Limnol. Oceanogr., 36(8), 1991, 1886-1898 0 199 1, by the American Society of Limnology and Oceanography, Inc. Rates of phytoplankton cell division in the fib iron enrichment experiments Karl Banse School of Oceanography WB-10, University of Washington, Seattle 98 195 Abstract Increases in chlorophyll with time for contained coastal plankton, expressed as daily division rates, are on average about as high as rates for nutrient-replete cultures at similar temperatures, when daylength is considered. In offshore areas with persistent high nutrients but low chlorophyll, division rates from increased chlorophyll and cumulative NO, uptake in the controls of Fe enrichments are on average also high and do not suggestmarked Fe deficiency. The normally observed phytoplankton growth in the controls is interpreted as due to release from grazing. Addition of Fe in the treatments leads to blooms and exhaustion of NO,. Differences between controls and treatments in rates of chlorophyll increase and NO, removal, however, as well as shifts in species composition toward rare species in the treatments, also indicate direct effects of Fe on phytoplankton. To clarify the issues, especially in respect to medium- and large-celled phytoplankton, I recommend measurements of species-specific division rates. In the large offshore areas of the subarctic Pacihc, the eastern equatorial Pacific, and the subantarctic, circumpolar Southern Ocean, concentrations of macronutrients (N, P, Si) are always high, but those of chlorophyll are low even in summer; for the most part, phytoplankton biomass resides in, and most of the primary production rate rests on, small cells (Cullen 199 1; Frost 199 1). The primary production rate is the product of phytoplankton concentration and the specific growth rate averaged over the community, and the specific growth rate is proportional to the rate of cell division. The phytoplankton concentration is maintained by a balance between the rates of gain and loss of cells. To help in understanding the processes underlying such a balance in the sea, my paper compares newly derived cell division rates in control bottles of shipboard enrichment experiments, planned to show limitation of phytoplankton growth from Fe deficiency, with published values from coastal waters where Fe limitation of Acknowledgments My attendance at the symposium was fully supported and the preparation of the manuscript partially supported by NASA (Oceanic Productivity Program) grant NAGW- 1007. J. R. Postel helped with the computations. The suggestions by two reviewers and D. P. Henry regarding the content and form are also gratefully acknowledged. Contribution 1916 from the School of Oceanography, llniversity of Washington. division rates presumably did not prevail. The measurements are normalized by temperature. This variable is chosen because it appears to be the principal environmental difference for phytoplankton from nutrientrich subpolar and equatorial seas; at least in summer, daylength at the higher latitudes largely compensates for the greater hourly incident irradiance of the lower latitudes. I do not address the maintenance of the biomass level presently found or the change in biomass and, hence, production rate, which may result from an eventual field manipulation. In the absence of animals, healthy phytoplankton illuminated in a bottle with replete nutrients will grow exponentially following X = &expW (1) w‘here X, and X0 are the concentrations at times t and zero and p is the insta.ntaneous growth coefficient (time-l; unit of d-l here). The underlying dX/dt is called herein the population growth rate when referring to single species and the bulk population growth rate when the increase of the phytoplankton as a whole is considered in the absence of grazing or other losses; the latter term is applied also to natural plankton when the effect of grazing is experimentally excluded, as in dilution or chlorophyll-labeling techniques. As long as nutrients do not terminate population growth, X, is time-de- 1886 Phytoplankton division rates pendent, while p over a wide range of nutrient concentrations is not, so that p is the preferred criterion of physiological condition. With zooplankton present in the bottle, as in the natural water used in the published Fe enrichment assays, Eq. 1 becomes X = &expK~ - m>tl where m is the instantaneous death rate of the phytoplankton from grazing and enclosure effects. The underlying dX/dt is called herein the net population growth rate when referring to individual species and the phytoplankton community growth rate when addressing the phytoplankton as a whole (e.g. as chlorophyll). The difference (p - m) is the instantaneous net community growth coefficient. In