The average composition of the dissolved and particulate products of early degradation of marine phytoplankton has been established for the first time in a coastal upwelling system using a mixing analysis along isopycnal surfaces combined with a stoichiometric model. About 17–18% of the mineralized organic matter is derived from the decomposition of organic particulates, and 16–35% is from the dissolved organic matter. The remaining 50–70% is derived probably from large fast sinking particles. On average, the mineralized material on large particles has the closest composition to the Redfield formula. The ratio of dissolved saccharides to dissolved organic matter respiration is >40% higher than expected from a material of Redfield composition. Finally, the ratio of lipid to particulate organic matter respiration is >80% larger than expected from a material of Redfield composition. Regarding the decomposition of hard structures, biogenic silica dissolves predominantly in the inner shelf, where organic carbon oxidation is more intense, and diatom deposition occurs preferentially.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The composition of the products of early degradation of phytoplankton photosynthesis has been a matter of controversy over the last 50 years, after the seminal study of Redfield et al. . On the basis of that work, Richards  established an average composition of C106H264O110N16P. Subsequently, Anderson  and Fraga et al.  proposed revised formulas, C106H175O42N16P and C106H171O42N16P respectively, which basically corrected the overestimated H and O proportions of the original Richard's  formula.
 Biogenic matter production is enhanced in the coastal zone because of intensified nutrient fluxes from the ocean, the continents and the atmosphere [Walsh, 1991]. Average primary production per unit area in the coastal zone, 250 g C m−2 yr−1, is more than twice that in the open ocean, 90 g C m−2 yr−1 [Wollast, 1998]. As a result, the decomposition of biogenic materials is also enhanced. About 83% of the ocean benthic respiration and 87% of the burial occurs in the sediments of the coastal zone [Middelburg et al., 1993]. Pelagic and benthic degradation processes are specially intensified in coastal upwelling areas, because of the magnified entry of nutrients.
 The aim of this work is to study the elemental and biochemical compositions of (1) the organic matter dissolved in seawater, (2) soft and hard particulates suspended in water and (3) biogenic materials that have been mineralized in the water column of the NW Iberian upwelling system. Our investigation was conducted in two contrasting environments: the system of large, V-shaped embayments of Galicia (NW Spain), known as Rías Baixas (Figure 1) and the adjacent open shelf. Although the rías are singular ecosystems, their large dimensions and their well-known hydrodynamic and biogeochemical behavior make them similar to any coastal upwelling system at comparable latitudes: off Oregon (NW America) or off Chile (SW America).
2. Study Area
 This study is focused on the NW Iberian upwelling system (42°–43°N), at the boundary between the temperate and subpolar regimes of the eastern North Atlantic (Figure 1). Wind stress/relaxation cycles of period 1–2 week take place from March–April to September–October, the upwelling season [Álvarez-Salgado et al., 1993]. Downwelling prevails the rest of the year, favoring a reversal of the coastal circulation with the arrival of warm and salty subtropical surface and central waters to our latitudes in the form of the well-defined slope Iberian Poleward Current (IPC) [Álvarez-Salgado et al., 2003]. The winter mixing period occurs at the time of the transition from the downwelling to the upwelling seasons.
 Two contrasting environments were sampled: (1) the Ría de Vigo, a 2.5 km3 large V-shaped embayment, which is primarily controlled by coastal winds and secondarily by continental runoff [Gilcoto et al., 2001]; and (2) the adjacent shelf, which exchanges water and materials with the rías and the open ocean and is affected by strong alongshore currents. Both environments are connected during the upwelling season, when the surface waters of the ría are exported to the adjacent shelf and the bottom waters of the shelf enters the ría. On the contrary, during the downwelling season a convergence front develops between the subtropical IPC waters transported onshore by the Ekman transport and the waters of the ría transported offshore by the continental runoff [Álvarez-Salgado et al., 2000].
3. Sampling and Analytical Methods
3.1. Sampling Strategy
 Two stations were sampled weekly from May 2001 to April 2002: station 00 was in the middle segment of the Ría de Vigo (45 m depth) and station 03 was in the middle shelf (150 m depth). Samples were taken with a rosette sampler equipped with twelve 10-L PVC Niskin bottles with stainless-steel internal springs. Salinity and temperature were recorded with a SBE 9/11 conductivity-temperature-depth probe attached to the rosette sampler. Conductivity measurements were converted into practical salinity scale values with the equation of UNESCO .
 Samples for the analyses of dissolved oxygen, pH, total alkalinity, nutrient salts, dissolved and particulate organic carbon and nitrogen were collected from 5, 15 and 40 m at station 00 and 5, 25, 40, 60, 75, 100 and 150 m at station 03 with a weekly periodicity. Samples for dissolved and particulate organic phosphorus and saccharides were taken at the same depths but with a fortnightly periodicity.
