Global Biogeochemical Cycles

Spatial and temporal variability in phytoplankton iron limitation along the California coast and consequences for Si, N, and C biogeochemistry

Authors


Abstract

[1] Iron limitation was investigated along the Northern California coast in the summer of 1999. Small-volume (1 l) shipboard iron addition bottle experiments were performed at 44 stations to gain the greatest possible temporal and spatial coverage of the area. Parameters measured in these 4-day incubations included size-fractionated chlorophyll a and particulate nutrients (C, N, and Si). Degrees of community iron limitation were quantified and compared using various iron limitation indexes, calculated as the ratio of chlorophyll a, particulate organic nitrogen, or particulate organic carbon produced in Fe-amended bottles to the amounts produced in controls. Iron limitation occurred most frequently and was most severe on stations off the continental shelf, away from sedimentary sources of iron, as well as during relaxation events in aged upwelled water. Size-fractionated chlorophyll a data did not suggest large Fe-mediated changes in phytoplankton community composition. Fe limitation reduced phytoplankton production of particulate organic nitrogen and carbon, but had much less effect on biogenic silica production. The result is an increase in particulate Si:N and Si:C ratios of control samples, which were frequently double those of Fe-amended samples. Particulate C:N ratios also decreased under Fe limitation, indicating that iron availability exerts a strong control on C, N, and Si elemental composition of phytoplankton, and thus on the biogeochemical cycling of these nutrients in the California upwelling region.

1. Introduction

[2] Areas of active coastal upwelling constitute only 1% of the ocean's surface area, but are responsible for the highest areal primary productivity rates in the ocean [Chavez and Barber, 1987; Toggweiler and Carson, 1995]. Upwelling off California is most intense during the spring and summer, when predominant winds blow southward. These seasonal winds displace the surface coastal water offshore and supply nutrients to the euphotic zone by Ekman transport [Dever, 1997]. Because upwelling events are not constant, the system goes through intercalating upwelling and relaxation events that can last up to 3 weeks [Dugdale and Wilkerson, 1989]. During upwelling, nutrients are plentiful in the euphotic zone, but turbulence and deep mixing can prevent phytoplankton blooms from occurring until relaxation occurs. Diatoms are usually the dominant phytoplankton group following an upwelling event, and these extensive blooms eventually exhaust nitrate, phosphate, and silicic acid from surface waters [Brzezinski et al., 1997; Bruland et al., 2001]. At this point, community structure often switches to a nanoplankton dominated phytoplankton community growing primarily on recycled nutrients [Hutchins and Bruland, 1998].

[3] Until recently, most work in coastal upwelling systems did not consider the effects of phytoplankton iron limitation because upwelling, river outflows, and continental atmospheric dust were assumed to supply ample inputs of iron to the system. We now know that iron limitation can affect productivity in both the California region [Hutchins and Bruland, 1998; Hutchins et al., 1998; Bruland et al., 2001] and the Peru upwelling [Hutchins et al., 2002] because iron inputs during the upwelling season are limited.

[4] Riverine outflow in the summer is at a minimum, and sedimentary iron from shelf areas entrained during upwelling is the primary source of iron. Iron supply thus depends on the degree of interaction between the upwelled water parcel and the continental shelf [Johnson et al., 1999, 2001; Bruland et al., 2001]. Wide ranges of Fe availability are expected off the California coast, since its bathymetry is not uniform. The shelf is relatively wide and shallow in areas such as Monterey Bay, off San Francisco Bay, and Point Arena (PA). In other areas the shelf is much narrower and drops off quickly, reaching 500 m depth within 3 km off the coast of Big Sur (BS). The water parcel being upwelled will have a limited interaction with narrow shelves compared with wide shelves, leading to large variations in the amount of iron being carried into lighted surface waters. Narrow shelf areas are likely to become Fe-limited once phytoplankton blooms develop and deplete the relatively small amount of iron available [Hutchins et al., 1998; Johnson et al., 2001; Bruland et al., 2001].

[5] In previous shipboard incubation experiments in parts of the Northern California coast where the shelf is narrow, phytoplankton growth rates and biomass in bottles with Fe added were higher than in controls. Molar Si:N (silicic acid:nitrate) uptake ratios in control samples were at least twice those of Fe-amended samples [Hutchins and Bruland, 1998; Hutchins et al., 1998]. Because Fe limitation appears to reduce nitrate uptake and carbon fixation rates much more than silicic acid uptake rates, Fe-limited diatoms often have much higher Si:N and Si:C uptake ratios in both laboratory studies and in Fe-limited regimes [Hutchins and Bruland, 1998; Takeda, 1998; Franck et al., 2000]. Iron-replete diatoms often have a molar Si:N uptake ratio close to 1 [Brzezinski, 1985], whereas iron-limited diatoms frequently have a much higher (2–3×) ratio [Hutchins and Bruland, 1998]. However, a quantitative relationship between the extent of diatom Fe limitation and variations in particulate Si:N and Si:C production ratios has not been established.

[6] The variability in phytoplankton Fe limitation and nutrient drawdown ratios in the California upwelling region is likely to be large, since the system is highly heterogeneous in both space and time. Hutchins et al. [1998] suggested that this region is a complex “mosaic” of iron availabilities that results in a broad range of biological effects, from iron-replete to severely iron-starved conditions. This previous incubation work demonstrating Fe limitation in the region, however, focused on only a few selected stations and could give little indication of how important this process is over larger scales. Here, we use a “mapping” approach to examine the varying degrees of iron limitation on much larger temporal and spatial scales, and the resulting consequences for phytoplankton community Si:N, Si:C, and C:N ratio variability along the California coast.

