Interaction of iron and major nutrients controls phytoplankton growth and species composition in the tropical North Pacific Ocean

Transcription

1 Limnol. Oceanogr.. 38(3), 1993, , by the American Society of Limnology and Oceanography, Inc. Interaction of iron and major nutrients controls phytoplankton growth and species composition in the tropical North Pacific Ocean Giacomo R. DiTullio Graduate Program in Ecology, 108 Hoskins, University of Tennessee, Knoxville David A. Hutchins Institute of Marine Sciences, University of California at Santa Cruz, Santa Cruz Kenneth W. B&and Institute of Marine Sciences, University of California at Santa Cruz Abstract A 6-d nutrient enrichment experiment was performed in the tropical North Pacific Ocean at 9 N, 147 W with ultraclean techniques. Changes in phytoplankton biomass, C and N assimilation rates, growth rates, and species composition were monitored with HPLC pigment analyses and flow cytometry techniques, as well as 14C fixation into particulate C, pigments, and protein. Prochlorophyte specific growth rates (from divinyl Chl a labeling) increased from an initial value of 0.15 d -I to 0.96 d-l following macronutrient addition (N, P, and Si). Diatoms, however, were unable to grow without added Fe. Diatom populations were severely colimited by Fe and macronutrients but achieved a specific growth rate of 2.5 d - following Fe and macronutrient additions. Results implied that grazing rates (g) on prochlorophytes were stimulated in approximate balance with prochlorophyte growth (p) after 6 d (g : p = O.SS), but that grazing processes were not as efficient (g: p = 0.40) at controlling the diatom standing stock. Our results suggest that grazing processes may be the most important factor regulating procaryotic biomass, but Fc limitation is the proximate control of diatom biomass and hence may limit the utilization of macronutrients in the equatorial Pacific Ocean. A full mechanistic understanding of the factors that control phytoplankton production and species composition in the open ocean still eludes ecologists. Traditionally, nutrient limitation has been considered a prime factor limiting autotrophic production in temperate and tropical oceanic waters because of the vanishingly low ambient concentrations of nutrients in the oligotrophic ocean (Eppley et al. 1973). In general, two major nutrients, N and P, have been suggested to limit oceanic phytoplankton production. Geochemists, primarily Acknowledgments We thank Michelle DuRand and Jeffrey Dusenberry for the flow cytometry analyses performed at sea. Their support was provided by grants from the EPA (IN-0826-NAEX to S. W. Chisholm) and from the Vetlesen Foundation (to R. J. Olson), We thank N. Welschmeyer for the use of some sampling equipment. Support for K. W. Bruland and D. A. Hutchins was provided by NSF (OCE ) and ONR NOOO14-92-J to K. W. Bruland. Support for G. R. DiTullio was provided by NSF (OCE ) to G. R. DiTullio and N. A. Welschmeyer. Finally, we acknowledge J. T. Hollibaugh and two anonymous reviewers whose comments improved the quality of the text. 495 arguing from the perspective of geological time scales, have suggested that P must ultimately limit marine phytoplankton production (Redfield et al. 1963) because of the abundance of atmospheric N2 and the potential for N2 fixation by some procaryotes in tropical oceanic regions. In contrast, biological oceanographers have postulated that most open-ocean phytoplankton communities are N limited (Thomas 1970; Eppley et al. 1973), at least on short time scales. Many researchers now believe that colimitation by macro- (N, P, or Si) and (or) micronutrients (trace metals) may simultaneously control phytoplankton growth and species composition in the open ocean (Morel and Hudson 1985). The idea that trace elements may limit phytoplankton growth in the sea is not new (Harvey 1947; Ryther and Guillard 1959). For instance, early Fe-enrichment bioassays demonstrated significant enhancement of phytoplankton biomass in Sargasso Sea communities (Ryther and Guillard 1959; Menzel and Ryther 196 1). All of the above studies, however, have been criticized because the levels

2 496 DiTullio et al. of Fe enrichment were orders of magnitude higher than the ambient Fe concentrations. Hence, perturbation in conjunction with contamination and insensitive analytical techniques probably confounded these earlier studies. With the advent of trace metal clean sampling techniques (Bruland et al. 1979) and sensitive analytical procedures for measuring oceanic Fe concentrations (Gordon et al. 1982), recent studies have suggested, once again, that Fe availability may control oceanic phytoplankton productivity and species composition (Martin and Gordon 1988; Martin et al ). However, the exact role of macro- and micronutrients in controlling phytoplankton species composition and production rates in the open ocean is still controversial (see Chisholm and Morel 199 1). We report here the results from a nutrient enrichment experiment performed in the tropical North Pacific Ocean. In order to facilitate comparisons with previous Fe-enrichment experiments performed at adjacent stations, we added nutrients to mimic the concentrations found in the high-nutrient equatorial Pacific Ocean. More importantly, this experimental design provided us with a unique perspective compared to typical equatorial enrichment experiments because of the low nutrient concentrations present at 9 N [e.g. N03-, Si(OH)4, and Fe concentrations at 20 m were 0.01 PM, 0.30 PM, and 0.03 nm, respectively]. Hence, we were able to vary both the major nutrients and Fe concentrations independently to investigate the interaction of Fe and major nutrients on phytoplankton growth and species composition, an option not possible in typical Fe-enrichment experiments carried out in highnutrient areas. Methods Seawater for the enrichment experiments was collected on 25 August at 9 N, 147 W in the North Pacific Ocean aboard the RV Moana Wave. Stringent anticontamination techniques were used so that the seawater was never exposed to the atmosphere (Bruland et al. 1979). Thirty-liter Teflon-coated Go-flo bottles were suspended on Kevlar line. Once on deck, the filled bottles were pressurized with high-purity N2 gas. Seawater was then forced through Beva-line tubing through the side of the van and into a laminar-flow clean working area where the.water was dispensed into incubation carboys. Four acid-cleaned 1 O-liter polycarbonate carboys were filled under low light with seawater from the surface mixed layer at a depth of 20 m (-40% I,) at 0300 hours local time. Because of logistical constraints we were unable to add replicate carboys for the various treatments. Small bottles can lead to artifactual bias, especially in long-term incubations (>24 h) due to wall effects, possible macrozooplankton exclusion, or both. For this reason we decided that it was more important to use large-volume incubations than to have many replicates of smaller bottles. We hypothesized that the variability within treatments in our experiment would be small relative to the variability between treatments. The results obtained from several independent analyses in the various treatments supported this assumption. For instance, our experiments did not apparently retard the overall phytoplankton growth rate. In fact, the absolute and relative growth rates estimated by both pigment and protein labeling, respectively, indicated that the total phytoplankton population was not severely inhibited after a 6-d incubation in our carboys, but rather was growing near hax (see results). Four treatments were compared. Ambient seawater with no additions served as the control. For simplicity, the treatment that received 6, 0.6, and 2.0 PM additions of sterile, NO-, P043-, and Si(OH), stock solutions is refc:rred to as the N treatment. The macronutrient stock solutions were previously eluted through a Chelex-100 column to remove trace metals. The Fe treatment received a 1 -nm addition of Fe. The Fe standard [atomic absorption standard stock solution of Fe (1,000 pg ml-- in 1 wt% HCl, Aldrich)] was chelated to EDTA in a 1 : 1 ratio. This small addition of ED,TA (1 nm) has no significant effect on the seawater trace metal speciation. The N, P, Si, and Fe carboy (FEN0 treatment) received both macronutrient and Fe additions at the same concentrations as the N and Fe treatments. The silicate concentration before Si addition was 0.3 PM. Phosphate concentrations were not determined, but based on the ambient NQ- concentrations were probably very low (e.g. co.01 PM). The carboys were incubated on-deck (40%