the seas of concern, the nearabsence of time dependence of X, (the accumulated biomass) indicates again the utility of studying p instead of terminal yield of cells or chlorophyll, as these might be measured in an enrichment experiment. It can be shown for the seas of concern that the physically caused losses of phytoplankton (from vertical advection, diffusion, and sinking) are small relative to division rates during most of the year, so that the principal process determining m must be grazing (Frost 199 1). Grazing is size-dependent, small (large) animals generally being unable to graze efficiently on large (small) plant cells, if at all (however, see Longhurst 199 1). Considering further that algal division rates are mildly size-dependent (see below), Eq. 2 can be expressed as sum xi(l) = sum X&eXp[(fli - T?Zi)t]. (3) In other words, the difference (b - m) in a mixed community cannot be profitably replaced by a variable r. Instead, it must be kept in mind that studying enrichment experiments in natural plankton means tackling one equation (Eq. 2) with two unknowns. In the following sections, maximal instantaneous rates of population growth [d-l, expressed as division (doubling) rates per day, div. d-l, from b/in 2 = p/0.693], as obtained in unialgal or pure cultures under saturating light with a full complement of nutrients, will be reviewed as dependent on 1887 temperature, daylength, cell size, taxonomic affiliation, and nutrients, including Fe. Interactions among these factors, however, are not considered. The results are compared with newly reviewed community and bulk population growth rates from coastal waters, which presumably are not Fe deficient, and with the division rates in the controls of bioassays from offshore waters where dissolved Fe is presumed to be in short supply. The community growth rates are expressed as community division rates (d-l) by dividing the instantaneous net community growth coefficients (h - d) by 0.693. The net population division rate is the homolog for individual species, and it and the bulk population division rate are obtained like the community division rate. Division rates in cultures This section treats published laboratory data on division rates of individual phytoplankton species. Temperature - Algal division rates in batch cultures, replete with light (mostly continuous), COZ, and nutrients, were plotted on temperature by Eppley (1972). The maximal values fell under an envelope that described an exponential dependence such that division rates almost doubled (1.88 times) upon an increase of temperature of 10°C. Numerous data points lay well below the envelope; Eppley showed that for some species these values reflected less than optimal temperatures but surmised that for most the cause could have been differences in cell size and concentrations of chlorophyll in the cells. The role of the last two variables for diatoms was treated theoretically by Geider et al. (1986; note that their figure 3 appears to present division rates), but only cell size is treated below. Earlier studies of division rates related to temperature among phytoplankton (some tending to slightly higher dependence, as mentioned by Eppley 1972; cf. Goldman and Carpenter 1974 for continuous cultures) were short on data for near-freezing temperatures. Also, the functional form of the dependence has been argued (e.g. Ahlgren 1987). New measurements for lightsaturated Antarctic diatoms (Fig. 1) indicate that Eppley’s temperature coefficient is rea- 1888 Banse I v I u >; .0 I I I I I I I 1.5- l.O- *0 A=? Y0 : TT 0.5- -4 --,/A-* a -AA I I I -2 0 co. 12 20 I 2 I I 4 I - I 6 “C h Fig. 1. Division rates of algal species in cultures as dependent on temperature with the “standard curve” for optimal growth (Eppley 1972) and half of this maximum (broken curve) added. Filled symbols-24 h of light; other estimated daylengths (h) shown below temperature scale. Double data points indicated by “2.” From left to right: O-diatoms (Rivkin and Putt 1987); ranges (bars) and medians (diamonds) for 10 diatoms (avg max rates, mostly for 24-h days, Gilstad and Sakshaug 1990) for 13 diatoms and 2 others (with clear size dependence within set, Sommer 1989); A-diatoms (Jacques 1983); O-diatoms (Fiala and Oriol 1990; note that p instead of division rates had been calculated). sonably accurate. These data are supported by observations on net population division rates in containers of natural Antarctic waters: Spies (1987) studied five diatom species in coastal samples; under continuous light at - 1°C 27 rates ranged from 0.38 to 1.33 div. d-l (median, 0.70). Buma et al. (199 1) investigated nine diatom species in offshore waters; during 16-h days at (terminally) 3. S”C, 13 rates