[7] Survey approaches have been used to examine phytoplankton Fe limitation over regional scales in the past. In particular, the diagnostic indicator Fv/Fm, representing the photosynthetic electron transfer efficiency of algal photosystem II, has been assessed by shipboard fast repetition rate fluorometry (FRRF) in order to map the extent of Fe limitation in surface waters [Kolber et al., 1994; Behrenfeld and Kolber, 1999]. Although Fv/Fm is a sensitive, noninvasive way to interrogate the phytoplankton community, it cannot predict the effects of varying degrees of iron limitation on major nutrient biogeochemistry and algal elemental stoichiometry. To do this, iron addition experiments are necessary.

[8] Our approach was to generate an extensive spatial and temporal coverage of the central California upwelling area by performing as many Fe addition experiments as possible during a 5-week occupation of the region. This mapping approach allowed us to examine fine-scale variability in Fe limitation effects, and to correlate these quantitatively with the observed response of major nutrient biogeochemistry. Experiments were carried out at both near-shore and offshore stations, and during active upwelling and full relaxation.

[9] It has become evident that phytoplankton iron limitation is not a simple “on and off” phenomenon but is more accurately represented as a wide gradient from completely iron-sufficient to severely iron-starved conditions [Hutchins et al., 1998]. In order to quantify the degree of iron limitation in our experiments and correlate this with nutrient utilization ratios, an index was designed to compare parameter values between +Fe and control treatments. The iron limitation index (ILI) was defined as the ratio of the production of chl a (ILIChl), PON (ILIPON), or POC (ILIPOC) in the +Fe bottles to the change in the same parameters in the controls. An ILI value of one represents no difference between amounts of chlorophyll (chl), particulate organic nitogen (PON), or particulate organic carbon (POC) produced in iron-amended and control bottles, and therefore no Fe limitation. Higher values of these three types of ILI indicate progressively greater degrees of iron limitation. The results of these experiments are intended to provide a much more comprehensive and quantitative view of coastal upwelling iron limitation than had been previously available. The relationship between ambient Fe concentrations and the extent to which phytoplankton communities are Fe-limited is examined over large temporal and spatial scales. Control of phytoplankton elemental ratios by iron availability has global biogeochemical consequences [Boyle, 1998], and this work thus also contributes to the emerging picture of the effects of iron on major nutrient utilization and carbon cycling throughout the world ocean.

2. Methods

[10] The California coast was studied for 5 weeks on two different occasions. During the first leg of the cruise (CIRCUS), from 3 to 17 June, 25 stations were sampled. The second leg of the cruise (SQUIRTS) took place between 23 June and 11 July, when 19 more stations were studied. To differentiate between experiments started at different stations, CIRCUS growouts are numbered sequentially and SQUIRTS experiments are designated GO (for growout) followed by the station number.

[11] The extensive spatial and temporal coverage of the area was achieved by simple 4-day, 1-l bottle incubations begun at as many as five stations each day, with up to 10 experiments carried out concurrently. Near-surface water (5–10 m) was collected for the incubation experiments using an acid-washed all-teflon Osmonic Bruiser diaphragm pump system [Hutchins et al., 1998; Bruland et al., 2001]. Collected water was homogenized in a 50-l acid-washed carboy using trace metal clean techniques, and dispensed into acid-washed 1 l polycarbonate control and +Fe bottles. During the first leg all experiments were performed in triplicates, and during the second leg duplicate control and Fe-amended samples were used. Iron was added at 2.5 nM as inorganic FeCl3 in 0.1 M Ultrex HCl. Experiments were incubated for 4 days in spectrally corrected flow-through deckboard incubators at 40% of incident light at ambient sea-surface temperatures [Laws et al., 1990; Hutchins et al., 1993]. At the beginning and end of each experiment, we measured size-fractionated chlorophyll a by filtering water through 0.2, 1.0, and 8.0 μm polycarbonate filters, followed by extraction in 90% acetone for 24 hours and shipboard fluorescence measurements [Yentsch and Menzel, 1963] using a Turner AU-10 fluorometer.

[12] Samples for dissolved nutrients (nitrate + nitrite and silicic acid) were filtered through a 0.2 μm filter and frozen for analysis on shore using a Technicon II Autoanalyzer [Friederich and Whitledge, 1972] during the first leg. During the second leg nutrients were measured on board using a Lachat autoanalyzer. Particulate nutrient (CHN) samples were filtered through precombusted 13-mm GF/F glass filters and the filters were frozen for analysis on shore according to the method described by Sharp [1991], using a Carlo Erba, Inc., CHNS-O EA 1108 Elemental Analyzer [Hutchins et al., 1998].

[13] Samples for biogenic silica (BSi) were filtered through 0.6-μm polycarbonate filters and dried at 65°C on board for 48 hours before being stored at room temperature [Franck et al., 2000]. Biogenic silica samples were analyzed on shore by digesting filters in 0.2 M NaOH at 95°C for 1 hour [Paasche, 1973] followed by the addition of 1N HCl to stop the reaction. The resulting dissolved silicic acid was then determined according to the method of Strickland and Parsons [1972], as modified by Brzezinski and Nelson [1995]. BSi:PON, BSi:POC, and PON:POC production ratios were then calculated and compared between treatments and stations [Hutchins and Bruland, 1998]. These data sets were also compared with those from the same experiments for ILI.

[14] ILIs were determined by calculating the increases in chlorophyll a, PON, and POC in the +Fe and control samples, and dividing the resulting +Fe values by the control values, yielding a +Fe/control production ratio for each parameter (ILIChl, ILIPON, and ILIPOC). Although any ILI value >1 indicates greater production in the +Fe bottles than in the controls, we considered only experiments with ILI values >1.3 to be significantly iron limited. This cutoff value was determined by statistical tests of the 44 experiments using ANOVA, and represents a 90% confidence level.

