lable at ScienceDirect

Estuarine, Coastal and Shelf Science 83 (2009) 126–132

Contents lists avai

Estuarine, Coastal and Shelf Science

journal homepage: www.elsevier .com/locate/ecss

Strong CO2 outgassing from high nutrient low chlorophyll coastal waters offcentral Chile (30�S): The role of dissolved iron

Rodrigo Torres a,*, Patricio Ampuero b

a Centro COPAS, Universidad de Concepcion, Casilla 160-C, Concepcion, Chileb Laboratorio de Geoquımica Organica Marina, Departamento de Oceanografıa, Universidad de Concepcion, Casilla 160-C, Concepcion, Chile

a r t i c l e i n f o

Article history:Received 7 March 2008Accepted 27 February 2009Available online 24 March 2009

Keywords:iron limitationair–sea CO2 exchangephytoplankton productivitycoastal upwellingChileCoquimbo (30�S)

* Corresponding author. Present address: Centro dede la Patagonia (CIEP), Bilbao 449, Coyhaique, Chile.

E-mail address: rtorres@ciep.cl (R. Torres).

0272-7714/$ – see front matter � 2009 Elsevier Ltd.doi:10.1016/j.ecss.2009.02.030

a b s t r a c t

Carbonate system parameters (pH and alkalinity) were used to estimate the coastal water CO2 fluxes offcentral Chile (30�S) during September 2007. Coastal waters rich in nitrate and silicate were strongly CO2

supersaturated and normally poor in chlorophyll a. MODIS satellite chlorophyll a data suggest thatphytoplankton biomass remained particularly low during September 2007 although coastal waters werehighly fertilized with nitrate and silicate. The phytoplankton gross primary productivity in macronu-trient-rich waters was very low with the exception of shallow waters (e.g. within or near bays). Severaliron-enrichment bottle experiments show that fCO2 rapidly decreases during iron-enrichment treat-ments compared to controls. This suggests that iron limitation of phytoplankton growth (mainly dia-toms) plays a role in maintaining high-CO2 outgassing by preventing rapid interception of upwelled CO2.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

The role of dissolved iron in the sequestration of atmosphericCO2 over large portions of the open ocean (Martin, 1990) has beengreatly debated during the last two decades (Jickells et al., 2005).However, much less attention has been paid to its effect on theintensity and variability of coastal air–sea CO2 fluxes. Coastal areasaffected by intense upwelling are often described as sources of CO2

(Borges et al., 2005). Certainly, the deep upwelling of normally coldand already CO2-supersaturated subsurface waters leads to animmediate outgassing of CO2 at the coastal divergence (Torres et al.,1999). However, the high macronutrient (nitrate and phosphate)content of the upwelled water can stimulate intense blooms ofphytoplankton that sequester inorganic carbon, reducing thefugacity of CO2 (fCO2) and thus reducing or even reversing the CO2

flux at the air–sea interface (Simpson and Zirino, 1980). High spatialand temporal variability in coastal CO2 air–sea fluxes has beenattributed to coastal circulation (Torres et al., 1999) and/or biolog-ical activity (Simpson and Zirino, 1980). Due to the ‘pulsed’ natureof the forces driving upwelling (often referred to as events, i.e.Upwelling Favorable Wind Events, UFWE), coastal upwelling is alsovariable and a rapid interception of the upwelled carbon by

Investigacion en Ecosistemas

All rights reserved.

phytoplankton may require a close coupling between macronu-trient fertilization and phytoplankton productivity. In the lastdecade, a growing amount of evidence suggests that a decouplingbetween macronutrient fertilization and the generation of phyto-plankton blooms (typically characterized by diatom proliferation)can occur in coastal upwelling areas due to non-optimal concen-trations of dissolved iron in the water that supplies the upwelling(Bruland et al., 2005, and references there in).

