Carbon and nitrogen dynamics in the coastal Japan Sea inferred from 15 years of measurements of stable isotope ratios of Calanus sinicus

Human activities have caused sometimes dramatic changes to the marine environment globally and locally during the last half century. We hypothesized that the carbon and nitrogen stable isotope ratios (δ13C and δ15N) of the copepod Calanus sinicus, one of the dominant secondary producers of North Pacific coastal waters, would record anthropogenic impacts on the coastal environment of the Japan Sea. We monitored these isotope ratios during the spring at four stations in the Japan Sea 15 from 2006 to 2020. The δ13C values ranged from −24.7‰ to −15.0‰ and decreased from the spring bloom (February/March) to the post-bloom (June/July). This monthly variation was attributed to changes in both the physiology of C. sinicus and phytoplankton δ13C. The negative correlation between the δ13C values of C. sinicus and their carbon:nitrogen ratios reflected lipid accumulation by the copepods; high δ13C values were associated with high sea surface chlorophyll a concentrations. The δ15N values ranged from 2.8‰ to 8.8‰. The tendency of the δ15N values to increase from the bloom to post-bloom was 20 attributable to an increase of the δ15N of the phytoplankton associated with nitrate depletion and Rayleigh fractionation. These monthly changes were synchronized among the four stations, but δ13C and δ15N differed significantly between stations. Interannual variations were statistically significant, but there were no significant monotonic trends. Interannual variations differed between δ13C and δ15N as well as among stations. These results suggest that local conditions rather than global-scale trends were the primary determinants of elemental cycles in this coastal ecosystem. 25


Introduction
The ecosystem services provided by coastal areas are more valuable than those provided by any other biome on Earth (Costanza et al., 1997). Coastal ocean ecosystems are important for human activities and have been greatly changed as a result of those activities (Halpern et al., 2008;Doney, 2010). The coastal waters of Japan are known to be among the areas most impacted by human activities (Halpern et al., 2008). The Japan Sea (the Sea of Japan), which has been greatly impacted by human activities 30 and global climate change, is considered to be an oceanic microcosm of the changing global ocean . ocean (Ishizu et al., 2019;Ono, 2021;Kodama et al., 2016). In the surface waters of the Japan Sea, pH and concentrations of phosphate and dissolved oxygen have been decreasing during the last few decades (Ishizu et al., 2019;Ono, 2021;Kodama et al., 2016), whereas anthropogenic inputs of nitrogen from the atmosphere to the Japan Sea have been increasing (Kitayama et 35 al., 2012).
Stable isotope ratios of carbon and nitrogen have been employed as tools to discern both elemental cycles and environmental changes in marine ecosystems (Ohman et al., 2012;Lorrain et al., 2020;Ren et al., 2017). During the 21st century, the 13 C: 12 C carbon isotopic ratio of tuna has been rapidly decreasing as a linear function with time, and the rate of decrease has been faster than would be expected based on the Suess effect and the increase of anthropogenic carbon emissions (Gruber et al., 1999). It 40 has thus been hypothesized that lower-trophic-level ecosystems have changed on a global scale (Lorrain et al., 2020). The input of anthropogenic nitrogen from the atmosphere to the ocean has been increasing (Duce et al., 2008), and in the marginal seas of East Asia, the 15 N: 14 N nitrogen isotopic ratio of the organic matter bound in coral skeletons has decreased with the increase of anthropogenic nitrogen deposition (Ren et al., 2017).