3 Fe and nutrient limitation 497 light level) under blue-light conditions (Rohm and Haas blue filter No. 2424) in order to mimic the spectral quality of the natural irradiance field (Laws et al. 1990). Circulating surface seawater was used to control the temperature in the incubators. There was no difference in temperature between the surface and 20 m. To prevent contamination from the circulating seawater, we wrapped the top covers of the carboys tightly with parafilm and tape. The carboys were then triple-bagged (polyethylene) and sealed with tape. Light attenuation by the bags was included in the light calibration of the incubators. The necks of the carboys were kept just above waterline to further minimize possible contamination from circulating seawater. The contents of each carboy were gently mixed before subsampling. The carboys were incubated for 6 d with subsamples taken at days 3 and 6. All subsampling procedures were performed inside a class- 100 laminar-flow hood. The subsampling procedure consisted of applying a small positive pressure on the carboy with 0.2-pm filtered, compressed air to facilitate transferring clean samples into the incubation bottles. Subsamples were removed for high performance liquid chromatography (HPLC) pigment analyses, nutrient measurements, fluorometric Chl a, flow cytometry, and 24-h 14C incubations. At the end of the 6-d experiment, a small portion from each carboy was analyzed on-board for Zn (a proxy for shipboard trace metal contamination) by anodic stripping voltametry (Bruland 1989). J. H. Martin (pers. comm.) has observed that Zn and Fe contamination generally occur together and that elevated Zn is a good diagnostic indicator of trace metal contamination. Results confirmed that the incubations were carried out without any spurious trace metal contamination. The 14C incubations were performed in acidcleaned 2.7-liter polycarbonate bottles. The H14C03- crystalline stock (ICN Chemicals) was diluted with a 2.8 mm solution (Fitzwater et al. 1982) (ph 10) of gold-label purity Na,CO, (Aldrich) in Milli-Q water. Incubations were started at dawn with (200 ~1) additions of 200 &i of H14C03- per bottle. Following a 24-h incubation, subsamples were filtered for sizefractionated primary productivity measurements (with 5-pm Poretics and filters). Estimates of autotrophic N assimilation and absolute growth rates were measured with the incorporation rate of 14C into protein (Di- Tullio and Laws 1986; DiTullio 1993) and Chl a (Redalje and Laws 198 l), respectively, after a 24-h incubation. The C-specific growth rate of other phytoplankton groups was estimated from the incorporation of 14C into xanthophylls (Gieskes and Kraay 1989; Goericke and Welschmeyer in press). Calculation of phytoplankton growth rates from pigment labeling was performed with the equations of Goericke and Welschmeyer (in press). Samples for HPLC pigment analyses were filtered (l-2 liters) onto filters and stored in liquid N2 until processing. Samples were run on a Beckman System Gold HPLC equipped with a model 126 PC-based controller and model 166 absorbance detector set at 440 nm. The samples were homogenized in 90% acetone and allowed to extract for -2 h at 4 C. Following centrifugation, subsamples were injected onto a C-l 8 column (Spherisorb ODS II, 30 cm) and eluted with an acetonitrilemethanol-acetone gradient (a modification of the unpublished method of Welschmeyer and Moreno) which gave a baseline separation of divinyl Chl a and normal Chl a. The method we used for quantifying pigment concentrations has been described (DiTullio and Laws 199 1). The C.V. for replicate injections of pure pigment standards on our system was ~2%. The concentration of divinyl Chl a (uniquely associated with prochlorophytes; Chisholm et al. 1988; Goericke and Repeta 1992) was calculated by assuming that it had the same absorption coefficient as normal Chl a. Fucoxanthin concentrations were assumed to be associated predominantly with diatoms and not as a trace accessory pigment in dinoflagellates or chrysophytes because peridinin concentrations were very low and because the accompanying increases in biomass (Chl a) and production ( C) were associated primarily with the >5-pm fraction. Concentrations of 19 - hexanoyloxyfucoxanthin (1 g-hex) and 19 -butanoyloxyfucoxanthin (1 g-but) were assumed to represent populations of prymnesiophytes (Arpin et al. 1976) and chrysophytes (Lewin et al. 1977). The labeled pigments were first concentrated onto Baker C-l 8 cartridges and washed with water to remove any residual H14C03-. The eluted pigments were injected onto a normal-

