μmolkg−1) Stable isotope analysis of bulk tissues Stable isotopes were measured using an elemental analyzer (Vario Select, Elementar, Germany) equipped with an isotope ratio mass spectrometer (Vision, Elementar, Germany). The entire body of a single or a few copepods per one sample (n = 1) were homogenized using mortar and pestle. Approximately 0.6 to 1.0 mg dry weight of the sample was used for bulk nitrogen isotope (δ15NBulk) analysis. The nitrogen stable isotope ratio is expressed in per mil (‰; versus atmospheric N2), calculated using the formulaδ=[(Rsample/Rstandard)–1]×1000where R = 15N/14N. IAEA-N-1 [ammonium sulfate (NH4)2SO4, δ15N = 0.4‰] was used as an international standard. The SD of the standard was less than 0.2‰. Nitrogen isotope analysis of nitrate Seawater samples were collected from the Bering Strait and the southern Chukchi Sea in 2017 (stations B1, B2, B3, B10, and B12). The seawater samples were immediately filtered through precombusted GF/F filter paper (25 mm diameter, 0.7 μm pore size). Nitrogen isotope analysis of nitrate was performed using the denitrifier method (93, 94). Briefly, denitrifying bacteria (Pseudomonas chlororaphis) were used to convert nitrate (NO3) into gaseous nitrous oxide (N2O). The resulting N2O is captured in 20-ml sealed vials. The nitrogen isotope ratio of nitrate (δ15N-NO3, ‰, versus atmospheric N2) was measured using an isotope ratio mass spectrometer equipped with a trace gas analyzer equipment (Trace gas–Isoprime 100; Isoprime Ltd., Cheadle, UK). IAEA-NO-3 (potassium nitrate KNO3, δ15N = 4.7‰), USGS34 (potassium nitrate KNO3, δ15N = −1.8‰), and USGS35 (sodium nitrate KNO3, δ15N = 2.7‰) were used as reference substances. The SDs of these standards were 0.1‰ for δ15N-NO3. Carbon and nitrogen isotopes of AA Three to four Calanus spp. were pooled to measure the carbon and nitrogen isotopes of AAs. Each sample was hydrolyzed for 12 to 24 hours with 6 M HCl at 110°C. Hydrophobic content was removed using 6:5 n-hexane:dichloromethane (v/v). The residual HCl was dried under N2 gas flow at 70°C. We used different methods for derivatizing AAs to achieve distinct chromatographic separation. We used N-acetyl methyl ester derivatization for the carbon isotope analysis, adding a few extra carbon atoms to decrease the uncertainty in estimating the original δ13CAA values. The N-pivaloyl-isopropyl ester method was used for nitrogen isotope analysis to achieve high stability and chromatographic resolution. We used a standard mixture of nine AAs [alanine, glycine, valine, leucine, norleucine, aspartic acid, methionine, glutamic acid, and phenylalanine; purchased from SHOKO-Science (Japan) and Indiana University (USA)] to verify the analytical precision of both δ13CAA and δ15NAA values. A review (95) describes the benefits and limitations of various derivatization methods. The AAs were derivatized for carbon isotope analysis with N-acetyl methyl ester using 1:6 acetyl chloride/methanol (v/v) and 1:2:5 acetyl anhydride/triethylamine/acetone (v/v/v). The AA derivatives were extracted with ethyl acetate and stored in a freezer at −20°C until analysis. The detailed experimental procedure was previously described (96). The AAs were derivatized for nitrogen isotope analysis with N-pivaloyl-isopropyl ester using 1:4 thionyl chloride/isopropanol (v/v) and 1:4 pivaloyl chloride/dichloromethane (v/v) as a protocol of the previous study (33). The AA derivatives were extracted with 6:5 n-hexane/dichloromethane (v/v) and redissolved in dichloromethane. Samples were stored in a freezer at −20°C until analysis. The δ13CAA and δ15NAA values (‰, versus VPDB and atmospheric N2, respectively) were measured with a gas chromatograph (HP 6890N, Agilent, USA) connected to a combustor (GC5 Interface, Elementar, Germany) and an isotope ratio mass spectrometer (Isoprime 100, Elementar, Germany). The detailed analytical conditions are described in table S2. The δ13C and δ15N values of the standard AA mixture had an SD of less than 1‰ during the calibration. Stable isotope mixing model LDA was conducted for stable isotope fingerprinting using MASS in R package version 7.3-65 (97). We used the mean-normalized δ13C values of five essential AAs (leucine, isoleucine, valine, threonine, and phenylalanine) of basal diet sources (eukaryotic algae, heterotrophic bacteria, N2-fixing cyanobacteria, and non–N2-fixing cyanobacteria, table S3) from previous studies (27, 31, 32, 98) and Calanus spp. (table S4). The percentage of reclassification success of diet source groups was computed to validate using cv argument in the LDA function (table S5). The accuracy of the LDA output was confirmed using leave-one-out cross-validation given by “caret” in R package version 7.0-1 (99), with reliable parameters (accuracy = 0.95, kappa = 0.93). We quantified the relative contributions of diet sources using the Bayesian stable isotope mixing model (MixSIAR) in R package version 3.1.11 (100, 101). The input datasets were the same with those of LDA. We used