Phytoplankton biomass and carbon assimilation were measured during four cruises to the southern Ross Sea, Antarctica during 1996 and 1997 in order to assess the details of the seasonal cycle of productivity. Phytoplankton biomass increased rapidly during the austral spring, and integrated chlorophyll reached a maximum during the early summer period (January 5), and decreased thereafter. Particulate matter ratios (carbon:nitrogen, carbon:chlorophyll) also showed distinct seasonal trends. Carbon assimilation increased rapidly in the spring, and reached a maximum of 162 mmol C m^-2 d^-1, 25 days earlier than the maximum observed biomass (during mid-December; year day 346). It decreased rapidly thereafter, and in austral autumn when ice was forming it approached zero. The time of maximum growth rate coincided with the maximum in C-assimilation, and at 0.9 d^-1, exceeded predictions based on laboratory cultures. Growth rates over the longer term, however, were generally much less. Deck-board incubations suggested that photoinhibition occurred at the greatest photon flux densities, but in situ incubations revealed no such surface inhibition. We suggest that due to the nature of the irradiance field in the Antarctic (greater contribution to reflected irradiance) that assemblages maintained in on-deck incubators received more light than those in situ, which resulted in photoinhibition, which in turn resulted in a 17% underestimate in on-deck productivity relative to in situ determinations. The phytoplankton bloom appeared to be initiated when vertical stability was imparted in austral spring, coincident with greater daily photon flux densities. Conversely, decreased productivity likely resulted from trace metal limitation, whereas biomass declines likely resulted from enhanced loss rates such as aggregate formation and enhanced vertical flux. The seasonal progression of productivity and biomass in the southern Ross Sea was similar to other areas in the ocean that experience blooms, although the relationship between phytoplankton and other food web components shows significant variations from the rest of the world's oceans.

• Introduction

In recent years the large-scale distribution of phytoplankton in the Southern Ocean, obtained by satellites, has been used extensively not only to quantify the temporal and spatial variations of carbon transformations observed (Sullivan et al., 1993; Comiso et al., 1993; Arrigo and McClain, 1994; Arrigo et al., 1998a; Moore et al., 1999), but to understand the relationship of primary productivity with environmental forcing such as wind, sea ice, and irradiance (Banse, 1996). Mitchell et al. (1991) questioned the use of older remote sensing algorithms to quantify pigments in polar waters, and the deployment and calibration of newer instruments confirmed the errors associated with older representations with CZCS imagery (Moore et al., 1999). Satellites are unable to estimate phytoplankton abundance in areas of extensive cloud and/or ice cover, and hence are of relatively little use during extended periods in the Southern Ocean, such as austral spring. However, Smith and Gordon (1997) clearly showed that phytoplankton biomass during the ice-covered, early spring (mid- November) was large (chlorophyll concentrations > 3 g l^-1), and thus growth must have been initiated during October. Austral spring contributes significantly to the annual production (e.g., Nelson et al., 1996; Arrigo et al., 1998b) and must be included in the assessment of seasonal carbon dynamics of the Ross Sea. Previous observations have shown that phytoplankton biomass declines throughout the summer from the maximum in December (Arrigo and McClain, 1994; Smith et al., 1996; Asper and Smith, 1999).

The seasonal cycle of polar phytoplankton productivity is even more poorly known than that of biomass. Most studies of productivity have been confined to shorter, often ice-free periods. For example, Holm-Hansen and Mitchell (1991) found that productivity ranged from 0.12 - 2.12 g C m^-2 d^-1 in the Antarctic Peninsula region. Smith et al. (1996) found that summer (January) production in the Ross Sea averaged 2.63 g C m^-2 d^-1 in mid-January, decreasing to 0.78 g C m^-2 d^-1 in February. Open ocean rates are substantially lower. El-Sayed et al. (1983) found the mean productivity in the Ross Sea sector (off the continental shelf) to be 0.33 g C m^-2 d^-1, or about 56% that found at similar times on the continental shelf. Hence both time and space variations confound the determination of seasonal production in the Antarctic. Arrigo et al. (1998b) estimated productivity of the Ross Sea using a bio-optical model, and their results were consistent with the direct observations made earlier.

Our objectives in this study were two-fold. First we wanted to determine the seasonal cycle of productivity in the southern Ross Sea, and also to determine the seasonal cycle of phytoplankton biomass and growth. Second, we wanted to determine the environmental controls of productivity in the region. It has been proposed that irradiance is the major limiting factor during the austral spring, whereas during austral summer growth is limited by micronutrient availability (Smith and Gordon, 1997; Sedwick and DiTullio, 1998). Changes in biomass have been suggested to result from grazing (DiTullio and Smith, 1996) and enhanced sinking due to aggregate formation (Asper and Smith, 1999); furthermore, the relative importance of these loss processes appears to be largely controlled by the composition of the surface phytoplankton assemblage. A further objective of our study, therefore, was to understand the mechanisms leading to the loss of biomass in the Southern Ross Sea.
 