4 498 DiTullio et al , m >5Ccm 0 Total m >5pm 0 Total 10-19% 0 19% 24% Control Fe N,P,Si - -13% 71% - JI 1 Fig. 1. Primary production rates measured after a 24-h 14C incubation at day 3 (A) and day 6 (B) of the enrichment experiment. The total phytoplankton assemblage was collected onto filters and the > 5-pm fraction was filtered onto 5-pm Poretics filters. Percentages represent the contribution (from replicate samples) made by the >J-prn fraction to the total primary production rate. phase Si column with a hexane-isopropanol gradient. The peaks of interest were collected, dried under NZ, resuspended in acetone, and rechromatographed on a reverse phase C-18 column with the acetonitrile-meoh-acetone gradient. Blanks were collected before and after the peaks of interest and the average of the blanks was subtracted from the activity of the collected pigments which were measured with a liquid scintillation counter. The method has been described in detail (see Goericke and Welschmeyer in press). N,P,Si,Fe I A B 1 A FACScan (Becton Dickinson) flow cytometer was used to determine cell concentrations of Synechococcus and Prochlorococcus. Samples were run live at sea (typically 0.1 ml was analyzed) and fluorescent beads (0.57 pm, Polysciences Fluoresbrite) were added as an internal standard (Olson et al. 1993). Results Carbon and nitrogen assimilation rates- Primary production rates in the N and FEN0 treatments were similar but - 500% higher than the control treatment at day 3, with both treatments having - 13% of the C production attributable to the > 5-pm fraction (Fig. 1 A). Primary production rates in the Fe treatment were very similar to those measured in the control treatment (Table 1). On day 6, the total primary production rates in the Fe, N, and FEN0 treatments were 14, 358, and 750% higher than the rate in the control bottle (Fig. 1B). In contrast to day 3, however, the contribution made by the >5-pm fraction to the total autotrophic production rate in the FEN0 treatment was dramatically higher than in all other treatments (Fig. 1B). For instance, although the primary production rate of the > 5-pm fraction on day 6 was not significantly different in the control and Fe treatments, those in the N and FEN0 treatments were 5-30-fold higher, compared to the control treatment (Table 1). The highest productivity indices (PI) were observed on day 3 in the N and FEN0 treatments (Table 1). Although both C and N as- similation rates in these treatments were higher on day 6 compared to day 3, the exponential increase in autotrophic biomass (Chl a) resulted in lowered PI on day 6 (Fig. 2). The > 5- pm Chl a fraction in the FEN0 treatment rep- resented 83% of the total community biomass on #day 7 (Fig. 3). This percentage was similar to the 7 1% contribution by the > 5-pm fraction to the total primary production rate (Fig. 1). In contrast, in the N treatment only 1% of the Chl. a but 24% of the production was attributed to the >5-pm fraction. Possibly, these large phytoplankton species were responsible for producing a disproportionate amount of the particulate C relative to their chlorophyll content. Autotrophic N assimilation rates as measured by protein labeling revealed a pattern

5 Fe and nutrient limitation 499 Table 1. Primary production rates (mg C m-3 d-l), Chl a concentrations (mg m-3) (measured by fluorometry), productivity indices (PI) (g C g Chl-l h-l), and NO,- concentrations (PM) are presented for the total ( filter) and > 5-pm populations in the four treatments (see methods). Values represent the mean of duplicate samples. The overall average difference between replicate values was 10%. Initial concentrations: N03-, 0.01 PM; Si(OH),, 0.30 PM; Fe, 0.03 nm. Phosphate concentrations were not determined. Day Treatment Size Primary Total production (W Chl a PI NO, Initial Control Fe N, P, Si (N) N, P, Si, Fe (FENO) Control Fe N, P, Si (N) N, P, Si, Fe (FENO) * The initial N03- concentration based on a stock addition :ll * among treatments that was very similar to that observed for the C assimilation data (Fig. 4A). The N assimilation rate at day 6 of the FEN0 treatment (6.7 mg N me3 d-l) was > 10 times higher than the rate observed in either the control (0.56 mg N m-3 d-l) or Fe treatments (0.6 1 mg N m-3 d- ) and 7 1% higher than the rate in the N treatment (3.9 mg N m-3 d-l). The N and FEN0 treatments had C : N assimilation ratios (Fig. 4B) that were nearly identical to the Redfield ratio of 5.7 (by wt), implying that these phytoplankton communities were growing at very high relative growth rates (P : prnax; Goldman et al. 1979). - W 3 0 Day 6 Control Fe N,P,Si N,P,Si,Fe Control Fe N,P,Si N,P,Si,Fe Fig. 2. Productivity indices at day 3 and day 6 of the Fig. 3. Size-fractionated (5-pm Poretics filter) and total enrichment experiment. The production rates and Chl a ( filter) Chl a biomass in the four treatments followvalues were obtained following a 24-h (dawn-to-dawn) 14C ing the 24-h incubation of seawater sampled on day 6. incubation. Values represent the average of duplicate samples.

6 r DiTullio A - et al. Table 2. Prochlorophyte cell numbers (Pro. No.) were meacured at sea by flow cytometry techniques (M. Du- Rand and J. Dusenberry). Divinyl Chl a (DChl a) was measured by HPLC. The percentage of total Chl a represented by divinyl Chl a and the divinyl Chl a concentratrion per cell were calculated for the four treatments. DChl a Pro. No. DChl a cell (x IO5 (ng (fs DChl a/ Day Treatment ml ) liter I) cell- ) ZChl a 0 Initial Control Fe N, P, Si (N) N, P, Si, Fe (FENO) 6 Control Fe N, P, Si (N) N, P, Si, Fe (FENO) Control Fe N,P,Si N,P,Si,Fe Fig. 4. A. Autotrophic N assimilation rate on day 3 and day 6 as estimated with the 14C protein-labeling method (DiTullio and Laws 1986; DiTullio 1993). B. The C : N assimilation ratio measured by the incorporation rate of Hi4C0,- into particulate C and protein. The dashed line represents the Redfield ratio of 5.7 (by wt). In contrast, the C : N assimilation ratio in the control and Fe treatments averaged 9.1 and 8.1 by weight. The allocation of 14C02 into protein implied that the phytoplankton in the control and Fe treatments were nutrient limited and growing only at and 63% of pmax, as estimated from the protein-labeling method (DiTullio and Laws 1986). Although the C : N assimilation ratio may not be a valid estimate of relative growth rate in severely Fe-limited diatoms (Greene et al. 199 l), the diatom populations in our experiment were colimited by Fe and macronutrients. The variability in C : N assimilation ratios under colimitation of Fe and macronutrients has not been previously investigated. The procaryotic algae in our experiments apparently were limited only by macronutrients (see Table 3). The higher relative growth rates observed in the N and FEN0 trealtments (relative to the control) were consistent with the increases noted in the phytoplankton absolute growth rate and pigment concentrations (see below). Phytoplankton species composition - Nutrient additions caused dramatic shifts in the phytoplankton species composition as determined by chemotaxonomic identification of algal pigments (Figs. 5-7) and flow cytometry (Table 2). The most significant changes in phytoplankton species composition were observed in the N and FEN0 treatments (Fig. 5). In contrast, the control and Fe treatments displayed very little increase in pigment concentrations during the incubation. The eucaryotic algal community at the start of the incubation was dominated by diatoms, prymnesiophytes, and chrysophytes as evidenced by the presence of the xanthophylls fucoxanthin, 19 -hex, and 1 g-but, respectively. In the FEN0 treatment, Chl a concentrations (as measured by HPLC) increased by 4513% from the start of the incubation to day 3 (Fig. 5A). Similarly, in the N treatment, Chl a increased by 432% during the first 3 d of inc:ubation. During the last 3 d of incubation,

