Optical backscattering in the Arabian Sea—continuous underway measurements of particulate inorganic and organic carbon

Optical backscattering in the Arabian Sea—continuous underway measurements of particulate inorganic and organic carbon

Deep-Sea Research I 48 (2001) 2423–2452 Optical backscattering in the Arabian SeaFcontinuous underway measurements of particulate inorganic and organ...

771KB Sizes 1 Downloads 82 Views

Deep-Sea Research I 48 (2001) 2423–2452

Optical backscattering in the Arabian SeaFcontinuous underway measurements of particulate inorganic and organic carbon William M. Balcha,*, David T. Drapeaua, Jennifer J. Fritzb, Bruce C. Bowlera, Jessica Nolanc a

Bigelow Laboratory for Ocean Sciences, McKown Point Road, P.O. Box 475, West Boothbay Harbor, ME 04575, USA Department of Meteorology and Physical Oceanography, University of Miami, 4600 Rickenbacker Causeway, Miami, FL 33155, USA c Marine Research Division, Scripps Institution of Oceanography, University of California, San Diego, La Jolla, CA 92093, USA

b

Received 26 May 2000; received in revised form 8 December 2000; accepted 2 April 2001

Abstract Continuous surface measurements of temperature, salinity, fluorescence and optical backscattering were made during R/V Thompson cruise no. TN053 in the northern Arabian Sea (‘‘Bio-Optical cruise’’; October– November, 1995). The cruise covered the early NE monsoon period. Optical measurements involved alternate estimates of total backscattering and acidified backscattering approximately every 1.5–2 min (measured after addition of a weak acid to dissolve calcium carbonate). The difference between total and acidified backscattering equals ‘‘acid-labile backscattering’’. Total and acid-labile backscattering were converted to the concentration of particulate organic carbon (POC) and particulate inorganic carbon (PIC; calcium carbonate), respectively, and discrete samples taken along the cruise track were used for calibration. Backscattering data were frequently coherent with temperature, salinity, and density variability. Acid-labile backscattering values revealed that calcium carbonate accounted for 10–40% of the total optical backscattering in the region, and the semi-continuous records demonstrated distinct patches of coccolith-rich water. The northern Arabian Sea had the highest acid-labile backscattering. Results suggest that PIC : POC ratios can vary over about four orders of magnitude. Highest surface values of PIC : POC approached one in several places. We also report qualitative observations of phytoplankton community structure made aboard ship, on fresh samples. r 2001 Elsevier Science Ltd. All rights reserved. Keywords: Coccolithophores; Optical backscattering; Arabian Sea; Continuous underway sampling

*Corresponding author. Tel.: +1-207-633-9600; fax: +1-207-633-9641. E-mail address: [email protected] (W.M. Balch). 0967-0637/01/$ - see front matter r 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 9 6 7 - 0 6 3 7 ( 0 1 ) 0 0 0 2 5 - 5

2424

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

1. Introduction One of the utilities of optics within biological oceanography is to provide information about phytoplankton abundance in space and time. There is little doubt that chlorophyll fluorescence measurements revolutionized the understanding of phytoplankton biomass distributions. Measurements of algal absorption also have been useful to infer the quantity of chlorophyll a and accessory pigments, extent of photoadaption, and pigment packaging. The absorptive properties of chlorophyll have made satellite estimates of phytoplankton possible, as well, on account of the relative constancy of phytoplankton absorption cross-sections (a*; m2 (mg chl a)@1) (Cleveland, 1995). Although chlorophyll is convenient to measure, the currency of choice for understanding ocean biogeochemistry is carbon. Unfortunately, the carbon : chlorophyll (C :chl) ratio of algae is highly variable because of its dependence on light, temperature, and nutrient conditions (Geider, 1987). This makes extrapolation of carbon biomass from chlorophyll difficult. Moreover, while chlorophyll a is useful for understanding the marine autotrophic community, the heterotrophic community is excluded from any chlorophyll-based estimate of total carbon biomass. Beam attenuation has shown great promise for understanding total particulate organic carbon (POC) concentration, because there is a well constrained relationship between beam attenuation and POC that holds across different water masses (Bishop, 1999; Bishop et al., 1999; Gardner et al., 1985; Stramski et al., 1999; Zaneveld, 1973). Beam attenuation (c) is the sum of absorption (a) and scattering (b), and usually b>a in seawater. Thus, strong POC vs. attenuation relationships are mainly due to the quasi-stable scattering per mole of POC (also known as ‘‘scattering cross-section’’ or b; m2 (mg C)@1). Organic carbon of phytoplankton can also be predicted from backscattering (volume scattering integrated in the backwards direction) because of the quasi-stable backscattering cross-section of organic matter (bb ; m2 (mg C)@1 (Balch et al., 1999). Knowledge of the backscattering cross-section of Antarctic phytoplankton, plus a good relationship between reflectance and backscattering, allowed Stramski et al. (1999) to estimate POC from satellite remote sensing reflectance measurements. Particulate biogenic carbon is found in two forms in the sea, POC and calcium carbonate (otherwise known as particulate inorganic carbon or PIC). The former has been best studied, beginning with the work of Sharp (1974). Nevertheless, PIC is also ubiquitous, mostly at concentrations ranging from 0.1–30 mg C l@1 (Balch et al., 2000; Balch and Kilpatrick, 1996), but occasionally it can exceed 350 mg C l@1 in coccolithophore blooms (Fern!andez et al., 1993). When particulate biogenic carbon is related to attenuation, scattering, or backscattering, the effect of PIC must be accounted for. This is particularly true when one considers the high refractive index, relative to water, of PIC (1.19) vs. organic matter (1.05–1.09) (Ackleson and Spinrad, 1988; Spinrad and Brown, 1986; Zaneveld et al., 1974). For example, the backscattering cross-section (bb ) of individually-sorted, detached coccoliths is B8 m2 mole C@1 (Balch et al., 1999), over an order of magnitude higher than the bb  of POC (B0.6 m2 mole C@1). Although there is some variability in scattering and backscattering cross-sections from different calcifying algal species, derivation of coccolithophore PIC from backscattering measurements is made possible by several important features of calcite particles: (a) other large calcifying organisms (foraminifera, pteropods) are vastly less abundant than coccolithophores, so in small volume optical measurements one is much more likely to sample coccolithophores, (b) most calcifying

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2425

algal species produce small coccoliths with diameters of a few micrometers, (c) coccolith shape is not terribly important to bb  (at least, not to a first order), and (d) two morphologically-similar species, Emiliania huxleyi and Gephyrocapsa oceanica dominate many coccolithophore assemblages in bloom (Holligan et al., 1983) and nonbloom (Reid, 1980) situations. Globally, coccolithophores are thought to contribute a majority of the calcium carbonate that reaches pelagic sediments (Bramlette, 1958; Lohmann, 1908). From a biogeochemical perspective, there is considerable interest in the distributions of biogenic CaCO3 due to the fact that CO2 is formed during calcification, leading to increases in atmospheric levels. This has been suggested over glacial to interglacial periods, where changes in the cycling of CaCO3 may have accounted for much of the 80 ppm increase in atmospheric CO2 (Archer and Maier-Reimer, 1994; Boyle, 1988; Broecker and Peng, 1989). Even at short time scales, surface calcification appears to increase atmospheric pCO2 by B25 ppm (Volk and Hoffert, 1985), or in coccolithophore blooms the shortterm pCO2 increase is up to 50 ppm (Holligan et al., 1993; Robertson et al., 1994). Little is known about the physical factors that enhance coccolithophore growth and calcification, although blooms of E. huxleyi frequently appear (but not always) in stratified regions where the water column shows a well-defined upper mixed layer, above a strong pycnocline (Balch et al., 1991; Brown and Yoder, 1994; Townsend et al., 1994). Even in nonbloom conditions of the Gulf of Maine, calcification is enhanced in low chlorophyll areas during the summer periods of long daylength (Graziano et al., 2000). Coccolithophores have been considered an important component of the northern Indian Ocean phytoplankton assemblage based on historical observations of large areas of sediments containing calcium carbonate coccoliths (Davies and Kidd, 1977). Honjo et al. (1999) described the broad divergent Arabian Sea as a ‘‘carbonate ocean’’, dominated by calcifying algae, and PIC : POC ratios of B0.9 in the top kilometer. Within this carbonate ocean, occasional bursts of diatom production occur with the passage of cyclonic eddies, lowering the carbonate production (changing to a ‘‘silica ocean’’). The sediment trap data of Honjo et al. (1999) provide a view through time of how the PIC : POC ratio varies. Questions still exist concerning the spatial patterns of PIC : POC in the surface ocean, and their biogeochemical relevance. Microscopic studies of Indian Ocean coccolithophores showed that the community of calcifying algae is significant and diverse (Kleijne, 1990; Norris, 1984; Norris, 1985); this work provided good insight into the composition and diversity of coccolithophore assemblages. Few data exist, however, for defining the role of coccolithophores in the biogeochemistry of the sinking particulate matter in the Arabian Sea. Fortunately, the quasistable bb  values for POC and PIC can provide a unique opportunity to examine their variability in space and time, with the help of continuous optical backscattering measurements (Claustre et al., 2000). One of the goals of the Bio-optics cruise of the Arabian Sea (process cruise 6) was to sample the optical properties of a diverse array of environments. Here, we show the results of a continuous survey of hydrographic and optical properties, measured along the cruise track through the Gulf of Oman and the Arabian Sea. Our focus was on biogeochemical variability at the meso- and sub-meso-scale spatial domain, and relating this to hydrographic variability. These data were ultimately used to estimate concentrations of bulk POC and PIC.