Stable isotope ratios are often used as metrics of the trophic positions of organisms in ecosystems (Aita et al., 2011) and as 45 indicators of changes in the chemical environment. The 15 N: 14 N ratio increases at successively higher tropic levels in a food chain. In this study, we focused on Calanus sinicus, one of the dominant copepod species in the coastal waters of the western North Pacific (Uye, 2000), including the Japan Sea (Kodama et al., 2018a). Copepods of the genus Calanus are the major component of the macrozooplankton in shelf and coastal ecosystems outside the tropics, and they are the major source of nutrition for pelagic fish (Uye, 2000). Coastal areas of the Japan Sea are spawning and nursery grounds for Japanese sardine 50 (Sardinops melanosticta), Japanese anchovy (Engraulis japonicus), and Pacific bluefin tuna (Thunnus orientalis) (Nishida et al., 2020;Furuichi et al., 2020;Ohshimo et al., 2017), and the larvae of these species prey on both the larval and adult stages of C. sinicus (Hirakawa et al., 1997;Hirakawa and Goto, 1996;Kodama et al., 2017a). Calanus sinicus therefore plays an important role as a major component of the second trophic level in the Japan Sea.
Studies of the long-term carbon and nitrogen dynamics in the Japan Sea (Ishizu et al., 2019;Ono, 2021;Kodama et al., 2016) 55 have suggested that monotonic changes are likely to occur throughout the coastal ecosystem of the Japan Sea as the climate changes on a global scale. In fact, the muscle of small pelagic fish in the Japan Sea and the East China Sea from 1996 to 2019 shows that annual mean 13 C: 12 C and 15 N: 14 N are monotonically decreased 0.08‰ year -1 and 0.05‰ year -1 , respectively (Ohshimo et al., in press). However, spatial heterogeneity of the chemical environment and ecosystem of the coastal Japan Sea have also been reported. For example, long-term trends in the amounts of anthropogenic inputs are not spatially uniform: since 60 1997 total nitrogen inputs from rivers to Toyama Bay have been decreasing (Terauchi et al., 2014b) and those to Wakasa Bay have been increasing (Sugimoto and Tsuboi, 2017). Such differences suggest that interannual variations of environmental conditions are unlikely to be spatially homogeneous. Evaluations of local marine ecosystems and comparisons based on observations will therefore be necessary to understand changes of the coastal environment and ecosystems of the Japan Sea, and thus to achieve Sustainable Development Goal 14 of the United Nations. The goal of this study was therefore to understand the spatiotemporal variations of lower-trophic levels in the coastal area of the Japan Sea through analysis of the carbon and nitrogen stable isotope ratios of C. sinicus.

Onboard observations
Onboard observations were conducted from pre-bloom to post-bloom in 2006-2020 during cruises of the R/V Mizuho-Maru, 70 R/V Yoko-Maru (Japan Fisheries Research and Education Agency), and R/V Dai-Roku Kaiyo-Maru (Kaiyo Engineering Co., Ltd.) in the territorial waters of Japan in the Japan Sea, a marginal sea of the western North Pacific (Fig. 1). Four stations for collection of stable isotope samples were chosen: in Toyama Bay (TB), Iida Bay (IB), north of the Noto Peninsula (NN), and Wakasa Bay (WB). These four stations are among 26 stations described in a previous study (Kodama et al., 2018a) that reported a clear west-east gradient of zooplankton community structure in this area during the month of May. Station (stn) TB was 75 located near the mouths of two rivers (the Sho River and Oyabe River, Fig. 1), and stn WB was in an area of restricted circulation in Wakasa Bay. The cruises were conducted during four time intervals: 1) the end of February and/or the beginning of March (described as March), 2) the end of April, 3) the middle or end of May, and 4) the end of June or beginning of July (described as June) (Fig. 1c). The March cruise corresponded to the early stage of the spring phytoplankton bloom, the April cruise was during the late stage of the bloom, and the May and June cruises occurred during post-bloom conditions according 80 to Kodama et al. (2018b). The observations after 2015 were conducted only in April and May (Fig. 1c).