7 ; the Chl a concentration in the FEN0 treat- b ment increased by an additional 135%, but the 4 Chl a concentration in the N treatment dei=l 400 creased by 48% (Fig. 5A). The increase in the Chl a concentration dur ing the first 3 d in the N and FEN0 treatments d was most likely due to the increase in prym- =I nesiophytes, as evidenced by the transient in- 2 crease in the 19 -hex concentration at day 3 g ZOO (Fig. 5C). Although Synechococcus spp. also increased during this time, Chl a is not the major light-harvesting pigment in these algae. It is also unlikely that this increase in Chl a 8 concentration at day 3 in the N and FEN a treatments could be attributed to either chrysophytes or diatoms since the concentra- Time (Days) tion of 19-but was about 5-6-fold lower (data not shown) than the 19 -hex concentration, and fucoxanthin concentrations at day 3 were not y* 60]]-J significantly different than those at the start of b the incubation (Fig. 5B). Hence, assuming that 2 a /v Fe,N,P,Si grazing processes were responsible for the 48% V decrease in the Chl a concentration on day 6 2 a of the N treatment, the decrease in Chl a was - 8o most likely associated with the 77% decrease.s in the 19 -hex concentration. A B Procaryotic pigments and cell numbers at 2.I5 day 6 decreased in both the control and Fe z loa <k : treatments relative to the start of the incuba- g tion (Table 2). Large increases in both auto trophic procaryotic cell numbers and pigment I& <: - concentrations were observed in the N and 0 I FEN0 treatments (Figs. 6, 7). In contrast to the eucaryotic pigments (which displayed the greatest increase in the FEN0 treatment), the Time (Days) greatest increase in both procaryotic cell num- zoo I I I I I I bers and pigment concentrations were noted 0 Control in the N treatment (Figs. 6, 7). For instance, h v N,P,Si _ l Fe the divinyl Chl a concentrations at day 6 in yk lea v Fe,N,P,Si the N and FEN0 treatments were 596 and ai 304% higher than at the beginning of the incubation. 2 Cell counts revealed the same general trend h Fig. 5. Phytoplankton pigment concentration during the experiment as measured by HPLC pigment analyses. A. Chlorophyll a concentration does not include divinyl Chl a (see Fig. 6). B. Fucoxanthin concentration (proxy for diatom abundance) during the enrichment experiment. Note scale break for fucoxanthin concentration in the FEN0 (Fe, N, P, Si) treatment on day 6 (491 ng liter- ). C. Concentration of 19 -hex, a chemotaxonomic indicator for the presence of prymnesiophytes. v 0 I e Time (Days)

8 502 DiTullio et al. I I I I I I 0 Control A ; E =: rn - : s m R I 0 v N,P,Si 0 Fe v Fe,N,P,Si v - j////j - I I I I I I 0 Control v N,P,Si 0 Fe I Fe,N,P,Si I I I I I I Time Fig. 6. A. Prochlorophyte cell abundance as measured by flow cytometry (courtesy of M. DuRand and J. Dusenberry). B. Concentration of the prochlorophyte pigment divinyl Chl a during the incubation. (Days) Time Fig. 7. A. Concentration of Synechococcus spp. as measured by flow cytometry (courtesy of J. Dusenberry and M. DuRand). B. Concentration of the procaryotic pigment zeaxanthin represents the contribution from both cyanobacteria and prochlorophyte species. (Days) with an increase of 122 and 63% in the N and FEN0 treatments at day 6 relative to the start of the incubation (Table 2). The close agreement between the divinyl Chl a concentration and the prochlorophyte cell numbers among the various treatments (Fig. 6A) suggested that no significant photoadaptation response occurred during the course of our incubation. For instance, the divinyl Chl a cell- l ratio in the control treatment after 6 d decreased by 34% while cell numbers decreased by 55% (Table 2). The decrease (relative to the initial ratio) in the divinyl Chl a cell- ratio in the control and Fe treatment on day 6 (Table 2) may have been due to severe N limitation as evidenced by the increase in the C : N assimilation ratio in these treatments (Fig. 4). Prochlorophyte growth rates-the absolute growth rate of the prochlorophyte population was estimated (Table 3) with the divinyl Chl a-labeling method of Redalje and Laws (198 1) as modified by Goericke and Welschmeyer (in press). Consistent with the trend in cell num-

