E ﬀ ects of Upwelling Intensity on Nitrogen and Carbon Fluxes through the Planktonic Food Web o ﬀ A Coruña (Galicia, NW Spain) Assessed with Stable Isotopes

: The input of new nutrients by upwelling in shelf waters, and the associated carbon and nitrogen ﬂuxes, can be traced by their stable isotope signatures in organic matter. Here, we analyze the relationships between upwelling intensity and natural abundance of stable carbon and nitrogen isotopes in seston, sedimented particles, and four plankton size fractions (200 to 5000 µ m) sampled monthly during 2010 and 2011 in a seasonal upwelling ecosystem. Upwelling modiﬁed the seasonal warming stratiﬁcation by introducing cold and nutrient-rich waters in subsurface layers, enhancing chlorophyll-a and diatom abundance. Seston and sedimented particles were enriched in heavy nitrogen (but not carbon) isotopes linearly with upwelling intensity, indicating a primary e ﬀ ect of upwelling on phytoplankton production. In contrast, all plankton fractions were enriched in heavy carbon isotopes with upwelling, mainly due to the consumption of diatoms. These results conﬁrm the di ﬀ erential e ﬀ ect of upwelling on nitrogen and carbon ﬂuxes in the plankton food web. Direct e ﬀ ects of the new nitrogen inputs on phytoplankton are less evident with the increase of plankton size as nitrogen is repeatedly recycled, while the enriched carbon of plankton suggests the consumption of diatoms during upwelling. We provide linear equations to assess the inﬂuence of changes in upwelling intensity on nitrogen and carbon ﬂuxes in seston and plankton in this ecosystem, as well as to estimate reference baseline values for food web studies.


Introduction
Coastal upwelling ecosystems are characterized by higher biological production than ecosystems at similar latitudes but with seasonal variability in nutrient inputs [1,2]. High production events of days to weeks occur through a favorable upwelling season, alternated with nutrient regeneration periods [3,4]. In these ecosystems, upwelling is generally the primary source of new (external) nutrients fueling primary production, but this can be influenced by inputs from continental origins, particularly near large rivers [5]. Upwelling nutrient inputs in the photic layer produce a primary response of phytoplankton growth (blooming of diatoms in a few days) closely followed by increases in bacteria and protozoans (in a few days), and later (in a few weeks) by increases in metazoan zooplankton [6]. The transfer of nutrients (mainly C and N) up the food web is assumed to be highly efficient in these systems because of the repetition of upwelling events and the presence of dense plankton populations leading to large populations of fish and other consumers [2].
Upwelling productivity is expected to increase with climate change because of an increase in favorable winds [1,7], but its effect on different components of the food webs requires further research. We collected mesoplankton (200-5000 µm) by double oblique tows of a Juday-Bogorov net (200 µm mesh size) between the surface and the bottom. Samples were fractionated using sieves to obtain four size-fractions (200-500, 500-1000, 1000-2000, and >2000 µm) and dried (50 °C, 48 h) immediately after collection.
Sinking particles were collected by means of a particle interceptor trap of 6 cm diameter deployed for 24 h at 50 m depth and anchored to the seafloor at a fixed point close to Stn. E2CO. Sediment trap deployments were made at the same dates of water and plankton sampling. Details of sampling and limitations of this approach can be found in Bode et al. [21].

Stable Isotope Anaysis
Dried (50 °C, 24 h) and finely ground (mortar and pestle) aliquots of plankton or whole filters in the case of seston were packed in tin capsules for conversion into CO2 and N2 in an elemental analyzer (Carlo Erba CHNSO 1108) coupled to an isotope ratio mass spectrometer (Finnigan Mat Delta Plus). Samples were not acidified to remove carbonates because zooplankton is only lightly calcified [22]. Nitrogen and carbon stable isotope abundance was expressed as δ 15 N and δ 13 C relative to atmospheric N2 isotope and VPDB (Vienna PeeDee Belemnite carbonate), respectively [23]. Standards of the International Atomic Energy Agency USGS40 and L-alanine, as well as internal acetanilide and sample standards, were analyzed with each batch of samples to ensure offsets between certified and measured values <0.1‰. Precision (± standard error) of replicate determinations of both C and N stable isotopes was <0.03‰. For comparison with water-column integrated plankton values, the weighted average of isotopic values for seston was computed for each sampling date by taking into account the depth distribution of C and N [24]. δ 13 C values for plankton fractions were normalized to correct for their variable lipid content using the sample C:N weight ratio following the empirical relationship derived from meta-analysis in aquatic animals [25]: corrected δ 13 C = uncorrected δ 13 C −3.32 + 0.99 C:N (1)
Sinking particles were collected by means of a particle interceptor trap of 6 cm diameter deployed for 24 h at 50 m depth and anchored to the seafloor at a fixed point close to Stn. E2CO. Sediment trap deployments were made at the same dates of water and plankton sampling. Details of sampling and limitations of this approach can be found in Bode et al. [21].