a minimum trophic discrimination factor (TDF; 0.1 ± 0.1‰) of the essential AAs between the diet and consumer as previously described (102). The models were run with uninformative Bayesian priors on the endmembers, and the chain length of the MCMC run was 300,000 iterations for the actual run and 200,000 iterations for the burn-in. MixSIAR calculated means, medians, and credible intervals from posterior distribution, and we used mean values for further statistics to consider wide ranges of credible intervals in several endmembers (see Results). We confirmed that the Gelman-Rubin diagnostic values for all variables were less than 1.01 before using the data. Both LDA and MixSIAR analyses were performed using R version 3.6.3 implemented in R Studio version 1.3. Ecological indices TP was estimated using the following equation (33)TP=(δ15NTr−δ15NPhe−βTDF)+1where δ15NTr corresponds to the δ15NGlu or δ15NAla, nitrogen isotope ratios of glutamic acid (Glu) and alanine (Ala), respectively; δ15NPhe is the nitrogen isotope ratio of phenylalanine (Phe); the β values are the empirical values of the isotopic difference between Glu and Phe (βGlu) as well as Ala and Phe (βAla) in primary producers. The TDF is the 15N elevation in Glu relative to Phe (TDFGlu) and of Ala relative to Phe (TDFAla) through trophic transfer. We used 7.6‰ for the TDFGlu value and 3.4‰ for the βGlu value, as previously reported (33). The βAla (3.2‰) and TDFAla (4.5‰) values were determined through multistep laboratory cultivations of ciliates fed on photosynthetic phytoplankton and copepods fed on heterotrophic dinoflagellates (44). The TP values were estimated using two different methods based on the β and TDF values from Glu and Ala. TP values determined using Glu (TPGlu) were used to estimate the trophic transfers involving only metazoans. In contrast, the TP values obtained using Ala (TPAla) included the trophic transfers through protist and metazoan organisms (43, 44). The contribution of protists to the food chain were estimated using the difference between the TPGlu and TPAla values (43)∆TP=TPAla−TPGlu The fractional contribution of protist trophic transfer to TPAla could be calculated as follows (59)microbial percentage(protistan dietary contribution,%)=∆TPTPAla×100 The nitrogen baseline (δ15NBase) was estimated using the TPGlu and δ15NBulk values of the Calanus spp. samples using the following equation (47)δ15NBase=δ15NBulk−(TPGlu−1)×2.3where 2.3 indicates the common TDF value in the δ15NBulk values (103). The δ15NBase values were close to the δ15NBulk values of the primary producers accumulated in the tissues of the consumers. We estimated the fraction of N2 fixers and other phytoplankton using DIN based on the simple mass-balance equation and the δ15NBase valuesδ15NBase×(fN2fixer+fDINfixer)=δ15NN2fixer×fN2fixer+δ15NDINfixer×fDINfixerfN2fixer=PropN2fixing cyanobacteriaPropTotal eukaryotic microalgae+PropN2fixing cyanobacteriafDINfixer=PropTotal eukaryotic microalgaePropTotal eukaryotic microalgae+PropN2fixing cyanobacteriafN2fixer+fDINfixer=1 The equations were transformed into the following equation to estimate the δ15NDINfixer values (δ15NDIN_ assimilated)δ15NDIN_assimilated=δ15NDINfixer=δ15NBase−δ15NN2fixer×fN2fixerfDINfixerwhere fN2fixer and fDINfixer are the proportions of N2-fixing cyanobacteria (PropN2-fixing cyanobacteria) and total eukaryotic microalgae (PropTotal eukaryotic microalgae), respectively, which we estimated using the stable isotope fingerprinting method. We used −1.5‰ as the δ15NN2fixer value, which was obtained from laboratory culture (104). Data analysis Our dataset covers multiple regions, but their sampling seasons were not consistent: oligotrophic ocean (tropical and subtropical Pacific) in summer, transition zone (Kuroshio Extension) in summer to autumn, HNLC region (subarctic Pacific), and relatively productive area (Bering Sea, Bering Strait, and Chukchi Sea) in summer. Although we cannot evaluate seasonal variations in our datasets from multiple regions and months, we could evaluate interannual variation of sampling sites where samples were collected in two different years. Two-way ANOVA was performed to understand the interannual and regional variations in datasets (inorganic nutrients, stable isotope ratios, trophic indices, and MixSIAR results) between 2015 and 2017 (ARA06A/06B and ARA08A/08B, respectively). For identifying differences in datasets (inorganic nutrients, microorganism abundance, Chl-a, stable isotope ratios, trophic indices, and MixSIAR results) among regions, homogeneity of variances was checked by Levene’s test before performing one-way ANOVA. Tukey’s post hoc comparisons were used in the case of significant differences in one-way ANOVA. The Kruskal-Wallis test was used when the variance among these regions was unequal. Spearman’s correlation was used to evaluate the latitudinal trend of the datasets mentioned above. These statistical analyses were computed by paleontological statistics (PAST) 4.03 software (105).