• Materials and Methods

We conducted four cruises to the southern Ross Sea in 1996 and 1997 on the R.V.I.B. Nathaniel B. Palmer as part of the U.S. Southern Ocean JGOFS study (AESOPS, or Antarctic Environment Southern Ocean Process Study; Table 1; Figure 1a, Figure 1b, Figure 1c, Figure 1d, Figure Legend). The first (NBP96-04A) sampled the very early spring period and was designed to investigate the causes of the initiation of phytoplankton growth in the spring. The second (NBP97-01) sampled during summer when phytoplankton biomass was large (albeit declining), and the third (NBP97-03) assessed the transition from late autumn to winter. The fourth cruise (NBP97-08) was conducted during the spring, the period of presumed most rapid growth and greatest productivity. Surface irradiance measurements as well as local meteorological conditions were recorded continuously from sensors mounted on the ship's mast. Photon flux densities were measured using a BioSpherical Instruments 4 sensor, and data were binned and averaged at 1 minute intervals.

Much of the sampling occurred along 7630'S at a series of eight stations, each separated by ca. 60 km (Table 1; Figure 1a, Figure 1b, Figure 1c, Figure 1d,  Figure Legend). Depths ranged from 324 m to > 700 m. The Ross Sea polynya generally opens near the ice shelf in austral spring and extends northward the area with reduced ice concentrations (Comiso et al., 1993), so that NBP96-04A sampled across the polynya. Complete ice cover was observed during NBP97-03, and extensive concentrations of ice (broken, with occasional extensive open areas and substantial mesoscale variability) were noted during NBP96-4A and NBP97-08. In contrast, little ice, except the pack ice to the west, was encountered during NBP97-01.

Profiles of temperature and salinity were collected throughout the water column using a CTD (SeaBird 911+) with dual sensors, and which also had a fluorometer (Chelsea Instruments) and an irradiance (PAR) sensor with a cosine collector (BioSpherical Instruments) interfaced with the CTD package. The CTD was mounted on a rosette of 24 10-liter Niskin bottles to collect discrete samples for nutrients and biomass (chlorophyll; particulate organic carbon and nitrogen) throughout the water column. The rosette frame was coated with an inert plastic resin to reduce trace metal contamination. Samples for productivity measurements were collected using a trace-metal clean rosette mounted with Go-Flo^© Niskin bottles (30-liter capacity) (Hunter et al., 1996), which also had a PAR sensor interfaced. The tops of all Go-Flo bottles were covered with plastic before and after deployment to reduce airborne contamination. Depths of collection were determined from the underwater irradiance profile collected during profile.

2.2 Analytical methods

Chlorophyll and nutrient concentrations were quantified using standard JGOFS procedures (JGOFS, 1996). Chlorophyll samples were filtered through 25 mm Whatman GF/F filters under low (< 1/3 atm) vacuum, folded in half, and placed in centrifuge tubes with 7 ml 90% acetone. All samples were stored in the dark at -20C for 24 h, and the extracted fluorescence read before and after acidification using a Turner Designs Model 10-AU fluorometer. Particulate carbon and nitrogen concentrations were determined by filtration through combusted (450C for 2 h) Whatman GF/F filters under low vacuum, after which the filters were placed in combusted glass tubes, capped with combusted aluminum foil, and dried at 60C. All samples were returned to the laboratory and processed using a Carlo-Erba Model 252 elemental analyzer. Acetanilide was used as a standard.

Primary productivity was measured as carbon assimilation with both in situ and on-deck incubations. Samples were collected from known depths within the euphotic zone and placed in tissue culture flasks (total volume 280 ml). They were immediately inoculated with ca. 20 Ci of trace-metal clean solution of NaH¹⁴CO₃ (pH 9.6), which was stored at 4C in a Teflon bottle Samples were placed in specially constructed plexiglass holders which were then attached to a plastic-coated, 0.25" steel rope. A 22.5 kg weight kept the line vertical, whereas the surface of the line was kept afloat by a large sphere, which was then connected to a spar-buoy with a navigational beacon. In addition to the samples, in situ PAR (photosynthetically active radiation) 4 sensors (BioSpherical Instruments Inc.) with internal recording devices were placed at selected depths to continuously record the irradiance within the water column. Generally four PAR sensors and samples from six depths were placed on each array. Samples were recovered after ca. 24 h. All samples were filtered through 25 mm GF/F filters under low vacuum. Filters were placed in scintillation vials with 0.1 ml 1N HCl (to remove adsorbed inorganic ¹⁴C and carbonates), and the radioisotope incorporation determined after another 24 h using liquid scintillation techniques. Values were corrected for abiotic absorption using time-zero controls. Total added radioactivity was measured by directly pipetting an unfiltered aliquot (0.1 ml) directly into 50 l -phenethylamine, to which fluor was directly added. All PAR data were down-loaded from each sensor immediately after each deployment.