9 Fe and nutrient limitation 503 bers and divinyl Chl a concentration (Fig. 6), the highest prochlorophyte growth rate was associated with the N treatment on day 6 (Table 3). In comparison, the diatom absolute growth rate as estimated by the specific activity of fucoxanthin was 2.5 d- l in the FEN0 treatment on day 6. Changes in prochlorophyte cell numbers [N, = No exp@] on day 6 indicated that grazing rates (g) in the N and FEN0 treatments were nearly in balance (g : p = 0.85) with prochlorophyte growth rates (Table 3). On the basis of the observed sea surface temperature at 9 N of 29 C, the predicted potential phytoplankton bmax is l (Eppley 1972). Hence, the diatom-specific growth rate measured was 68% of the maximum growth rate limit as defined (solely) by temperature under continuous light (Eppley 1972). The actual diatom pmax, however, probably will be somewhat lower because of the 16-h photoperiod. For instance, if we assume a linear relationship between P,,.,~~ and photoperiod, a 16-h photoperiod at 29 C would predict a pmax of 2.5 d-l. Our estimated diatom absolute growth rate (2.5 d- I) suggested that the diatom assemblage was growing at pmax -a growth rate supported by the relative growth rate estimate from protein labeling (Fig. 4). The diatom growth rate was 2.6 times higher than the estimated prochlorophyte growth rate (0.96 d-l) and presumably is at least partially responsible for the diatom blooms typically observed in Fe-enrichment incubation experiments at the equator (Martin et al. 1991). The divinyl Chl a cell- ratios in the N and FEN0 treatments on day 6 were 376 and 276% higher compared to the control (Table 2). Higher Chl per cell ratios are typically associated with increasing cellular growth rates (Goldman 1980). Hence, the increase in the prochlorophyte growth rate on day 6 is consistent with the increase in the divinyl Chl a cell- 1 ratio (Table 2). Discussion Our data suggest that phytoplankton growth and species composition in the tropical North Pacific can be controlled, at least sometimes, by the distribution of Fe and major nutrients. The phytoplankton biomass in these oligotrophic ocean waters is dominated by procaryotic picoplankton. Our results suggest that these procaryotes are not severely Fe limited, even Table 3. Prochlorophyte growth rates (II) were determined with the divinyl Chl a-labcling method of Redalje and Laws (198 1) as modified by Goericke and Welschmcyer (in press). The net specific prochlorophyte growth rate (d-l) was calculated [N, = iv,exp(&] from changes in cell number as estimated by flow cytometry analysis (M. DuRand and J. Dusenberry). Grazing rates (g, d-l) were calculated by assuming that the major cell loss rate was due to hcrbivory. Pro. No. Net Prochlorophyte (x 105 spccitic Day Treatment ml- ) P P g R:P 0 Initial Control Fe N, P, Si (N) N, P, Si, Fe (FENO) 6 Control Fe N, P, Si (N) N, P, Si, Fe (FENO) under low Fe conditions. Conversely, the diatom biomass is low and is apparently colimited by the lack of both Fe and macronutrients. The most significant increases (relative to the control) in procaryotic and eucaryotic biomass were observed in the N and FEN0 treatments, respectively (Figs. 5, 6). Although the N treatment did cause a 2.7-fold increase in fucoxanthin concentration on day 6, the FEN0 treatment resulted in a 49-fold increase relative to the initial concentration (Fig. 5B). In contrast, Fe addition alone produced no such increase in fucoxanthin concentration, but rather yielded a response similar to that in the control. The data suggested that these diatom populations were colimited by the supply of Fe and macronutrients. Supply of one of these essential nutrients without the other was insufficient to promote the substantial diatom growth observed when both nutrients were added simultaneously (i.e. FEN0 treatment). Thus, primary production by diatoms in our experiment was controlled by the combination of both Fe and macronutrients. The growth of diatoms in the FEN0 treatment (Fig. 5) is a striking contrast to the proliferation of procaryotes observed in the N treatment (Figs. 6, 7) and suggests that the addition of 1 nm Fe preferentially stimulated the diatom population. It appears that when sufficient Fe is avail-

10 504 DiTullio et a/. able, diatoms can successfully compete with heterotrophic marine procaryotes (Trick 1989), procaryotic algae for macronutrients. are more speculative. In contrast, the greatest increase in procary- If cell size were the only important factor, otic biomass occurred in the N treatment (Fig. then we would also expect to see a biomass 6). The sevenfold increase in divinyl Chl a increase in small eucaryotic algae which have concentration in the N treatment suggested that a lower Fe cell quota (Brand 199 1) than prothe prochlorophyte population was able to ob- caryotic algae. However, only a transient intain all of its Fe nutrition from either regen- crease (on day 3) in the prymnesiophyte pigeration processes (Frey and Small 1979; ment (19 -hex) was observed (Fig. 5C). Labeling Hutchins et al. 1993) or from stored Fe re- of the 19 -hex pigment indicated that the speserves. cific growth rate of the prymnesiophytes in- It is possible that previous luxurious Fe creased from 0.13 d-l at the start of the inuptake and subsequent growth dilution of ex- cubation to 0.43 and 0.40 d-l after 3 and 6 d isting cellular Fe pools could explain the growth in the FEN0 treatment. Although the prymof procaryotes in the N treatment. This expla- nesiophyte growth rate was similar during day nation, however, appears unlikely for the fol- 3 and day 6, the 19 -hex concentration delowing reasons. First, episodic dust events creased > 5-fold. (which supply 95-99% of the Fe input to the If we assume that grazing processes were pri- North Pacific, Martin and Gordon 1988) are marily responsible for decreasing the 19 -hex rare during the low dust season. Hence, during concentration, then it appears that grazing prostratified conditions in late summer very little cesses were more important in controlling the Fe is added to the upper photic zone of the prymnesiophyte biomass during the last 3 d North Pacific to provide the supply for luxu- than during the first 3 d of incubation. In suprious Fe uptake. Second, the integral average port of this hypothesis, prochlorophyte grazing absolute growth rate during the experiment indicated that the prochlorophyte population doubled -6 times. If only stored Fe reserves were responsible for the Fe nutrition required for prochlorophyte growth, then the last cell doubling would have proceeded with only 1% of the initial cell Fe concentration. The fastest prochlorophyte growth rate was observed at the end of the experiment. Hence, we suspect that Fe recycling (see Hutchins et al. 1993), via protozoan grazing, was probably more important than stored Fe reserves in fueling prochlorophyte growth during our experiment. On the basis of our results, the autotrophic procaryotic community at 9 N, 147 W was predominantly limited by macronutrients, and the C : N assimilation ratio (Fig. 4) suggested that it could apparently grow near P,,,~~ without any additional input of Fe. The physiological mechanism for this ability to meet Fe nutritional requirements under such low Fe conditions is unclear. Certainly, picoplankton have a size advantage relative to larger algae because their small size leads to both a greater surface area : cell volume ratio and a thinner diffusion layer thickness, both of which should enhance Fe uptake under Fe-limiting conditions. Other possibilities, such as the ability to use siderophores for Fe uptake, as demonstrated for rates were higher in the last 3 d compared to the first 3 d of incubation (Table 3). Prymnesiophyte growth could also have been affected by interspecific competition for nutrient:;. Hence, the differential ability to compete for other resources (e.g. N03-) and not cell size per se may be responsible for the increase in procaryotic biomass relative to small eucaryotic species such as prymnesiophytes. Regardless of the mechanism, it appears that autotrophic procaryotes are somehow more capable of meeting their cellular demand for Fe in low Fe environments than are larger eucaryotic species such as diatoms. This result see ms paradoxical because procaryotic phytoplankton reportedly have a higher Fe cell quota than do eucaryotic algae (Brand 199 1). In fact, oceanic eucaryotic algae appear to require only a hundredth as much Fe as cyanobacteria in the open ocean (Brand 199 l), and the low Fe requirements (e.g. Sunda et al ) of some of these small oceanic eucaryotes may explain their presence in low Fe regions. The fact that algal biomass in oligotrophic oceanic waters is frequently dominated by procaryotic picoplankton cells (e.g. Pefia et al. 1990; Chavez et al ) suggests that they can outcomper;e eucaryotic phytoplankton for recycled nutrients such as NH,+ and Fe.