[15] Dissolved iron concentrations for the first leg of the cruise (CIRCUS) were measured by flow injection analysis [Weeks and Bruland, 2002]. During SQUIRTS, total dissolved iron was measured by cathodic stripping voltammetry methods [Rue and Bruland, 1995, 1997]. In both cases, seawater samples for total dissolved Fe measurements were 0.2 μm-filtered, acidified to pH 1.7–1.8, and microwaved prior to analysis [Bruland and Rue, 2001], releasing even relatively refractory (i.e., colloidal) Fe for analysis.

3. Results

3.1. Iron Limitation Indexes

[16] ILI values versus dissolved iron concentrations for the CIRCUS leg are presented in Figure 1 and for SQUIRTS in Figure 2. There is a strong relationship between dissolved iron concentrations and ILIChl (Figures 1a and 2a), ILIPON (Figures 1b and 2b), and ILIPOC (Figures 1c and 2c). In all three cases there were no differences between treatments at iron concentrations >0.6 nM on the CIRCUS leg and >0.2 nM on the SQUIRTS leg. However, when concentrations were <0.6 nM (CIRCUS) or <0.2 nM (SQUIRTS), Fe-amended samples sometimes produced as much as 5× more chl a, 4× more PON, and 10× more POC than did control samples. Not all samples from stations where dissolved Fe was less than <0.6 nM (CIRCUS) or less than 0.2 nM (SQUIRTS) were Fe-limited, but no greatly elevated ILI values were observed in any case from stations with Fe levels higher than these values.

Figure 1.

Iron limitation indexes (ILI) using (a) chlorophyll a, (b) particulate organic nitrogen, and (c) particulate organic carbon versus ambient dissolved iron concentration during the CIRCUS cruise. The ILI is defined as the ratio of the change in these parameters in the +Fe treatments to the change in the controls. Note the scale difference in Figure 1c.

Figure 2.

ILI using (a) chlorophyll a, (b) particulate organic nitrogen, and (c) particulate organic carbon versus ambient dissolved iron concentration during the SQUIRTS cruise. Note the scale difference in Figure 2a.

[17] There was a difference between the iron concentration at which an effect was seen in CIRCUS and SQUIRTS. During CIRCUS, the inflection point for iron limitation was found to be ∼0.6 nM. During SQUIRTS, 0.1–0.2 nM was found to be the range at which the phytoplankton community responded to iron deprivation. We do not believe this difference is due to an analytical artifact or contamination. There may have been actual differences in iron concentrations between the two occasions the California coast was sampled, and/or in biological communities and their response to Fe additions. For instance, higher total Fe values similar to those of CIRCUS were observed during the second leg of SQUIRTS (data not shown), although incubation experiments were not carried out during this later part of the cruise, since there was insufficient time to complete the incubations.

3.2. Physical Variability

[18] Fifteen out of 25 stations showed iron limitation at the 90% confidence level during the first leg of the cruise according to ILIChl, 14 according to ILIPOC, and 13 according to ILIPON (Figure 3a and Table 1). The sites studied varied from close to the shore, on the continental shelf, to stations in deep water within the offshore California Current; and from as far north as PA, to as far south as the BS coast (see bathymetry contours, Figure 3a). With the exception of stations 10 and 18, considered slightly Fe-limited by ILIChl (1.89 and 1.38, respectively) all stations north of San Francisco Bay (between Pt Reyes and Pt Arena) that were iron-limited were off the relatively broad and shallow continental shelf.

Figure 3.

(a) Stations where iron limitation experiments were carried during CIRCUS. Gray circles represent Fe replete stations, and black circles Fe limited stations, according to ILIChl. (b) Stations where iron limitation experiments were carried during SQUIRTS. Gray circles represent Fe replete stations, and black circles Fe limited stations, according to ILIPOC.

Table 1. Station Numbers, Initial Properties of the Water (Temperature, Salinity, Nutrient Concentrations), ILI, Particulate Nutrients Produced and Particulate Nutrient Production Ratios of CIRCUS Stationsa
CIRCUSInitial NutrientsIron Limitation IndexParticulate Nutrients ProducedParticulate Nutrients Produced Ratio
StationTemp, °CSalinityFe, nMNO3, μMSi(OH)4, μMPONChl aPOCPONBSiPOCBSi:PONBSi:POCPOC:PON
ControlFeControlFeControlFeControlFeControlFeControlFe
  • a

    Numbers in bold indicate significant iron limitation. Asterisks: not determined because of low (<1.0 μM) particulate nutrient production. ND: no data. Initial silicic acid values may be underestimated because of incomplete depolymerization of frozen samples during thawing. Particulate organic carbon may be elevated due to detritus.