Copepods for stable isotope analyses were collected by towing a Bongo net (500-μm mesh and 70-cm mouth diameter) obliquely at 0.5 m s −1 from a depth of 75 or from 10 m above the bottom to the surface at each station. Collected samples were frozen and preserved at a temperature below −20°C until sorting at an onshore laboratory. Vertical profiles of temperature, salinity, and chlorophyll a concentrations were measured by using a conductivity-temperature-depth (CTD) sensor (Seabird,85 SBE9plus or SBE19plus) with an in vivo chlorophyll fluorescence sensor from a depth of 200 m to the surface. Temperature and salinity of the sensors were calibrated by the manufacturer every year. The surface seawater was sampled with a bucket for measurement of sea surface temperature (SST), sea surface salinity (SSS), and sea surface chlorophyll a concentration (SSC). SST and SSS were measured with a calibrated mercury thermometer and a salinometer (Autosal, Guildline Instruments), respectively. For measurement of chlorophyll a concentrations, particles in 300 ml of water were collected on a glass fiber 90 filter (GF/F, Whatman) and the pigments were extracted with N,N-dimethylformamide. The chlorophyll a concentrations were estimated based on the fluorescence of the extract, which was measured with a fluorometer (10-AU, Turner Designs) (Holm-Hansen et al., 1965). The chlorophyll fluorescence sensor was calibrated using these discrete samples during each cruise.
Nutrient concentrations at the surface were measured during some cruises after 2015. The procedures used for the nutrient analyses have been described by Kodama et al. (2015).

Stable isotope analyses
The frozen-preserved Bongo net samples were thawed at room temperature. From every single thawed sample, 7-14 individuals of Calanus sinicus (Copepoda; Calanoida) adults or copepodite stage V were sorted as quickly as possible to avoid alteration of their condition under the dissecting microscope. We collected 1-8 C. sinicus individuals from 94 net samples; in 100 some samples, we could not find enough C. sinicus individuals for stable isotope analysis. The sorted copepods were dried at 60°C in a drying oven for 36-48 h. A total of 274 dried samples were wrapped in a tin disk. Table 1 provides details about the numbers of samples. At some stations, several samples were prepared, and we treated each sample separately, i.e., we measured the stable isotope ratios of every sample, and we did the statistical analyses without averaging data from the same station. The carbon and nitrogen stable isotope ratios of the samples were measured with a stable isotope ratio mass spectrometer 105 (IsoPrime100; Elementar) coupled with an elemental analyzer (vario MICRO cube; Elementar). Stable isotope ratios of carbon and nitrogen were calculated as the per mil (‰) deviations from the corresponding standards using the following equation: where R is the 13 C/ 12 C or 15 N/ 14 N ratio. The standards for carbon and nitrogen were L-alanine, and the references were Pee Dee Belemnite and atmospheric N2, respectively. The precisions of the analyses were within 0.2‰ for both δ 13 C and δ 15 N. 110 Based on the precision of analyses of L-alanine, the δ 13 C and δ 15 N values for each sample were rounded off to one decimal place.
The amounts of carbon and nitrogen in each sample were also recorded, but that information was missing for 69 samples when the database was created. The carbon:nitrogen ratio (C/N ratio) of C. sinicus was calculated using the C/N ratios in the database.

Statistical analyses 115
The relationships between the δ 13 C and δ 15 N of C. sinicus and environmental parameters were analyzed using general linear models (GLMs) with the linear link function in R software (R Core Team, 2020). The structures of the GLMs were based on concepts articulated by Kodama et al. (2021). The δ 13 C and δ 15 N errors were assumed to be normal distributions, and explanatory variables were expressed as linear functions in the GLMs, with the exception of SSC, which was log transformed.
Two types of GLMs were applied as follows: 120 where δX and stn represent the target δ 13 C or δ 15 N and station (WB, NN, IB and TB), respectively. The arguments of the f functions were categorical variables used to simulate nonlinear relationships. The second argument of the poly functions (i.e., 2) indicates that the first argument was incorporated into a quadratic expression in the model, i.e., dome-shaped responses 125 were included in the models.