Parallel incubations were conducted in on-deck incubators. The incubators were constructed of clear plexiglass with lids to help reduce freezing, and all but one covered with neutral density screens that reduced irradiance to 50, 23, 16, 7, 5 and 2% of the surface value. These reductions were repeatedly checked using a hand-held quantum meter. All sample collection, incubation, filtration and quantification procedures were the same as those used for the in situ incubations. Chlorophyll concentrations were measured from all trace-metal clean casts as well.

2.3 Data analysis

All data from on-deck incubations were processed using the procedures described by Barber et al. (1997), since the largest source of error in estimates of primary productivity occurs in assigning depths (Barber et al., 1996). We used the observed chlorophyll distributions to model the in situ irradiance attenuation after Morel (1988), and then recalculated the productivity-irradiance profiles, which in turn allowed accurate estimates of integrated production of the water column. All productivity data (actual observations and modeled values) are available via http://www1.whoi.edu/jg/dir/jgofs/southern/. Growth rates were calculated from productivity and particulate carbon concentrations (Eppley, 1967). Mixed layer depths were determined from the _t values calculated from the 1-m averaged data from the CTD casts. The mixed layer was defined by a change of 0.01 _t unit from the stable, surface value.
 

• Results

Early spring biomass was extremely low, with surface concentrations of chlorophyll averaging 0.11 ± 0.06 g l^-1 during the first week of NBP96-4A (October 17-24). Chlorophyll concentrations increased markedly in November and December (Table 2; Fig. 2a), and on year day 347 (December 12) the maximum observed surface value was reached (14.1 g l^-1). The maximum integrated (through 100 m) chlorophyll concentration occurred (377 mg m^-2 ) on January 5. Similarly, particulate organic carbon values also increased rapidly in spring (euphotic zone values ranged from 0.95 to 108 mol l^-1; Fig. 2b). Particulate nitrogen (PN) values also increased concomitantly, reaching 566 mmol ^--2 in early January (Table 2). A second order polynomial trend line generated from the C/N ratio of surface particulate organic matter predicted minimal values on January 3. The minimum observed value of 5.24 mol/mol occurred on year day 343 (December 9) and the maximum during the austral autumn/winter when PN concentrations were lowest (Fig. 2c). Particulate carbon/chlorophyll ratios were lowest in late spring (Table 2), with the observed minimum of 24.8 occurring on year day 324 (November 20). The minimum predicted from a second order polynomial regression would occur on year day 322 (November 22; Fig. 2c). Values increased markedly during the autumn cruise and weekly means exceeded 1,150 (Table 2).

Primary productivity measured by on-deck incubations also increased rapidly in austral spring (Table 2, Fig. 3a), concurrent with the integrated daily photon flux density (Fig. 4). Maximum productivity values were observed on year day 341 (December 6) and equaled 162 mmol C m^-2 d^-1. Minimum values were recorded during the autumn cruise (0.63 mmol C m^-2 d^-1). Maximum in situ productivity values equaled 182 mmol C m^-2 d^-1 on year day 346 (December 12). Integrated daily irradiance varied by three orders of magnitude among the cruises, with minima in daily PAR occurring during the autumn-winter cruise. Photoperiods reached 24 h on year day 293 (October 20) and did not decrease until after the summer cruise (mid-February). During the autumn cruise photoperiods decreased from more than 8 h to less than 4 h (Fig. 4). Productivity was weakly correlated with daily irradiance when data from all cruises are pooled (PP = 0.498PAR + 11.8; R² = 0.287, n = 83, p < 0.001), but there was a large amount of variability in this relationship for individual cruises as well as the entire series. P^B_opt (the maximum chlorophyll-specific productivity within the water column) increased even more rapidly than chlorophyll or productivity, and tripled from 0.62 to 1.92 mg C (mg chl)^-1 h^-1 in 13 days in early spring (Fig. 3b), suggesting rapid photoadaptation to the increasingly favorable irradiance field. Maximum P^B_opt values were found on day 322 (3.41 mg C (mg chl)^-1 h^-1).

In situ measurements of productivity generally were greater than those obtained from on-deck incubations. This difference often resulted from photoinhibition that on-deck incubations exhibited at the highest irradiance levels (Fig. 5). The mean underestimate (determined by regression) of integrated productivity obtained by on-deck incubations relative to the in situ measurements incubated at the same irradiance percentage was 17%. These underestimates varied temporally as well, with the greatest underestimates occurring during the summer when photoinhibition was most pronounced, and the least during the late autumn.