3.2. Nutrient Salts (NH4+, NO2−, NO3−, HPO42−, and SiO4H4)
 Water samples were collected in 50-mL polyethylene bottles; they were kept cold (4°C) until analysis in the laboratory using standard segmented flow analysis (SFA) procedures. The precisions are ±0.02 μM for nitrite, ±0.1 μM for nitrate, ±0.05 μM for ammonium, ±0.02 μM for phosphate and ±0.05 μM for silicate.
3.3. Dissolved Oxygen (O2)
 It was directly collected into calibrated 110 mL glass flasks and, after fixation; they were kept in the dark until analysis in the laboratory 24 h later. O2 was determined by Winkler potentiometric endpoint titration using a Titrino 720 analyzer (Metrohm) with a precision of ±0.5 μmol kg−1. The apparent oxygen utilization, AOU = O2sat − O2, was calculated using the algorithm proposed by Benson & Krause [UNESCO, 1986] for oxygen saturation (O2sat).
 The corrected dissolved oxygen (O2C) is the O2 concentration in the water column when all the inorganic nitrogen is in the oxidation state of nitrate. Since 0.5 mole of oxygen are necessary to oxidize 1 mole of nitrite to nitrate and 2 moles of oxygen are required to oxidize 1 mole of ammonium to nitrate, O2C is calculated as [Ríos et al., 1989]
3.4. Total Alkalinity (TA) and Total Inorganic Carbon (CT)
 Samples for TA and pH (total hydrogen concentration scale, 25°C) were collected into 500-mL glass flasks and analyzed within a few hours in the base laboratory. Seawater pH was measured spectrophotometrically following Clayton and Byrne . The precision is ±0.003 pH units. TA was determined by titration to pH 4.4 with HCl, according to the potentiometric method of Pérez and Fraga  with a precision of ±2 μmol kg−1. The potential alkalinity (TAP) was calculated following Fraga and Álvarez-Salgado :
 Total inorganic carbon (CT) was calculated from pH and TA with the carbonic and boric acid dissociation constants of Lueker et al. . The estimated precision of this calculation is ±3 μmol kg−1.
 The influence of the precipitation/dissolution of CaCO3 on CT can be corrected by subtracting of TAP to CT [Broecker and Peng, 1982]. The resulting variable is called corrected CT (CTC):
3.5. Dissolved Organic Carbon (DOC) and Nitrogen (DON)
 Samples were taken into 500-mL acid-cleaned flasks and filtered through precombusted (450°C, 4 hours) 47 mm diameter Whatman GF/F filters (nominal pore size, 0.7 μm) in an acid-cleaned glass filtration system, under low N2 flow pressure. Aliquots for the analysis of DOC/DON were collected into 10 mL precombusted (450°C, 12 hours) glass ampoules. After acidification with H3PO4 to pH < 2, the ampoules were heat-sealed and stored in the dark at 4°C until analysis. DOC and DON were measured simultaneously with a nitrogen-specific Antek 7020 nitric oxide chemiluminescence detector, coupled in series with the carbon-specific Infrared Gas Analyzer of a Shimadzu TOC-5000 organic carbon analyzer [Álvarez-Salgado and Miller, 1998]. The precision is ±0.7 μmol C L−1 for carbon and ±0.2 μmol N L−1 for nitrogen. Their respective accuracies were tested daily with the reference materials provided by Prof. D. Hansell (Univ. of Miami). We obtained an average concentration of 45.7 ± 1.6 μmol C L−1 and 21.3 ± 0.7 μmol N L−1 (n = 26) for the deep ocean reference (Sargasso Sea deep water, 2600 m) minus blank reference materials. The nominal value for TOC provided by the reference laboratory is 44.0 ± 1.5 μmol C L−1; a consensus total dissolved nitrogen (TDN) value has not been supplied yet, but a mean ± SD value of 22.1 ± 0.8 μmol N L−1 for four HTCO systems and 21.4 μmol N L−1 for one persulphate oxidation method has been provided by Sharp et al.  as a result of the Lewes intercalibration exercise. DON was obtained by subtracting NT (=ammonium + nitrite + nitrate) from TDN.
3.6. Dissolved Organic Phosphorus (DOP)
 Samples were taken and filtered as indicated for DOC/DON. The filtrate was collected into 50 mL polyethylene containers and frozen at −20°C until analysis. It was measured by the SFA system for phosphate, after oxidation with Na2S2O8/borax and UV radiation [Armstrong et al., 1966]. Only the organic mono-phosphoric esters are analyzed because poly-phosphates are resistant to this oxidation procedure. Daily calibrations with phosphate, phenyl phosphate and adenosine 5′-monophosphate (AMP) in seawater were carried out. Standards of AMP were determined in order to calculate the mono-phosphoric esters recovery (∼80%). The precision of the method is ±0.04 μmol P L−1.