211.533.320.66NDND1.01.03.111.311.112.212.277.7240.41.081.100.160.056.9021.62
311.733.100.64NDND3.53.49.52.89.84.64.514.7140.01.660.460.310.035.3314.28
412.233.070.23NDND2.52.47.02.87.03.54.017.2120.51.240.570.210.036.0517.11
512.332.980.25NDND1.32.82.02.73.62.02.527.955.50.720.680.070.0410.2315.22
612.932.980.19NDND2.04.42.13.26.50.21.443.291.4****13.4014.07
712.632.900.17NDND2.42.29.42.86.71.11.28.680.90.400.180.130.023.1112.03
89.733.501.62NDND1.02.91.113.413.028.830.4286.6319.12.162.340.100.1021.4524.55
99.033.521.15NDND0.91.21.018.016.332.423.9311.8303.71.801.460.100.0817.3018.61
108.833.552.07NDND1.01.91.019.819.032.038.8354.6354.31.622.040.090.1117.9318.67
118.233.598.00NDND1.51.11.711.117.222.419.482.8139.22.021.130.270.147.468.10
128.533.502.10NDND0.91.20.910.79.910.59.461.558.20.990.950.170.165.775.86
139.733.200.6621.919.31.22.12.111.313.921.822.3102.2212.11.931.600.210.109.0215.25
149.533.180.4922.318.92.11.71.85.711.614.918.752.894.72.631.600.280.209.358.13
159.233.380.9327.726.61.41.81.310.614.713.817.272.697.61.301.170.190.186.856.64
168.733.510.9730.529.71.11.81.112.213.712.013.194.5107.70.980.960.130.127.737.89
179.433.390.7825.524.41.31.51.215.119.831.230.6199.8230.62.061.550.160.1513.2011.67
188.733.532.9029.931.61.11.41.119.622.142.138.9254.4274.42.151.760.170.1512.9712.42
199.933.536.1024.326.51.11.31.116.719.021.219.3132.9152.51.261.010.160.147.948.01
2010.533.49ND17.117.1ND1.2NDNDND23.222.3NDNDNDNDNDNDNDND
219.433.481.2029.527.0ND1.1NDNDNDNDNDNDNDNDNDNDNDNDND
2211.633.120.1312.87.11.791.21.23.15.55.04.036.744.21.630.880.140.1111.857.97
2311.433.330.0812.310.91.11.12.17.18.110.514.545.694.51.471.780.230.156.4211.63
2412.033.120.1410.13.02.01.82.03.97.62.82.832.865.00.730.360.090.048.498.56
2512.332.800.228.23.51.91.63.81.63.14.14.414.254.42.461.390.290.088.6317.32
2611.033.300.3414.212.31.51.01.67.811.55.43.881.9126.90.700.330.070.0310.5511.07

[19] In contrast, stations south of Monterey Bay in the BS area, where the shelf is narrow and drops off quickly, presented iron limitation very close to the shore (<6 km) as well as offshore. Total dissolved iron concentrations ranged from 0.08 nM off the BS coast, to 8.0 nM south of PA in an area of strong winds and active upwelling (Table 1).

[20] On the second leg of the cruise (SQUIRTS), nine of the 19 stations studied showed iron limitation using the ILI calculated using chlorophyll a, 13 according to ILIPOC (Figure 3b), and 10 according to ILIPON. To examine gradients in Fe concentrations and availability with distance from the shore, we carried out an onshore/offshore transect off the coast of PA (stations GO3–GO9 and GO11). The first two stations were very close to shore and were iron replete according to ILI using chl a, PON, or POC as parameters, but three stations off the shelf showed iron limitation by all indexes (GO5, GO7, and GO9), whereas stations GO8 and GO11 were limited according ILIPOC only (Table 2). Winds were very high and sea state was extremely rough during the latter part of this particular transect. These conditions increase the chances of Fe contamination from the ship at the pump intake, as well as during water collection in the clean area. It is possible that the results from stations 11 and 12 were therefore compromised by slight contamination.

Table 2. Station Numbers, Initial Properties of the Water (Temperature, Salinity, Nutrient Concentrations), ILI, Particulate Nutrients Produced and Particulate Nutrient Production Ratios of Experiments at SQUIRTS Stationsa
SQUIRTSInitial NutrientsIron Limitation IndexParticulate Nutrients ProducedParticulate Nutrients Produced Ratio
StationTemp (°C)SalinityFe, nMNO3, μMSi(OH)4, μMPONChl aPOCPONBSiPOCBSi:PONBSi:POCPOC:PON
ControlFeControlFeControlFeControlFeControlFeControlFe
  • a

    Numbers in bold indicate significant iron limitation. Asterisks: not determined because of low (<1.0 μM) particulate nutrient production. ND: no data. Initial silicic acid values may be underestimated because of incomplete depolymerization of frozen samples during thawing. Particulate organic carbon may be elevated due to detritus.

GO110.533.410.06NDND2.51.33.04.912.422.622.339.8118.14.641.800.570.198.179.51
GO212.533.290.07NDND1.61.42.53.86.22.73.127.269.20.700.490.100.047.0911.09
GO310.733.640.7NDND1.11.10.821.623.236.135.8300.2232.31.671.540.120.1517.5615.49
GO410.133.420.620.721.30.71.21.18.110.819.421.694.0105.91.802.660.210.248.6813.06
GO510.033.45ND24.326.81.91.63.88.015.221.621.984.1320.12.711.450.260.0710.5721.12
GO612.433.300.0712.91.71.00.90.85.75.73.64.272.058.50.630.730.050.075.0510.29
GO712.433.200.0513.46.01.61.51.82.03.33.83.930.956.11.891.190.120.074.326.49
GO812.833.190.0511.22.40.01.13.10.04.11.41.219.660.1**0.070.02*14.60
GO912.533.210.06NDND1.71.72.52.13.52.92.824.962.31.410.790.120.0411.9517.73
GO1011.632.450.0511.012.91.41.92.12.73.93.05.820.542.11.121.510.150.147.5710.91
GO1111.233.350.116.713.01.01.01.65.45.59.58.933.855.01.751.620.170.2610.1219.41
GO1210.633.370.0821.022.71.01.11.28.58.915.915.2110.2127.51.861.710.140.1212.9514.37
GO1312.233.040.058.95.14.01.31.61.87.33.23.140.963.91.750.420.080.0515.588.77
GO1412.033.280.214.913.71.91.01.85.310.211.811.458.4106.42.211.120.200.1110.9310.40
GO1512.932.980.088.26.41.01.32.22.12.24.25.536.781.61.982.540.110.0717.3837.33
GO1613.333.190.06NDND4.11.92.90.83.20.94.713.338.8****17.2512.21
GO1713.133.270.059.411.21.11.41.54.14.76.47.439.961.81.571.590.160.129.7513.28
GO1816.333.170.80.05.02.80.91.10.51.50.00.38.79.1**0.040.07*6.16
GO1916.133.250.4NDND0.9ND1.31.41.20.40.511.915.0****8.7012.53