Stable Isotope Anaysis
Dried (50 • C, 24 h) and finely ground (mortar and pestle) aliquots of plankton or whole filters in the case of seston were packed in tin capsules for conversion into CO 2 and N 2 in an elemental analyzer (Carlo Erba CHNSO 1108) coupled to an isotope ratio mass spectrometer (Finnigan Mat Delta Plus). Samples were not acidified to remove carbonates because zooplankton is only lightly calcified [22]. Nitrogen and carbon stable isotope abundance was expressed as δ 15 N and δ 13 C relative to atmospheric N 2 isotope and VPDB (Vienna PeeDee Belemnite carbonate), respectively [23]. Standards of the International Atomic Energy Agency USGS40 and L-alanine, as well as internal acetanilide and sample standards, were analyzed with each batch of samples to ensure offsets between certified and measured values <0.1% . Precision (± standard error) of replicate determinations of both C and N stable isotopes was <0.03% . For comparison with water-column integrated plankton values, the weighted average of isotopic values for seston was computed for each sampling date by taking into account the depth distribution of C and N [24]. δ 13 C values for plankton fractions were normalized to correct for their variable lipid content using the sample C:N weight ratio following the empirical relationship derived from meta-analysis in aquatic animals [25]: corrected δ 13 C = uncorrected δ 13 C −3.32 + 0.99 C:N (1)

Upwelling Intensity
Upwelling intensity (UI) was estimated by calculating the Ekman transport from surface winds (m 3 s −1 km −1 ) computed by the Instituto Español de Oceanografía (http://www.indicedeafloramiento.ieo.es/) in a cell of 1 • × 1 • centered at 44 • N, 9 • W from atmospheric pressure at sea level [26]. This location is representative of the subregional wind regime affecting upwelling in northern Galicia. Positive values of this index indicate net upwelling periods when surface water is transported offshore, while negative  15 , and UI 30 , respectively) preceding each sampling event of E2CO station, as previous studies indicated that Chla was correlated with upwelling events occurring several days before sampling [27].

Statistical Analysis
The relationships between upwelling and other variables were analyzed by correlation (Pearson coefficient) and linear regression (ordinary least squares). All variables were normally distributed after logarithmic transformations (Shapiro-Wilk normality test, P < 0.05). Constant values were added to UI and δ 13 C values prior to transformation to avoid negative numbers. Differences between mean values of isotope abundance in the different compartments were assessed with ANOVA and Bonferroni post-hoc tests. Differences between regression slopes were determined with ANCOVA. Corrections for inequality of variances were applied as required. As the observations were obtained along a gradient of time, the residuals of the regressions were tested for autocorrelation (Durbin-Watson test, Supplementary Table  S1). Statistical analysis was made using the package Past version 4.0 [28]. All raw data used in this study can be accessed through the PANGAEA repository (https://doi.pangaea.de/10.1594/PANGAEA.911575).

Upwelling and Nutrient Enrichment
Positive values of UI were more frequent between March and September in both 2010 and 2011, while negative values occurred mainly in October-November and February ( Figure 2). However, interannual variations were also observed as the upwelling events in December 2010 and January 2011 or the downwelling events in June and July 2011. All variations of the index were significantly correlated (UI 7 with UI 15 , r = 0.742; UI 7 with UI 30 , r = 0.706; and UI 15 with UI 30 , r = 0.718, all with P < 0.001); therefore, UI 15 was employed in all subsequent analysis. Observations were classified as "upwelling" when UI 15 > 0 or "downwelling" when UI 15 < 0. affecting upwelling in northern Galicia. Positive values of this index indicate net upwelling periods when surface water is transported offshore, while negative values indicate an accumulation of surface water against the coast (downwelling). For the purpose of the present study, 6-hourly UI values were averaged over the 7, 15 or 30 days (UI7, UI15, and UI30, respectively) preceding each sampling event of E2CO station, as previous studies indicated that Chla was correlated with upwelling events occurring several days before sampling [27].