3.7. Dissolved Monosaccharides and Total Saccharides
 Sampling and storage procedures are identical to for DOP samples. Mono and total saccharides were determined by the oxidation of the free reduced sugars with 2,4,6-tripyridyl-s-triazine (TPTZ) followed by spectrophotometric detection at 595 nm [Myklestad et al., 1997; Hung et al., 2001]. The system was standardized daily with D-glucose. The precision is ±0.6 μmol C L−1 for monosaccharides and ±0.7 μmol C L−1 for total saccharides (DCho), and the detection limit was ∼2 μmol C L−1. See Nieto-Cid et al.  for further details.
3.8. Particulate Organic Carbon and Nitrogen (POC and PON)
 Suspended organic matter was collected under low vacuum on precombusted (450°C, 4 hours) 25 mm diameter Whatman GF/F filters of 0.7 μm nominal pore size (POC/PON, 0.5–1.5 L of seawater). All filters were dried overnight and frozen (−20°C) before analysis. Measurements of POC and PON were carried out with a Perkin Elmer 2400 CHN analyzer. An acetanilide standard was used daily. The precision of the method is ±0.3 μmol C L−1 and ±0.1 μmol N L−1.
3.9. Particulate Organic Phosphorus (POP)
 It was determined by H2SO4/HClO4 digestion at 220°C of the particulate material collected from 250–500 mL of seawater on Whatman GF/F filters. The phosphoric acid resultant from the digestion was analyzed with the SFA method for phosphate. The precision is ±0.02 μmol P L−1.
3.10. Particulate Saccharides (PCho)
 About 250–500 mL of seawater were filtered and stored as indicated for POC, PON and POP. PCho was determined by the anthrone method [Ríos et al., 1998]. It is based on the quantitative reaction of sugars with anthrone in a strongly acid medium at 90°C, to give an intensely colored compound. The absorption was measured at 625 nm. The system was calibrated daily with D-glucose. The precision of the method is ±0.1 μmol C L−1
3.11. Chlorophyll (Chl)
 Between 100 and 200 mL of seawater were filtered through GF/F filters and the filters were immediately frozen (−20°C) until analysis. Chl was determined with a Turner Designs 10000R fluorometer after 90% acetone extraction [Yentsch and Menzel, 1963]. The precision is ±0.05 μg L−1.
4. Results and Discussion
4.1. Hydrography of NW Iberian Shelf Waters
Figure 2 identifies the seven hydrographic periods defined by Nieto-Cid et al.  during the sampling period, on basis of the meteorological conditions (offshore Ekman transport, continental runoff) and the water column response (salinity and temperature):
4.1.1. Period 1 (15 May to 21 August)
 Period 1 is characterized by spring and summer upwelling events separated by short intervals of wind calm, which cause a marked thermal stratification with warm waters in the surface layer and cold and salty Eastern North Atlantic Central Water (ENACW) in the bottom (Figures 3a and 3b). Strong gradients of the chemical variables are observed, with high O2C and low NT, CTC, TAp, and SiO4 levels in seasonal thermocline waters and low O2C and high NT, CTC, TAp and SiO4 in shelf bottom waters (Figures 3c, 3d, and 4a–4c). Accumulations of DOC, DCho, POC and PCho in the surface layer were observed during the periods of wind calm (Figure 5). Maximum primary production rates in the study area are commonly observed in association with this succession of wind stress/relaxation events: Arístegui et al.  have proposed an average value of 2.5 g C m−2 d−1 during the upwelling season off NW Spain. It contrasted with the low organic matter content of ENACW on the shelf (<60 μM DOC, <5 μM DCho, <2.5 μM POC and <0.5 μM PCho), except that in the bottom nepheloid layer, where POC values >5 μM C were commonly observed.
4.1.2. Period 2 (28 August to 18 September)
 Period 2 is characterized by a late summer strong upwelling event, which produced the sudden uplift of the cold, salty, O2C and organic-poor matter and nutrient, CTC and TAp rich ENACW to the surface layer, where temperatures <15°C were recorded.
4.1.3. Period 3 (25 September to 30 October)
 Period 3 is characterized by autumn downwelling, forced by the predominant southerly winds and low continental runoff (Figure 2), provoked the entry of warm (>17°C), O2C and organic-rich matter and nutrient, CTC and TAp poor oceanic surface water.
4.1.4. Period 4 (6–20 November)
 Period 4 is the transition from stratification to vertical homogenization enhanced by strong northerly winds (Figure 2).