2426

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2. Methods This work was performed aboard the R/V Thompson during Process Cruise 6 (cruise no. TN053) of the JGOFS Arabian Sea Program from 29 October to 25 November 1995. The cruise track and times of stations are given in Balch et al. (2000) and in Fig. 1. In several instances, the

Fig. 1. Map of the northern Indian Ocean, showing the location and month that high reflectance features were observed at 550 nm in the CZCS browse files. High reflectance is defined here as water-leaving radiance >0.5 mW m@2. Typical water-leaving radiance values at 550 nm are B0.1–0.2 mW m@2. The cruise track of TN053 is also shown for reference. Several features are marked with symbols for clarity: }=Gulf of Oman; %=Ras al Hadd frontal region (B22.421N Lat  B60.331E Lon); ’=shelf region SW of the JGOFS southern line (18.131N Lat  58.01E Lon); and m=cyanobacterial eddy (18.271N Lat  60.321E Lon).

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2427

cruise track deviated from the standard JGOFS cruise track in order to investigate interesting physical features detected with AVHRR satellite imagery. Those features included the Ras al Hadd front (22.421N Lat  60.331E Lon), the shelf region SW of the JGOFS southern line (18.131N Lat  58.01E Lon), and an eddy containing a dense bloom of cyanobacteria (18.271N Lat  60.321E Lon (Bidigare et al., 1997)).

2.1. Sampling system Water was sampled from the ship’s non-toxic stainless steel seawater line and plumbed into a continuous sampling system. An earlier version of this system was described in Balch and Kilpatrick (1996). Basically, temperature and salinity were measured with a Tempsal (InterOcean) flow-through thermal and conductivity sensor, then in vivo fluorescence from chlorophyll a was measured with a Turner 111 fluorometer equipped with a flow-through door. Discrete chlorophyll measurements were made according to JGOFS protocols (JGOFS, 1996). Underway fluorescence measurements were calibrated with discrete chlorophyll measurements taken every two hours from the seawater system. Water flow in the continuous underway system ran through a debubbler (to eliminate bubbles and their associated light scattering). A metering pump then pushed the de-bubbled water through a Wyatt Technologies (Santa Barbara, CA) Dawn light scattering photometer equipped with a flow-through cell and 5 mW helium–neon laser (l ¼ 632:8 nm) (Balch et al., 1999). This instrument measured the volume scattering function at 15 angles from 21.71– 158.31. Average volume scattering functions were fit with a Beardsley–Zaneveld function (Beardsley and Zaneveld, 1969), which was then integrated in the backwards direction in order to calculate total backscattering. Complete details of the Wyatt light scattering photometer can be found in Balch et al. (1999; see the legend to their Table 2). Briefly, absolute calibration of the instrument was achieved by pushing 0.2 mm filtered, analytical grade, methanol through the optical cell, correcting for differences in the index of refraction between water and methanol. The flow was stopped routinely and the flow-cell cleaned to minimize bio-fouling. To correct for biofouling or instrument drift, vicarious calibrations were run throughout the cruise in which 0.2 mm filtered Milli-Q distilled water was pushed through the flow cell. Post cruise, distilled water calibration results were used to back correct all backscattering data. LabVIEW software was used to control all aspects of sampling. Complete volume scattering functions of the untreated seawater were measured at 200 Hz, and averaged at 1 s intervals for calculation of backscattering. These averages were recorded for 50 s (bb tot ). In order to estimate the backscattering of CaCO3, a peristaltic pump was activated, which delivered 0.05% glacial acetic acid to the seawater stream. Flow through a mixing coil ensured adequate mixing of the acid and seawater. A micro-pH probe downstream of the optical cuvette monitored the pH. The conventional solubility product for calcite (KspðcalcÞ ) is 4.467  10@7, at 251C in seawater with salinity of 35 (practical salinity scale; UNESCO, 1987). This value reasonably assumes the activity of the pure CaCO3 is 1, and the calcite is in its chemically pure form (Stumm and Morgan, 1981). When the pH of the seawater flow was o5.8 (below the pKspðcalcÞ of 6.35), CaCO3 dissolved and volume scattering data was measured, integrated and averaged for another 50 s (bb acid ). Next, the acid pump was stopped, and pH allowed to equilibrate to seawater values >7.8 for the next bb tot measurement.

2428

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2.2. Assumptions underlying optically-derived POC and PIC measurements There are two implicit assumptions about the above acidification treatment. The first is that the sequential measurements of raw and acidified volume scattering over several minutes on a moving ship indeed represented the same water parcels. The time for one complete raw/acid cycle of our system, plus associated volume scattering measurements, was 3–4 min. Thus, at 10 knots steaming speed, this translated to a total and acidified measurement of bb every 0.9–1.2 km. While there can be variability in the backscattering signal over such small horizontal scales, the variance is negligible relative to phytoplankton variance at larger horizontal scales. Power spectra of horizontal phytoplankton records have demonstrated almost three orders of magnitude more variance at the 10 km scale than at the 1 km scale and five orders of magnitude more variance at 100 km than at 1 km scales (Denman and Platt, 1975; Gower et al., 1980). This also is readily apparent in ocean color imagery of CZCS and SeaWiFS. Thus, we conclude that we are essentially viewing the same water throughout a 3–4 min measurement cycle, except in those instances when we crossed a frontal boundary during the measurement cycle (rare). To evaluate this over small spatial scales, our software calculated the standard deviation over the 50 s that the signals were averaged. The other assumption in the scattering measurements of acidified seawater is that lowering the pH to 5.8 did not oxidize organic particles to CO2 (and the only effect was on the inorganic CaCO3 particles). Previous controls have been performed for calcifying and noncalcifying algae that demonstrated lowering the pH to 5.0 had no effect on the concentration of coccolithophore cells (Balch et al., 1991). There also was no significant loss of POC (as CO2) upon acidification of non-plating coccolithophores to pH 5.8 (Balch et al., 1993). Mague et al. (1980) previously showed that treatment of marine phytoplankton cells (on filters) with 10% acetic acid did not result in the oxidation of cellular POC. He also showed that reduction of pH to 2.8 (with 2 N sulfuric acid plus CO2 bubbling for 15 min) did not cause significant oxidation of 14C-POC and 14 C-DOC. In total, reasonably, it can be assumed that the acidification treatment here (decrease pH to 5.8) had no significant effect on the concentration of POC or DOC, or numbers of cells. Discrete samples for POC and nitrogen were taken when the ship was stopped on station. Sample analysis was according to JGOFS protocols (JGOFS, 1996). The coefficient of variation of particulate carbon and nitrogen analyses has been previously cited as B712% (Winn et al., 1991). Discrete samples for particulate calcium were taken underway and while on station (0.3 l), and they were filtered and processed according to Fernandez et al. (1993), as described in Balch et al. (2000). Ca concentrations were determined ashore with a Perkin Elmer model Z5100 graphite furnace atomic absorption spectrometer. Analytical sensitivity of the technique was B0.03 mg C l@1, and the coefficient of variation was B73% at a PIC concentration of B1 mg C l@1. Given that there can be a strong, but variable, relationship between bb acid and POC (Balch et al., 1999), we used direct measurements to follow changes in the POC backscattering cross-section (bb acid =POC). In effect, we calibrated bb acid to POC, just as continuous fluorescence records are calibrated to chlorophyll. That is, fluorescence : chlorophyll regressions are often highly time/ space variable with poor correlation coefficients due to changing species, fluorescence quenching, physiological state, etc. Calibrations must therefore be adjusted to account for changing

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2429

fluorescence efficiencies. Ratios of bb acid : POC ratios behave similarly; POC backscattering crosssections vary with particle size and type, phase of cell cycle, presence of detrital carbon, etc. Our calibration technique began with any two successive stations where bb acid and discrete POC were measured. The ratio bb acid : POC was calculated exactly at these stations, after which the bb acid : POC ratio was linearly interpolated between them. The intervening POC concentrations were then derived as the measured bb acid divided by the interpolated bb acid : POC ratio. Upon arrival at the next station, bb acid : POC values were exactly calculated once again, and interpolated to the next station. This type of underway calibration maximized the correlation of measured POC vs. bb acid (r2 of 1.0), and the calibration took into effect changes in the backscattering cross-sections, composition and refractive index of the organic particles. We used the same approach to derive continuous underway PIC as we did for POC; in this case, however, we converted acid-labile backscattering (b0b ) to suspended PIC concentrations by using the frequent discrete PIC samples for calibration. Microscopy of fresh phytoplankton samples was performed aboard ship with an Olympus BH2 light microscope equipped with phase, epi-fluorescence, and polarization optics. While we performed some 800 counts of settled preserved material (Balch et al., 2000), the purpose of these microscope examinations of non-preserved material was to gain a qualitative sense of the relative numbers of fluorescent diatoms, dinoflagellates, coccolithophores (plated cells, empty coccospheres and detached coccoliths), Synecococcus, and Prochlorococcus. Two fresh samples were viewed each day on station, usually one from the surface, and the other from either the deep chlorophyll maximum or base of the mixed layer. Our epifluorescence microscope was equipped with 490 nm excitation (515 nm barrier filter, and a 590 nm dichroic filter) for chlorophyll a fluorescence, or 545 nm excitation (590 nm barrier filter and 580 nm dichroic filter) for biliprotein fluorescence. Samples were prepared by filtering 50–100 ml samples onto 0.2 mm poresize filters. The filtered material was then transferred to glass microscope slides with the ‘‘filter-transferfreeze’’ technique (Hewes and Holm-Hansen, 1983). Counts were first done with background illumination at three magnifications (100  , 200  , and 400  ; all phase contrast). This ensured adequate counts of large, rare cells and small abundant cells. Next, the sample was examined with 400  cross-polarized light (lower magnification was usually not necessary with polarized light, since most of the suspended calcium carbonate particles were o20 mm diameter). Combined epifluorescent and polarized light scattering measurements allowed best viewing of plated coccolithophores. Lastly, the magnification was increased to 1000  and the sample sequentially illuminated with blue and green excitation to allow enumeration of picoeukaryotes and cyanobacteria, respectively.