Growth rates of surface assemblages, derived from standing stock estimates (particulate organic carbon concentrations) and productivity estimates, were maximal in austral spring (Figure 6), with the maximum rate (0.94 h^-1) occurring on day 321. Mean surface growth rates during early and late spring averaged 0.15 and 0.32 h^-1, respectively. The four greatest growth rates observed occurred within a four-day period from November 19 to 23. Growth rates during the summer cruise were extremely low, and averaged 0.06 h^-1, despite the favorable irradiance fields and macronutrient concentrations. That is, under conditions that are seemingly favorable for phytoplankton growth (vertical stratification, little wind mixing, large amounts of macronutrients, no ice and large amounts of incident irradiance), phytoplankton growth in summer was close to zero. In contrast, during early spring when ice was present, irradiance was lower, and the mixed layers were relatively deep, growth rates were 2.5 times greater.
 
 
 

Discussion

Waters of the Ross Sea are striking for the seasonal variations that they exhibit with regard to phytoplankton biomass and productivity. Previous studies as well as recent satellite images have established that this region is the site of the Antarctic's most spatially extensive phytoplankton bloom (Sullivan et al., 1993; Comiso et al., 1993; Smith et al., 1996; Arrigo et al., 1998b), and its peak concentrations of biogenic material rival those of highly eutrophic coastal systems (Smith and Nelson, 1985; Lancelot and Mathot, 1985; Platt and Sathyendranath, 1988). However, the details of the entire seasonal cycle have until this time been poorly resolved. Our results show that the Ross Sea is characterized by a unimodal peak in both primary productivity and phytoplankton biomass, although the timing of the maximum is different for each. That is, biomass is maximal in early January, whereas productivity was greatest approximately 25 days earlier (Table 2, Figs. 2a, 2b, 2c, 3a, 3b). Additionally, the maximum observed chlorophyll-specific photosynthetic rate was observed ca. 3 weeks prior to the integrated productivity maxima. The rapid increase in P^B_opt reflected the increase in carbon assimilation early in the season and the temporal delay in biomass accumulation. These lags suggest a distinct cascade of events within the assemblage that enable phytoplankton to maximize its growth. First, the photophysiological status of the phytoplankton is optimized (as evidenced by the low C;Chl ratios in austral spring, as well as the rapid photoadaptation to enhanced photon flux densities; Fig. 3b), allowing for rapid growth per unit biomass to occur under in situ irradiance and nutrient conditions. Second, water column productivity becomes maximal under optimal photophysiological state (created by adaptation to increased vertical stratification). Finally, productivity ultimately decreases due to limitation by an environmental factor. Biomass increases as productivity becomes maximal because loss rates are low at this time, and then decline when losses become greater. As such, we suggest that growth is limited by one factor, whereas biomass is limited by a completely independent process.

We further suggest that the factor limiting phytoplankton productivity and growth during late spring and summer is trace metal availability. Sedwick and DiTullio (1997) and Coale et al. (submitted) both found a strong growth response by Ross Sea phytoplankton during early summer upon addition of trace amounts ( 1 nM) of iron. In addition, Olson and Sosik (in preparation) found that the variable fluorescence of single cells measured during pump-during-probe flow cytometry experiments, which is a sensitive measure of the limitation of cells by inorganic nutrients, was low in unenriched controls but increased significantly upon iron addition. Irradiance clearly did not limit phytoplankton growth in the surface mixed layer, as mean photon flux densities in late spring and early summer were always greater than those required to saturate photosynthesis (Hiscock et al., in preparation). Macronutrient concentrations also remained elevated throughout the study (greater than measured half-saturation constants; Smith and Harrison, 1991) and could not have contributed to growth rate reductions. Hence, it appears that the most likely cause for the reduction in productivity during late December was the depletion of iron.

In contrast, during early spring productivity was likely limited by irradiance. During NBP96-04A, NBP97-01 and NBP97-08, the mixed layer depths were always less than the calculated critical depths (Nelson and Smith, 1990; Table 2). The mean irradiance [calculated after Riley (1957)] within the mixed layer also increased rapidly during the spring and was much greater than the photon flux density required to drive positive photosynthesis. However, this analysis used the irradiance levels that impinged upon the ice, and up to 98% of surface irradiance can be attenuated by a 1-m thick slab of ice (Marra and Boardman, 1984; Smith and Sakshaug, 1990). The ice in our region in spring was largely thinner, and even if one-half of surface irradiance was attenuated by ice, enough energy would remain to force photosynthesis. However, it is impossible to quantify the effect of ice on water column productivity in our study, but it is likely that even in early spring, irradiance was sufficient to support phytoplankton growth, albeit not at maximum rates.