4.1.5. Period 5 (27 November to 13 February)
 Period 5 is the arrival of the Iberian Poleward Current (IPC) carrying warm and salty subtropical surface and central waters (Figures 3a and 3b) to our latitudes, producing a strong impact in the water column: NT levels remained <1 μM throughout the water column (Figure 3d) and dissolved and particulate organic matter reached minimum levels (Figure 5).
4.1.6. Period 6 (20 February to 26 March)
 Period 6 is characterized by winter mixing, the period of maximum vertical homogenization, when ventilation of the water column occurred (Figure 3c) and organic matter levels remained very low (Figure 5) because of the low primary production rates (<0.2 g C m−2 d−1 [Álvarez-Salgado et al., 2003]).
4.1.7. Period 7 (2–24 April)
 Period 7 is characterized by incipient spring stratification under dominant upwelling-favorable winds (Figure 2), which produced a surface accumulation of dissolved and particulate organic materials (Figure 5) in the first spring bloom peak of the season.
Figure 5 shows the seasonal evolution of the DOC, DCho, POC and PCho profiles at the mid shelf station. DON and DOP are significantly correlated with DOC: r = +0.78 for DOC versus DON (n = 284, p < 0.001) and r = +0.41 for DOC versus DOP (n = 160, p < 0.001). PON and POP also correlate significantly with POC: r = +0.95 for POC versus PON (n = 325, p < 0.001) and r = +0.90 for POC versus POP (n = 213, p < 0.001). In addition, DCho correlates with DOC (r = +0.70, n = 208, p < 0.001) and PCho with POC (r = +0.87, n = 215, p < 0.001).
 Despite the correlations referred above, average profiles of the DOC/DON, POC/PON, DOC/DOP and POC/POP molar ratios (Figures 6a–6d) exhibit a conspicuous vertical structure, characterized by a general increase with depth: the C/N/P ratios of the bottom samples are significantly different for the C/N/P ratios of the surface samples at p < 0.001 either for the dissolved or the particulate materials. This is consistent with previous results obtained from marginal seas and open ocean waters [Williams et al., 1980; Hopkinson et al., 1997; Sanders and Jickells, 2000; Hung et al., 2003] and can be explained by the preferential oxidation of organic nitrogen and phosphorus compounds at shallower levels in the water column [Suess and Müller, 1980; Shaffer et al., 1999]. The elemental C/N/P ratios of DOM are greater than the Redfield ratios, ranging from average C/N values of 12.6 in the surface layer to 14.5 in the bottom layer, and average C/P of 615 in the surface layer to 1090 in the bottom layer. It should be noticed again that the method for the determination of DOP is only able to analyze monophosphoric esters. By contrast, the elemental C/N/P ratios of POM are much closer to the Redfield values: from 6.8 and 100 for C/N and C/P in the surface layer to 9.5 and 140 in the bottom layer.
4.2. Stoichiometry of the Degradation of Biogenic Materials in Coastal and Shelf Waters of NW Spain
 Chemical data of waters below the upper mixed layer (AOU > 0) were analyzed along isopycnal surfaces to obtain respiratory ratios. Subsurface waters of the NW Iberian upwelling consists of a mixing of the subtropical and subpolar branches of ENACW with ENACW modified in the surface layer by heat exchange with the atmosphere and continental runoff from the Rías Baixas [Álvarez-Salgado et al., 1997]. Four isopycnal ranges (σ0) were defined: 26.7–26.8, 26.8–26.9, 26.9–27.0 and 27.0–27.1 (Figure 7a). Samples with σ0 < 26.7 were discarded, to minimize the ENACW samples affected by continental runoff/heat exchange with the atmosphere. Therefore for each isopycnal range, a simple two end-member mixing problem has to be solved to obtain an anomaly (ΔY) for each nutrient (Y)
where a0 and a1 are the coefficients of the linear regression of Y with temperature. Temperature was chosen as the conservative variable that retains the effect of the two ENACW types mixing because it presents a ratio of the precision of the analytical determination (er) to the standard deviation of the set of field measurements (SD) lower than salinity (see equation (6)). ΔY retains only the variability associated to the biogeochemical processes that occur within a given isopycnal range, i.e., the decomposition of organic matter and the dissolution of calcareous and siliceous structures.