11 Fe and nutrient limitation 505 Regenerated NH4+ production fuels much of the nitrogenous nutrition of phytoplankton in the oligotrophic ocean and in the equatorial Pacific, despite the high N03- concentrations present at the latter locale (Price et al. 1991). In general, phytoplankton biomass in the tropical and subtropical North Pacific appears to be dominated by picoplankton. For instance, 42% of the total Chl a biomass at 9 N could be attributed to prochlorophytes (Table 2). During the nutrient enrichment experiment, the contribution made by divinyl Chl a to the total Chl a concentration was highest (65%) in the N treatment on day 6 (Table 2). Although procaryotic algae growing in oligotrophic waters are thought to predominantly use regenerated NH4+, they nevertheless are capable of substantial rates of new production (P,,,) when exposed to transient increases in N03- concentration (Glover et al. 1988). The ability of natural populations of Synechococcus to respond quickly to transient N03- events implies that they are not severely Fe limited even in low Fe waters. The in situ response of Synechococcus to a N03- pulse (Glover et al. 1988) was similar to the increases we observed in Synechococcus cell numbers in the N treatment (Fig. 6). These results suggest that procaryotic algae may be responsible for a significant fraction of the Pnew in the North Pacific central gyre, unless atmospheric Fe deposition coincides with the N03- pulse (e.g. the high dust spring season) and stimulates Pnew by the eucaryotic algal community (DiTullio and Laws 199 1). Based on the observations that the prochlorophyte cell cycle may be controlled by N availability (Vaulot and Partensky 1992) and that prochlorophyte populations were observed to track the nitracline in the Sargasso Sea (Olson et al. 1990), we suspect that N03- - addition was the major nutrient that stimulated prochlorophyte growth in our N treatment. In our N treatment, procaryotic algal growth was stimulated without any additional input of Fe. However, after 6 d phytoplankton in both the N and FEN0 treatments were unable to deplete the added N03- to any considerable extent (Table 1). It is conceivable that the low initial phytoplankton biomass (Chl a = 0.05 mg m-3) resulted in a long lag time and that the day 6 sampling may actually represent the beginning of the exponential phase of nutrient drawdown. Despite the fact that only a small fraction of the N03- was used up in both the N and FEN0 treatments (Table l), the phytoplankton in the FEN0 treatment used 1.8 times as much N as the phytoplankton in the N treatment. This result implies that atmospheric dust deposition to high nitrate-low chlorophyll waters may cause a more rapid increase in Pnew than vertical transport of N03- alone (on a mole equivalent basis) across the pycnocline in such low-fe oligotrophic waters. The 3-d lag period observed in fucoxanthin concentrations was not observed for either the 19 -hex concentration or for the procaryotic phytoplankton pigments. Possibly, the 3-d lag observed in fucoxanthin concentrations is related to a small inoculum of Fe-stressed diatom cells and (or) to the induction time required for the Fe-enzyme systems of severely Fe-limited cells. Natural populations of prymnesiophytes may be more abundant than diatoms or may not be as Fe limited as diatoms because of a lower Fe requirement for growth or both (Sunda et al ). For instance, high relative growth rates were observed for the prymnesiophyte Emiliania huxleyii grown at very low Fe concentrations (e.g. pfe = 21.7; DiTullio unpubl). Procaryotes may have circumvented this problem by recycling Fe more efficiently compared to larger eucaryotic algae (Hutchins et al. 1993). Grazing control (see Banse 1992) has been implicated as the major process in controlling phytoplankton populations in the subarctic (Frost 199 1; Welschmeyer et al ) and equatorial Pacific Ocean (Cullen et al. 1992). Our results indicated that at the end of our experiment, prochlorophyte growth rates in the N and FEN0 treatments were nearly in balance with the cell loss rate (g : p = 0.85) as estimated by changes in cell numbers (Table 3). The near balancing of prochlorophyte growth and grazing rates that we observed is similar to the general ratios (0.80) reported in the subarctic (Frost 199 1) and equatorial Pacific Ocean (Chavez et al ). We believe our data have important implications for the equatorial Pacific ecosystem. Although the phytoplankton species present at the equator and 9 N may be different, it seems likely that seed stocks of typical high-nu-