Coastal upwelling HNLC (High Nutrient Low Chlorophyll) areasand/or periods associated with low levels of dissolved iron havebeen reported for California (Hutchins and Bruland, 1998; Hutchinset al., 1998; Johnson et al., 1999; Bruland et al., 2001; Firme et al.,2003), Peru (Hutchins et al., 2002; Bruland et al., 2005), andrecently off Coquimbo, Chile (Torres et al., submitted for publica-tion). Suboptimal levels of dissolved iron for diatom growth havebeen ultimately linked to physical features that prevent additionaliron inputs into the upwelled water, for example: a narrow conti-nental shelf, little or no fresh water input, and air mass trajectoriesthat would minimize dust inputs from surrounding land masses(Hutchins et al., 1998; Johnson et al., 1999; Bruland et al., 2005). TheCoquimbo upwelling system (w30�S, Central Chile) exhibits all theaforementioned characteristics as well as consistently low levels ofdissolved iron in nitrate-rich recently upwelled waters (w1 nMtotal dissolved iron; Torres et al., submitted for publication).

In this paper, we will suggest that suboptimal levels of dissolvediron in highly CO2-supersaturated coastal water off Central Chile(30�S) prevent a rapid interception of the upwelled CO2.

R. Torres, P. Ampuero / Estuarine, Coastal and Shelf Science 83 (2009) 126–132 127

2. Methods

2.1. Oceanographic stations

Discrete surface seawater samples and SST data were collectedfrom the Stella Maris II during the austral spring of 2007(September) off the coast of central Chile (29.8�S–30.3�S). Thesampling was divided into two parts: nine stations during earlySeptember (September 11th and 13th, 2007; Stations 1–9) and 10stations during late September (September 28th, 2007; Stations10–19; See Fig. 1). Surface-water samples (collected using an acid-cleaned, all plastic ‘sipper’ system) were taken for chemical analysis(pH, alkalinity, chlorophyll a, nutrients and others measurements)and for bottle incubation experiments.

2.2. Surface-water carbonate system parameters and CO2 net fluxat the air–sea interface

Samples for pH and Total Alkalinity (AT) were analyzedfollowing the DOE (1994) potentiometric method and the methodof Haraldsson et al. (1997), respectively. The pH, AT and hydro-graphic data were used to calculate the seawater fCO2 using CO2SYSsoftware (Lewis and Wallace, 1998). Seawater fCO2 was calculatedusing Mehrbach solubility constants (Mehrbach et al., 1973) refittedby Dickson and Millero (1987).

The net flux of CO2 (FCO2) was calculated as FCO2¼ kaDfCO2,where k is the CO2 transfer velocity (k), a is the solubility of CO2 inseawater (Weiss,1974), and DfCO2 is defined as the difference in fCO2

between the surface water and atmosphere. In order to estimate k(Wanninkhof and McGillis,1999), daily QuikSCAT Mean Wind Fields(for an area of 0.5� � 0.5�, centered at 29.75�S and 71.75�W) wereobtained from CERSAT (http://www.ifremer.fr/cersat). In order toestimate DfCO2, an approximate value of 389 matm for atmosphericfCO2 was extrapolated from measurements of the atmospheric molefraction of CO2 in Patagonia from September 1994 to September2002 (NOAA GMD Carbon Cycle Cooperative Global Air SamplingNetwork; Tans and Conway, 2005; data obtained from http://cdiac.ornl.gov/trends/co2/cmdl-flask/tdf.html).

Fig. 1. The study area and oceanographic stations. The dashed line depicts the 200 misobath.

2.3. Iron-enrichment bottle experiments

Unfiltered surface seawater for bottle incubation experimentswas collected from 11 stations during September 2007 (Stations 1, 4,5–9, 13–17; see Fig. 1). Water was transferred into polycarbonateacid-cleaned bottles. The bottles were either spiked with inorganiciron – to produce a nominal increase of 5 nM in iron concentration –or left as controls. The bottles were incubated for 5–7 days atsurface-water temperature (and natural illumination) in a coastallaboratory at Coquimbo (Universidad Catolica del Norte).

2.4. Nutrients, chlorophyll a, dissolved oxygen and gross primaryproduction

Nitrate and silicate were analyzed in unfiltered seawatersamples within 24 h after collection, following the methods ofStrickland and Parsons (1968). Samples for chlorophyll a analysiswere filtered through 0.7 mm glass fiber filters and maintainedunder liquid nitrogen until analysis following Holm-Hansen et al.(1965). Additionally, Moderate Resolution Imaging Spectroradi-ometer (MODIS) derived chlorophyll a data were obtained fromhttp://las.pfeg.noaa.gov/oceanWatch/. Gross Primary Production(GPP) was estimated using the oxygen incubation method (24 hincubation in w300 ml Wheaton BOD bottles). Dissolved oxygenwas measured by a manual method based on the Winkler proce-dure using a piston burette (Dickson, 1995).