[21] As with the previous leg, all Fe-limited stations were located off the shelf, however, not all stations located off the shelf were Fe-limited. Besides stations discussed above, station GO12 offshore of Pt Reyes did not show iron limitation by any parameter. Station GO14 along the BS coast was considered iron limited as determined by both ILIPON and ILIPOC, station GO15 by ILIPOC only, and station GO18 north of the Channel Islands by ILIPON only. The California coast is a very hydrodynamically complex system, where upwelling plumes are susceptible to currents, eddies, and variable degrees of mixing with low-nutrient offshore California Current water [Johnson et al., 1999]. The off-shelf iron-replete stations could have had enough residual available iron from previous mixing with freshly upwelled or nearshore water.

3.3. Temporal Variability

[22] The BS area was visited on two occasions; first when a relaxation event was taking place, characterized by low iron values (0.17–0.64 nM Fe), relatively high water temperature (11.7°–12.9°C), and low salinity (32.9–33.1). All stations were iron limited during relaxation according to all parameters, although to different degrees (Table 1). When visited a week later, an upwelling event had just taken place, as evidenced by lower water temperature (9.4°–12.3°C), higher salinity (33.12–33.48), and higher iron values (up to 1.2 nM Fe). During active upwelling, four stations did not show iron limitation (Figure 3a), and the two that did were iron limited to a much lesser degree than during the previous week. Chlorophyll ILI (ILIChl) are presented in Figure 4 for experiments at four of these BS stations that were occupied twice, once during relaxation and once during upwelling. These ILIChl values demonstrate that phytoplankton become much more Fe-limited as upwelled surface waters age and upwelling-supplied Fe is depleted during relaxation events.

Figure 4.

Degrees of Fe limitation as assessed by chl a production (ILIChl) at the same stations along the Big Sur coast visited in two different occasions, once during relaxation (white bars) and once during active upwelling (black bars). During relaxation iron had been depleted from the system and phytoplankton were much more iron-limited than during active upwelling.

3.4. Chlorophyll a Size Fractionation

[23] Unlike many previous open-ocean iron addition experiments [Martin et al., 1994; Coale et al., 1996; Boyd et al., 2000], there was no apparent shift to larger phytoplankton species when iron was added to the bottles. As observed previously in this region by Hutchins et al. [1998], both control and iron treated samples during CIRCUS were dominated by the large size class (>8.0 μm), as estimated by size-fractionated chlorophyll a. Large phytoplankton dominated at 20 of the 25 stations, comprising 35–84% of total chlorophyll. Although chl a biomass increased when iron was added to iron limited samples, the proportion of the different size classes usually did not change significantly. As in the CIRCUS experiments, there was little obvious relationship between size-fractionated chlorophyll a distributions and iron availability in the two treatments during the SQUIRTS leg of the cruise.

3.5. Elemental Ratios

[24] To examine the effects of iron limitation on phytoplankton elemental production ratios, Si:N, Si:C, and C:N ratios were calculated from changes in BSi, PON, and POC concentrations in both treatments over the course of the incubations. These ratios are presented only when particulate production values were higher than 1.0 μmol l−1, because calculation of production ratios from very small changes in BSi and PON values is subject to large errors due to minor analytical variability.

[25] In most iron limited stations, control samples had a BSi:PON ratio approximately double that of the Fe-treated counterpart, indicating that lack of iron hindered the uptake of nitrate by the phytoplankton community. This difference between treatments, and its effect on BSi:PON production ratios is represented in Figure 5a, where selected typical iron limited and iron replete stations from BS and PA are shown. However, ratios were often below one, for instance, in all samples from the BS region (Table 1), indicating perhaps that diatoms did not dominate the phytoplankton community at the time of sampling. Due to the large number of experiments, detailed community composition analyses such as microscopic cell counts were not attempted, but size-fractionated chlorophyll data showed that at most BS stations (stations 21–26) the small size class (0.2–1.0 μm) was initially dominant at the time of sampling, and no stations were initially dominated by the large cell size class (>8.0 μm).

Figure 5.

Selected stations off Big Sur (BS) and Point Arena (PA) showing typical nutrient production ratios in controls and Fe-amended treatments under iron-limited and iron replete conditions (a) BSi:PON, (b) BSi:POC, and (c) POC:PON.

[26] Production ratios of BSi:PON ranged from 0.18 to 2.6 in Fe-treated samples, and from 0.4 to 4.6 in control samples, indicating considerable variability in relative nitrate and silicic acid uptake among the stations (Tables 1 and 2). At all but two stations considered iron limited by the ILIChl, production of PON was considerably higher in iron treated samples than in controls. However, Si:N ratios did not double at all stations considered iron limited by the ILIChl. Of the 24 stations that were significantly iron-limited according to the ILIChl, 15 had BSi:PON production ratios that were approximately twice as high in the control samples as in the Fe-treated bottles. When the degree of Fe limitation was assessed by PON (ILIPON) or POC (ILIPOC), four more of these stations showed a doubling of the BSi:PON production ratio in controls compared with +Fe treatments (Tables 1 and 2). Because of the large variability in absolute BSi:PON ratios between stations, BSi:PON ratios of the control samples were divided by the BSi:PON ratios of the +Fe samples to quantify the relative change in elemental ratios between treatments. These values were then plotted versus iron concentration for the CIRCUS (Figure 6a) and SQUIRTS samples (Figure 7a). If a particular station is iron replete, addition of iron should not affect particulate nutrient production by phytoplankton, and therefore the ratio of control over Fe-amended BSi:PON will be one. However, if the addition of iron affects either BSi or PON production, there will be a deviation from the 1:1 relationship. Figures 6a and 7a show that at iron concentrations below 0.6 and 0.1 nM, respectively, BSi:PON production ratios increase sharply.