3. Results The cruise track is shown in Fig. 1. To aid in interpreting our optical data, the Coastal Zone Color Scanner data base (1977–1986) was searched for high reflectance patches with normalized water-leaving radiance at 550 nm>0.5 mW m@2 (note: typical values are 0.1–0.2 mW m@2). Such areas of high reflectance could have been due to coccolithophores, blue–green algae such as Trichodesmium (Subramaniam et al., 1999), or high concentrations of other suspended particulate matter. The location and ‘‘month of occurrence’’ of high 550 nm reflectance observations are

2430

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

given in Fig. 1. It is clear that high reflectance features were most common around the perimeter of the northern Indian Ocean, especially during the Fall months along the SE coast of Oman. The underway data are shown either as a function of time (Fig. 2) or space (Fig. 3). Because of the difficulty in clearly plotting all of the data in three-dimensional plots, the data of Fig. 3 show only the north leg, transit to the beginning of the southern leg, and the southern leg. Inclusion of the remaining part of the cruise (from the end of the southern leg, back to Muscat) made these plots difficult to interpret. Hydrographic data (Fig. 2A–C and Fig. 3A–C) showed the influence of warm (>291C), high salinity (>37) water emanating from the Gulf of Oman. This water was not observed at the end of the northern leg in the Arabian Sea, where salinities were B36.4. Surface conditions showed no large-scale changes from the end of the northern leg to the 101N station, only some small spikes that would be discussed later. Coolest, freshest water was observed in shelf waters at the shoreward end of the southern leg. Upon the re-crossing of the Ras al Hadd front at the end of the cruise, the salinity increased, almost to previously observed levels in the Gulf of Oman, but temperatures had cooled B1.51C since measurements made during the previous month (Fig. 2B). The highest density water was found at the Ras al Hadd front, at the end of the cruise. Gulf of Oman and Ras al Hadd water (beginning of the cruise) had relatively high density, too. Lowest density water was observed in well-defined patches between 101N and the end of the southern leg (Figs. 2C and 3C). Highest chlorophyll a values were observed in the Gulf of Oman, north of the Ras al Hadd front (Figs. 2D and 3D). The one exception was in a cyclonic open-ocean eddy dominated by Synechococcus sp. (Bidigare et al., 1997), where chlorophylls reached B0.7 mg l@1 (Fig. 2D). A vertical profile inside this eddy revealed total chlorophyll a as high as 1.8 mg l@1 at 11 m, with the bulk of the pigment confined to the top 24 m (Bidigare et al., 1997). The dynamic range of chlorophyll over the entire cruise was B4  . Patchiness in the chlorophyll record was very apparent, with sharp discontinuities of 0.1–0.2 mg chl a l@1 (Figs. 2D and 3D).

3.1. Phytoplankton species observations One striking example of the significant patchiness was on the transit between the end of the northern line and beginning of the southern line. Virtually, all of the water samples taken during this part of the cruise showed minimal algal biomass in surface and deep waters, dominated by cyanobacteria. The exceptions were in two patches of high algal fluorescence that were B6 miles wide, associated with sharp discontinuities of temperature and salinity, at B16.251N and 66.51W (see arrows in Figs. 2D and 3D). These patches contained many diatom species, including Rhizosolenia sp., Chaetoceros sp., multiple pennate species, armoured dinoflagellates, and relatively high concentrations of cyanobacteria. The most striking aspect of the surface algal assemblage of this patch was the numerous aggregates of pennate diatoms arranged radially around clumps of debris, often containing plated coccolithophore cells (Fig. 4). Other ancillary measurements were taken in this patch; the chlorophyll concentration was 1.12 mg l@1. Nitrate, nitrite, ammonium, phosphate, and silicate in this patch were 0.24, 0.06, 0.1, 0.42, and 0.6 mM, respectively. Total detached coccolith concentrations were only B100 ml@1 and plated coccolithophore concentrations were B2 ml@1. There was the possibility that the clumps were an artifact of the filter-freeze-transfer concentration technique (Hewes and Holm-Hansen, 1983),

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2431

Fig. 2. Underway surface measurements shown as a function of calendar day (1995). (A) Temperature; (B) salinity; (C) density; (D) chlorophyll a; (E) total backscattering at a wavelength of 632.8 nm; (F) acid-labile backscattering at 632.8 nm; (G) fraction of total backscattering at 632.8 nm which is acid-labile. Geographic points of note are shown below time axis. Arrow in panel D designates high fluorescence, high chlorophyll patch on calendar day 311, mentioned in text.

2432

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

Fig. 3. Underway surface measurements shown as a function of geographic position: (A) temperature; (B) salinity; (C) density; (D) chlorophyll a; (E) total backscattering at a wavelength of 632.8 nm; (F) acid-labile backscattering at 632.8 nm; (G) fraction of total backscattering at 632.8 nm which is acid-labile. Arrow in panel D designates high fluorescence, high chlorophyll patch on calendar day 311, mentioned in text.

but this was probably not the case since samples from outside the patch did not show these aggregates. Surface water at the station furthest offshore along the southern line was still dominated by cyanobacteria and picoeukaryotes, but the fluorescence maximum (75 m) had healthy populations of diatoms (of the genera Hemiaulus, Rhizosolenia, Chaetoceros, Corethron, Thalassionema, Striatella, Navicula, and Eucampia), the dinoflagellate Prorocentrum sp., and one naked dinoflagellate species. Midway along the southern line, the surface waters were dominated by cyanobacteria with a few small pennate diatom species; waters from the fluorescence maximum showed a few species of rare large diatoms, and mostly picoeukaryotes and cyanobacteria. Still closer to shore along the southern line (17.181N  59.8151E), the surface water contained numerous coccolithophores and detached coccliths, a few small pennate diatoms, one species of

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2433

Fig. 4. Light micrographs of three clumps of radially arranged pennate diatoms, some with coccoliths at the center. Cells were observed in an 11 km-wide, high-fluorescence feature located at 16.2491N  66.4981E on 7 November 1995, at 1600 h GMT (calendar day 311). Each group of panels (A–C, D–F, and G–I) is from a different cell aggregate. Images in panels A, D, and G were taken at 100  magnification under phase-contrast illumination. The scale bar is 100 mm. Panels B, E, and H were taken at 400  magnification under phase-contrast illumination. Panels C, F, and I were taken at 400  magnification under cross-polarized illumination (the white spots are birefringent CaCO3 coccoliths). The scale bars in panels B, C, E, F, H, and I are 25 mm. It can be seen that the clumps of cells in panels A–C and G–I had a significant numbers of coccoliths and coccolithophores at the center of the radial clusters of pennate diatoms, but panels D–F showed a clump of pennate diatoms with only one or two coccoliths at the center.

naked dinoflagellate, cryptomonads, and a combination of eukaryotes and cyanobacteria. Water from 44 m at this same station was quite similar to the surface water. The two stations closest to the coast (18.11N  581E and 18.51N  57.31E) were dominated by cyanobacteria and coccolithophores at the surface, and a combination of pico-eukaryotes (including Prochlorococcus) and cyanobacteria at depth. There were few large diatom species in these cool, low salinity waters.

2434

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

Waters from the Ras al Hadd front contained a wide variety of large and small phytoplankton: armored dinoflagellates (e.g. Ceratium) along with large diatoms (Rhizosolenia, Navicula, Nitzschia), naked dinoflagellates, some coccolithophores (plus their detached coccoliths), cryptomonads and considerably higher concentrations of cyanobacteria (with the cyanobacteria outnumbering picoeucaryotes by as much as 14  ). 3.2. Backscattering Total backscattering (bb tot ) showed a dynamic range of B10  over the entire cruise (Figs. 2E and 3E) with high values in the Gulf of Oman, and occasional patches with high values (>0.01 m@1) in the northern leg. Note the log scale in Fig. 3E. Such high values were not observed again until we crossed into the cyclonic eddy containing large concentrations of cyanobacteria (Fig. 2D). Upon re-crossing the Ras al Hadd front, bb tot increased to values seen at the beginning of the cruise. Values of b0b , acid-labile backscattering, showed a dynamic range of B40  , with most variability again near Ras al Hadd and the northern leg of the cruise track. Patches with b0b > 0:001 m@1 were observed several times during the latter half of the cruise (Figs. 2F and 3F). The fraction of total scattering that was ‘‘acid-labile’’ was determined as b0b =bb tot This had a dynamic range of B9  , with highest b0b =bb tot values (B45%) observed in the Gulf of Oman, Ras al Hadd frontal region, and northern leg of the cruise, and lowest values (5%) en route to 101N (Figs. 2G and 3G). Values of 10–20% were most common throughout the entire cruise. The cyclonic cyanobacterial eddy had low values of b0b =bb tot approaching 5%. A plot of bb acid vs. bb tot demonstrated the same point, but more quantitatively (Fig. 5A); one can see that the data distribution falls below the 1 : 1 line, indicative of a decrease in backscattering following acidification. Statistics of the scattering data were collected over the 50 s that the pH was below 5.8 during each acidification cycle. Because of a software problem, statistics on the total backscattering samples were lost. Nevertheless, for the acid-resistant particles, the standard deviation of each mean increased in a curvilinear fashion when plotted on a log–log plot. Put another way, the coefficient of variation for the largest mean bb acid values was B2, and for lower mean bb acid values, the coefficient of variation was B0.1 (Fig. 5B). Total backscattering (which included POC and PIC) increased with increasing chlorophyll, but at any one chlorophyll concentration, overall variability of bb tot was about 5  (Fig. 6A) and correlation coefficients were low. The mean trend of the backscattering data in this plot (n ¼ 3484) is shown along with Eq. (12) of Gordon et al. (1988), which uses results from his Table 1, assumes a l@1 dependence of particulate backscattering, and adds this to bb values for pure seawater (Gordon et al., 1979). The backscattering data for acidified seawater (bb acid ; where CaCO3 had been dissolved) is shown in Fig. 6B along with the Gordon et al. (1988) line. Slight displacement of the data below the Gordon et al. (1988) line can be clearly seen in Fig. 6B. Data for suspended calcite (as the mass of calcite carbon per unit volume) were plotted against the values of bb 0 , on linear (Fig. 7A) and log scales (Fig. 7B). To aid in interpretation, data from the Arabian Sea were plotted along with data from other field experiments. Data are included from a bloom of the coccolithophore, Emiliania huxleyi, in the North Atlantic during 1992 (Balch et al., 1996b; Balch et al., 1996a; Holligan et al., 1993), and the Straits of Florida over carbonate