Figure 7 also shows a selection of the property versus temperature and property anomaly versus property anomaly plots obtained after applying the isopycnal mixing analysis to the NW Iberian subsurface waters data set. Table 1 summarizes the results obtained analyzing the linear correlation between pair of anomalies of the chemical variables measured during this study for the mid ría (station 00) and mid shelf (station 03) sites. The anomalies obtained for each of the four isopycnal surfaces were analyzed together, to obtain average water column respiratory ratios. The best fit between any pair of nutrient anomalies (ΔX, ΔY) was obtained minimizing the function:
where Δ and Δ are the expected values of ΔX and ΔY from the linear regression equation respectively, i.e., Δi = m × ΔXi and Δi = ΔYi/m, with m been the slope of the correlation between ΔX and ΔY; wX and wY are weights for variables X and Y respectively, with wX, wY ≥ 0 and wX + wY = 1. The weight factors were estimated as a function of the precision of the analytical determination of the variable (er) compared with the standard deviation of the set of measurements of that variable for water samples with AOU > 0 (SD). For a given couple of variables X and Y
Table 1. Regression Coefficient (r), Slope, and Standard Error of the Slope of the Correlation Between Selected Pairs of Nutrient Anomalies for Samples With σ0 > 26.7 at Stations 00 and 03a
The value of wX, weight for variable X (see equation (6)), is also included. When the notation “%” is used, it means that the slope of the linear regression has been multiplied times 100.
 Minimizing equation (5) considering the weight factor of equation (6) ensures that the slopes of the linear regression equations account for the relative precision (er/SD) of the pairs of nutrient anomalies that are correlated each time. The resultant slopes are an intermediate case between (1) a type I regression, which should be applied when wX = 0, wY = 1 (or wX = 1, wY = 0) and (2) a type II regression, when wX = wY = 0.5 [Sokal and Rohlf, 1995]. The slope of a linear regression equation (Y on X) depends strongly on the regression type used: it ranges from a value of (ΔXi × ΔYi)/ΔXi2 for type I to a value of for type II. Table 1 summarizes also the values of wX for the corresponding X/Y pairs.
 Especially remarkable in Table 1 are the high correlations found between ΔO2C, ΔCTC, ΔNT and, ΔP and ΔSiO4, both in the ría and the shelf waters. The correlations are also significant between the anomalies of the dissolved and the particulate organic materials. These nutrient anomalies can be converted into proportions of saccharides (PCho), lipids (Lip), proteins (Prt), photosynthetic pigments (Chl) and phosphorus compounds (Pho). Table 2 summarizes the average composition of these groups of biomolecules and the relative contribution of each group to the average composition of marine phytoplankton proposed by Fraga et al. .
Table 2. Chemical Composition of the Main Organic Products of Synthesis and Early Degradation of Marine Phytoplankton According to Fraga et al. a
Percentages (in weight (w)) of each group correspond to the average composition of marine phytoplankton.
 Considering the chemical formulas in Table 2, the following set of five mass balance equations can be written for the particulate organic material:
where ΔPOC, ΔPOH, ΔPOO, ΔPON and ΔPOP are the anomalies (equation (4)) of particulate organic carbon, hydrogen, oxygen, nitrogen and phosphorus and ΔPrt, ΔPCho, ΔLip, ΔPho and ΔChl are the anomalies of particulate proteins, saccharides, lipids, phosphorus compounds and pigments, respectively. Since particulate C, N, P, Cho and Chl have been measured, the system can be solved to obtain the average chemical formula and the proportions of the different biomolecules. Unknowns are ΔPOH, ΔPOO, ΔPrt, ΔLip and ΔPho.
 The same set of equations can be used for the case of the dissolved organic matter assuming that the DOM that oxidized in the water column is composed of the same biomolecules as the particulate organic matter. This assumption is supported by (1) the similar results of Hopkinson et al.  in the Middle Atlantic Bight; (2) the reduced flushing time of shelf waters of the NW Iberian shelf, 1–2 weeks [Rosón et al., 1999]; and (3) the low ΔC/ΔN/ΔP ratios of DOM (Table 1). Another necessary assumption is that ΔChl = 0 for dissolved organic matter, which is very reasonable since the porphyrin groups of the chlorophylls are among the most resistant biomolecules in nature [McCarthy et al., 1997].