12 506 DiTullio et al. trient phytoplankton species will be present at 9 N because of northward surface Eckman transport away from the equatorial divergence zone. It is also possible that seed stocks of highnutrient species are present in the surface waters at 9 N because of the proximity of the strong, shallow nutricline (40 m). In support of these arguments, the phytoplankton species composition at the equator following an Feenrichment experiment was very similar to that observed at 9 N (Chavez et al. 1991). Although our data indicated that diatoms from the tropical North Pacific could respond to macronutrient additions (at concentrations similar to equatorial waters), there must be some other factor that is also responsible for preventing diatom populations from exhausting the N03- supply in the equatorial Pacific. For instance, the diatom-specific growth rate (2.5 d-l) we measured in the FEN0 treatment must have been substantially higher than the diatom-specific grazing rate to account for the 43-fold net increase in fucoxanthin concentrations between day 3 (11.4 ng liter-l) and day 6 (49 1 ng liter-l). If we assume that the fucoxanthin : cell ratio in the diatom population did not change significantly (i.e. no photoadaptive response) during the last 3 d of the incubation (exponential growth phase), then the fucoxanthin-specific growth rate [based on pigment concentration; Fucoday 6 = Fucoday 3 exp(@)] equals 1.25 d-l. This fucoxanthin-specific growth rate was equal to the net specific growth rate (1.25 d-l) and was 50% of the diatom absolute growth rate. The diatom net growth rate was + 12 times higher than the prochlorophyte net growth rate in the FEN0 treatment (Table 3). Thus, protozoan grazing processes appeared to be more important in controlling the prochlorophyte population (g : p = 0.8 5) than the macrograzers were at controlling the diatom population (g :,u = 0.50). We can obtain an additional estimate of the diatom g : h ratio by calculating the amount of diatom chlorophyll expected in the FEN0 treatment over the 6-d incubation based on the chlorophyll in the > 5-pm fraction (0.62 pg liter-l; Table 1 and Fig. 3), the utilization of N03- during the experiment (1.8,uM; Table I), the phytoplankton C : Chl ratio (58; based on Chl labeling), and the C: N assimilation ratio (5.7 by wt; Fig. 4). The calculated diatom Chl concentration expected in the FEN0 treatment is 2.1 pg liter-. Thus, the actual Chl concentration measured in the > 5-pm fraction (0.6,2 bg liter- ) was 30% of the expected Chl concentration (the remainder of the Chl, presumably, being grazed) or g : p = In light of the possible errors associated with botlh g : p calculations, the values are relatively close and suggest that the diatom growth and grazing processes were not in balance in our FEN0 treatment (g : p = ). Although the two sets of calculations are in approximate agreement, we should emphasize that these values represent only indirect estimates of grazing rates on diatoms. Hence, our calculated ratio of g : p (N 0.40) for diatoms should be viewed with caution. Diatom blooms in Fe-incubation experiments may be simply an artifact of subculturing due to macrograzer exclusion (Dugdale and Wilkerson 1990; Welschmeyer et al ). Our grow-out experiments were performed in lo- 1ite:r carboys in an attempt to minimize possible macrograzer exclusion which could potentially cause an upper bias in the final yield of diatom biomass. In a separate experiment, we obtained results (i.e. diatom blooms following Fe enrichment) with 22-liter carboy incubations at the equator that were similar to those obtained by investigators who used 2.7-liter bottles (data not. shown). It is entirely possible that present seawater collection methods (Go-flo bottles) do not sample the macrograzer community representatively. If macrograzers (and not Fe), however, were solely responsible for controlling diatom populations in situ and if we assume that there was an equal probability that we excluded these macrograzers from each of our treatments, then we should still have observed the same increase in fucoxanthin concentration in the N tre*atment as we did in the FEN0 treatment. Thus, our results did not support the macrogra.zer exclusion hypothesis as the sole cause for diatom blooms in these types of experiments. Furthermore, other equatorial enrichment experiments have reported diatom blooms despite the presence of significant numbers of copepods and diatom-grazing protists (Chavez et al ). Based on our results, it seems more likely that Fe affects macrograzer biomass indirectly by bottom-up control of

13 Fe and nutrient limitation 507 diatom growth rather than macrograzers exerting top-down control of diatom biomass. Conclusions We conclude from our data that grazing was the proximate control on prochlorophyte biomass but that diatom populations in the tropical North Pacific were colimited by Fe and macronutrients. Paradoxically, procaryotic algae were not limited by Fe even though they possess a higher Fe cell quota than eucaryotic algae (Brand 199 1). Our data corroborated the recent report suggesting that, in these low-fe oceanic regimes, procaryotic algae may be more efficient at using recycled Fe than are eucaryotic algae (Hutchins et al. 1993). Grazing rates on prochlorophytes were more nearly in balance with growth rates (g : p = 0.85) than they were for the diatom community (g : or ). Our data imply that during episodes of atmospheric dust deposition in the equatorial Pacific, protozoan grazing processes may be more important in controlling the procaryotic biomass than macrograzers are in regulating the diatom community, which appears to be predominantly Fe limited. Our data support Martin s hypothesis that Fe is vital in controlling diatom growth in the equatorial Pacific. On the basis of the diatom growth rate in our FEN0 treatment, P,,,, rates in the equatorial Pacific seem to be limited by episodic Fe inputs. It appears that the procaryotic contribution to P,,, in high nitrate waters is limited because of the efficiency of the grazing community in controlling the procaryotic phytoplankton biomass. Because larger diatoms have more rapid settling velocities, export C flux may well be affected by atmospheric depositional events. Microbial loop recycling processes and Fe limitation of diatom growth may be responsible for the relatively lowf-ratios reported from the equatorial Pacific region. Only in situ evidence, however, will confirm or disprove these hypotheses and reveal to what extent Fe controls global new production. References ARPIN, N., W. A. SVEC, AND S. LIAAEN-JENSEN New fucoxanthin-related carotenoids from Coccolithus huxleyi. Phytochemistry 15: BANSE, K Grazing, temporal changes of phyto- plankton concentrations, and the microbial loop in the open sea, p In P. G. Falkowski and A. D. Woodhead [eds.], Primary productivity and bio- geochemical cycles in the sea. Plenum. BRAND, L. E Minimum iron requirements of marine phytoplankton and the implications for the biogeochemical control of new production. Limnol. Oceanogr. 36: 1756-l BRULAND, K. W Oceanic Zn speciation: Complexation of Zn by natural organic ligands in the central North Pacific. Limnol. Oceanogr. 34: , R. P. FRANKS, G. A. KNAUER, AND J. H. MARTIN, Sampling and analytical methods for the determination of copper, cadmium, zinc and nickel at the nanogram per liter level in seawater. Anal. Chim. Acta 105: CHAVEZ, F. P., AND OTHERS Growth rates, grazing, sinking, and iron limitation of equatorial Pacific phytoplankton. Limnol. Oceanogr. 36: CHISHOLM, S. W., AND OTHERS A novel free-living prochlorophyte abundant in the oceanic euphotic zone. Nature 334: , AND F. M. M. MOREL [EDS.] What controls phytoplankton production in nutrient-rich areas of the open sea? Limnol. Oceanogr. 36: CULLEN, J. J., M. R. LEWIS, C. 0. DAVIS, AND R. T. BARBER Photosynthetic characteristics and estimated growth rates indicate grazing is the proximate control of primary production in the equatorial Pacific. J. Geophys. Res. 97: DITULLIO, G. R Incorporation of 14C0, into protein as an estimate of phytoplankton N-assimilation and relative growth rate, in press. Zn P. Kemp et al. [eds.], Current methods in aquatic microbial ecology. Lewis. AND E. A. LAWS Diel periodicity of C and N assimilation in 5 species of marine phytoplankton: Accuracy of methodology for predicting N-assimilation rates and N/C composition ratios. Mar. Ecol. Prog. Ser. 32: 123-l 32. -, AND Impact of an atmosphericoceanic disturbance on phytoplankton community dynamics in the North Pacific central gyre. Deep-Sea Res. 38: DUGDALE, R. C., AND F. P. WILKERSON Iron addition experiments in the Antarctic: A reanalysis. Global Biogcochem. Cycles 4: 13-l 9. EPPLEY, R. W Temperature and phytoplankton growth in the sea. Fish. Bull. 70: , E. H. RENGER, E. L. VENRICK, AND M. M. MULLIN A study of plankton dynamics and nutrient cycling in the central gyre of the North Pacific Ocean. Limnol. Oceanogr. 18: FITZWATER, S. E., G. KNAUER, AND J. H. MARTIN Metal contamination and its effect on primary production measurcmcnts. Limnol. Oceanogr. 27: FREY, B. F., AND L. F. SMALL Recycling of metabolized iron by the marinc dinoflagellate dinium carterue. J. Phycol. 15: Amphi- FROST, B. W The role of grazing in nutrient-rich areas of the open sea. Limnol. Oceanogr. 36: GIESKES, W. W. C., AND G. W. KRAAY Estimating