The biomass of phytoplankton did not reach its maximum until some 25 days after the productivity maximum (Figs. 2a, 2b, 2c, 3a, 3b). We interpret this to mean that although phytoplankton growth and productivity had declined due to micronutrient limitation, the biomass (the balance between growth and losses) continued to slowly increase until loss processes exceeded growth, causing biomass to decrease. Loss processes operative in the Ross Sea include zooplankton grazing and ingestion (and subsequent enhanced vertical flux of fecal material), increased sinking of large colonies of phytoplankton due to micronutrient limitation, production of large particles via aggregation and accelerated vertical flux, increased rates of viral lysis of phytoplankton cells (and enhanced rates of sinking of detrital material), and losses due to increased vertical mixing. We cannot effectively discriminate among all of these processes, but believe some are more likely explanations for the observed decreases in biomass than others. For example, grazing appears to be a function of phytoplankton community composition (e.g., Verity and Smetacek, 1996). As large portions of the southern Ross Sea are dominated by Phaeocystis antarctica, and because P. antarctica in the Ross Sea has extremely low losses as a result of grazing by microzooplankton (Caron and Dennett, in preparation), losses due to microzooplankton removal appear to be small. However, in other regions Phaeocystis has been reported to be effectively removed by mesozooplankton (e.g., Holm-Hansen and Huntley, 1984; Estep et al., 1990, Haberman, 1998), and we cannot discount the potential for removal by larger zooplankton in the Ross Sea. However, Dunbar et al. (1999) found that the contribution of fecal pellets to the organic matter collected in sediment traps in areas dominated by Phaeocystis was small, suggesting that Phaeocystis in large part was not being grazed by mesozooplankton. No data are available on the rates of viral lysis on Phaeocystis in the Ross Sea, although viruses have been shown to infect P. pouchetii (Jacobsen et al., 1996). Mixed layer depths (and hence mixing losses) are not significantly greater throughout summer or between summer and late spring (Table 2), although at specific locations (e.g., 17630'S, 176E) the mixed layer did deepen during late summer. The broad temporal trend, however, suggests that physical dispersion of the bloom alone cannot account for the decline in biomass.

Asper and Smith (1999) found that vertical flux rates, as measured by short-term deployments of floating sediment traps, increased in summer relative to spring, and suggested that loss processes exhibited a temporal signal that was substantially uncoupled from phytoplankton productivity and biomass accumulation. They further suggested that this resulted from the temporal pattern of aggregate formation and enhanced vertical flux, and camera studies showed that the abundance and sinking rates of aggregates (> 0.5 mm) increased during periods of enhanced flux (Asper et al., unpublished). They did not, however, rule out simple enhanced sinking rates of colonies of P. antarctica, since these particles were smaller than the camera system's resolution, and our data likewise cannot preclude that possibility.

The in situ measurements of primary productivity were quantitatively different from the on-deck measurements both in the absolute rates and the vertical distribution of carbon incorporation. Comparison of paired, productivity values showed that in situ measurements were ca. 17% greater than the estimates derived from on-deck incubations. This difference can be attributed to the strong photoinhibition exhibited in the on-deck samples at the isolumes with substantial photon flux densities (Fig. 5). Photoinhibition was most marked during periods when irradiance was greatest. We suggest that the light environment which samples experience in on-deck incubators is significantly different than that in situ, since the surface irradiance includes a strong contribution of reflected and low angle light (for example, from the ship's deck, as well as surrounding snow and ice) which does not occur within the water column Barber et al., 1996). Measurements of PAR using a cosine collector showed that the photon flux densities measured by a 4 sensor were ca. 45% greater (Reynolds and Mitchell, unpublished), but that the temporal patterns were consistent between the two types of sensors. Hence the in situ irradiance environment was quantitatively less than that observed on deck, and but this reduction led to the reduction of photoinhibition and increased measured rates of photosynthesis. We also made a some paired comparisons between samples with the spectral quality changed (by using blue filters) and those with only the quantity changed (by neutral density filters), and no significant difference was observed (data not shown). Hence it appears that in turbid, eutrophic regions such as the Ross Sea, light quality effects are of less importance than light quantity effects, in contrast to the results of oligotrophic tropical regions (Laws et al., 1991).

Growth rates during the cruises varied dramatically (Fig. 6). Modest rates were observed in early spring (mean growth rate was 0.15 h^-1) compared to the temperature-limited rate of 0.52 h^-1 (Eppley, 1972). Late spring growth rates were greater, and averaged 0.32 h^-1 during NBP97-03. Maximum growth rates were observed during NBP97-08, with growth during one four-day period in November exceeding the maximum predicted by the Eppley (1972) equation. It is possible that the error due to detrital contribution to POC concentrations varied with time, but the magnitude of the error is likely much less than the trends we observed. Few data are available to provide a quantitative description of temperature-limited growth below 2C (the lower limit of the data available to Eppley), and it is impossible to know if the large rates observed are overestimates due to unbalanced growth or if they represent attainable growth rates. Smith et al. (submitted) found that diatom growth rates (determined by ³²Si uptake rates) were always low in the Ross Sea (< 0.2 h^-1), whereas carbon-based growth rates occasionally approached 0.5 h^-1. Clearly both environmental factors and community composition can influence net growth rates in Antarctic waters, and our results show the extreme variability that can occur within a restricted region. They also suggest that a better understanding of the maximum attainable growth rate is needed to improve our predictive capability in the event of future increases in surface layer temperatures.