Statistical Analysis
The relationships between upwelling and other variables were analyzed by correlation (Pearson coefficient) and linear regression (ordinary least squares). All variables were normally distributed after logarithmic transformations (Shapiro-Wilk normality test, P < 0.05). Constant values were added to UI and δ 13 C values prior to transformation to avoid negative numbers. Differences between mean values of isotope abundance in the different compartments were assessed with ANOVA and Bonferroni post-hoc tests. Differences between regression slopes were determined with ANCOVA. Corrections for inequality of variances were applied as required. As the observations were obtained along a gradient of time, the residuals of the regressions were tested for autocorrelation (Durbin-Watson test, Supplementary Table S1). Statistical analysis was made using the package Past version 4.0 [28]. All raw data used in this study can be accessed through the PANGAEA repository (https://doi.pangaea.de/10.1594/PANGAEA.911575).

Upwelling and Nutrient Enrichment
Positive values of UI were more frequent between March and September in both 2010 and 2011, while negative values occurred mainly in October-November and February ( Figure 2). However, interannual variations were also observed as the upwelling events in December 2010 and January 2011 or the downwelling events in June and July 2011. All variations of the index were significantly correlated (UI7 with UI15, r = 0.742; UI7 with UI30, r = 0.706; and UI15 with UI30, r = 0.718, all with P < 0.001); therefore, UI15 was employed in all subsequent analysis. Observations were classified as "upwelling" when UI15 > 0 or "downwelling" when UI15 < 0. Upwelling events modified the seasonal stratification caused by the warming of the surface layer ( Figure 3a). In most cases, this upper layer narrowed because of the input of cold waters (<15 °C) near the bottom. In turn, downwelling periods were characterized by the warming of the whole water column because of the accumulation of surface waters towards the coast. The input of cold  Upwelling events modified the seasonal stratification caused by the warming of the surface layer ( Figure 3a). In most cases, this upper layer narrowed because of the input of cold waters (<15 • C) near the bottom. In turn, downwelling periods were characterized by the warming of the whole water column because of the accumulation of surface waters towards the coast. The input of cold subsurface water was mirrored by consequent increases in nitrate (Figure 3b), while surface waters displayed increases in ammonium concentrations mainly after the upwelling events and during downwelling ( Figure 3c). Not all upwelling events produced large phytoplankton blooms, as indicated by Chla subsurface water was mirrored by consequent increases in nitrate (Figure 3b), while surface waters displayed increases in ammonium concentrations mainly after the upwelling events and during downwelling ( Figure 3c). Not all upwelling events produced large phytoplankton blooms, as indicated by Chla ( Figure 3d). However, Chla values were generally high in the upper 50 m during, or immediately after, upwelling periods. Considering all observations, UI 15 was negatively correlated with temperature and positively correlated with nitrate and phosphate concentrations in bottom waters (Figure 4a-c). All these relationships could be described by linear regressions illustrating the changes in temperature or concentrations for a given intensity of upwelling (or downwelling). The same occurred with the Diversity 2020, 12, 121 6 of 14 increase in 50 m-integrated Chla, but in this case, the effect of upwelling is less clear because of the increase in variance when UI 15 > 0 (Figure 4d). Integrated Chla was also positively correlated with the abundance of diatoms (r = 0.444, P < 0.05), which was also correlated with UI 15 (r = 0.505, P < 0.05).
Diversity 2020, 12, 121 6 of 14 relationships could be described by linear regressions illustrating the changes in temperature or concentrations for a given intensity of upwelling (or downwelling). The same occurred with the increase in 50 m-integrated Chla, but in this case, the effect of upwelling is less clear because of the increase in variance when UI15 > 0 ( Figure 4d). Integrated Chla was also positively correlated with the abundance of diatoms (r = 0.444, P < 0.05), which was also correlated with UI15 (r = 0.505, P < 0.05).
with standard errors of the coefficients indicated in parenthesis (r = 0.525, P < 0.01, n = 24). Using this equation we can estimate the δ 15 N value for UI 15 = 0 (5.13% ) as a mean reference δ 15 N value of integrated seston for non-upwelling situations.
Diversity 2020, 12, 121 7 of 14 Neither δ 15 N or δ 13 C showed a clear seasonal pattern for seston and sediment traps, but mean δ 13 C values for all plankton fractions were significantly higher during the March-September period when compared to those observed during October-February ( Figure 5, ANOVA, p < 0.05). However, seston and sediment trap δ 15 N values were linearly enriched with upwelling and isotopically diluted with downwelling (Figure 6a). Regression slopes of δ 15 N with UI15 were not significantly different for seston and for sediment trap particles (ANCOVA, p > 0.05). In this way, the regression equation for seston was: log(δ 15 N) = −0.663 (±0.468) + 0.395 (±0.137) log(UI15 + 3000) (2) with standard errors of the coefficients indicated in parenthesis (r = 0.525, P < 0.01, n = 24). Using this equation we can estimate the δ 15 N value for UI15 = 0 (5.13‰) as a mean reference δ 15 N value of integrated seston for non-upwelling situations Similarly, there was a positive linear increase in δ 13 C of all plankton fractions with the abundance of diatoms with non-significantly different slopes (ANCOVA, P > 0.05), while δ 13 C of other compartments did not display significant relationships with diatoms ( Figure 7).