14 508 DiTullio et al. the carbon-specific growth rate of major algal species in eastern Indonesian waters by 14C labcling of taxonspecific carotenoids. Deep-Sea Res. 36: 1127-l 139. GLOVER, H.E.,B.B, PRÉZELIN, L. CAMPBELL, M. WYMAN, AND C. GARSIDE A nitrate-dependent Synechococcus bloom in surface Sargasso Sea water. Nature 331: GOERICKE, R., AND D. J. REPETA The pigments of Prochlorococcus marinus: The presence of divinyl chlorophyll a and b in a marine procaryote. Limnol. Oceanogr. 37: AND N. A. WELSCHMEYER. In press. The carotenoid labeling method: Measuring specific rates of carotenoid synthesis in natural phytoplankton communities. Mar. Ecol. Prog. Ser. GOLDMAN, J. C Physiological processes, nutrient availability, and the concept of relative growth rate in marine phytoplankton ecology, p. 179-l 94. Zn Primary production in the sea. Brookhaven Symp. Biol Plenum. -, J. J. MCCARTHY, AND D. G. PEAVEY Growth rate influence on the chemical composition of phytoplankton in oceanic waters. Nature 279: 2 1 O GORDON, R. M., J. H. MARTIN, AND G. A. KNAUER Iron in northeast Pacific waters. Nature 299: 6 1 l GREENE, R. M., R. J. GEIDER, AND P. G. FALKOWSKI Effect ofiron limitation on photosynthesis in a marine diatom. Limnol. Oceanogr. 36: 1772-l 782. HARVEY, H. W Manganese and the growth of phytoplankton. J. Mar. Biol. Assoc. U.K. 26: HUTCHINS, D. A., G. R. DITULLIO, AND K. W. BRULAND Iron and regenerated production: Evidence for biological iron recycling in two different marine environments. Limnol. Oceanogr. 38: in press. LAWS, E. A., G. R. DITULLIO, K. L. CARDER, P. R. BETZER, AND S. HAWES Primary production in the deep blue sea. Deep-Sea Res. 37: LEWIN, J., R, E. NORRIS, S. W. JEFFREY, AND B. E. PEARSON An aberrant chrysophycean alga Pelagococcus subviridis gen. nov. et sp. nov. from the North Pacific Ocean. J. Phycol. 13: MARTIN, J. H., AND R. M. GORDON Northeast Pacific iron distribution in relation to phytoplankton productivity. Deep-Sea Res. 35: AND S. E. FITZWATER The case fo; iron. Limnol. Oceanogr. 36: 1793-l 802. MENZEL, D. W., AND J. H. RYTHER Nutrients limiting the production of phytoplankton in the Sargasso Sea, with special reference to iron. Deep-Sea Res. 7: MOREL, F. M. M., AND R. J. M. HUDSON The geobiological cycle of trace elements in aquatic systems: Rcdfield revisited, p. 25 l Zn W. Stumm [ed.], Chemical processes in lakes. Wiley. OLSON, R. J., S. W. CHISHOLM, E. R. ZETTLER, M. A. ALTABET, AND J. A. DUSENBERRY Spatial and temporal distributions of prochlorophyte picoplankton in the North Atlantic Ocean. Deep-Sea Res. 37: , E. R. ZETTLER, AND M. D. DURAND Phytoplankton analysis using flow cytometry, in press. Zn P. Kemp et al. [eds.], Current methods in aquatic microbial ecology. Lewis. PEG,\, M. A., M. R. LEWIS, AND W. G. HARRISON Primary productivity and size structure of phytoplankton biomass on a transect of the equator at 135 W in the Pacific Ocean. Deep-Sea Res. 37: PRIC:E, N. M., L. F. ANDERSON, AND F. M. M. MOREL Iron and nitrogen nutrition of equatorial Pacific plankton. Deep-Sea Res. 38: 136 l-l 378. REDALJE, D. G., AND E. A. LAWS A new method for estimating phytoplankton growth rates and carbon biomass. Mar. Biol. 62: REDFIELD, A. C., B. H. KETCHUM, AND F. A. RICHARDS The influence oforganisms on the composition of seawater, p Zn M. N. Hill [ed.], The sea. V. 2. Interscience. RYTHER, J. H., AND R. R. L. GUILLARD Enrichment experiments as a means of studying nutrients limiting to phytoplankton production. Deep-Sea Res. 6: SUNDA, W. G., D. SWIFT, AND S. HUNTSMAN Low iron requirement for growth of oceanic phytoplankton. Nature 351: THC,MAS, W. H On nitrogen deficiency in tropical Pacific oceanic phytoplankton: Photosynthetic parameters in poor and rich water. Limnol. Oceanogr. 15: TRKK, C. G Hydoxamate-siderophore production and utilization by marine eubacteria. Curr. Microbiol. 18: VAIJLOT, D., AND F. PARTENSKY Cell cycle distributions of prochlorophytes in the north western Mediterranean Sea. Deep-Sea Rcs. 39: WEI-SCHMEYER, N. A., R. GOERICKE, S. STROM, AND W. PETERSON Phytoplankton growth and herbivory in the subarctic Pacific: A chemotaxonomic analysis. Limnol. Oceanogr. 36: 163 l-l 649. Submitted: 31 July 1992 Accepted: 3 February 1993 Revised: 1 March 1993