ranged from 0.36 to 0.98 div. d-’ (median, 0.63). Daylength -Among papers treating the effect of daylength on division rates are the comprehensive treatments by Brand and Guillard ( 198 1) and Langdon ( 198 8) and a study on Arctic diatoms by Gilstad and Sakshaug (1990). No taxonomic pattern is obvious among the responses, which range from insensitivity of division rates to daylength to rates at a combination of, for example, 10 h x - 50 pmol mm2 s-l being higher than at 4 h of saturating irradiances of 500 pmol mm2 s-’ (division rates may differ more than twofold; example from Gilstad and Sakshaug) and to inability to grow under continuous light. Thus, the average division rate of mixed phytoplankton under replete light and nutrients might not come close to the value expected from temperature alone. cell size (mass)-In contrast to most other organisms, the dependence of division rates on cell size is mild among phytoplankton-the maximal (optimal) rates in diatoms and dinoflagellates at 20°C approximately being halved upon a lOO-fold to 200-fold increase in cell C (from linear predictive regression) or upon an increase of -30-fold [from model-2 regression (divide the former mass dependence by the correlation coefficient); both ranges calculated from Banse 19821. The mild mass dependence was confirmed by Sommer (1989) for diatoms at 0°C. Even so, large-celled phytoplankton species, growing under saturating light with replete nutrients, cannot come close to the division rates expected from temperature alone, and mixed diatom assemblages also may not be able to do so. Note: Schlesinger et al. (198 1) and Langdon (1988) in equations that lumped large taxa, calculated steeper exponents of dependence of growth (division) rates on cell C than reported by, for example, Banse (1982) and Sommer (1989) for diatoms or dinoflagellates. Clearly in Schlesinger’s data set (their figure 1), the relation was unduly weighted (steepened) by the inclusion of many small diatoms and several large non-diatoms. Because of the differences in maximal division rates among major algal taxa (see below), a re-evaluation of the equations of Schlesinger et al. and Langdon is in order. Taxonomic afiliation -Earlier I showed (Banse 1982) that, at the same cell volume (C content) and at 20°C division rates of medium-sized diatoms are -3 times those of dinoflagellates of the same size. 1 mentioned that those of 19 species from other algal groups and the Cyanophyceae fell largely between diatoms and dinoflagellates, with a few growing even more slowly than the average dinoflagellate of the same size. The field data of Furnas (1982, 1991) and Landry et al. (1984), as well as the review by Furnas (1990), support this generalization. Again, the average division rate of mixed phytoplankton under replete light and nutrients might not come close to the value expected from temperature alone. Phytoplankton division rates 1889 Macronutrients- Because this paper fo- ever, Rich and Morel (1990) stated that, if growth in the open sea is cuses on nutrient-rich seas, little needs to phytoplankton be said about nutrients vs. division rates. limited by Fe, slow kinetics of Fe solubilization must be the reason. In view of the suggested Fe fertilization project in the Southern Ocean (Martin et al. 1990a), however, two points will be noted. Division rates in the field This section treats experiments in coastal First, Zentara and Kamykowski ( 1977) found that Si04, rather than N03, is apt to waters where I assume Fe is not limiting. Methods-Division rates were calculated, be the limiting nutrient at intermediate and when incubation temperatures were availhigher latitudes of the western South Pacific. was Similarly, low values of SiO, and Si04 : NO3 able and limitation by macronutrients unlikely, from observed concentrations or ratios were reported offshore from the Indian sector of the Southern Ocean (Jacques independent evidence (as by Laws et al. and Minas 198 l), and SiO, exhaustion was 1984). Division rates were found either from growth rates (chlorophyll infound to set-in offshore before that of NO3 community in bioassays in the Scotia-Weddell Seas creases, simple symbols in Fig. 2; usually, samples were strained through coarse net(Atlantic sector; Buma et al. 1991). In contrast, near the coast of Antarctica, Spies ting to remove large zooplankton before incubations) or bulk population growth rates (1987) observed that PO, was exhausted first (grazing experimentally excluded; circled in two of five large incubation experiments. symbols in Fig. 2). Average relations for the Second, Jacques (1983) and, more fully, Sommer (1986) showed for Antarctic pe- two types of rates were calculated by linear regression analysis from