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2435

Fig. 5. (A) Regression of acidified backscattering vs. total backscattering. A 1 : 1 line is shown for reference. Least2 squares fit to data; bb acid ¼ 0:2783  b0:801 b tot (r =0.675; n=3333; F=6929; SE of exponent=0.00962; SE of the constant =0.015; Fig. 5). (B) Standard deviation of bb acid vs. mean bb acid value. For a 50 s measurement, SE 2 bb tot ¼ ð0:2974Þ  b1.0392 b acid (r ¼ 0:24; n ¼ 3311; F ¼ 1029; SE of exponent=0.032; SE of constant=0.0532).

banks (cruises in 1994 and 1995). By plotting these data on linear axes, one can see the effect of the North Atlantic bloom results (from a mesoscale coccolithophore bloom) and flow cytometer data that drive the regression correlations. Plotting with log axes allows one to better discern the data

2436

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

Fig. 6. (A) Plot of total backscattering vs. chlorophyll a concentration. Best fit statistical line shown as dashed line (Y ¼5.612  10@3 X 0.599; r2 =0.15). Eq. (12) of Gordon et al. (1988) relating backscattering to chlorophyll a is shown as a solid line. (B) Plot identical to panel A except acidified backscattering is plotted vs. chlorophyll a. Best statistical fit is shown with the dashed line (Y=5.426  10@3) X 0.651; r2 =0.19).

points at low calcite concentration. The Arabian Sea data were wholly restricted to low PIC concentrations. 3.3. Errors in optically-derived POC and PIC measurements The error in estimating PIC from b0b (Fig. 7) was estimated by standard methods of error propagation (Meyer, 1975). Acidified backscattering was modeled from total backscattering results and the fitted relation given in Fig. 5. The standard error of the bb acid values was directly measured during this cruise (and was modeled as a function of bb acid by the relation given in Fig. 5b). To model the standard error of bb tot , we used values from 20 cruises in the Gulf of Maine (since the Arabian Sea data on standard error of bb tot were inadvertently lost). The relationship 2 for a 50 s bb tot measurement was: Std Err bb tot ¼ ð0:2321Þ* b1:369 b acid (r ¼ 0:413; n ¼ 2843; F=1999;

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2437

Fig. 7. (A) Suspended calcite concentration plotted against acid-labile backscattering (b0b ). Results shown on linear axes. Data from several regions are shown: nFArabian Sea; mFStraits of Florida, 1994; ~FStraits of Florida, 1995; *F1992 coccolithophore bloom south of Iceland. (B) Same data plotted with logarithmic axes. Line represents least squares linear fit (bb ¼ 1:37 m2 mole CaCO@1 ; n ¼ 224; SE of bb ¼ 0:0346; r2 ¼ 0:847; F ¼ 1266). Broken lines 3 represents 71SE about the mean, as calculated using standard error propagation (Meyer, 1975).

SE of exponent =70.0306; SE of constant=0.0457). Note that bb tot , bb acid and SE bb acid values from these other cruises were in the same range as those found in the Arabian Sea. The standard error of the mean backscattering cross-section for CaCO3 was derived from the slope of the data in Fig. 7A. In a statistical sense, most of the error in estimating PIC was in errors associated with the b0b measurement since it was based on the difference between two relatively large numbers, bb tot and bb acid , each of which had relatively large variance. After propagation of all the errors, the total standard error in PIC increased non-linearly as CaCO3 concentration increased (based on modeling 50 s average bb measurements, see Fig. 7). 3.4. Hydrographic observations and their relation to bio-optical variables A plot of surface temperature vs. salinity showed three distinct water types (Fig. 8). Water having temperatures of 27.6–29.51C and salinities of 36.5–37 characterized Gulf of Oman water at the beginning of the cruise (end of October). The 26–281C water with salinity of 35.5–36.5 was from approaches to the Omani southern coast (JGOFS ‘‘southern line’’ and nearby stations). The density of these water masses roughly fell along the 23.5 sy isopycnal, suggesting mixing along a continuum, between two end members, the Gulf of Oman and and Oman SE coastal water. The small pool of data centered at 26.31C and salinity of 36.6 was from the transit north of the Ras al Hadd front into the Gulf of Oman at the end of November (where surface waters had cooled since our first transit through this area).

2438

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

Fig. 8. (A) Temperature vs. salinity plot from underway surface results. (B) Contour diagram of chlorophyll a concentration plotted in temperature-salinity space. (C) Contour diagram of b0b =bb tot plotted in temperature-salinity space. In all panels, isopleths of sy are shown with dashed lines.

Chlorophyll concentrations plotted in temperature-salinity space showed that the colder water had lower chlorophylls, while highest values were observed above 281C. The percentage of total backscattering that was acid-labile (b0b =bb tot ) was also lowest in the cooler, low chlorophyll, southern waters (10–20%), and values increased up to 30% approaching the Ras al Hadd front, and Gulf of Oman, in water having sy values of 23.5. Careful inspection of Figs. 8B and C shows that regions of high chlorophyll usually were not regions of high b0b =bb tot .

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2439

4. Discussion It is striking how the majority of the high reflectance features in the Arabian Sea during the nine years of CZCS occurred in coastal waters (Fig. 1), with 28 of the 41 features (or 68%) north of the equator, during the last four months of the year. The transition from SW monsoon to fall intermonsoon, and NE Monsoon period, is associated with a relative strengthening of stratification, which most likely helps the persistence of high reflectance features. Highest chlorophyll concentrations were seen in north of Ras al Hadd during our work. The CZCS results were in agreement with the backscattering data (Figs. 2E–G and Figs. 3E–G), which showed some of the highest backscattering values (>0.006 m@1) between Muscat and the end of the north leg, then again in the return leg along the coast of Oman. High beam attenuation, which is typically dominated by scattering, was also observed throughout the mixed layer at Ras al Hadd during TN nos. 050 and 054 (Gundersen et al., 1998). Many of the peaks in total backscattering were associated with major peaks in acid-labile backscattering (b0b ), such that >30% of the total backscattering was acid-labile, presumably CaCO3. During the remainder of the trip, b0b =bb tot was typically B10–20%, suggesting lower relative standing stock of CaCO3. The one high-reflectance, high-scattering, feature that did not have a high b0b was the cyanobacterial eddy observed on day 327 (Bidigare et al., 1997). It should be noted that coccolithophore growth rates in this eddy (estimated from coccolith detachment rates) were some of the highest measured during the entire cruise (Fritz, 1997). High-reflectance patches dominated by cyanobacteria have been observed before in other oceans (Dupouy et al., 1988; Morel, 1997). 4.1. Patchiness in light scattering Our bio-optical data showed strong influence of eddies in the Arabian Sea, as noted by many others involved in this JGOFS study (Bohm et al., 1999; Dickey et al., 1998; Flagg and Kim, 1998; Gundersen et al., 1998; Latasa and Bidigare, 1998; Manghnani et al., 1998). The along-track chlorophyll data showed evidence of patches of several hundred kilometers (greater than the Rossby Radius in this region) along N and S lines, in agreement with Campbell et al. (1998), who cited length scales of 300–400 km for Synechococcus patches. There were relatively small changes in total backscattering and chlorophyll concentration between the end of the northern line to waters off the SE Omani coast, yet there were significant changes in SST (several degrees), salinity (B1) and density (B1/2 sigma theta unit). This does not mean that algal populations were not responding to hydrography. Instead, there probably were multiple sources of backscattering (phytoplankton, bacteria, detritus, etc), and multiple chlorophyll-containing species, that continually changed with hydrography such that the net backscattering remained relatively constant. Our species data support this interpretation since there were clear changes in species assemblages. In contrast, results from the northern line demonstrated much more variability in bb tot , b0b and b0b =bb tot , with patch sizes on the order of several hundred kilometers. It is tempting to hypothesize that this was due to one primary source, patches of calcifying algae. As long as calcium carbonate particles represented a large fraction of the total backscattering (e.g. along the N line), then b0b and bb tot , showed strong patchiness. From a bio-optical perspective, two regions, the Ras al Hadd front and northern leg, were the most dynamic (Fig. 3). Clearly, variability at Ras al Hadd was related to the strong barotropic