A seasonal study was previously conducted in the Bransfield Strait region (e.g., Mitchell and Holm-Hansen, 1990), and this project attempted to relate quantitatively the local mixing regime with phytoplankton biomass and growth. The AESOPS results represent a much more comprehensive data set with which to test these relationships. We found that significant chlorophyll concentrations were not observed in mixed layers greater than 50 m, similar to the suggestion of Sakshaug and Holm-Hansen (1984). Furthermore, surface chlorophyll concentrations were inversely related to mixed layer depths (Fig. 7; CHL = 77.2z_m^-1.31; R = 0.722; p <0.001). The implication of this correlation is that biomass accumulates when stratification occurs, not only because the in situ environment is optimized for phytoplankton growth, but because phytoplankton biomass can in turn accumulate when vertical mixing losses are minimized. Trace-metal limitation occurs after phytoplankton demand exceeds the supply of additional iron. Conversely, deep mixing will disrupt stratification and reduce growth while replenishing trace metals to the euphotic zone. Only those processes which can supply micronutrients and re-establish vertical stratification (e.g., ice ablation, mesoscale eddies and their associated vertical transports) can provide a semi-continuous, optimal environment for phytoplankton growth. Martin et al. (1990) and Sedwick and DiTullio (1997) suggested that ice melt provided iron which in turn stimulates water column productivity, but surface layers always show very low iron concentrations rather than pulses of metals in melt water (Coale et al., submitted). As such, the role of ice ablation in trace metal stimulation of phytoplankton remains equivocal.

The Ross Sea is the Antarctic's most productive sea (Arrigo et al., 1998), and our results confirm and extend this general conclusion. Phytoplankton photosynthesis, productivity, growth and biomass accumulation in the southern portion respond rapidly to the modest vertical stratification and increasing surface solar radiation found in early spring, and the bloom is initiated. This bloom initiation appears to occur in early October (Smith and Gordon, 1997; Arrigo et al., 1998a), which is earlier than most, if not all, ice-covered waters in the Antarctic, but the reason for this rapid onset of growth is unclear. Rapid growth occurs at sub-zero temperatures until early December (coincidentally the time of complete ice retreat), at which time productivity rapidly decreases, most likely due to trace metal limitation. Biomass remains high, and declines more slowly, as loss processes are only weakly coupled to phytoplankton productivity and biomass. Virtually no autotrophic activity occurs in late autumn. Although the seasonal cycle of growth, biomass accumulation and decline is limited to at most five months, carbon transformations during this period are substantial and are unlike many other areas of the world's oceans (e.g., Karl, 1993). As such, the Ross Sea can serve as a model to which to compare all other highly productive, continental shelf areas in the Southern Ocean.
 
 
 

Acknowledgments. This research was supported by the National Science Foundation (Grants OPP-9531990, OPP-9531981 and OPP-9530611). MRH was supported by a graduate fellowship from the American Meteorological Society and NASA's "Misson to Planet Earth" program. A.-M. White, S. Polk, C. Knudsen, E. Barber, L. Borden, J. Borden and J. Seward provided expert field assistance. We thank J. Morrison and L. Codispoti for hydrographic data, and H. Ducklow for constructive comments. This is JGOFS Contribution No. xxxx and Virginia Institute of Marine Sciences Contribution No. xxxx.

References

Arrigo, K.R. and C.R. McClain, 1994. Spring phytoplankton production in the western Ross Sea. Science 266, 261-263.

Arrigo, K.R., Weiss, A.M. and W.O. Smith, Jr., 1998a. Physical forcing of phytoplankton dynamics in the western Ross Sea. Journal of Geophysical Research 103, 1007-1022.

Arrigo, K.R., Robinson, D.H., Worthen, D.LO., Schieber, B. and M.P. Lizotte, 1998b. Bio-optical properties of the southwestern Ross Sea. Journal of Geophysical Research 103, 21,683-21,696.

Asper, V.L. and W.O. Smith, Jr., 1999. Particle fluxes during austral spring and summer in the southern Ross Sea (Antarctica). Journal of Geophysical Research 104 (in press).