3. Results

3.1. Strong CO2 outgassing in HNLC coastal waters off Coquimboduring September 2007 (austral spring)

Winds in the Coquimbo upwelling area were upwelling favor-able (equatorward) during September 2007; typically, a week longUFWEs occur between short periods (1 or 2 days) of upwellingrelaxation (Fig. 2). The cold coastal waters (ranging from 11.5 �C to13 �C; data not shown) were normally very rich in nitrate andsilicate (w17 mM; Fig. 3a–b), low in pH at 25 �C (w7.6 pH units;Fig. 3c), and poor in chlorophyll a (normally <1 mg Chla m�3;Fig. 3d). We will define the surface waters containing less than 1 mgChla m�3 and more than 16 mM of nitrate as High Nutrient LowChlorophyll-Recently Upwelled Water (HNLC-RUW), which was thecase for Stations 1–3, 5–9 and 13–16. The HNLC-RUW was invari-ably high in fCO2 (735� 56 matm, n¼ 12; Fig. 3e) and with partic-ularly low levels of primary production (0.3� 0.1 mmol C kg�1 h�1,n¼ 4; Fig. 3f). Exceptions to the HNLC pattern occurred at stationslocated very close to the shore or in front of the bay system(Stations 4, 10–12, 17–19; Fig. 1) where the surface-water fCO2 wasrelatively low (591�185 matm, n¼ 7; Fig. 3e) and primaryproduction was relatively high (2.4�1.1 mmol C kg�1 h�1, n¼ 4;Fig. 3f). High wind speeds from September 4th–29th, 2007(8� 3 m s�1) drove high values of k (25� 21 cm h�1), which,combined with the strong CO2 supersaturation in HNLC-RUW (theCO2 saturation ranged from 156% to 209%, with a mean value of188% equivalent to DfCO2¼ 346 matm, see methods), led to a veryintense estimated carbon flux from the HNLC-RUW to the atmo-sphere (83� 69 mmol m�2 day�1). The lower levels of surface fCO2

found in chlorophyll-rich coastal waters (more than 1 mgChla m�3) resulted in a less intense estimated CO2 flux from theocean to the atmosphere (50� 41 mmol m�2 day�1).

Strongly CO2-supersaturated HNLC-RUW was detected in bothsampling periods: early September (September 11th–13th, 2007;Stations 1–3 and 5–9) and late September 2007 (September 28th,2007; Stations 13–16). MODIS chlorophyll a data available forSeptember 2007 confirmed that the coastal waters off the

Fig. 2. Meridional wind speed off Coquimbo during September 2007. Positive values depict equatorward velocities (upwelling favorable). Numbers on the bars depict theoceanographic stations sampled on the 11th, 13th and 28th of September 2007.

R. Torres, P. Ampuero / Estuarine, Coastal and Shelf Science 83 (2009) 126–132128

continental shelf were low in chlorophyll a (<1 mg Chla m�3)compared with areas over the continental shelf (mainly withinbays, see Fig. 4a–b). On October 1st, higher levels of chlorophylla were observed off the continental shelf at latitudes lower than30.1�S; however, south of 30.1�S, chlorophyll a was as low as levelsmeasured a couple of weeks before (Fig. 4c).

3.2. Relationship between nitrate, silicate and fCO2 in surfacewaters

The nitrate and silicate concentrations were positively corre-lated in both HNLC-RUW and high-chlorophyll coastal waters(R2¼ 0.55, p-value¼ 0.006, n¼ 12 and R2¼ 0.83, p-value¼ 0.004,n¼ 7, respectively; Fig. 5a). The fCO2 in HNLC-RUW was notcorrelated with nitrate or silicate (p-value> 0.05, n¼ 12).Conversely, in high-chlorophyll waters (i.e. Chla> 1 mg m�3), fCO2

was highly correlated with nitrate and silicate (R2 was 0.86 and0.98, respectively; p-value <0.005; n¼ 7).