2440

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

front/jet system (Bohm et al., 1999). Lack of coherence between the chlorophyll and backscattering data implied that other factors besides chlorophyll were driving the optical backscattering. Data of Fig. 6 showed that chlorophyll accounted for only 15% of the variance in total backscattering, and 19% of the variance in acidified backscattering (see legend for least squares fits). This is in good agreement with the observations of Kitchen and Zaneveld (1990), who showed a lack of correlation between light scattering and chlorophyll and best correlations between beam attenuation and particle volume. 4.2. Deconvolving factors that affect particle scattering As noted in the introduction, there is compelling evidence for a good correlation between attenuation, scattering, or backscattering vs. POC (Balch et al., 1999; Kitchen and Zaneveld, 1990; Morel, 1988; Stramski and Reynolds, 1993; Stramski et al., 1999). Such results suggest that several factorsFchanging C : chl ratios, presence of other non-chlorophyll containing organic matter or minerogenic particlesFcan degrade the relationship between chlorophyll and backscattering. Our results certainly agree with this. The Gordon et al. (1988) relationship for predicting backscattering from chlorophyll was probably originally derived with field measurements of bb tot (which would have included both organic and minerogenic scatterers). This was apparent when we applied the Gordon et al. relationship to our data, since the function better intersected the bb tot data than the bb acid data (Fig. 6). Note, however, that when bb acid was plotted against chlorophyll a, it showed a slightly higher overall correlation than did bb tot vs. chlorophyll a. Presumably, this was due to the elimination of highly scattering calcium carbonate particles, which would have partially masked the chlorophyll-specific scattering. It is also worthy of note that there was no discernible relationship between the backscattering : chlorophyll ratio and time of day. Thus, diurnal variation in chlorophyll can be eliminated as a factor causing the high variance in the backscattering : chlorophyll ratio, as shown in Fig. 6. These data lead us to conclude that CaCO3 particles were often an important, but variable, component of the total backscattering. The implication of high values of b0b =bb tot over relatively large areas of the Arabian Sea is that a substantial portion of the remote sensing reflectance originates from particulate inorganic carbon (PIC), not POC. Equally interesting were the statistics associated with the backscattering data. The scattergram of bb acid variance vs. its mean (Fig. 5B) suggested two general trends. (1) In more oligotrophic waters of high clarity and low POC content, increases in scattering were associated with occasional, rare, highly scattering particles, along with a monotonous background of lowscattering particles. In these regions, coefficients of variation were generally o1, and an increase in the mean bb acid was associated with a steep increase in its variance. Such areas contained many small picoplankton, with occasional large species like diatoms, which have been shown to account for a large fraction of the new production in oligotrophic regions (Goldman, 1993; Goldman et al., 1992). It has been demonstrated that in natural populations of phytoplankton, cell volume (hence cell POC) is log normally distributed, which means that rare, large particles are mixed with more abundant smaller ones (Campbell, 1995; Campbell et al., 1989; Yentsch and Campbell, 1991). (2) Eutrophic waters of lower clarity and greater productivity had greater mean bb acid (i.e. higher POC), with higher variance and coefficients of variation up to B3.5. Increases in backscattering in these more productive regions were associated with both high- and low-scattering particles, such

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2441

that the overall variance did not increase substantially. We have seen this pattern in eutrophic waters of the Gulf of Maine (results to be published elsewhere). 4.3. PIC and acid-labile backscattering The relationship between b0b and suspended calcite in several different regions showed a backscattering cross-section (bb ) of 1.37 (SE=70.0.03) m2 (mole C)@1, which was strikingly divergent from results from several flow cytometry experiments in which individual coccoliths were sorted free of organic matter (Balch et al., 1999). In those experiments, suspensions of pure calcite coccoliths sorted from lab cultures and field samples (and separated from organic matter) showed significantly higher bb  values (6.4 (70.46) m2 (mole C)@1). One explanation is that light scattered through pure CaCO3 particles was not subsequently attenuated by organic matter, hence greater backscattering resulted (Balch et al., 1999). Another explanation was that when coccoliths and organic matter were mixed together (as coccoliths surrounding a cell), then the highly overlapping coccoliths ‘‘self shaded’’ each other. The analogy is to a calcium carbonate ‘‘package effect’’ (Morel and Bricaud, 1981) in which scattering per mole of carbon was reduced when only the exterior coccoliths were responsible for the bulk of the light scattering. This same effect can be described by anomalous diffraction theory (Van de Hulst, 1957), as discussed in Balch et al. (1996b), and also in Bricaud et al. (1992). Nevertheless, there was no obvious relationship between the measured bb  and the coccolith/plated coccolithophore ratio of the seawater samples. Thus, the mechanism responsible for the high bb  of sorted CaCO3 coccoliths is more complicated than hypothesized here. Error propagation calculations shown in Fig. 7 demonstrated that the backscattering cross-section (bb ) of CaCO3 particles was sufficiently variable that the PIC estimates based on b0b were good to B74.8 mg C l@1 in nonbloom situations, and B719 mg C l@1 in dense coccolithophore blooms (but with a significantly lower coefficient of variation in the latter). 4.4. Hydrography, chlorophyll, optics, and species assemblages Our chlorophyll and scattering data provide interesting contrast when plotted in temperaturesalinity space. Most of the temperature/salinity data were centered roughly on the 23.5 sy isopleth (except for the Gulf of Oman water, which, upon cooling by the end of November, became significantly more dense). It is reasonable to hypothesize mixing along a gradient between two endpoints: warm, salty Gulf of Oman water and cool, low salinity, water off the SE coast of Oman. Highest chlorophyll values (>0.6 mg l@1) were just NE of the Ras al Hadd front. Careful inspection of Figs. 8B and C showed that the two regions of highest chlorophyll were associated with low acid-labile backscattering (b0b =bb tot ). Two possible reasons for this were: (1) the high chlorophyll was accompanied by high bb tot , which would have lowered the b0b =bb tot , ratio, and (2) coccolithophores were predominant in low chlorophyll environments, while other groups (e.g. diatoms) would have been more successful in high chlorophyll regimes. We believe that both of the above factors were responsible for the data distribution shown in Fig. 8. Cell counts of living material, done aboard ship, showed that nearshore water from the southern line was dominated by small cells (with cyanobacteria typically out-numbered 4–9  by picoeukaryotes). In contrast, waters from the Ras al Hadd front contained a wide variety of large

2442

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

and small phytoplankton, with cyanobacteria dramatically outnumbering picoeucaryotes. Differences in population structure that we observed are interesting in light of the observations of McCarthy et al. (1999). They showed a dominance of small cells nearshore on the southern line during TN053 which was associated with extremely low f ratios (Eppley and Peterson, 1979) of o5%. Moreover, integrated euphotic zone nitrate concentrations were actually lower nearshore than offshore along the southern line (see Fig. 4 of McCarthy et al., 1999). This, combined with the higher ammonium concentrations and uptake rates nearshore, suggested that grazing was significant along the coast during the early NE Monsoon period, and this fueled a significant fraction of ‘‘regenerated’’ primary production. Thus, dominance of small cells may have been due to intense, preferential grazing of the large phytoplankton cells. Caron and Dennett (1999) observed (albeit during the spring intermonsoon, not the fall) that coastal stations showed grazing rates that were a significant fraction of the total phytoplankton growth rates, and mortality rates 2–4  higher nearshore than offshore. Thus, grazing cannot be eliminated as a factor affecting algal community structure, nitrogen dynamics, and optics (see Yentsch and Phinney (1989) for more discussion of the links between algal nutrition and optics). The small patches of high chlorophyll concentration are difficult to explain. There is no doubt from the microscope examination of these waters that they contained unique assemblages of phytoplankton (Fig. 4). Although the small patches were strongly associated with ship-measured spikes in temperature and salinity (Fig. 3), AVHRR images within one day of the shipboard observations showed no indication of temperature anomalies or frontal boundaries, which might have given rise to the high-fluorescence feature. Thus, such small, isolated patches were likely old enough that any SST signal would have been obliterated by surface heating effects, while shipmeasured temperature at 1 m depth still showed a significant change. 4.5. Derivation of underway POC and POC : Chl To estimate the error in the POC estimate, it was regressed against bb acid , and least-squares fits calculated (Fig. 9). For comparison, we show our discrete data along with the Antarctic circumpolar data of Stramski et al. (1999), one of the few studies where data of this type exist. It can be seen that, with the exception of three anomalous data points, which can be fit by a different regression, the trend of our Arabian Sea data is most similar to the Antarctic Polar Front Zone (APFZ) results, not those of the Ross Sea (Stramski et al., 1999). Based on the data of Fig. 9, the standard error of the POC estimates (based on 50 s measurements of bb acid ) was 730 mg POC l@1. With an average POC of 105 mg POC l@1, then the average coefficient of variation was 28.6% for optically-derived POC; these estimates (Fig. 10A) were divided by fluorometrically-derived chlorophyll values (calibrated to discrete chlorophyll) to estimate the bulk POC : chl ratio (POC : Chl; Fig. 10B). Defining the error of the underway chlorophyll and POC : Chl terms are non-trivial. Fluorescence data were calibrated exactly at each station with discrete chlorophyll measurements, which accounted for daily changes in the fluorescence efficiency of chlorophyll. Between discrete chlorophyll measurements, the fluorescence per unit chlorophyll was interpolated. Consequently, a plot of fluorescence vs. chlorophyll concentration derived from discrete measurements had a perfect correlation. Unfortunately, we did not have an independent data set to define the total error of fluorescence-derived chlorophyll. All that can be concluded is that discrete chlorophyll

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2443

Fig. 9. Discrete measurements of particulate organic carbon vs. particulate backscattering at 632.8 nm (black circles; 2 least-squares fit: POC=10641*b0:794 bp ; r ¼ 0:618; n ¼ 16). Three samples were reasonably fit by another relationship 2 (open diamonds; POC=11591*b1:097 ; r =0.927). The data are shown along with the average relationships of Stramski bp et al. (1999), estimated from the Ross Sea and Antarctic Polar Front Zone (APFZ). The relationships for each region are shown as two sets of three parallel dotted lines (with the central line representing the least-squares average fit, and upper and lower parallel lines show 71SE; see their Fig. 1 for original data and relationships). It can be seen that Arabian Sea average fit is most similar to their APFZ results, and the variance that we observed about the average lines is similar to that described by Stramski et al. (1999).