Banse, K., 1996. Low seasonality of low concentrations of surface chlorophyll in the subantarctic water ring: underwater irradiance, iron, or grazing? Progress in Oceanography 37, 241-291.

Barber, R.T., Borden, L., Johnson, Z., Marra, J., Knudsen, C. and C.C. Trees, 1997. Ground truthing modeled k_PAR and on deck primary productivity incubations with in situ observations. SPIE 2963, 834-839.

Barber, R.T., Sanderson, M.P., Lindley, S.T., Chai, F., Newton, J., Trees, C.C., Foley, D.G. and F.P. Chavez, 1997a. Primary productivity and its regulation in the equatorial Pacific during and following the 1991-92 El Niño. Deep-Sea Research 43, 933-970.

Coale, K., Gordon, M., Fitzwater, S. and S. Tanner. Trace metal concentrations in the Antarctic Polar Front zone and the Ross Sea. Deep-Sea Research (submitted).

Comiso, J.C., McClain, C.R., Sullivan, C.W., Ryan, J.P. and C.L. Leonard, 1993. Coastal Zone Color Scanner pigment concentrations in the Southern Ocean and relationships to geophysical surface features. Journal of Geophysical Research 98, 2419-2451.

Cushing, D.H., 1981. Temporal variability in production systems. In: Analysis of Marine Ecosystems, A.R. Longhurst, editor, Academic Press, New York, Pp. 443-471.

DiTullio, G.R. and W.O. Smith, Jr., 1996. Spatial patterns in phytoplankton biomass and pigment distributions in the Ross Sea. Journal of Geophysical Research 101, 18,467-18,478.

Ducklow, H.W. and C.A. Carlson, 1992. Oceanic bacterial production. Advances in Microbial Ecology 12, 113-181.

Dunbar, R.B., Leventer, A.R. and D.A. Mucciarone, 1999. Water column sediment fluxes in the Ross Sea, Antarctica (I): Atmospheric and sea ice forcing. Journal of Geophysical Research (in press).

El-Sayed, S. Z., Biggs, D. C. and O. Holm-Hansen, 1983. Phytoplankton standing crop, primary productivity, and near-surface nitrogenous nutrient fields in the Ross Sea, Antarctica. Deep-Sea Research 30, 871-886.

Eppley, R.W. 1967. An incubation method for estimating the carbon content of phytoplankton in natural samples. Limnology and Oceanography 13, 574-582.

Eppley, R.W. 1972. Temperature and phytoplankton growth in the sea . Fisheries Bulletin 7, 1063-1085.

Estep, K.W., Nejstgaard, J.C., Skjoldal, H.R. and F. Rey, 1990. Grazing of copepods upon Phaeocystis colonies as a function of the physiological state of the prey. Marine Ecology Progress Series 67, 235-249.

Haberman, K.L., 1998. Feeding ecology of the Antarctic krill, Euphausia superba: the role of phytoplankton community composition in the krill's diet. Ph.D. Dissertation, University of California, Santa Barbara. 180 pp.

Holm-Hansen, O. and B.G. Mitchell, 1991. Spatial and temporal distribution of phytoplankton and primary production in the western Bransfield Strait region. Deep-Sea Research 38, 961-980.

Holm-Hansen, O. and M. Huntley, 1984. Feeding requirements of krill in relation to food sources. Journal of Crustacean Biology 4, 156-173.

Hunter, C.N., Gordon, R.M., Fitzwater, S.E. and K.H. Coale, 1996. A rosette system for the collection of trace metal clean seawater. Limnology and Oceanography 41, 1367-1372.

Jacobsen, A., Bratbak, G. and M. Heldal, 1966. Isolation and characterization of a virus infecting Phaeocystis pouchetii (Prymnesiophyceae). Journal of Phycology 32, 923-927.

JGOFS, 1996. Protocols for the Joint Global Ocean Flux Study (JGOFS) core measurements. Report No. 19, Intergovernmental Oceanographic Commission, Bergen, Norway. 170 pp.

Karl, D.M., 1993. Microbial processes in the Southern Ocean. In: Antarctic Microbiology, I. Friedmann, editor, Wiley-Liss, New York, Pp. 1-63.

Lancelot, C., Keller, M.D., Rousseau, V., Smith, W.O. Jr. and S. Mathot, 1998. Autecology of the marine haptophyte Phaeocystis sp. In: Physiological Ecology of Harmful Algal Blooms, D.M. Anderson, A.D. Cembella and G.M. Hallegraeff, editors, NATO ASI Series, Vol. G41, Springer-Verlag, Heidelberg, Pp. 211-224.

Lancelot, C. and S. Mathot, 1985. Biochemical fractionation of primary production by phytoplankton in Belgian coastal waters during short- and long-term incubations with ¹⁴C-bicarbonate. I. Mixed diatom population. Marine Biology 86, 219-226.