Figure 6.

(a) Biogenic silica to particulate organic nitrogen (BSi:PON) production ratios in controls over the same ratios in +Fe treatments versus ambient dissolved iron concentrations during CIRCUS. (b) Linear regression plot showing the relationship between changes in BSi:PON production ratios in the two treatments, and ILIPON.

Figure 7.

(a) Biogenic silica to particulate organic nitrogen (BSi:PON) production ratios in controls over the same ratios in +Fe treatments versus ambient dissolved iron concentrations during SQUIRTS. (b) Linear regression plot showing the relationship between changes in BSi:PON production ratios in the two treatments, and ILIPON.

[27] Although PON production increased in iron-amended samples, BSi production remained relatively constant (Tables 1 and 2). Particulate nitrogen production therefore, is the controlling factor for the observed increases in BSi:PON ratios under Fe limitation. This is evidenced by the linear (r2 = 0.87, CIRCUS and r2 = 0.95, SQUIRTS) relationship between BSi:PON production ratios in controls relative to +Fe samples, and ILIPON (Figures 6b and 7b).

[28] BSi:POC production ratios ranged from 0.02 to 0.26 in iron treated samples and from 0.03 to 0.57 in control samples. Iron limited stations had consistently higher BSi:POC ratios in control samples than did Fe-amended samples (Tables 1 and 2). Examples from selected stations are shown on Figure 5b, illustrating the difference in ratios between BS iron limited stations and PA iron replete stations. Figures 8a and 9a show that just as with BSi:PON ratios, at dissolved iron concentrations below 0.6 and 0.1 nM, respectively, BSI:POC production ratios increase. POC production was the determining factor in the variation observed in BSi:POC production ratios between control and +Fe samples. This is verified by the strong linear relationship (r2 = 0.98, CIRCUS and r2 = 0.68, SQUIRTS) between relative BSi:POC production ratios in the two treatments, and ILIPOC (Figures 8b and 9b). In some cases +Fe samples produced over 9× more POC than control samples (see ILIPOC for stations 3 and 7, and Table 1).

Figure 8.

(a) BSi:POC production ratios (control/+Fe) versus ambient dissolved iron concentrations during CIRCUS. (b) Linear regression plot showing the relationship between the change in BSi:POC production ratios between the two treatments, and ILIPOC.

Figure 9.

(a) BSi:POC production ratios (control/+Fe) versus ambient dissolved iron concentrations during SQUIRTS. (b) Linear regression plot showing the relationship between the change in BSi:POC production ratios between the two treatments, and ILIPOC.

[29] POC:PON production ratios ranged from 3.1 to 21.5. In control samples and 5.9–37.3 in iron treated samples. At iron replete stations the addition of iron did not affect the relative ratio of particulate carbon:nitrogen production, whereas at iron limited stations POC:PON production ratios were frequently double those of control treatments. This is illustrated by the bar graph depicting results from experiments at selected stations off BS and PA (5C), as well as the data presented in Tables 1 and 2. Because both particulate carbon and particulate nitrogen production are affected by iron availability, the relationship between POC:PON ratios in the two treatments is less straightforward than for BSi:PON and BSi:POC ratios. Figure 10 shows a comparison between POC:PON ratios of the control samples over POC:PON ratios of +Fe samples, versus iron concentration for CIRCUS (Figure 10a) and SQUIRTS stations (Figure 10b). When dissolved iron concentration was less than 0.6 nM (CIRCUS) and 0.1 nM (SQUIRTS), the control:+Fe ratios of POC:PON produced were most often less than one, sometimes considerably less.

Figure 10.

POC:PON production ratios (control/+Fe) during (a) CIRCUS and (b) SQUIRTS, and their relationship to ambient dissolved iron concentrations.

4. Discussion

[30] Our results suggest that coastal iron limitation occurs throughout large portions of central California during the summer, when upwelling provides the major source of iron [Johnson et al., 1999; Bruland et al., 2001]. Of the 44 sites studied, 34 stations presented some form of iron limitation. Iron limitation was widespread in many areas between PA and Point Conception, with degrees of iron limitation ranging from slight to acute. At these Fe-limited stations, ILIChl values ranged from 1.3 to 4.1, ILIPON from 1.3 to 4.4, and ILIPOC from 1.3 to 9.5. Most iron limited stations were found off the continental shelf, corroborating earlier studies that cite the availability of shelf iron as a determining factor for the high biological productivity of coastal upwelling ecosystems [Hutchins et al., 1998; Johnson et al., 1999, 2001; Fung et al., 2000; Bruland et al., 2001].

[31] The BS coast was the most Fe limited of all the regions visited. This is indicated by the ambient dissolved Fe concentrations, as well as by the biological ILI (Figures 3a and 3b). Because of its steep and narrow continental shelf and the absence of riverine inputs this region is usually deprived of iron, except when strong summer upwelling events episodically provide iron-enriched waters and allow the phytoplankton community to bloom [Bruland et al., 2001].

[32] We observed temporal variability in iron limitation in BS waters when the same stations were visited during a relaxation (stations 3–7) and an upwelling event (stations 21–25). When an upwelling event occurs, the water parcel replacing the surface water interacts with the narrow shelf of the region, bringing deep colder water enriched in nutrients and a limited amount of iron to promote phytoplankton production. When winds subside, a bloom ensues and iron is quickly removed from the system. As the relaxation event continues and the upwelled water parcel ages, phytoplankton become increasingly more iron limited, as shown by the large differences observed in ILI values between relaxation and upwelling events. All stations were iron limited according to ILIChl (2.2–4.4), ILIPON (1.3–3.6), and ILIPOC (2.0–9.5) during the relaxation event. After upwelling, however, ILI values decreased considerably, ranging from 1.10 to 1.82 (ILIChl), 1.14 to 1.97 (ILIPON), and 1.21 to 3.82 (ILIPOC).