values were good to at least 710% (Parsons et al., 1984) and probably closer to 75% (Winn et al., 1991). The total error in chlorophyll derived from in vivo fluorescence was likely to be larger. As for the POC : Chl calculation, propagating the error of both optically-derived POC and chlorophyll estimates gives a minimum coefficient of variation of B29%, but we caution that it is likely to have been larger for reasons outlined above. This is why the optically-derived POC and C : Chl data were plotted along with the discrete values for comparison. Most of the C : Chl values fell between 15 and 300, much the same range cited by others (see Geider, 1987). There were some higher values that we cannot explain. Geider (1987) cited an inverse relationship between C : Chl ratio of algal cultures versus temperature and a strong positive relationship between C : Chl and photon flux. We applied his relationship (his Eq. (7)) to our Arabian Sea observations of daily averaged photosynthetically available radiation and surface temperature. All predicted C : Chl values fell between 40 and 80, far below the majority of data in Fig. 10B. The rather constant and low C : Chl predictions of the Geider equation are not surprising, especially given the high SST (26–301C) and daily PAR values (40–47 Ein m@2 d@1) during the

2444

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2445

early NE monsoon period. The observed high bulk POC : Chl values measured aboard ship were likely due to the presence of nonphytoplanktonic POC, such as from bacteria, protozoa, and detritus. An ecosystem dominated by non-algal POC is probably representative of a post-bloom period, in which production is declining as nutrients become limiting. We hypothesize that (1) during the time of active growth (SW Monsoon), there was a tighter relationship between POC and chlorophyll, with low POC : Chl ratios, and (2) with the onset of the intermonsoon, stratification increased, causing nutrient limitation and a proportionate increase in heterotrophic and detrital POC. Total POC : Chl ratios presented by Gundersen et al. (1998) along the southern line showed high surface values during the spring intermonsoon (>240 g C g Chl a@1). Unfortunately, no fall intermonsoon data were available. Their surface POC : Chl ratios during the southwest monsoon (TN049) were o50 g C g Chl a@1, in support of our hypothesis, and there was clear development of a subsurface maximum of particles with high chlorophyll content (Gundersen et al., 1998). Garrison et al. (2000) also showed highest autotrophic cell POC : Chl ratios during the spring intermonsoon. Eppley et al. (1977) suggested that low concentrations of refractory POC (such as seen during the SW Monsoon) indicated rapid turnover via consumption, decomposition, or horizontal/vertical export. Our data, combined with the Geider (1987) and Eppley et al. (1977) predictions suggested that during the fall intermonsoon/early NE Monsoon, there was a slower turnover of POC, a net build-up of heterotrophic/detrital POC, and an increase in the POC : Chl ratio. This can also be seen in Fig. 2 of Garrison et al. (2000), which shows the dominance of large diatoms during the NE and especially SW monoons. These large-celled autotrophic assemblages gave way to communities dominated by nano- and pico-autotrophs during the spring intermonsoon and early NE monsoon, with relatively larger contributions of heterotrophic picoplankton. 4.6. Derivation of underway PIC and PIC : POC PIC estimates based on b0b demonstrated a similar standard error, based on several field campaigns. Discrete PIC measurements, however, had a much lower standard error (70.03 mg PIC l@1; Fig. 7) and thus, were used to calibrate the underway optical estimates (such that a plot of atomic absorption vs. optically-derived PIC estimates had a perfect correlation). PIC data (measured, or derived from b0b ) showed large areas of lower PIC concentration with small patches containing significant concentrations of PIC (Fig. 11A). However, note that optical derivation of PIC from b0b had the largest coefficient of variation at low PIC concentrations, and these data should be interpreted cautiously. Results from the intense patches, however, were significant and probably represented small blooms. ’––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Fig. 10. (A) Time course of POC concentration derived from continuous bb acid measurements (K) or measured analytically using standard CHN combustion techniques (2). Error bars would make this plot too busy, but the error of the optical estimates is 730 mg POC l@1 (CVB28.6%) See text for details. (B) POC : Chl data plotted against time. Underway data (K) are derived from continuous POC estimates in panel (A) and fluorescence-derived chlorophyll estimates (Fig. 2C or 3D). Square symbols (2) represent POC : Chl ratios measured in discrete samples. Note, since the underway data are calibrated to discrete measurements, each calibration datum (2) in panel A exactly covers an underway datum (K).

2446

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

Fig. 11. (A) Time course of PIC concentration, as derived from b0b data (K) or measured analytically using atomic absorption (2). Error bars make this plot too busy, so the right Y-axis shows standard error of 50 s optical PIC estimates, based on error propagation estimates (Meyer, 1975). Standard deviation of discrete PIC measurements, B70.03 mg C l@1, is smaller than the symbols on the plot; the discrete samples were used to calibrate the opticallyderived PIC values. Note, since the underway data are calibrated to discrete measurements, each calibration datum (2) in panel A exactly covers an underway datum (K). Horizontal line shows approximate boundary above which the predicted coefficient of variation is o1. (B) PIC : POC time course estimated using results of panel A and Fig. 10A showing optically-derived values (K) or discrete measurements (2) for PIC : POC. Coefficient of variation for PIC : POC, derived from discrete analytical measurements, is 70.12. See text for discussion of standard error of optically-based PIC : POC data points.

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2447

The results from Fig. 10A were combined with Fig. 11A to produce a continuous record of PIC : POC (Fig. 11B). The standard error of this optically-based ratio could not be easily shown on this figure without making it uninterpretable because the error was so variable. For example, at PIC and POC values of 1 and 100, respectively (PIC : POC=0.01), the optically-based PIC : POC ratio would have had a coefficient of variation of 75. At PIC and POC values of 10 and 1000 (same PIC : POC), the optically-based ratio would have had a coefficient of variation of 70.6. In contrast, the analytical measurements of the PIC : POC ratio had a coefficient of variation of 70.12. The most notable feature about Fig. 11B was the dynamic range of the PIC : POC results. The PIC : POC ratios measured in discrete water samples varied over 100  ; the continuous, optically-based ratios showed a dynamic range of B10,000  . Undoubtedly, some of this can be attributed to the larger error of the optically-based PIC measurements, especially at low PIC concentrations. High dynamic range of PIC : POC is of interest when one considers the rain ratios of B1 (with a few instances of 0.5) measured in the 800 m JGOFS sediment traps, regardless of season (Honjo et al., 1999; their Fig. 9). This shows that the variance in PIC : POC is almost completely stabilized in the top kilometer of ocean, probably by remineralization of both carbon fractions. Evidence cited elsewhere (Balch et al., 2000) suggested that B75% of the surface-produced CaCO3 dissolved in the top kilometer, best explained by CaCO3 dissolution associated with biological processes (Milliman et al., 1999). Even so, POC fractions still showed higher remineralization (98%) than PIC in the top kilometer. The semi-continuous PIC : POC data illustrate in space what the JGOFS sediment trap data illustrated in time. Results from the 800 m sediment traps in the western Arabian Sea (Honjo et al., 1999) typically showed the influence of several ‘‘pulsing’’ events as eddies or jets moving nutrient rich water by the traps. This transiently increased the ratio of particulate silica : CaCO3, indicative of increased diatom abundance relative to coccolithophores and foraminifera. The influence of hydrographic features (probably eddies) on surface PIC : POC variability can be seen in the sawtooth nature of the data in Fig. 11B, as well as chlorophyll and b0b data plotted in temperaturesalinity space (Fig. 8). We would not have expected particulate carbon (PC) to be a conservative tracer, but one obvious question was whether PC standing stocks showed any relationship to water mass. That is, was there any relationship between PC and seawater density? We observed no such relationship, however (data not shown), in agreement with Claustre et al. (2000), who made scattering-based POC measurements in the Mediterranean. The same was true for Arabian Sea PIC, which also showed no relationship with seawater density. While it is tempting to conclude that coccolithophores are equally likely to grow at the Ras al Hadd front as 101N, care should be taken in using PIC as a proxy for coccolithophore cell abundance. Factors affecting PIC in the sea are not just the abundance of coccolithophores, but the rate of calcification, coccolith detachment, grazing, and dissolution (Balch et al., 2000; Fritz, 1997; Fritz and Balch, 1996; Milliman et al., 1999). For example, it is possible to have abundant, non-plated coccolithophores in seawater with no PIC. It is reasonable to use PIC as a proxy for coccolith abundance, however, since our PIC estimates were made in volumes of water ranging from B10 (optical measurement) to 300 ml (discrete samples), and the probability of sampling CaCO3 tests of terrapods or foraminifera in such small volumes is extremely low. It is worthy of note that as we crossed several fronts along the southern line, we noted distinctly higher values of acid-labile backscattering as we moved into

2448

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

the warm, stratified sides of fronts, sharply declining as we traversed back into cooler mixed waters. This variability was apparent at spatial scales of B10 km (sub-mesoscale), and shows up as sharp spikes in the PIC : POC record of Fig. 11B. Honjo et al. (1999) characterized the W. Arabian Sea as a ‘‘carbonate ocean’’, with punctuated, eddy-driven, diatom-dominated peaks (‘‘silica ocean’’), which rapidly reverted back to the ‘‘carbonate ocean’’ after eddy passage. Regions of high PIC : POC ratios would have been expected to have high backscattering: absorption ratios, likely regions of high reflectance. Indeed, the Oman coastal waters and Ras al Hadd frontal region were precisely where we observed high reflectance features in the historical survey of the CZCS archive (Fig. 1). Note, however, we also saw high PIC : POC values in transit to 101N, but only a few observations of high reflectance water were observed in this offshore region in our CZCS survey. The sediment trap data of Honjo et al. from the 101N oligotrophic station showed the greatest constancy in terms of PIC and POC fluxes. As we described above, oligotrophic regions were generally characterized by a monotonous background of low-scattering particles, punctuated with occasional, rare, highly scattering, particles. We can only surmise that at 101N, such rare, highly-scattering particles (e.g. foraminifera?) contained sufficient CaCO3 to raise the PIC : POC ratio, but that the large size of the CaCO3 particles lowered their effective bb  ( Balch et al., 1996a), and low numerical abundance guaranteed low normalized water-leaving radiance. The other inference that one can make from waters containing particulate matter with high PIC : POC ratios is that there is greater potential for ballasting of organic matter with CaCO3, which may lead to elevated sinking rates of carbon to the sediments. New improvements in light scattering measurements will allow us to estimate POC and PIC much more accurately (to be published elsewhere). Such semi-continuous measurements of POC and PIC will allow mapping of particulate carbon standing stocks at meso- and sub-meso-scales, a better understanding of hydrographic factors affecting PC variability (including short, episodic events), and better modelling of vertical carbon fluxes in the sea.