Discussion
This study revealed differential effects of upwelling on the carbon and nitrogen isotopic composition across various compartments of particulate organic matter. Such differences are indicative of the processes affecting carbon and nitrogen fluxes across the food web. First, the new nitrogen inputs by the upwelling were tracked by seston and sediment trap particles but not in larger size-fractions of plankton. Second, the carbon produced by the upwelling, mainly in the form of diatoms, can be tracked in plankton fractions, but not in seston or recently-sedimented particles.

Nitrogen Isotopes and Fluxes
Reported changes in δ 15 N of seston and small (<100 µm) plankton organisms during upwelling were inconsistent, as there were examples of decreases [29,30], no change from the δ 15 N of subsurface nitrate [31], or even increases in δ 15 N [32,33]. This variability in δ 15 N can be attributed to rapid changes in the isotopic composition of the upwelled nitrate, as the phytoplankton assimilates Similarly, there was a positive linear increase in δ 13 C of all plankton fractions with the abundance of diatoms with non-significantly different slopes (ANCOVA, P > 0.05), while δ 13 C of other compartments did not display significant relationships with diatoms ( Figure 7).

Discussion
This study revealed differential effects of upwelling on the carbon and nitrogen isotopic composition across various compartments of particulate organic matter. Such differences are indicative of the processes affecting carbon and nitrogen fluxes across the food web. First, the new nitrogen inputs by the upwelling were tracked by seston and sediment trap particles but not in larger size-fractions of plankton. Second, the carbon produced by the upwelling, mainly in the form of diatoms, can be tracked in plankton fractions, but not in seston or recently-sedimented particles.

Nitrogen Isotopes and Fluxes
Reported changes in δ 15 N of seston and small (<100 µm) plankton organisms during upwelling were inconsistent, as there were examples of decreases [29,30], no change from the δ 15 N of subsurface nitrate [31], or even increases in δ 15 N [32,33]. This variability in δ 15 N can be attributed to rapid changes in the isotopic composition of the upwelled nitrate, as the phytoplankton assimilates

Discussion
This study revealed differential effects of upwelling on the carbon and nitrogen isotopic composition across various compartments of particulate organic matter. Such differences are indicative of the processes affecting carbon and nitrogen fluxes across the food web. First, the new nitrogen inputs by the upwelling were tracked by seston and sediment trap particles but not in larger size-fractions of plankton. Second, the carbon produced by the upwelling, mainly in the form of diatoms, can be tracked in plankton fractions, but not in seston or recently-sedimented particles.