Equation (2) was intended to consider environmental and geographical effects, and Eq. (3) was intended to take into consideration monthly and interannual variations at every station (δXstn). We could theoretically have evaluated monthly and interannual variations at every station using interaction terms such as f(month):f(year):f(stn) in Eq. (2), but when we included those interaction terms, the generalized variance inflation factors (GVIFs) became infinite. We therefore used two models. 130 Environmental parameters were not included in Eq. (3) because we wanted to calculate climatological values that would reflect monthly and interannual variations of environmental parameters. The times of sample collection were to some extent opportunistic (Fig. 1C), and thus comparisons of simple mean values could be misleading if interannual or monthly variations were significant. An interaction term f(month):f(year) was considered in Eq. (3), but because the GVIFs were infinite, we did not include an interaction term in the model. The explanatory variables in the final version of Eq. (2) were determined on the 135 basis of values of GVIFs and Akaike information criteria (AICs). We required that the GVIFs of explanatory variables be less than 10, and the model with the smallest AIC was accepted as the final model.
The "gglsmean" function in the "ggeffects" package (Lüdecke, 2018) was used to calculate the least squares mean (lsmean) values and standard errors based on the AIC-selected GLMs and to visualize the effect of the explanatory variables on the δ 13 C or δ 15 N values. When the categorical variables (i.e., month and year) remained in the least-AIC models, their values were fixed 140 for purposes of calculating the lsmean values. We note that we considered the nonlinear effects of environmental variables using generalized additive models (GAMs) instead of Eq. (2). However, the AIC values were not greatly improved in the GAMs. In this study, we have therefore reported only the GLM results. All the data and program code are in Mendeley data (Nakamura et al., 2021).

Environmental factors
SST increased from March to June at every station (Fig. 2a). The SST at WB, the westernmost station, was the highest among the four stations on every cruise. The SST ranges were 8.8-23. 1°C, 9.5-21.9°C, 9.3-23.2°C, and 10.1-23.4°C at stations TB, IB, NN, and WB, respectively. SSS was low at the TB station (Fig. 2b). In this less-saline water, the nitrate concentration was several micromoles even in June, whereas it was depleted at the other three stations (data not shown). SSS was highest in May 150 at the other three stations. Variations of SSC at TB differed from those at the other stations. SSC was higher at TB than at the other stations during every month. In addition, SSC at TB was higher in June than in May (Fig. 2c). The SSC values fell in the ranges 0.02-7.45, 0.08-5.46, 0.05-9.07, and 0.08-6.60 µg L −1 at TB, IB, NN, and WB, respectively.

Stable isotope ratios of C and N
The means ± standard deviations of all of the C. sinicus δ 13 C and δ 15 N values were −20.7 ± 1.8‰ (−24.7‰ to −15.0‰) and 155 6.9 ± 1.2‰ (2.8-8.8‰), respectively. Significant monthly variations were observed for both δ 13 C and δ 15 N at every station (ANOVA, p < 0.001, Fig. 3 all four stations, although there was only one sample at WB in May (Fig. 3). In June, monthly mean δ 13 C values were similar to those in May, except at NN. The mean δ 13 C at NN was significantly lower in May (−22.4 ± 2.2‰) than in June (−21.2 ± 160 1.3‰) (ANOVA with Tukey's HSD test, p = 0.028). The monthly mean values were significantly different among stations in April and March (ANOVA, p < 0.006). In all four months, the δ 13 C at NN was the lowest among the four stations, and there were significant differences between the δ 13 C values at WB and at NN in April and between those at TB and at NN in May (ANOVA with Tukey's HSD test, p < 0.006).
The lowest monthly mean δ 15 N was observed in March at all four stations (TB: 5.1 ± 0.8‰; IB: 5.1 ± 1.4‰; NN: 5.6 ± 1.3‰; 165 and WB: 4.5 ± 1.0‰). The δ 15 N values were significantly elevated in April (TB: 7.4 ± 1.3‰; IB: 7.3 ± 0.5‰; NN: 7.5 ± 0.5‰; and WB: 7.0 ± 0.7‰), and they were stable from April to June (ANOVA with Tukey's HSD test, p > 0.2) at all four stations (Fig. 3b). The differences of the monthly values among stations were significant during all four months (ANOVA, p < 0.037): monthly δ 15 N values at NN were highest among the four stations in every month. In March, April, and June, the δ 15 N values were significantly lower at WB than at NN (ANOVA with Tukey's HSD test, p < 0.03), but there were no significant 170 differences between other pairs of stations. In May, the values were significantly lower at TB and IB than at NN (ANOVA with Tukey's HSD test, p < 0.003).