[33] This study introduces a novel approach for quantifying iron limitation, the ILI. This index provides insight into the degree of iron limitation of the phytoplankton community by comparing physiological and biogeochemical responses in the experimental treatments, using commonly measured parameters such as chlorophyll a and particulate nutrient production. In the California upwelling region, where iron availability and supply are highly variable, the simple “iron limited/iron replete” dichotomy is insufficient to understand the complexities of the system. The ILI approach could be used to quantify degrees of limitation estimated from dissolved nutrient uptake, carbon fixation rates, flavodoxin/ferredoxin ratios, algal accessory pigments, or any other parameter used to examine the biological effects of iron limitation during incubation experiments. In this study, during the CIRCUS leg, ILIPOC was most clearly dependent on dissolved iron concentrations (Figure 1), whereas during SQUIRTS ILIChl a and ILIPON were more closely correlated to iron levels (Figure 2). Because chlorophyll a per cell can vary due to changes in iron, light, nutrients, etc, ILIChl a may prove to be a less reliable diagnostic of iron limitation than ILIPOC or ILIPON. Definition of the ILI using multiple parameters rather than a single one will likely offer the most accurate indication of phytoplankton iron status, however.

[34] Biogenic silica production in our experiments was not affected by iron availability, in contrast to recent work indicating possible Fe limitation of Si uptake in the Southern Ocean HNLC region [Franck et al., 2000]. Particulate nitrogen and carbon production were suppressed at iron limited stations in unamended samples, usually resulting in high Si:N and Si:C production ratios. Si:N and Si:C production ratios were always low when sufficient iron was present. However, in a few cases when dissolved iron values were low, ratios did not increase, a pattern also observed in a small number of ILI.

[35] Without knowledge of the prior history of the stations, it is not possible to know if in these cases the phytoplankton community had been supplied with upwelled water rich in iron and depleted it previous to the water collections. Phytoplankton are capable of storing iron in excess of levels needed to meet immediate metabolic needs [Hutchins, 1995]. This appeared to be the case at BS station 23, where phytoplankton were iron replete despite very low dissolved iron levels (0.08 nM), and at most of the BS stations following the upwelling event. This ability is termed “luxury uptake,” and it is considered an iron acquisition strategy “most advantageous for species living in coastal waters where iron concentrations are high, but spatially and temporally variable” [Sunda and Huntsman, 1995], such as the California coastal upwelling region. We also do not know the details of community composition at the stations we occupied. If the phytoplankton community is largely composed of species that do not require silicic acid, Si:N and Si:C ratios will not be representative of effects of iron limitation on nutrient production ratios, as observed in Si-depleted Subantarctic waters by Hutchins et al. [2001].

[36] Luxury uptake and storage of iron prior to our collections is also one possible reason for the observation that large phytoplankton (presumably diatoms) often dominated in both our controls and +Fe treatments (although biomass and growth rates were always much higher in the latter in Fe-limited areas). This dominance of both treatments by large diatoms in California Fe addition experiments has been observed in the past [Hutchins et al., 1998], but is in strong contrast to oceanic HNLC areas, where experimental controls and the ambient community are usually composed largely of nano- and picoplankton [Price et al., 1991; Hutchins, 1995]. Sporadic inputs of iron are undoubtedly more frequent in the coastal regime than in open ocean areas, and storage of iron by diatoms may allow them to continue to dominate the community even when iron becomes limiting. Another likely explanation for these differences in phytoplankton community composition between coastal and oceanic HNLC areas is that seed stocks of large diatoms are much higher in coastal areas. Many large coastal bloom-forming diatoms such as Chaetoceros form benthic resting spores that can be later brought up with upwelled water [Grimm et al., 1997]. These benthic sources of diatom inoculum are not available in deep-water HNLC areas [Chavez, 1989]. However, the reason(s) for diatom dominance even under iron-limited conditions in coastal upwelling areas remain unknown, and will require further research.

[37] C:N production ratios in our experiments were somewhat variable, since both carbon and nitrogen production are dependent on iron supply [Greene et al., 1991; Hutchins, 1995]. Generally, however, C:N production ratios were low when iron concentrations were low, and increased when iron was added. This suggests that carbon fixation is more sensitive to iron stress than nitrate assimilation. At lower dissolved iron levels, the difference in C:N ratios between treatments was more pronounced. Although iron deficiency has been shown to decrease both cellular C and N content [Greene et al., 1991; Price et al., 1991; Sunda and Huntsman, 1995; Maldonado and Price, 1996], to our knowledge this is the first time iron availability has been observed to systematically affect phytoplankton C:N production ratios.

[38] Our results suggest that iron availability has large implications for phytoplankton elemental composition and nutrient biogeochemistry in the California coastal upwelling region, with altered biological N, Si, and C stoichiometry at limiting iron conditions. During this investigation, the phytoplankton community showed iron deprivation when iron concentrations fell between 0.1 and 0.6 nM, depending on the cruise. Recent work in the Southern Ocean HNLC region found 0.5 nM to be the inflection point at which Si(OH)4 uptake increases relative to NO3 [Franck et al., 2000], although effects on Si:C and C:N ratios were not examined. Our study supports the findings of Hutchins and Bruland [1998] and Takeda [1998], and their suggestions that carbon export should decrease relative to sinking Si fluxes under iron-deplete conditions, potentially biasing paleoproductivity estimates from sedimentary biogenic opal deposits.