 For the case of the dissolved inorganic nutrients, the chemical composition of the oxidized material and the proportions of the different biomolecules are calculated from the O2, CT, NT and HPO42− anomalies. Equations describing the oxidation of proteins, saccharides, lipids and phosphorus compounds can be written
 The corresponding linear system of mass balance equations is
 Introducing the nutrient ratios of Table 1 in equations (7)–(11) and (16)–(19) it is possible to obtain the average biochemical composition of the organic and inorganic materials mineralized in bottom waters of the ría and the shelf of the NW Iberian upwelling system during a complete seasonal cycle. The RC ratio (=−ΔO2C/ΔCTC) cannot be directly used because it is affected by the dissolution of anthropogenic CO2. Therefore we tested the whole range of possible RC values. It varies from 1.00 (100% saccharides) to 1.59 (100% proteins). Following Anderson , we arbitrarily established that only the RC values that produce a percentage >5% for saccharides and lipids would be valid. Results are shown in Table 3; it ranges from 1.28 to 1.45 in the ría and from 1.39 to 1.49 in the shelf water. Note that the calculated RC values are 1.22 and 1.29 for the ría and the shelf, respectively (Table 1), i.e., out of the permitted range. Table 3 also shows the biochemical composition of the materials mineralized as obtained from the inorganic nutrients, the DOM and the POM anomalies. The results presented in Table 3 also allow us to write the following biochemical reactions for the decomposition of the biogenic material:
Table 3. Average Chemical Composition of the Products of Early Degradation of Marine Phytoplankton Photosynthesis as Obtained From Inorganic Nutrients, Dissolved Organic Matter, and Particulate Organic Mattera
RC is in mol O2 (mol C)−1; phosphorus compounds, proteins, lipids, and saccharides are in percent (w/w).
 Mid ría, from nutrients anomalies (using an average RC = 1.37)
 Mid ría, from DOM anomalies
 Mid ría, from POM anomalies
 Mid shelf, from nutrients anomalies (using an average RC = 1.44)
 Mid shelf, from DOM anomalies
 Mid shelf, from POM anomalies
 The H+ produced during the oxidation of biogenic materials is mainly consumed by the carbonic/boric acids buffer system of seawater.
4.2.1. Respiration of Organic Biomolecules
 Sinking, particulate and dissolved organic matter has different gravity properties, they cycle differently in the ocean, and they have different elemental ratios [e.g., Loh and Bauer, 2000; Hopkinson et al., 1997, 2002]. At station E00, the variation of CO2 caused by the oxidation of organic matter (CTC) was 25.6 μmol C L−1, whereas the variation of DOC and POC observed at this station was 4.0 and 4.4 μmol C L−1, respectively. This indicates that only 33% (=(4.0 + 4.4)/25.6) of the observed CTC increase can be accounted for by the combined loss of DOC (16%) and POC (17%). Similarly, CTC, DOC and POC variations of 10.0, 3.5 1.8 μmol C L−1 were observed at station E03, respectively. This indicates that about 53% (=(3.5 + 1.8)/10.0) of the CTC increase in the water column can be accounted for (35% from DOC and 18% from POC). These observations suggest that 50–70% of the change observed in the water column CTC might have been derived from the oxidation of organic matter, like large fast sinking particles, which could not be captured by our method (representative for particles in the 1–200 μm size range).
 Remarkable differences were found between the average composition of the dissolved, particulate and missing sinking organic materials that are mineralized in the waters below the upper mixed layer of the NW Iberian upwelling system. Regarding the dissolved material, there are not differences between the mid ría and mid shelf sites and the most conspicuous characteristic of this material is that saccharides are preferentially consumed (32–36%; Table 3) as compared with a material of Redfield composition (22.7%; Table 2). As for the case of the dissolved material, the composition of the oxidized particulate organic matter did not differ between the ría and the shelf (p < 0.05), but in this case lipids are preferentially oxidized: 31% (Table 3) as compared with 17.5% for a material of Redfield composition (Table 2). Finally, assuming average RC values of 1.37 and 1.44 for the mid ría and mid shelf stations respectively, it results that the missing fraction of mineralized organic material experiences a preferential consumption of saccharides in the mid ría (31% as compared with 22.7% in Table 2) and of proteins in the mid shelf (55% as compared with 46% in Table 2).
Hopkinson et al.  were able to differentiate between labile and recalcitrant materials in the mesotrophic Middle Atlantic Bight. The C/N/P ratios of the labile material were similar to the values obtained in the NW Iberian Peninsula and indicate that the stoichiometry of DOM respiration is similar to that for particles [Garber, 1984]. The study of Hopkinson et al.  showed that DOP was more labile than DON and DON more than DOC. This result was also observed by Lucea et al.  in a different environment: the oligotrophic NW Mediterranean. They both confirmed previous conclusions reached by Jackson and Williams  and Hopkinson et al.  about preferential degradation of phosphorus. Abell et al.  merit an special mention since they were able to differentiate the organic matter respiration in upper thermocline waters of the subtropical (oligotrophic) and temperate (mesotrophic) North Pacific Ocean: DOC+POC respiration exhibited a C/N molar ratio of 30 ± 10 in the subtropical and 8 ± 1 in the temperate North Pacific. Hung et al.  obtained a C/N molar ratio of DOM respiration of 8.4 and a N/P molar ratio of 19 in the East China Sea, suggesting that they are due to the recently produced fractions. Therefore there is a body of evidence in the literature supporting the fact that the labile dissolved organic material consumed in mesotrophic environments, especially in coastal areas, has a composition similar to the particulate and sinking organic particles.