There were significant linear relationships between the δ 13 C and δ 15 N values and SST, logarithm-transformed SSC, and SSS, but the nature of the relationships differed between δ 13 C and δ 15 N (Fig. 4). The Pearson's correlation coefficient was positive between δ 13 C and logarithm-transformed SSC (p < 0.001; r = 0.478), negative between δ 13 C and SST (p < 0.001; r = −0.564), 175 and negative between δ 13 C and SSS (p < 0.001; r = −0.202) (Fig. 4a-c). In contrast, the Pearson's correlation coefficient was positive between δ 15 N and both SST (p < 0.001; r = 0.361) and SSS (p < 0.001; r = 0.228), and it was negative between δ 15 N and logarithm-transformed SSC (p < 0.001; r = −0.298) ( Fig. 4e-g). When an outlier value of SSS was removed (SSS was 28 in March at TB), the correlations with SSS were still significant. Although there were relatively few C/N ratios of C. sinicus, they were negatively and positively correlated with δ 13 C and δ 15 N, respectively (both p < 0.001, Fig. 5d, h). The monthly 180 means of the C/N ratios of C. sinicus were high during April at every station.
The responses of δ 15 N were the mirror images of the δ 13 C responses (Fig. 5d, e). The least-AIC models produced concave graphs of δ 15 N as a function of SST or logSSC (Fig. 5d, e). The δ 15 N maxima occurred at approximately 18°C and 1 µg L −1 in the case of SST and SSC (Fig. 5d, e). Inter-station comparisons indicated that δ 15 N values were 0.4-8.5‰ lower at WB and TB than at IB and NN (Fig. 5f). 195 The GLM described by Eq. (3) indicated that there were significant interannual and monthly variations of δ 13 C (Fig. 6). The Interannual and monthly variations of lsmean δ 15 N values were also significant (Fig. 7). The monthly lsmean δ 15 N values 205 increased significantly from March to April, and the values in June were similar to those in April at all four stations (Fig. 7).

Discussion 220
Calanus sinicus is one of the important secondary producers that connect primary producers with higher trophic levels in the coastal area of the Japan Sea, although C. sinicus is known to prey on heterotrophic plankton in addition to phytoplankton (Hirai et al., 2018;Yi et al., 2017;Uye and Yamamoto, 1995). The δ 15 N values of C. sinicus (6.9 ± 1.2‰) are consistent with this scenario: their δ 15 N values are intermediate between those of particulate organic matter (POM) in coastal areas of the Japan Sea (around 2-6‰) (Kogure, 2004;Antonio et al., 2012) and predatory fishes such as Japanese anchovy (8.9-11.7‰) 225 (Tanaka et al., 2008) and Japanese sardines (9.4 ± 0.7‰) (Ohshimo et al., 2019).
The δ 13 C and δ 15 N values of C. sinicus at a station record the variations of the δ 13 C and δ 15 N of the POM at the same station.
The in situ environment, however, is not immediately reflected in the δ 13 C and δ 15 N values of secondary producers. The turnover of carbon and nitrogen must also be considered. Zooplankton production has not been reported in the Japanese coastal waters of the Japan Sea, but in the Kuroshio area, Kobari et al. (2018) have reported zooplankton production to be 0.7-1.0 mg 230 C m −3 day −1 and zooplankton dry-weight (not carbon-weight) biomass to be 9.3-13.4 mg m −3 . If the carbon content of C.