Nitrogen Isotopes and Fluxes
Reported changes in δ 15 N of seston and small (<100 µm) plankton organisms during upwelling were inconsistent, as there were examples of decreases [29,30], no change from the δ 15 N of subsurface nitrate [31], or even increases in δ 15 N [32,33]. This variability in δ 15 N can be attributed to rapid changes in the isotopic composition of the upwelled nitrate, as the phytoplankton assimilates preferentially light nitrogen atoms and leaves the remaining nitrate enriched in heavy isotopes [34]. Therefore the relative enrichment or depletion of seston δ 15 N compared to the initial nitrate upwelled will depend on the timing of the phytoplankton bloom produced by the upwelling. In the present study, seston and trap particles showed a significant increase in δ 15 N with upwelling intensity (and conversely a decrease with downwelling) integrated over the preceding 15 days. Even when temporal autocorrelation may affect Type 1 statistical errors in both cases, similar increases were also found in the small plankton fractions that did not show significant autocorrelation (Supplementary Table S1). Such a timescale is similar to the time required for reaching maximum uptake rates by phytoplankton after an upwelling pulse [3]. Therefore, the isotopic enrichment is consistent with the rapid uptake and assimilation of the nitrate produced in subsurface waters through nitrification, as this process has a relatively large isotopic fractionation effect [10]. In contrast, the uptake of isotopically-light inorganic nitrogen regenerated in the surface layer, mostly in the form of ammonium and accumulated during downwelling, produced less enriched particles during non-upwelling periods. Upwelling of subsurface nitrate was also claimed as the main cause of high values of δ 15 N in seston in open waters of the Atlantic, as opposed to the low values caused by the fixation of nitrogen [32,35,36].
Other sources of nitrogen could also cause the elevation or decrease of δ 15 N in organic matter. For instance, the input of anthropogenic nitrogen from residual waters may affect coastal ecosystems, as detected in the study region [18,37]. However, the δ 15 N values observed in suspended and sedimented particles at the studied station were closer to those of nitrate at 70 m depth in the same station (6.08 ± 0.28% , n = 24) than to those typical in the effluent of a water treatment plant (16.6 ± 1% , n = 3), as reported in a previous study [18]. Furthermore, denitrification or other nitrogen transformation processes can be expected to be of low importance at the studied station because the water column remains well-oxygenated through the year [38]. The value of the intercept of the regression line between seston δ 15 N and UI 15 (equation 2) is almost equivalent to the overall mean (Table 1), thus supporting that the upwelled nitrogen is conserved in the water column. Alternatively, atmospheric nitrogen fixation would decrease δ 15 N, but, although with measurable rates, this process has a low impact on nitrogen inputs for phytoplankton in this station [39]. The prevalence of eukaryotic phytoplankton (mainly diatoms and dinoflagellates) in local primary production and biomass [21,40] would also explain the increase of δ 15 N in particles with upwelling, as phytoeukaryotes showed a larger affinity for upwelled nitrate than prokaryotes [33].
Furthermore, the similarity in the magnitude and variation of δ 15 N in seston and sediment trap particles observed in this study suggests the rapid (i.e., within the 15 d of integration of the upwelling event) and complete uptake of the nitrate provided by the upwelling with a minimal process in the water column, as found in other studies of coastal upwelling ecosystems [41]. Meanwhile, in oceanic environments characterized by deeper water columns, the enrichment in heavy isotopes of the sedimented particles at deep water layers is the norm [31,42]. All these results are indicative of a rapid response of phytoplankton to upwelling nitrate (and other nutrient) inputs, but at the same time suggest also rapid changes to regenerated nutrient sources during downwelling, with minimal influence of additional nutrient sources. Intensive and repeated remineralization of the organic matter in the shelf causes the upwelled water in this region to contain a mixture of new (oceanic) and regenerated nitrogen [4], thus affecting the variability in isotopic composition observed in the particles.
Another implication of the observed variability in seston δ 15 N is that annual or seasonal averages of δ 15 N may underestimate the reference baseline for trophic position estimations because most consumers in the food web would have different integration times [43]. Therefore, alternative baselines must be defined by primary consumers [12] as zooplankton. In this study, all plankton fractions larger than 200 µm displayed lower variability in δ 15 N than seston, thus allowing for the use of mean values as representative baselines for trophic position estimates in the pelagic food web of this ecosystem. However, the observed mean isotopic enrichment of plankton relative to seston δ 15 N was much lower than the commonly-accepted mean (± sd) value of 3.4 ± 1.0% increase per trophic level [11]. One explanation could be the presence of phytoplankton in the various fractions, but previous studies in the same station indicate that the fraction larger than 200 µm is composed mainly by zooplankton (up to 8000 indiv m −3 during spring and autumn peaks), with a dominance of copepods for most of the year [44]. While the exact taxonomic composition of plankton in all the fractions considered in this study was not determined, the influence of phytoplankton in the δ 15 N signal of these fractions is expected to be low, although the contribution of some large diatom cells and colonies in the 200-500 µm fraction cannot be totally excluded. Alternatively, our results suggest that plankton consumers had lower isotopic fractionation per trophic level than other consumers. Even when there are reports of the variability in trophic δ 15 N fractionation caused by differences in diet, environmental factors, or the taxonomic composition and size of consumers [12,45,46], few studies highlighted the low fractionation observed in plankton. This low fractionation can be attributed in part to the consumption of protists with similar δ 15 N values to their phytoplankton and bacterial prey [47].
In absence of clear seasonal patterns, the δ 15 N of all plankton fractions in our study reflects a balanced integration of upwelling-derived and regenerated nitrogen during non-upwelling periods, as the dominance of upwelling nitrogen sources would have produced an enrichment at least equivalent to the values observed in seston during upwelling. This is due to the longer time required for integration of the isotopic signal in the fractionated plankton than in seston, where the dominance of phytoplankton assimilation and heterotrophic remineralization processes alternate [48]. These results highlight the importance of the repeated cycling of the nitrate initially provided by the upwelling in the Galician shelf, almost doubling the effect of the upwelling nitrogen on primary production [4].
To overcome the short-term variability in δ 15 N, studies in other upwelling regions focused on the analysis of benthic organisms, notably filter-feeder mollusks integrating the signal at seasonal time scales. This approach successfully revealed geographic gradients in the effect of upwelling in the coasts of South Africa [49], northern Chile [50], and California [51]. Nevertheless, the analysis of the long series of δ 15 N in zooplankton allowed the detection of variations in the upwelling source waters, as exemplified in the California Current [52].