Acknowledgements Sharon Smith (U. Miami, RSMAS) is acknowledged for her hard work in organizing the US JGOFS Arabian Sea Program. I would like to thank Charles Trees (CHORS, San Diego) who helped formulate and implement the original plan for a dedicated JGOFS Bio-Optical Cruise. Robert Arnone (NRL, Stennis) kindly provided access to his shipboard AVHRR data, which we used to guide the R/V Thompson to interesting hydrographic features. Capt. Glen Gomes and crew of the R/V Thompson are thanked for their expert ship handling and deck operations. Katherine Kilpatrick and Charlie Byrne (U. Miami, RSMAS) helped with pre-cruise logistics as well as other backscattering measurements from the Straits of Florida (Fig. 7). Martin Bowen (WHOI) provided logistical support in Oman. Mike Sieracki (Bigelow Laboratory) helped process the images in Fig. 4. Larry Mayer (U. Maine Darling Marine Center) kindly allowed us to use his graphite furnace atomic absorption spectrometer. Amanda Ashe (Oregon State University) plotted several figures. Joaquim Goes (Bigelow Laboratory) and two anonymous reviewers commented on earlier drafts of the manuscript. This paper is dedicated to the memory of Dr. Maureen Keller (1954–1999), research scientist of the Bigelow Laboratory, who exemplified how

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2449

to integrate taxonomic observations into field studies of the physiological ecology of phytoplankton. This is Bigelow Laboratory contribution number 200101, and U.S. JGOFS contribution number 665. References Ackleson, S.G., Spinrad, R.W., 1988. Size and refractive index of individual marine particulates: a flow cytometric approach. Applied Optics 27, 1270–1277. Archer, D., Maier-Reimer, E., 1994. Effect of deep-sea sedimentary calcite on preservation on atmospheric CO2 concentration. Nature 367, 260–263. Balch, W.M., Drapeau, D.T., Cucci, T.L., Vaillancourt, R.D., Kilpatrick, K.A., Fritz, J.J., 1999. Optical backscattering by calcifying algaeFseparating the contribution by particulate inorganic and organic carbon fractions. Journal of Geophysical Research 104, 1541–1558. Balch, W.M., Drapeau, D., Fritz, J., 2000. Monsoonal forcing of calcification in the Arabian Sea. Deep-Sea Research II 47, 1301–1337. Balch, W.M., Holligan, P.M., Ackleson, S.G., Voss, K.J., 1991. Biological and optical properties of mesoscale coccolithophore blooms in the Gulf of Maine. Limnology and Oceanography 36, 629–643. Balch, W.M., Kilpatrick, K., 1996. Calcification rates in the equatorial Pacific along 1401W. Deep-Sea Research II 43 (4-6), 971–993. Balch, W.M., Kilpatrick, K.A., Holligan, P.M., 1993. Coccolith formation and detachment by Emiliania huxleyi (Prymnesiophyceae). Journal of Phycology 29, 566–575. Balch, W.M., Kilpatrick, K., Holligan, P.M., Harbour, D., Fernandez, E., 1996a. The 1991 coccolithophore bloom in the central north Atlantic. II. Relating optics to coccolith concentration. Limnology and Oceanography 41, 1684– 1696. Balch, W.M., Kilpatrick, K.A., Holligan, P.M., Trees, C., 1996b. The 1991 coccolithophore bloom in the central north Atlantic. I. Optical properties and factors affecting their distribution. Limnology and Oceanography 41, 1669–1683. Beardsley, G.F., Zaneveld, J.R.V., 1969. Theoretical dependence of the near-asymptotic apparent optical properties on the inherent optical properties of seawater. Journal of the Optical Society of America 59, 373–377. Bidigare, R.R., Latasa, M., Johnson, Z., Barber, R.T., Trees, C.C., Balch, W.M., 1997. Observations of a Synechoccusdominated cyclonic eddy in open-oceanic waters of the Arabian Sea. Ocean Optics, Proceedings of the Society of Photo-Optical Instrumentation Engineers 2963, 260–265. Bishop, J.K.B., 1999. Transmissometer measurement of POC. Deep-Sea Research I 46, 353–369. Bishop, J.K.B., Calvert, S.E., Soon, M.Y.S., 1999. Spatial and temporal variability of POC in the northeast subarctic Pacific. Deep-Sea Research II 46, 2699–2734. Bohm, E., Morrison, J.M., Manghnani, V., Kim, H.-S., Flagg, C.N., 1999. The Ras al Hadd Jet: remotely sensed and acoustic doppler current profiler observations in 1994–1995. Deep Sea Research II 46, 1531–1549. Boyle, E., 1988. The role of vertical chemical fractionation in controlling late quaternary atmospheric carbon dioxide. Journal of Geophysical Research 93, 15701–15714. Bramlette, M.N., 1958. Significance of coccolithophorids in calcium carbonate deposition. Bulletin of the Geological Society of America 69, 121–126. Bricaud, A., Zaneveld, J.R.V., Kitchen, J.C., 1992. Backscattering efficiency of coccolithophorids: use of a three-layered sphere model. Ocean Optics XI, Proceedings of the Society of Photo-Optical Instrumentation Engineers 1750, 27–33. Broecker, W.S., Peng, T.-H., 1989. The cause of the glacial-interglacial atmospheric CO2 change: a polar alkalinity hypothesis. Global Biogeochemical Cycles 3, 215–239. Brown, C.W., Yoder, J.A., 1994. Coccolithophorid blooms in the global ocean. Journal of Geophysical Research 99 (C4), 7467–7482. Campbell, J.W., 1995. The lognormal distribution as a model for bio-optical variability in the sea. Journal of Geophysical Research 100 (C7), 13237–13254. Campbell, L., Landry, M.R., Constantinou, J., Nolla, H.A., Brown, S.L., Liu, H., Caron, D.A., 1998. Response of microbial community structure to environmental forcing in the Arabian Sea. Deep-Sea Research II 45, 2301–2325.

2450

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

Campbell, J.W., Yentsch, C.M., Cucci, T.L., 1989. Variance within homogeneous phytoplankton populations, III: Analysis of natural populations. Cytometry 10, 605–611. Caron, D.A., Dennett, M.R., 1999. Phytoplankton growth and mortality during the 1995 northeast monsoon and spring intermonsoon in the Arabian Sea. Deep Sea Research II 46, 1665–1690. Claustre, H., Fell, F., Oubelkheir, K., Prieur, L., Sciandra, A., Gentili, B., Babin, M., 2000. Continuous monitoring of surface optical properties across a geostrophic front: biogeochemical inferences. Limnology and Oceanography 45 (2), 309–321. Cleveland, J.S., 1995. Regional models for phytoplankton absorption as a function of chlorophyll a concentration. Journal of Geophysical Research 100 (C7), 13333–13344. Davies, T., Kidd, R., 1977. Sedimentation in the Indian Ocean through time. In: Heirtzler, J., Bolli, H., Davies, T., Saunders, J., Sclater, J. (Eds.), Indian Ocean Geology and Biostratigraphy. American Geophysical Union, Washington, pp. 61–85. Denman, K.L., Platt, T., 1975. Coherences in the horizontal distribution of phytoplankton and temperature in the upper ocean. M!emoires Soci!et!e Royale des Sciences de Li"ege 7 (6), 19–30. Dickey, T., Marra, J., Sigurdson, D.E., Weller, R.A., Kinkade, C.S., Zedler, S.E., Wiggert, J.D., Langdon, C., 1998. Seasonal variability of bio-optical and physical properties in the Arabian Sea: October 1994–October 1995. DeepSea Research II 45, 2001–2025. Dupouy, C., Petit, M., Dandonneau, Y., 1988. Satellite detected cyanobacteria bloom in the southwestern tropical Pacific: implication for oceanic nitrogen fixation. International Journal of Remote Sensing 9 (3), 389–396. Eppley, R.W., Harrison, W.G., Chisholm, S., Stewart, E., 1977. Particulate organic matter in surface waters off southern California and its relationship to phytoplankton. Journal of Marine Research 35, 671–696. Eppley, R.W., Peterson, B., 1979. Particulate organic matter flux and planktonic new production in the deep ocean. Nature 282, 677–680. Fern!andez, E., Boyd, P., Holligan, P.M., Harbour, D.S., 1993. Production of organic and inorganic carbon within a large scale coccolithophore bloom in the northeast Atlantic Ocean. Marine Ecology Progress Series 97, 271–285. Flagg, C.N.K., Kim, H.-S., 1998. Upper ocean currents in the northern Arabian Sea from shipboard ADCP measurements collected during the 1994–1996 US JGOFS and ONR programs. Deep-Sea Research II 45, 1917– 1959. Fritz, J.J., 1997. Growth dependence of coccolith detachment, carbon fixation and other associated processes by the coccolithophore Emiliania huxleyi. Ph.D. Thesis, University of Miami, Miami, FL., 179pp. Fritz, J.J., Balch, W.M., 1996. A coccolith detachment rate determined from chemostat cultures of the coccolithophore Emiliania huxleyi. Journal of Experimental Marine Biology and Ecology 207, 127–147. Gardner, W.D., Biscaye, P.E., Zaneveld, J.R.V., Richardson, M.J., 1985. Calibration and comparison of the LDGO nephelometer and the OSU transmissometer on the Nova Scotia Rise. Marine Geology 66, 323–344. Garrison, D.L., Gowing, M.M., Hughes, M.P., Campbell, L., Caron, D.A., Dennett, M.R., Shalapyonok, A., Olson, R.J., Landry, M.R., Brown, S.L., Liu, H.-B., Azam, F., Steward, G.F., Ducklow, H.W., Smith, D.C., 2000. Microbial food web structure in the Arabian Sea: a US JGOFS study. Deep Sea Research II 47, 1387–1422. Geider, R.J., 1987. Light and temperature dependence of the carbon to chlorophyll a ratio in microalgae and cyanobacteria: Implications for physiology and growth of phytoplankton. New Phytologist 106, 1–34. Goldman, J.C., 1993. Potential role of large oceanic diatoms in new primary production. Deep-Sea Research I 40 (1), 159–168. Goldman, J.C., Hansell, D.A., Dennett, M.R., 1992. Chemical characterization of three large oceanic diatoms: potential impact on water column chemistry. Marine Ecology Progress Series 88, 257–270. Gordon, H.R., Brown, O.B., Evans, R.H., Brown, J.W., Smith, R.C., Baker, K.S., Clark, D.K., 1988. A semianalytic radiance model of ocean color. Journal of Geophysical Research 93, 10909–10924. Gordon, H.R., Smith, R.C., Zaneveld, J.R.V., 1979. Introduction to ocean optics. Ocean Optics VI, Proceedings of the Society of Photo-Optical Instrumentation Engineers 208, 14–15. Gower, J.F.R., Denman, K.L., Holyer, R.J., 1980. Phytoplankton patchiness indicates the fluctuation spectrum of mesoscale oceanic structure. Nature 288, 157–159. Graziano, L., Balch, W., Drapeau, D., Bowler, B., Dunford, S., 2000. Organic and inorganic carbon production in the Gulf of Maine. Continental Shelf Research 20, 685–705.