3.3. Rapid CO2 reduction in off-shelf HNLC-RUW experimentallyenriched with dissolved iron

Iron-enrichment bottle experiments performed with HNLC-RUW (Stations 1, 5–9, 13–15; Fig. 6) showed that at the end ofincubation the reduction in fCO2 was particularly intense for theiron-enriched treatment compared with controls, with the excep-tion of Station 5 (Fig. 6). The largest fCO2 difference (absolute value)between control and iron-enriched treatment corresponded to off-shelf stations (Sts. 8, see also 14, 9, 15, 1, 13 and 7 at Table 1) whichwere invariably characterized by initial HNLC-RUW conditions(Table 1). Differently, the small fCO2 difference (absolute values)between control and iron-enriched treatment at the end of theincubation occurs at Sts. 5, 17, 6 and 4 (Table 1). Note that HNLC-RUW located over the continental shelf (at Station 6, Fig. 6) does notdevelop large differences between treatments (Table 1), contrastingwith most HNLC-RUW located off the continental shelf that dodevelop large differences between treatments (e.g. Sts. 8, 14, 9, 15and 1; see Table 1). In general, while the incubations of watercollected off the continental shelf showed statistically significantdifferences between control and Fe-enriched treatments (WilcoxonMatched Pair Test, p-level¼ 0.02, n¼ 8; see also Fig. 6), no differ-ences between treatments was observed in the water collected overthe continental shelf (p-level¼ 0.3, n¼ 3).

Due to the slow reduction in observed fCO2 in most of the ‘‘off-shelf’’ control bottles, CO2 supersaturation remained longer incontrols compared with the iron-enriched treatments (e.g. the CO2

levels in control treatments remained supersaturated for up to 6days for Stations 8 and 9; Fig. 6). An experiment using chlorophyll-

rich, ‘‘over shelf’’ CO2-undersaturated surface water (Station 17)was characterized by extremely low fCO2 (Fig. 6) and the smallestdifference between the control and the iron-enrichment treatment(see absolute values at Table 1).

4. Discussion

4.1. The role of the continental shelf on CO2 sequestration bydiatoms in the Coquimbo upwelling system during September 2007

Wind-forced upwelling was intense at 30�S during September2007 (Fig. 2). A consistent deep (intense) upwelling can explain theobserved high levels of surface nutrients (NO3

�, CO2) and the lowlevels of chlorophyll in RUW. MODIS chlorophyll data forSeptember 2007 (Fig. 4a–b) confirmed that phytoplankton biomasswas low in nitrate-rich waters off the continental shelf (Fig. 3a; notethat the continental shelf is extremely narrow at latitudes>30.25�Sor in front of Coquimbo at w30�S). High levels of chlorophyll a andprimary production were confined to the bay areas and/or areaswithin the continental shelf (Figs. 3 and 4). The high correlationbetween the concentrations of nitrate, silicate, and fCO2 in high-chlorophyll surface waters (Fig. 5) suggests that diatoms were thefunctional group driving the consumption of CO2 and NO3

�. Indeed,the analysis of water samples (by SEM and optic microscopy)showed that chain-diatoms dominate the phytoplankton abun-dance in coastal waters over the continental shelf (Ampuero, 2007).Conversely, low levels of photosynthesis off the continental shelfduring September 2007 (Fig. 3f) were unable to rapidly reduce theintense CO2 outgassing. Consistent with this, diatom abundance atHNLC-RUW stations was low and the plankton community wasdominated by small flagellates (Ampuero, 2007).

The availability of iron is expected to be critical for the shift fromhigh nitrate, small flagellate dominated RUW to a diatom domi-nated system (Hudson and Morel, 1990). During September 2007,chlorophyll a and opal were found to be positively correlated(Ampuero, 2007), suggesting that the maximum inner-bay chlo-rophyll a shown by MODIS (Fig. 4) corresponds to diatom prolif-eration. The continental shelf appears to be the dominant factor fordetermining elevated iron levels in RUW (Johnson et al., 1999), andthus, the continental shelf may constitute a critical factor deter-mining the rapid production of diatom biomass and a rapidreduction in inorganic nutrients (including carbon). Since ironenrichment promotes inorganic carbon drawdown in open oceaniron-limited waters (e.g. Southern Ocean; Bakker et al., 2001), weexpect that the iron enrichment of RUW on the continental shelf isa critical step for rapid interception of the upwelled CO2 as weargue based on our iron-enrichment bottle experiments (seebelow).