 Since most experimental work on nutrients production from particulate and dissolved organic matter has dealt with laboratory decomposition studies, this work is likely the first estimation in the field of the molecular composition of the mineralized dissolved organic matter in a coastal environment. This type of analysis is more accurate than the estimation of respiration ratios based on the vertical profiles of the bulk C/N/P ratios for DOM and POM as shown in Figure 6, because DOM and POM concentrations can contain refractory components and preformed contributions.
4.2.2. Dissolution of Siliceous and Calcareous Structures
 The ratio between soft and hard biogenic material decomposition rates varies with depth in the oceans. Since the upper ocean is CaCO3 over saturated [Takahashi et al., 1981; Broecker and Peng, 1982], carbon regeneration at that level is predominantly due to the oxidation of organic carbon [Honjo et al., 1982; Honjo and Manganini, 1993]. However, a recent review of the global carbonate budget showed evidences that 60–80% of the biogenic CaCO3 dissolves in the upper 1000 m, above the lysocline, as a result of biological mediation [Milliman et al., 1999]. In deeper layers, the dissolution of hard structures predominates because of the increased pressure, decreased temperature and longer residence times [Broecker and Peng, 1982]. These authors, using the GEOSECS data set, estimated a ΔCaCO3/ΔCTC ratio of ∼0.1 for the permanent thermocline and ∼0.5 for deep waters. Takahashi et al.  obtained a ΔCaCO3/ΔCTC ratio of 0.05 for the upper and 0.08 for the lower permanent thermocline of the North Atlantic. Finally, Ríos et al.  found a ΔCaCO3/ΔCTC ratio of 0.08 for the upper 2300 m and 0.52 for deeper waters of the eastern North Atlantic Ocean.
 A ΔCaCO3/ΔCTC ratio of 0.05 ± 0.01 was estimated for shelf subsurface waters of the NW Iberian upwelling system (Table 1). This value is within the range mentioned for the upper thermocline waters. A higher ΔCaCO3/ΔCTC value of 0.08 ± 0.01 was calculated for the subsurface waters of the Ría de Vigo. The water column below the mixed layer is always supersaturated with respect to calcite and aragonite (average ± SD, 312 ± 42% and 200 ± 28% respectively). Therefore CaCO3 dissolution should occur preferentially at or near the top of sediments [Archer et al., 1989]. The high organic matter content of the sediment, either in the rías [Vilas et al., 2005] or the inner shelf [Lopez-Jamar et al., 1992], creates an environment where the dissolution of CaCO3 is favored by strong acidification. In fact, the sediments inside the rías are characterized by low proportions of CaCO3 where the organic matter content is high [Vilas et al., 2005].
 Regarding biogenic silica, it is also exported to the deep ocean as sinking particles, where dissolution occurs. As for the case of CaCO3, most of the silica dissolution takes place below the main thermocline. The high covariation between alkalinity and silicate profiles in open ocean deep waters [Brewer et al., 1995; Broecker and Peng, 1982; Ríos et al., 1995] is probably due to the biologically mediated dissolution of these hard structures in the microenvironments created by marine snow, zooplankton guts, etc. According to Milliman et al.  and Tréguer et al. , the open ocean CaCO3 and opal decomposition rates in the water column are 47·1012 mol yr−1 and 91·1012 mol yr−1 respectively. Therefore a mean ΔSiO4/ΔCaCO3 molar decomposition ratio of ∼2 can be proposed. For the North Atlantic, an average ΔSi/ΔCaCO3 ratio for the whole water column of 1.4 was calculated by Pérez et al. . According to Berger and Herguera , a ΔSiO4/ΔCaCO3 ratio of 1.4 is expected for an area with a mean organic carbon flux of 10 mmol m−2 d−1, which is in close agreement with their productivity value for the eastern North Atlantic Ocean [Martin et al., 1993]. Lower ΔSiO4/ΔCaCO3 ratios, around 1.05, were measured in sediment traps deployed at the North Atlantic Bloom Experiment (NABE) site below 3100 m [Newton et al., 1994].