sinicus is half its dry weight (Omori, 1969), the reported production rate is approximately 15% of the C. sinicus biomass per day. If the metabolism of carbon is ignored, one week is necessary for turnover of the carbon of C. sinicus. This turnover time agrees with that of Calanus finmarchicus, 5-10 days (Mayor et al., 2011). This turnover time implies that monthly variations of the δ 13 C and δ 15 N of C. sinicus correspond to monthly variations of the environmental climatology. 235 The monthly variations of δ 13 C and δ 15 N were similar among the stations, with the exception of δ 15 N in May. This similarity among the four stations likely reflected the influence of the spring phytoplankton bloom. In the southern Japan Sea, the spring phytoplankton bloom occurs from the beginning of March to the beginning of April (Kodama et al., 2018b;Ishizaka and Yamada, 2019). In Wakasa Bay, the spring bloom causes the δ 15 N of POM to increase from spring (3-5.5‰) to summer (5.5-7‰) (Antonio et al., 2012). In the Kuroshio Current, which flows through Japanese coastal waters on the Pacific side of Japan, 240 the δ 15 N of POM increases from winter to spring because of Rayleigh fractionation as the inorganic nitrogen concentrations in the ambient water decrease (Kodama et al., 2021). The δ 15 N of C. sinicus therefore increases with the increase of the δ 15 N of their prey. The δ 15 N of the phytoplankton increases because of Rayleigh fractionation as their uptake of nitrate draws down the ambient nitrate concentration in the Japan Sea during the decline of the spring phytoplankton bloom.
In the present study, the monthly δ 13 C of C. sinicus declined from the early stage to the late stage of the bloom or the post-245 bloom period. This trend has not been clearly apparent during seasonal monitoring of POM in the coastal area around Japan (Antonio et al., 2012;Kodama et al., 2021), but the δ 13 C of POM increases in the western North Pacific when the chlorophyll a concentration increases (Kodama et al., 2021) and the phytoplankton are growing rapidly (Laws et al., 1995). The relationships between environmental parameters and the δ 13 C of C. sinicus revealed by the GLMs in this study are similar to the relationships between the δ 13 C of POM and environmental conditions in the western North Pacific (Kodama et al., 2021). 250 The relationship between chlorophyll a concentrations and the δ 13 C values of C. sinicus may be impacted by the turnover rate, but the monthly variations of the δ 13 C of the prey of C. sinicus may drive the monthly variations of the δ 13 C of C. sinicus. We also considered the possibility that the decline of the monthly δ 13 C values of C. sinicus during the bloom period was attributable to physiological changes of C. sinicus. Calanus sinicus is capable of storing oil in a sac (Zhou et al., 2016), and the monthly variations of C/N ratios and δ 13 C values at our observation sites indicated that C. sinicus stored lipids in April. The δ 13 C values 255 of copepods decrease with increases of their fatty acid content (Smyntek et al., 2007). This tendency was evidenced in this study by the relationship between the δ 13 C values and C/N ratios of C. sinicus. The decline of δ 13 C values from March to April was therefore likely to have been the result of lipid accumulation in C. sinicus. However, the relatively low δ 13 C values of C. sinicus in May and June are not attributable to lipid accumulation because the C/N ratios of C. sinicus were not high at that time. We suspect that lipids with a low δ 13 C that had accumulated in April were assimilated into tissues beginning in May, 260 while the δ 13 C of the POM in the ambient water may be low in May and June comparing to April.