Fig. 3. Surface levels of (a) Nitrate, (b) Silicate, (c) pH at 25 �C, (d) Chlorophyll a, (e) fCO2 and (f) GPP during September 2007. The diameters of the circles are proportional to thevalues. Maximum and minimum values are shown in the right corner of each plot.

R. Torres, P. Ampuero / Estuarine, Coastal and Shelf Science 83 (2009) 126–132 129

4.2. Interpreting changes in fCO2 from iron-enrichment bottleexperiments

The large decrease in fCO2 at the end of a 5 to 7-day incubationperiod (Fig. 6) was most probably an artifact caused by severalfactors, e.g. absence of vertical diffusion, absence of CO2 air–seaexchanges, higher levels of light during incubation (i.e. absence of

vertical mixing), and lower predation on primary producers (i.e.absence of migratory predators). We will discuss only the differ-ences between the control and the iron-enriched treatment. Thesmall differences (absolute value) between treatments occurredmainly in stations located over the continental shelf (Table 1),suggesting that these locations were iron replete (i.e. chlorophyll-rich water at Sts. 17 and 4, and HNLC-RUW at St.6; see Table 1). Most

Fig. 4. MODIS Chlorophyll a surface field during September and early October 2007.

R. Torres, P. Ampuero / Estuarine, Coastal and Shelf Science 83 (2009) 126–132130

probably shallow waters were already enriched in dissolved irondue Fe inputs from sediments (see Bruland et al., 2005 and refer-ences there in). We expect that a rapid consumption of CO2 in ironrich - RUW, will rapidly reduce and revert the initial CO2 flux,consistently iron replete upwelled waters off Oregon (Chase et al.,2005) are typically described as CO2 undersaturated (Hales et al.,2005).

Differently, at stations located off the continental shelf:

(1) Most of the experiments showed that a larger reduction in fCO2

occurred during the iron-enrichment treatment than incontrols (Table 1);

(2) One experiment (Station 8; Fig. 6) showed that while thecontrol treatment remained CO2 supersaturated, the iron-enriched treatments was strongly undersaturated in CO2 after6 days of incubation.

While (1) and (2) suggest that iron enhances the capacity of thephytoplankton community to rapidly lower CO2 surface levels, (2)suggests that the lack of dissolved iron can result in CO2 outgassingfor up to a week (although we expect that at the ocean surface CO2

supersaturation will remain far longer due to light limitation andhigh vertical carbon flux in poorly stratified RUW). In the Coquimboupwelling system, Upwelling Favorable Wind Events (UFWE) aretypically on a scale of 5–7 days (e.g. Fig. 2; see also Rutllant, 1994),

Fig. 5. Relationship between (a) Nitrate and Silicate, (b) fCO2 and Nitrate, and (c) fCO2 andcircles depict high Chlorophyll a (>1 mg m�3 Chla). The solid line in (b) and (c) shows a re

so (2) is important because it suggests the lack of iron can allow forcontinuous CO2 outgassing.

4.3. Uncoupling of nutrient fertilization and primary production inareas of intense coastal upwelling off the west coast of SouthAmerica: the connection between strong CO2 outgassing and ironlimitation

The prevalence of the high-CO2 HNLC condition describedbefore (Section 3.1) could be related to the intensity and/orpersistence of upwelling: The areas covered by HNLC-RUW areexpected to increase in size with the intensity of upwelling, but theareas expected to be enriched with dissolved iron remain constantin size, since they depend on the extension of the continental shelf.In the Coquimbo upwelling system, as well as many othertemperate coastal upwelling systems, upwelling intensifies duringthe spring (Bakun, 1990), so it is during these periods when weexpect more intense CO2 outgassing and a more extended condi-tion of iron limitation.