 In this study, the ΔSiO4/ΔCaCO3 ratio was found to be variable, with a value of 2.5 ± 0.6 in the Ría de Vigo and 1.3 ± 0.2 in the shelf (Table 1). The latter value is in close agreement with that found at the NABE site. As observed in the open ocean [Berger and Herguera, 1992], the ΔSiO4/ΔCaCO3 ratio increases from the shelf to the ría as the organic matter flux increases. Along the middle shelf of the NW Iberian Peninsula, Álvarez-Salgado et al.  found a large silicate accumulation in bottom waters due to rapid opal dissolution. Opal dissolution was high compared with organic matter decomposition: the ΔNT/ΔSiO4 molar ratio in the ría (1.1 ± 0.1) is about half than in the shelf, (1.8 ± 0.1; Table 1). In this sense, maximum percentages of opal and diatom valve numbers in sediments were recorded in the main channel of the Ría de Vigo [Prego et al., 1995], suggesting a strong downward flux of diatom frustules there.
4.3. Contribution of DOM to Respiration in the NW Iberian Upwelling
 A huge effort has been made over the last decade by marine biogeochemist to understand the role played by DOM in carbon, nitrogen and phosphorus cycles in the oceans. Most of this knowledge has been reviewed in a recent book edited by Hansell and Carlson .
 One of the key questions to answer is the true contribution of DOM to the apparent oxygen utilization of subsurface ocean waters. According to the results of the biogeochemical general circulation model of Yamanaka and Tajika , ∼70% of the biogenic organic matter exported from the surface (<100 m) to the central (100–500 m) waters of the World Ocean are sinking particles. The remaining ∼30% is in the dissolved form. DOM below 1000 m is extremely refractory, confirming that recycling times of this pool range from years to thousands of years [Hansell and Carlson, 1998] and that oxygen consumption is almost exclusively due to large sinking particles [Jahnke, 1996].
 The analysis of the variation of calculated ΔCTC, ΔDOC and ΔPOC values on the previous section yielded that from 16% to 35% of the organic carbon regenerated in bottom waters of the ría and the shelf, respectively, was in the dissolved form. The increase in the offshore direction is consistent with the larger particle fluxes recorded within the rías [Varela et al., 2004] than over the shelf [Olli et al., 2001]. It is also remarkable the low correlation between DOC and CTC anomalies (lower than −0.33), which confirms the general statement that the variability observed in the AOU of the oceans is mainly linked to the flux of large sinking organic particles [Anderson and Sarmiento, 1994]. DON/NT and DOP/PO4 anomalies reproduced the same pattern observed for DOC/CTC: the contribution of dissolved organic matter to the mineralization of organic nitrogen and phosphorus increased from the ría to the shelf, ranging from 7% to 30%, and the correlation coefficients were also very low, ranging from −0.20 to −0.44. In the continental shelf of Georges Bank, a region of nearly constant upwelling of nutrient rich deep water that can be compared with the NW Iberian shelf, Hopkinson et al.  obtained that 19% and 15% of the mineralized N and P that accumulates in deeper water is due to dissolved organic matter respiration. In open ocean waters of the North Pacific, DON contributed from 10% to 25% to the mineralized nitrate in deep waters according to different estimates [Jackson and Williams, 1985; Maita and Yanada, 1990]. As indicated in the previous section, Abell et al.  differentiated the organic matter respiration in upper thermocline waters of the subtropical (oligotrophic) and temperate (mesotrophic) North Pacific Ocean: dissolved+particulate organic carbon contributed 70% to organic mater respiration in the subtropical North Pacific and only 20% in the temperate North Pacific.
 Four main conclusions can be extracted from this work about the decomposition of biogenic materials in a coastal upwelling system:
 1. More than 50% of the oxygen consumption in shelf subsurface waters is due to the respiration of large, fast sinking organic matter, 17–18% to particulate organic matter and 16–35% to dissolved organic matter.
 2. The products of early degradation of the sinking particulate and dissolved organic matter can be expressed as a linear combination of the main groups of biomolecules: lipids, saccharides, proteins and phosphorus compounds.
 3. Proteins are preferentially oxidized in the sinking organic matter, saccharides in the dissolved organic matter and lipids in the particulate organic matter, as compared with a biogenic material with the reference Redfield composition.
 4. Biogenic opaline silica preferentially dissolved in the middle ría.
 The authors wish to thank the captain and crew of R/V Mytilus and the members of the Department of Oceanography of the Instituto de Investigacións Mariñas and the Group of Physical Oceanography of the University of Vigo for their valuable help. Special thanks go to T. Rellán for dissolved oxygen analyses and J. L. Cortijo and S. Piedracoba for the thermohaline and meteorological data. Financial support for this work came from MCYT grant numbers MAR1999–1039–C02–01 and MAR2000–0880–C02–01 and XUGA grant number PGIDT01MAR40201PN. Fellowships of the Spanish MCYT, the I3P CSIC programme, the MCYT grant REN2002-00503/MAR and the Fundación Provigo funded M.N.-C. and S.B. to carry out this work. C.G.C. developed this work with a contract of the Ramon y Cajal programme of the MCYT.