Eppley’s ( 1972) lagic diatoms that half-saturation constants model, for SiO,-controlled growth may be so high (ranging between 6 and 89 IAM, median 22, log div. (d-l) = a + b x T(“C). for five spp.) that some species exhibit greatly reduced growth rates under common, Plots of the residuals on daylength showed “nutrient-rich” conditions. Sommer and no trend for community and a nonsignifiStabel(1986) reported from the same cruise cant trend (increase with time) for populaas Sommer (1986) that the observed horition division rates. Also, linear regressions zontal distribution relative to the ambient of log division rate on hours were nonsigSi04 : NO3 ratios of three dominant diatoms nificant for either group (r2 = 0.003 and and “other, non-silicious algae” (principal0.0 1, respectively). Therefore, time was nely, cryptophycean flagellates) reflected these glected and the new average relations of diphysiological properties. In the last-menvisions are dependent on only temperature tioned data set, then, division rates (and in Fig. 2. I believe, however, that daylength hence competition for nutrients) overrode (incubation period) should play a role and sinking and grazing losses in determining that its generally small range in the present community composition. Thus, the effect of material (Fig. 2), combined with other macronutrients on cell division cannot be sources of variability, obscures the effect of neglected even in the nutrient-rich Southern time. To prevent premature generalization Ocean. of weak equations, I chose not to report the Micronutrients (Fe) - Fe uptake and Fe- relations (regressions) numerically. limited growth by marine phytoplankton Note: Figure 2 does not incorporate division rates were studied by, among others, Brand et al. that were estimated by dividing 14C uptake by (1983), Sunda et al. (199 l), and Morel et al. initial Chl concentration and multiplying with a (199 1). Because Michaelis-Menten kinetics Chl: C ratio nor cell C calculated from microreign and growth even for oceanic species scopical counts. Both approaches eliminate the may be reduced severely or become nil at effect of grazing to the extent that the 14Cmeasurement is not affected by it (but see Laws 1984), oceanic concentrations of dissolved Fe, the but they are afflicted by other error sources. For element might exert a very large effect on high growth rates especially, the appropriate didivision rates. Noting the prevailing, relavisor is not initial Chl but mean Chl over the tively high concentrations of total Fe, howincubation period; further, accurate Chl : C ratios 1890 Banse 3 6 12-13 12-15 8 10 12 14 16 109 91312 1616 v8 18 20 q 22 24 26 28°C 14 121212 h 14 Fig. 2. Division rates for natural marine plankton in enclosures as dependent on temperature, with the “standard curve” for optimal growth (Eppley 1972) and half of this maximum (broken curve) added. Estimated daylength (h) shown below temperature scale. Filled symbols- two stations. Dashed line (from regression, see text) and circled or enclosed symbols-bulk population division rates [results without grazing, from dilution method, or Chl labeling (the latter for Welschmeyer and Lorenzen 1984; Downs and Lorenzen 1985; Laws et al. 1984)]; thin line and other symbols- community division rates (from Chl increases). From left: x -Spies 1987; A, A-Paranjape 1987 (samples with excess NO,); O-Taylor and Haberstroh 1988; V-Sakshaug and Holm-Hansen 1986 (highest values at intermediate n-radiances); + - Kuiper et al. 1983 (examples of high rates, No. 24, 25; temperature from Jahnke et al. 1983); Cl-Gifford 1988; x -Welschmeyer and Lorenzen 1984 [table 3; @-highest values except for March (second highest); x -means of two highest values for each date; temperatures from original log books]; V-Downs and Lorenzen 1985 [table 3; means of two highest values for November, January and April, when NO3 and incident irradiance were high (from original log books)]; OWilkerson and Dugdale 1987 (No. 58B 2); V-Riemann et al. 1988 (enclosure C, 12-19 September); AMcAllister et al. 196 1 (days 4-14) Antia et al. 1963 (days 8-l 5) both at low irradiance; A-Eppley et al. 197 1 (from figure 2, 1l-l 3 July average of NO, and NH, additions; “20“-25”C”); 0-Landry et al. 1984 (temperature estimated); + -Cullen et al. 1992 [mean of several stations, under low irradiance (moved from Fig. 3, not used in regression)]; v, V-Laws et al. 1984 (without offshore station F). are normally not available. In the microscopical approach, all cells must have been preserved and counted, and there is variance in the equations converting cell volumes to C. Results-Considering the effects of daylength, cell size, and taxonomic affiliation on cell division rates in algal