Borges et al. (2005) pointed out that coastal upwellingsystems characterized by high upwelling index values tend to besources of CO2. The intensity of upwelling-favorable winds alongthe west coast of South America peaks at approximately at 15�Sand 30�S (Pizarro, 2000). Consistently deep upwelling of DIC(Dissolved Inorganic Carbon) enriched water results in strongly

Silicate at stations 1–19 during September 2007. The dots depict HNLC-RUW and thegression model derived from high-Chlorophyll a data (depicted by circles).

Fig. 6. fCO2 changes in iron-enrichment bottle experiments during September 2007. The Y-axis corresponds to fCO2 in matm. The base line depicts atmospheric fCO2. The three barsdepict, from left to right: fCO2 at the beginning of incubation (i.e. t¼ 0 days), fCO2 in the control treatment at the end of incubation (i.e. C; at t¼ 5–7 days) and fCO2 in the iron-enriched treatment at the end of incubation (i.e. þFe, at t¼ 5–7 days). Error bars at Stations 1, 5, 6, 7, 15 and 17 depict one standard deviation (two replicas). Note that while barsbelow the base line depict CO2 undersaturation, the bars above the base line depict CO2 supersaturation.

R. Torres, P. Ampuero / Estuarine, Coastal and Shelf Science 83 (2009) 126–132 131

CO2-supersaturated coastal waters at 15�S (Simpson and Zirino,1980) and at 30�S (Torres et al., 1999, 2002).

The coastal upwelling that fertilizes (with macronutrients,including CO2) the euphotic layer also reduces the stratification ofthe water column. Thus, light rather than macronutrients could belimiting primary production and the consumption of nutrients inrecently upwelled waters (e.g. MacIsaac et al., 1985). Since ironlimitation can be enhanced at low levels of light (iron–light co-limitation; see revision by Watson, 2001), areas of deep (intense)upwelling can be particularly sensitive to iron limitation (i.e. ifRUW is not enriched in dissolved iron from sediments; see Section4.2.) and thus inefficient in rapidly intercepting upwelled CO2 bymeans of primary productivity. On the other hand, outgassing of

Table 1fCO2 difference between the control and the Fe-enriched, at the end of each incu-bation experiment.

Station fCO2 iron enriched�fCO2control (matm)

Characteristics Location

8 �234 HNLC-RUW Off-Continental Shelf14 �126 HNLC-RUW Off-Continental Shelf9 �117 HNLC-RUW Off-Continental Shelf15 �114 HNLC-RUW Off-Continental Shelf1 �110 HNLC-RUW Off-Continental Shelf13 �97 HNLC-RUW Off-Continental Shelf7 �66 HNLC-RUW Off-Continental Shelf4 �49 High Chlorophyll Over-Continental Shelf6 �32 HNLC-RUW Over-Continental Shelf17 16 High Chlorophyll Over-Continental Shelf5 35 HNLC-RUW Off-Continental Shelf

CO2 from poorly stratified upwelling waters can be very intense atthese windy latitudes (Torres et al., 2002). Since the continentalshelf is particularly narrow in both areas (15�S and 30�S), a reducedinput of dissolved iron from sediment to the upwelled volume isexpected (see Bruland et al., 2005), which is consistent withexperimental results suggesting that suboptimal levels of dissolvediron do limit diatom growth in both areas (Hutchins et al., 2002 andTorres et al., submitted for publication). We suggest that the largeCO2 outgassing in coastal water at 15�S and 30�S is not independentof the availability of iron, and thus spatial and temporal variabilityin dissolved iron inputs must be considered in order to understandthe variability of coastal air–sea CO2 fluxes.

Acknowledgements

We would like to thank Praxedes Munoz at Universidad Catolicadel Norte for providing coastal laboratory facilities; Ivan Peric, Headof the Department of Analytical and Inorganic Chemistry atUniversity of Concepcion for providing laboratory space for ourtrace metal work; Carmen Morales and Silvio Pantoja for providinglaboratory facilities at Dichato Marine Station; and the captain andcrew of the Stella Maris for their helpful assistance at sea. This studywas funded by Fondecyt 1060694.