In the Japan Sea, the interannual variations of δ 13 C and δ 15 N are possibly affected by 1) global-scale or regional-scale events that occur throughout the Japan Sea, and 2) local-scale events that occur in every bay. Increases of anthropogenic carbon and nitrogen inputs from the atmosphere to the ocean have been monotonically changing the δ 13 C and δ 15 N of marine organisms on a global scale (Lorrain et al., 2020;Ren et al., 2017). In our study, however, the temporal trends of the annual lsmeans of 265 the δ 13 C and δ 15 N values were not significant at any station. The duration of our study was shorter than those of previous studies that have reported long-term monotonic trends. Lorrain et al. (2020) have reported that the δ 13 C values of tuna decreased monotonically by 0.12‰ year −1 , that is 1.8‰ in 15 years. Such a trend would have been detectable if it had occurred at our study sites. Some previous longer-term studies of zooplankton stable isotope ratios have noted that the trends have not been strictly linear (Christensen and Richardson, 2008;Chiba et al., 2012). 270 Regional-scale interannual variations possibly impacted the interannual variations of the δ 13 C and δ 15 N values of C. sinicus, but such variations were not detected in our study. Ohshimo et al. (in press) indicated the monotonic decreasing trends of δ 13 C and δ 15 N values of small pelagic fish in the Japan Sea and the East China Sea. Much nitrate was discharged by the Changjiang River into the Japan Sea during the summer of 2013 (Kodama et al., 2017b), and lowering of surface salinity by inflow from the Changjiang River occurred in the summer or autumn of 2010, 2012, and 2015 (Kosugi et al., 2021). The spatial distribution 275 of the impact of these interannual variations, however, differed from that of the monthly variations of the δ 13 C and δ 15 N values of C. sinicus. Local-scale variations were therefore more important than global-scale trends and basin-scale variations at our study sites in the coastal area of the Japan Sea, even at stations highly impacted by human activities.
Carbon and nitrogen dynamics are closely coupled, even in coastal areas where anthropogenic impacts are high (Doney et al., 2007). Our GLM results with environmental parameters as independent variables revealed mirror-image responses of δ 13 C and 280 δ 15 N. We therefore hypothesized that there would be a negative correlation between interannual variations of δ 13 C and δ 15 N to the extent that they were controlled by environmental parameters such as temperature and chlorophyll a concentration, but the correlation between the interannual variations of δ 13 C and δ 15 N was not significant. The implication is that the nitrogen and carbon dynamics varied in different ways at our sites. We cannot clarify the causes of the local interannual variations, but local-scale variations have been reported in the Japan Sea. For example, at TB, where the circulation is restricted, summer 285 phytoplankton blooms exploit nutrients delivered by rivers, but fluvial nutrient inputs have little influence on the phytoplankton blooms at IB in the mouth of Toyama Bay (Terauchi et al., 2014a;Terauchi et al., 2014b). The long-term variations of total nitrogen inputs from rivers have been increasing in Wakasa Bay (Sugimoto and Tsuboi, 2017) but decreasing in Toyama Bay (Terauchi et al., 2014b). In addition, the δ 15 N values of riverine nitrate differ as a function of their source: fertilizer, sewage, or forest soil (Sugimoto et al., 2019). In our study, the causes of the interannual variations of the δ 13 C and 290 δ 15 N of C. sinicus were unclear, but local-scale events accounted for much of the variation.

Conclusions
We used data of a 15-year study of the δ 13 C and δ 15 N of the calanoid copepod C. sinicus to indirectly examine spatiotemporal variations of local ecosystems. Calanus sinicus was a "secondary" producer in the area. Monthly variations of δ 13 C and δ 15 N were affected by spring phytoplankton blooms. The interannual variations of δ 13 C and δ 15 N differed among the areas and 295 showed no evidence of monotonic trends. We concluded that undocumented, local-scale events accounted for most of the changes of carbon and nitrogen dynamics in the coastal areas.
The most important and informative finding of this study is that the temporal changes of coastal carbon and nitrogen dynamics were not monotonic, and they were spatially heterogenous. Global-scale and regional-scale studies have indicated that human activities have led to temporally monotonic and spatially homogenous impacts on coastal environments and ecosystems 300 (Halpern et al., 2008;Ono, 2021;Kodama et al., 2016;Ishizu et al., 2019), but coastal areas are not spatially homogenous.
This spatial heterogeneity is a very important contributor to the high value of the ecosystem services provided by coastal areas, including their role as nursery grounds for many commercially valuable fish. Management of watersheds must be carried out at a local scale, and recruitment to local fisheries resources is sensitive to local disturbances. Global-scale and regional-scale environmental changes are very important, but local-scale changes are also important. More comparative studies that address 305 local-scale effects are needed.