Carbon Isotopes and Fluxes
As opposed to δ 15 N, there was a linear increase in δ 13 C with upwelling intensity in plankton fractions, but not in seston or sediment trap particles. These results evidence the differential routing of carbon and nitrogen provided by the upwelling. While a large fraction of nitrogen is recycled in situ, most of the carbon produced by the upwelling is directed to the biomass of zooplankton consumers, leaving a clear imprint of diatom carbon, as inferred from the significant relationship between diatom abundance and plankton δ 13 C. Actively-growing diatoms are a source of enriched δ 13 C for zooplankton, and this feature was used to ascertain diatom importance in plankton food webs [13]. Upwelling events in Galicia, as in other upwelling ecosystems [14], are associated with a large abundance of diatoms [16,53] that in turn are related to the increase in primary production [4,17,38]. Other studies at the same sampling station confirmed the dominance of diatoms in total phytoplankton abundance [41] and biomass [54] through the year, thus supporting their importance for zooplankton feeding. However, because seston contains other organic particles in addition to diatoms, the observed δ 13 C in seston and sediment trap particles represented a mixture of various carbon sources. A comparison of δ 13 C values of suspended particles and coastal macrophytes concluded that almost 70% of the carbon collected by sediment traps at this station was provided by macrophytes, mainly in the form of detritus [37]. For this reason, the increase of δ 13 C in plankton could be due also to the consumption of detrital particles enriched in heavy carbon isotopes because of the loss of light isotopes due to microbial reworking [42], but this is unlikely as a large fraction the material collected in coastal environments by sediment traps is composed of aged, and presumably biologically-refractory, organic matter [55]. The significant relationship between the δ 13 C of the smallest plankton size fractions and the abundance of diatoms can be explained by the dominance of copepod grazers, like Acartia clausi, during upwelling events [44].
The persistence of the diatom signal in the plankton found in this study allows for the use of δ 13 C values as baseline reference values to estimate the input of carbon sources derived from the upwelling in upper trophic levels, such as fish and marine mammals. In this way, the influence of upwelling could be better addressed using geographic distributions of plankton δ 13 C [56]. This approach will improve the resolution of previous studies of plankton consumers based on isotopic values averaged over most of the Galician shelf region, including upwelling and non-upwelling situations [57,58]. In addition, the comparison of lipid-corrected δ 13 C values of plankton samples from previous or future studies in this ecosystem with the values estimated from the regression line computed in this study will provide a quantitative framework to analyze the influence of upwelling on carbon fluxes through the pelagic food web.

Conclusions
The studied station showed the influence of wind on key environmental properties (lowering the temperature and increasing nutrient concentrations of subsurface waters, and increasing water-column integrated chlorophyll and diatom abundance) characteristic of all upwelling ecosystems. However, there was a differential influence on nitrogen and carbon fluxes in plankton and suspended particles. The upwelled nitrogen left an isotopically-enriched signal in seston and sedimented particles and was conserved in the plankton likely by repeated remineralization in the water during alternating upwelling and downwelling sequences. Plankton between 200 and 5000 µm integrated the different nitrogen sources better than short-living seston particles and showed relatively homogeneous δ 15 N values through the year. In contrast, δ 13 C patterns indicated a variable mixture of carbon sources in seston and sedimented particles but a clear signal of carbon produced by upwelling events in the form of diatoms that can be tracked in all plankton fractions. The computed linear regressions between seston δ 15 N or plankton δ 13 C and the upwelling index integrated over the preceding 15 days, will provide reference baseline values for food web studies and new quantitative procedures for estimating the impact of past and future variations in upwelling intensity in the carbon and nitrogen fluxes in this ecosystem.