References

Ampuero, P., 2007. Efectos del hierro disuelto sobre la productividad primaria yestructura fitoplanctonica en el centro de surgencia costera de Coquimbo(w30�S). Tesis, Universidad de Concepcion, Chile, unpublished manuscript.

Bakker, D., Watson, A., Law, C., 2001. Southern Ocean iron enrichment promotesinorganic carbon drawdown. Deep-Sea Research II 48, 2483–2507.

R. Torres, P. Ampuero / Estuarine, Coastal and Shelf Science 83 (2009) 126–132132

Bakun, A., 1990. Global climate change and intensification of coastal oceanupwelling. Science 247, 198–201.

Borges, A.V., Delille, B., Frankignoulle, M., 2005. Budgeting sinks and sources of CO2in the coastal ocean: diversity of ecosystems counts. Geophysical ResearchLetters 32, L14601. doi:10.1029/2005GL023053.

Bruland, K.W., Rue, E.L., Smith, G.J., 2001. Iron and macronutrients in Californiacoastal upwelling regimes: implications for diatom blooms. Limnology andOceanography 46, 1661–1674.

Bruland, K.W., Rue, E.L., Smith, G.J., DiTullio, G.R., 2005. Iron, macronutrients anddiatom blooms in the Peru upwelling regime: brown and blue waters of Peru.Marine Chemistry 93, 81–103.

Chase, Z., Hales, B., Cowles, T., Schwartz, R., van Geen, A., 2005. Distribution andvariability of iron inputs to Oregon coastal waters during the upwelling season.Journal of Geophysical Research 110, C10S12. doi:10.1029/2004JC002590.

Dickson, A.G., Millero, F.J., 1987. A comparison of the equilibrium constants for thedissociation of carbonic acid in seawater media. Deep-Sea Research 34,1733–1743.

Dickson, A.G., 1995. Determination of Dissolved Oxygen in Seawater by WinklerTitration. WOCE Operations Manual. Part 3.1.3 Operations & Methods. WHPOffice Report WHPO 91-1.

DOE, 1994. Handbook of Methods for Analysis of the Various Parameters of theCarbon Dioxide System in Seawater. Version 2. In: Dickson, A.G., Goyet, C. (Eds.)unpublished manuscript.

Firme, G.F., Bruland, K.W., Rue, E.L., Weeks, D.A., Hutchins, D.A., 2003. Spatial andtemporal variability in phytoplankton iron limitation along the California coastand consequences for Si, N and C biogeochemistry. Global BiogeochemicalCycles 17, 1016. doi:10.1029/2001GB001824.

Hales, B., Takahashi, T., Bandstra, L., 2005. Atmospheric CO2 uptake by a coastalupwelling system. Global Biogeochemical Cycles 19, GB1009. doi:10.1029/2004GB002295.

Haraldsson, C., Anderson, L.G., Hassellov, M., Hulth, S., Olsson, K., 1997. Rapid, highprecision potentiometric titration of alkalinity in ocean and sediment porewaters. Deep-Sea Research I 44, 2031–2044.

Holm-Hansen, O., Lorenzen, J., Holmes, R.W., Strickland, D.H., 1965. Fluorometricdetermination of chlorophyll. Journal du Conseil 30, 3–15.

Hudson, R.J.M., Morel, F.M.M., 1990. Iron transport in marine phytoplankton:kinetics of cellular and medium coordination reactions. Limnology andOceanography 35, 1002–1020.

Hutchins, D.A., Bruland, K.W., 1998. Iron-limited diatom growth and Si:N uptakeratios in a coastal upwelling regime. Nature 393, 561–564.

Hutchins, D.A., DiTullio, G.R., Zhang, Y., Bruland, K.W., 1998. An iron mosaic in theCalifornia upwelling regime. Limnology and Oceanography 43, 1037–1054.

Hutchins, D.A., Hare, C.E., Weaver, R.S., Zhang, Y., Firme, G.F., DiTullio, G.R.,Alm, M.B., Riseman, S.F., Maucher, J.M., Geesey, M.E., Trick, C.G., Smith, G.J.,Rue, E.L., Conn, J., Bruland, K.W., 2002. Phytoplankton iron limitation in theHumboldt Current and Peru. Limnology and Oceanography 47, 997–1011.