[39] The results of our study, earlier work in the California system, and new results from the Peru upwelling [Hutchins et al., 2002] can be synthesized into a conceptual model of the biological and biogeochemical evolution of upwelled water, as depicted in Figure 11. As newly upwelled water parcels are advected offshore and age, or as the system moves from active upwelling into relaxation, iron is preferentially exported from the system by phytoplankton community utilization and sinking particles. Unlike major nutrients, iron is a particle-reactive element that is removed by passive scavenging. Iron is incorporated into biomass at Fe:C ratios that depend largely on external Fe concentrations [Sunda et al., 1991], so biological Fe removal is also rapid at the relatively high initial levels in freshly upwelled water. Thus, developing communities will deplete Fe faster than major nutrients. Increasing iron stress on the diatom community then eventually reduces the ability of the phytoplankton to take up nitrate and carbon while the uptake of silicic acid is much less affected. Iron limitation greatly increases the efficiency of the “silicate pump” [Dugdale et al., 1995] since cell Si:C and Si:N production ratios increase under iron-deplete conditions [this study; Hutchins and Bruland, 1998; Takeda, 1998; Hutchins et al., 1998]. The result is the preferential export of Si relative to N as the water ages. Because nitrogen is regenerated in the euphotic zone much more efficiently than silicic acid [Dugdale et al., 1995], nonsiliceous phytoplankton will likely dominate in the end-member phytoplankton community of this process [Sommer, 1994; Hutchins and Bruland, 1998]. This is indeed observed in Si-depleted, N-replete systems such as the summertime Subantarctic Southern Ocean, where Si:N production ratios are extremely low and are decoupled from Fe availability [Hutchins et al., 2001]. The net result of this biogeochemical evolution process in the California system should be the offshore export of surface water with low Fe and Si levels, but with residual N.

Figure 11.

Postulated biogeochemical evolution of an upwelled water parcel as it ages and advects offshore along the California coast. Initially, newly upwelled, nutrient-rich water is relatively iron-replete, depending on the degree of prior interaction with shelf sediments (1). The developing phytoplankton bloom begins to deplete the water of dissolved iron and to export iron relative to major nutrients (2). Increasing iron stress reduces phytoplankton nitrate uptake much more than uptake of silicic acid, leading to preferential drawdown and export of silicon relative to nitrogen (3). Finally, silicic acid becomes limiting with “excess” nitrate left over (4). The result is a shift in community structure from a diatom-dominated to a nanoplankton-dominated community, and export of high nitrate, low-silicate water into the offshore oligotrophic California Current.

[40] This model is supported by changes in nitrate and silicic acid we measured along the onshore/offshore transect at PA (Table 2). Initial nitrate and silicic acid concentrations were comparable at nearshore stations GO5 and GO6 (20.7–21.3 and 24.3–26.8 μM, respectively). However, at the next three stations further offshore, silicic acid concentrations decreased to 1.7–6.0 μM, while nitrate concentrations were much less depleted (11.2–13.4 μM). Similar transect data from both the California and Peru upwelling areas show the same trend, with Si depletion relative to N during aging of upwelled water in Fe-limited regions, while drawdown is usually close to 1:1 in Fe-replete upwelling plumes (K. W. Bruland, unpublished data, 2000).

5. Conclusions

[41] This is the first large-scale spatial and temporal study examining coastal iron limitation and the consequences for phytoplankton nutrient biogeochemistry. We used a simple method by which as many as five stations could be sampled daily, for a total of 44 incubation experiments. The large data set produced by this survey allowed the creation of a useful method for quantifying the extent of phytoplankton iron limitation, the ILI.

[42] Varying degrees of iron limitation are widespread in the California coastal upwelling region during the summer. Phytoplankton iron limitation is most common and severe in regions removed from continental shelf sources of iron, or in aged upwelled water where the phytoplankton community has depleted the dissolved iron concentration. Size fractionated chlolorophyll a does not, however, show any obvious trends with iron availability. We found no evidence for iron being a co-factor in silicic acid uptake, but iron limitation suppressed the production of particulate carbon, nitrogen, and chlorophyll a, confirming previous studies [Greene et al., 1991; Sunda and Huntsman, 1995]. The result is elevated Si:N and Si:C production ratios, as has been shown by other laboratory [Takeda, 1998] and field experiments in iron limited HNLC areas [Hutchins et al., 1998; Takeda, 1998, Hutchins and Bruland, 1998; Franck et al., 2000]. Iron availability was also found to affect C:N production ratios, which generally decreased with declining ambient iron levels, a trend not previously observed.

[43] This study proposes a model for the evolution of upwelled water parcels in coastal upwelling regimes, where iron is rapidly removed from the system as the primary limiting nutrient, contributing to the “silicate pump” and thus driving the preferential export of Si out of the euphotic zone relative to C and N. In California waters, the result is a nanoplankton-dominated community and low iron, low silicic acid water with residual nitrate being advected offshore to mix with the oligotrophic California Current.

[44] This research provides evidence that iron availability controls phytoplankton elemental ratios, with large implications for global biogeochemical cycles. This intertwined relationship between Fe, Si, C, and N biogeochemistry should be further examined, in particular the response of phytoplankton community composition to iron additions. Understanding the linkages between Fe and other biogeochemically significant elements is especially important, in light of current proposals for large-scale ocean iron fertilizations.

Acknowledgments

[45] We thank Y. Zhang, G. Smith, and J. Conn for their assistance during sampling, as well as the captain and crew of the R. V. Pt Sur. We also thank J. Sharp and G. Luther III for providing valuable advice during the preparation of this manuscript. This work was supported by NSF grants OCE 9811062 and 0094535 to D. Hutchins and OCE 9811114 to K. Bruland.

Ancillary