Laws, E.A, DiTullio, G.R., Carder, K.L., Betzer, P.R. and S. Hawes, 1991. Primary production in the deep blue sea. Deep-Sea Research 37, 715-730.

Marra, J. and D.C. Boardman, 1984. Late winter chlorophyll a distributions in the Weddell Sea. Marine Ecology Progress Series 19, 197-205.

Martin, J.H., Fitzwater, S.E. and R.M. Gordon, 1990. Iron deficiency limits phytoplankton growth in Antarctic waters. Global Biogeochemical Cycles 4, 5-12.

Mitchell, B.G., Brody, E.A., Holm-Hansen, O., McClain, C.R. and J. Bishop, 1991. Light limitation of phytoplankton biomass and macronutrient utililization in the Southern Ocean. Limnology and Oceanography 36, 1662-1677.

Mitchell, B.G. and O. Holm-Hansen, 1991. Observations and modeling of the Antarctic phytoplankton crop in relation to mixing depth. Deep-Sea Research 38, 981-1007.

Morel, A., 1988. Optical modeling of the upper ocean in relation to its biogenous matter content (Case I waters). Journal of Geophysical Research 93, 10749-10768.

Moore, J.K., Abbott, M.R., Richman, J.G., Smith, W.O., Cowles, T.J., Coale, K.H., Gardner, W.D. and R.T. Barber, 1999. SeaWiFS satellite ocean color data at the U.S. Southern Ocean JGOFS line along 170W. Geophysical Research Letters (submitted).

Nelson, D.M. and W.O. Smith, Jr. 1991. Sverdrup revisited: critical depths, maximum chlorophyll levels and the control of Southern Ocean productivity by the irradiance/mixing regime. Limnology and Oceanography 36, 1650-1661.

Nelson, D.M., Smith, W.O. Jr., Gordon, L.I. and B. Huber, 1987. Early spring distributions of nutrients and phytoplankton biomass in the ice-edge zone of the Weddell/Scotia Sea. Journal of Geophysical Research 92, 7181-7190.

Nelson, D.M., DeMaster, D.J., Dunbar, R.B. and W.O.Smith, Jr., 1996. Cycling of organic carbon and biogenic silica in the Southern Ocean: estimates of water column and sedimentary fluxes on the Ross Sea continental shelf. Journal of Geophysical Research 101, 18,519-18,532.

Platt, T. and S. Sathyendranath, 1988. Oceanic primary production: estimation by remote sensing at local and regional scales. Science 241, 1613-1620.

Riley, G.A., 1957. Phytoplankton of the North Central Sargasso Sea. Limnology and Oceanography 2, 252-270.

Sakshaug, E. and O. Holm-Hansen, 1984. Factors governing pelagic production in polar oceans. In: Marine Phytoplankton and Productivity, O. Holm-Hansen, L. Bolis and R. Giles, editors, Springer-Verlag, Berlin. Pp. 1-18.

Sarmiento, J.L., Hughes, T.M.C., Stouffer, R.J. and S. Manabe, 1998. Simulated response of the ocean carbon cycle to anthropogenic climate warming. Nature 395, 245-249.

Sedwick, P.N. and G.R. DiTullio, 1997. Regulation of algal blooms in Antarctic shelf waters by the release of iron from melting sea ice. Geophysical Research Letters 24, 2515-2518.

Smith, W.O., Jr. and L.I. Gordon, 1997. Hyperproductivity of the Ross Sea (Antarctica) polynya during austral spring. Geophysical Research Letters 24, 233-236.

Smith, W.O., Jr. and W.G. Harrison, 1991. New production in polar regions: the role of

environmental controls. Deep-Sea Research 38, 1463-1479.

Smith, W.O., Jr. and D.M. Nelson, 1985. Phytoplankton bloom produced by a receding ice edge in the Ross Sea: Spatial coherence with the density field. Science 227, 163-166.

Smith, W.O., Jr., Nelson, D.M., DiTullio, G.R. and A.R. Leventer, 1996. Temporal and spatial patterns in the Ross Sea: Phytoplankton biomass, elemental composition, productivity and growth rates. Journal of Geophysical Research 101, 18,455-18,466.

Smith, W.O., Jr. and E. Sakshaug, 1990. Autotrophic processes in polar regions. In: Polar Oceanography, Part B, W.O. Smith, Jr., editor, Academic Press, San Diego. Pp. 477-525.

Sullivan, C.W., Arrigo, K.R.,McClain, C.R., Comiso, J.C. and J. Firestone, 1993. Distributions of phytoplankton blooms in the Southern Ocean. Science 262, 1832-1837.

Verity, P.G. and V. Smetacek, 1996. Organism life cycles, predation, and the structure of marine pelagic ecosystems. Marine Ecology Progress Series 130, 277-293.