Jickells, T.D., An, Z., Baker, A.R., Bergametti, G., Brooks, N., Boyd, P.W., Duce, R.A.,Hunter, K.A., Junji, C., Kawahata, H., Kubilay, N., Krogh-Andersen, K., LaRoche, J.,

Liss, P., Mahowald, N., Prospero, J.M., Ridgwell, A.J., Sulzberger, B., Tegen, I.,Torres, R., 2005. Global iron connections between desert dust, ocean biogeo-chemistry and climate. Science 308, 67–71.

Johnson, K.S., Chavez, F.P., Friederich, G.E., 1999. Continental shelf sediment asa primary source of iron for coastal phytoplankton. Nature 398, 697–700.

Lewis, E., Wallace, D.W.R., 1998. Program Developed for CO2 System Calculations.ORNL/CDIAC-105. Carbon Dioxide Information Analysis Center, Oak RidgeNational Laboratory, U.S. Department of Energy, Oak Ridge, Tennessee.

MacIsaac, J.J., Dugdale, R.C., Barber, R.T., Blasco, D., Packard, T.T., 1985. Primaryproduction cycle in an upwelling center. Deep-Sea Research 32, 503–529.

Martin, J., 1990. Glacial–interglacial CO2 change: the iron hypothesis. Paleo-ceanography 5 (1), 1–13.

Mehrbach, C., Culberson, C.H., Hawley, J.E., Pytkowicz, R.M., 1973. Measurement ofthe apparent dissociation constants of carbonic acid in seawater at atmosphericpressure. Limnology and Oceanography 18, 897–907.

Pizarro, O., 2000. Low Frequency Fluctuations in the Eastern Boundary Current OffSouth America: Remote and Local Forcing. Ph.D. thesis, Goteborg University,Sweden, unpublished manuscript.

Rutllant, J., 1994, Modelling the Day to Day Wind Variability Offshore Central Chileat About 30�S. Internal Report IC/94/166, IAEA, UNESCO and InternationalCentre for Theoretical Physics, 22 pp., unpublished manuscript.

Simpson, J.J., Zirino, A., 1980. Biological control of pH in the Peruvian coastalupwelling area. Deep-Sea Research 27, 733–744.

Strickland, J.D.H., Parsons, T.R., 1968. A practical handbook of seawater analysis.Fisheries Research Board of Canada Bulletin 167, 311.

Tans, P., Conway, T.J., 2005. Monthly Atmospheric CO2 Mixing Ratios from the NOAACMDL Carbon Cycle Cooperative Global Air Sampling Network, 1968–2002. InTrends: a Compendium of Data on Global Change. Carbon Dioxide InformationAnalysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, OakRidge, TN, USA.

Torres, R., Turner, D., Silva, N., Rutllant, J., 1999. High short-term variability of CO2fluxes during an upwelling event off the Chilean coast at 30�S. Deep-SeaResearch I 46, 1161–1179.

Torres, R., Turner, D., Rutllant, J., Sobarzo, M., Antezana, T., Gonzalez, H., 2002. CO2outgassing off Central Chile (31–30�S) and northern Chile (24–23�S) duringaustral summer 1997: the effect of wind intensity on the upwelling andventilation of CO2-rich waters. Deep-Sea Research I 49, 1413–1429.

Torres, R., Moffet, J., Ulloa, O., Pantoja, S., Morales, C., Antezana, T. Iron deficiency ina coastal upwelling area of central Chile (30�S). Marine Chemistry, submittedfor publication.

Wanninkhof, R., McGillis, W.R., 1999. A cubic relationship between air–sea CO2exchange and wind speed. Geophysical Research Letters 26, 1889–1892.

Watson, A., 2001. Iron limitation in the ocean. In: Turner, D.R., Hunter, K.A.(Eds.), The Biogeochemistry of Iron in Seawater. IUPAC Series on Analyticaland Physical Chemistry of Environmental Systems, vol. 7. John Wiley andSons, 396 pp.

Weiss, R.F., 1974. Carbon dioxide in water and seawater: the solubility of a non-idealgas. Marine Chemistry 2, 203–215.