cultures, one might surmise that the bulk population or community division rates of natural, mixed phytoplankton, which grows with replete nutrients near light saturation, might not nearly approach the temperature-set maximal rates. In fact, however, as reviewed by Furnas (1990) and quantified in Fig. 2, natural assemblages initially enclosed in containers may double fast relative to the maximal rate when considering the shorter periods of illumination (Eppley’s 1972 envelope for maximal division rates refers to continuous light). The field results suggest that, on the average, only phytoplankton species well suited to the environmental controls and the simultaneous, size-dependent grazing pressure tend to dominate natural communities; the other species tend to lose out because of the relentless loss of cells to the animals. Further, because several data points in Figs. l-3 fall above the line for half of Eppley’s maximal rate, although the incubation periods were not overly long, Eppley’s curve may underestimate very high community division rates where, for example, assemblages dominated by small diatoms grow under a suitable day-night cy- 1891 Phytoplankton division rates cle. As to be expected, however, Fig. 2 shows that the community division rates (i.e. some grazing present) tend to be lower than the bulk population rates. The intercepts of the regression-based curves in Fig. 2 differ significantly on the P = 0.05 level, but the slopes (curvatures in Fig. 2) do not. The data in Fig. 2 are supported by other observations that did not meet all selection criteria. Of Furnas’ (1982) 29 temperate coastal assemblages, which were restricted to cells < 10 pm, two-fifths grew at half the optimal rate (Eppley’s 1972 curve) or faster, even though exposed to only 13-15 h of light (16”-22”C, from Chl change; median rate, 1.25 div. d-l, but 1.55 div. d-l after omission of four dates without appreciable growth). Bienfang and Takahashi (1983) reported 1.3,2.2, and 2.5 div. d-l for the <3pm fraction in subtropical coastal seawater (25”C, 11 h of light; from Chl change). These two data sets suggest that the curvature of the line for the community division rates in Fig. 2 is too low. Using the Chl-labeling technique, Laws et al. (1984) estimated 2.9 div. d-l (27”C, 12 h of light) at a tropical, oligotrophic, offshore station. Based on the same technique, Laws et al. (1987) reported light-saturated bulk population rates of - 1.7kO.4 div. d-l (25”C, 13 h of light) for several nutrient-depleted offshore stations; these data are included here because independent evidence suggested absence of growth limitation. Thus, Laws’ rates support the line for bulk population growth rate in Fig. 2 well. Overall, it appears that the Eppley (1972) relation of maximal division rates, derived from cultures, approximately holds for natural communities once the shorter exposure time is discounted; in view of the unsatisfactory statistics on time dependence, more work on the latter is clearly desirable. Division rates in the published Fe assays In this section, recent Antarctic enrichment experiments data are re-evaluated before division rates from reputedly Fe-limited offshore waters are compared with the coastal observations shown in Fig. 2. Re-evaluation -Results from Fe bioassays were published by Martin et al. (1990a) for the Ross Sea (Pacific Sector of the Southern Ocean) and de Baar et al. (1990) for the Scotia and Weddell Seas (Atlantic Sector). Both papers reported time changes of Chl and NO3 in control bottles and in samples with Fe additions under natural illumination in the presence of grazers. The pigment change results from the balance between the addition and loss (grazing) of cells and ought to be evaluated by Eq. 2; without a grazing estimate, however, the instantaneous rate of pigment change must be estimated from Eq. 1. In contrast, the doubling rate of cumulative NO3 removal results from a specific activity of phytoplankton times its concentration; the activity is apt to be constant during exponential growth when there is no time change of division rate (straight lines in semilog plots). This division rate is a measure of the rate of bulk population growth because N03, once assimilated, is gone (no release by grazing) while newly formed Chl in the assays is being grazed. The rate is evaluated by Eq. 1 (cf. Banse 199 1). The doubling rate of cumulative NO3 uptake should be faster than that of Chl increase, which in fact tends to be the case [Table 1 and table 1 of Banse 199 1 from the Gulf of Alaska; also McAllister et al. 196 1; Spies 1987 (newly read from their figures)]. The data by Martin et al. (1990a) were taken from their figure 2 and those by de Baar et al. (1990) from their tables. For