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

2451

Gundersen, J.S., Gardner, W.D., Richardson, M.J., Walsh, I.D., 1998. Effects of monsoons on the seasonal and spatial distributions of POC and chlorophyll in the Arabian Sea. Deep-Sea Research II 45, 2103–2132. Hewes, C.D., Holm-Hansen, O., 1983. A method for recovering nanoplankton from filters for identification with the microscope: The filter-transfer-freeze (FTF) technique. Limnology and Oceanography 28, 389–394. Holligan, P.M., Fernandez, E., Aiken, J., Balch, W., Boyd, P., Burkill, P., Finch, M., Groom, S., Malin, G., Muller, K., Purdie, D., Robinson, C., Trees, C., Turner, S., van der Wal, P., 1993. A biogeochemical study of the coccolithophore, Emiliania huxleyi, in the north Atlantic. Global Biogeochemical Cycles 7 (4), 879–900. Holligan, P.M., Viollier, M., Dupouy, C., Aikens, J., 1983. Satellite studies on the distributions of chlorophyll and dinoflagellate blooms in the western English Channel. Continental Shelf Research 2 (2/3), 81–96. Honjo, S., Dymond, J., Prell, W., Ittekkot, V., 1999. Monsoon-controlled export fluxes to the interior of the Arabian Sea. Deep-Sea Research II 46, 1859–1902. JGOFS, 1996. Protocols for the Joint Global Ocean Flux Study (JGOFS) core measurements. Report no. 19 of the Joint Global Ocean Flux Study, Scientific Committee on Oceanic Research, International Council of Scientific Unions, Intergovernmental Oceanographic Commission, Bergen, Norway, 170pp. Kitchen, J., Zaneveld, R., 1990. On the noncorrelation of the vertical structure of light scattering and chlorophyll a in case I waters. Journal Geophysical Research 95, 20237–20246. Kleijne, A., 1990. Distribution and malformation of extant calcareous nanoplankton in the Indonesian seas. Marine Micropaleontology 16, 293–316. Latasa, M., Bidigare, R.R., 1998. A comparison of phytoplankton populations of the Arabian Sea during the Spring Intermonsoon and Southwest Monsoon of 1995 as described by HPLC-analyzed pigments. Deep-Sea Research II 45, 2133–2170. Lohmann, H., 1908. On the relationship between pelagic deposits and marine plankton. Internationale Revue der Gesamten Hydrobiologie 1, 309–323. Mague, T.H., Friberg, E., Hughes, D.J., Morris, I., 1980. Extracellular release of carbon by marine phytoplankton: a physiological approach. Limnology and Oceanography 25, 262–279. Manghnani, V., Morrison, J.M., Hopkins, T.S., Bohm, E., 1998. Advection of upwelled waters in the form of plumes off Oman during the Southwest Monsoon. Deep-Sea Research II 45, 2027–2052. McCarthy, J.J., Garside, C., Nevins, J.L., 1999. Nitrogen dynamics during the Arabian Sea Northeast Monsoon. DeepSea Research II 46, 1623–1664. Meyer, S.L., 1975. Data Analysis for Scientists and Engineers. John Wiley, New York. Milliman, J., Troy, P.J., Balch, W., Adams, A.K., Li, Y.-H., MacKenzie, F.T., 1999. Biologically-mediated dissolution of calcium carbonate above the chemical lysocline? Deep-Sea Research 46, 1653-1669. Morel, A., 1988. Optical modeling of the upper ocean in relation to its biogenous matter content (case 1 waters). Journal of Geophysical Research 93, 10749–10768. Morel, A., 1997. Consequences of a Synechococcus bloom upon the optical properties of oceanic (case 1) waters. Limnology and Oceanography 42 (8), 1746–1754. Morel, A., Bricaud, A., 1981. Theoretical results concerning light absorption in a discrete medium, and application to specific absorption of phytoplankton. Deep-Sea Research 28, 1375–1393. Norris, R.E., 1984. Indian ocean nanoplankton. I. Rhabdosphaeraceae (Prymnesiophyceae) with a review of extant taxa. Journal of Phycology 20, 27–41. Norris, R.E., 1985. Indian ocean nanoplankton. II. Holococcolithophorids (Calyptrosphaeraceae, Prymnesiophyceae) with a review of extant genera. Journal of Phycology 21, 619–641. Parsons, T.R., Maita, Y., Lalli, C.M., 1984. A manual of chemical and biological methods for seawater analysis. Pergamon Press Inc., New York, 173pp. Reid, F., 1980. Coccolithophorids of the North Pacific Central Gyre with notes on their vertical and seasonal distribution. Micropaleontology 26, 151–176. Robertson, J.E., Robinson, C., Turner, D.R., Holligan, P., Watson, A.J., Boyd, P., Fernandez, E., Finch, M., 1994. The impact of a coccolithophore bloom on oceanic carbon uptake in the northeast Atlantic during summer 1991. Deep-Sea Research I 41 (2), 297–314. Sharp, J.H., 1974. Improved analysis for particulate organic carbon and nitrogen from seawater. Limnology and Oceanography 19, 984–989.

2452

W.M. Balch et al. / Deep-Sea Research I 48 (2001) 2423–2452

Spinrad, R.W., Brown, J.F., 1986. Relative real refractive index of marine microorganisms: a technique for flow cytometric estimation. Applied Optics 25, 1930–1934. Stramski, D., Reynolds, R.A., 1993. Diel variations in the optical properties of a marine diatom. Limnology and Oceanography 38, 1347–1364. Stramski, D., Reynolds, R.A., Kahru, M., Mitchell, B.G., 1999. Estimation of particulate organic carbon in the ocean from satellite remote sensing. Science 285, 239–242. Stumm, W., Morgan, J.J., 1981. Aquatic Chemistry. John Wiley, New York. Subramaniam, A., Carpenter, E.J., Falkowski, P.G., 1999. Bio-optical properties of the marine diazotrophic cyanobacteria Trichodesmium spp. II. A reflectance model for remote sensing. Limnology and Oceanography 44 (3), 618–627. Townsend, D.W., Keller, M.D., Holligan, P.M., Ackleson, S.G., Balch, W.M., 1994. Coccolithophore blooms in the Gulf of Maine. Continental Shelf Research 14, 979–1000. UNESCO, 1987. Thermodynamics of the carbon dioxide system in seawater. Report by the carbon dioxide sub-panel of the joint panel on oceanographic tables and standards. United Nations Educational, Scientific and Cultural Organization, Paris, Vol. 51, 54pp. Van de Hulst, H.C., 1957. Light Scattering by Small Particles. John Wiley, New York, 470pp. Volk, T., Hoffert, M.I., 1985. Ocean carbon pumps: analysis of relative strengths and efficiencies in ocean-driven atmospheric pCO2 changes. In: Sundquist, E.T., Broecker, W.S., (Eds.), Carbon Dioxide and the Carbon Cycle, Archean to Present, Vol. 32. American Geophysical Union, Washington, DC, pp. 99–110. Winn, C., Chiswell, S., Firing, E., Karl, D., Lukas, R., 1991. Hawaii Ocean Time-series Data Report 2- 1990. SOEST92-1, University of Hawaii, Honolulu, Hawaii, 175pp. Yentsch, C.M., Campbell, J.W., 1991. Phytoplankton growth: perspectives gained by flow cytometry. Journal of Plankton Research 13, 83–108. Yentsch, C.S., Phinney, D.A., 1989. A bridge between ocean optics and microbial ecology. Limnology and Oceanography 34, 1694–1705. Zaneveld, J.R.V., 1973. Variation of optical sea parameters with depth, Lecture Series on Optics of the Sea. Advisory Group for Aerospace Research and Development, NATO, Nevillyi sur Seine, France, pp. 1–22. Zaneveld, J.R.V., Roach, D.M., Pak, H., 1974. The determination of the index of refraction distribution of oceanic particulates. Journal of Geophysical Research 79, 4091–4095.