both sets, the periods of exponential growth were judged from semilog plots (the criterion being straight lines) and regressions calculated with the model P = ln([X,/&]/At where X, and X0 are Chl concentration or cumulative NO3 uptake at times t and zero. For Martin et al. (1990a), I evaluated only the controls and the treatments with 5 nmol Fe kg-l, but did not utilize the NO3 data from Sta. 1 because of an apparently too high initial (day 0) value, as well as nonlinearity in the semilog plot, when day 1 was used as the starting date. Material death (i.e. disappearance of Chl) occurred in all experiments early on, especially at Sta. 2, where about three-fourths of the initial pigment vanished in the first 2 d. Finally, extrapolation of the NO, regressions to time zero 1892 Table 1. Division rates (d-l, from instantaneous rates of change) for Chl increase and cumulative NO, uptake in bioassays. C-controls; T-Fe treatments (5 nmol kg-’ for Martin et al. 1990a, l-10 nM for de Baar et al. 1990); days in parentheses, with bracketed days preceding digit, if referring to C, and following, if referring to T; means for Sta. 145 and 158; single values for remainder. -Martin et al. 1990~ Days Chl NQ (l-9) CW110) Chl NO, Sta. 1 0.13 Sta. 2 0.24 0.23* (2-13181) Chl NO, (6[31-13) (6-l 3) Chl NQ de Baar et al. 1990 c Sta. 3 0.16 0.27(*) Sta. 4 (3-l 1) (3-l 1) 0.25 0.53 T 0.18 (2-7) (2-6) WW’I) WWI) 0.49 0.42* 0.26 0.35* 0.66 0.82 (l-lO[ll]) (2-l 1) (2-9) (4-9) W-8) Chl NO, * Intercept Days (0.4-8) s 1 rmol kg-’ (PM) (i.e. large NO, uptake C Sta. 145 0.64 0.72 Sta. 158 0.22 0.29* Sta. 159 0.45 0.40 Sta. 169 0.49 0.52 Sta. 182 0.48 0.48* I-T 0.69 0.65 0.29 0.5 1* 0.56 0.46 0.60 0.59 0.51 0.68 by time zero). often yielded intercepts (i.e. appreciable cumulative uptake at time zero) which for Sta. 2 and 3 were close to 1 ,umol kg-l (see also the mentioned initial value at Sta. 1). Actually, cumulative NO3 uptake at time zero is zero by definition; such erroneous intercepts lead to lower calculated rates of cumulative uptake and, hence, division rates (Banse 199 1, table 2). Note that Martin et al. (1990a) estimated the community division rates under the assumption that the decrease of NO3 was linear in time and, by implication, that linear algal growth prevailed. Their figure 2, however, shows exponential time changes for NO3 and Chl, as do my semilog plots. As stated by Martin et al. (1990a), all (P < 0.05) treatments differ significantly from the controls (Table l), i.e. Fe addition enhances division rates. Since I use parametric statistics in spite of the small number of points, the significance level of the enhancement shown is apt to be overestimated (cf. Sta. 1; perhaps also Sta. 3). My reevaluation (Table 1) of the data of de Baar et al. (1990; see also Buma et al. 199 1) fully supports their interpretations: Growth rates in controls are high, but Fe addition can enhance division rates further (see also next section). Note: Dugdale and Wilkerson (1990) in their reevaluation of the data of Martin et al. (1990~~) emphasized the Chl-specific uptake rates of NO,. On the basis of a model involving several assumptions (e.g. conversion factors), they found no significant difference between controls and treatments and, thus, did not agree with, Martin et al. (1990a). However, I wish to advise caution, since my more straightforward, hence parsimonious, approach yields ambiguous results regarding Chl-specific uptake rates. Dividing the predicted NO, uptake for 1 d (from regressions, cf. Table 1, Martin et al. 1990a data) by the mean pigment content during that day, after multiplying the initial pigment of that day by [exp(pCLI) l]lk.t, (with t = l), yields satisfactory ratios [in pmol N (pg Chl x d)-‘1 of 0.77 and 0.66 for the controls and 0.90 and 0.66 for the treatments for days 2-3 and 7-8 at Sta. 2; these may or may not be indistinguishable. At Sta. 4, however, the ratios vary greatly with time even within treatments becausethe slopes (exponents) differ greatly among the NO, and pigment regressions; such variation within a flask is not to be expected with ordinary exponential growth. Rates in Fe-poor ofshore areas and coastal waters- For comparing the averages of Chl- and NO,-based division rates from bioassays with the coastal data, it is appropriate to use the range between bulk population and community division rates (from Fig. 2). Figure 3 shows that on average, di- vision rates in the controls of the bioassays from subantarctic and subarctic ojfshore wa- Phytoplankton division rates 1893 A ters are about as high as ratesfrom the coastl.5' ' ' I"' ' al-andpresumably Fe replete-stations depicted in Fig. 2, once temperature is accounted for. Therefore, community division rates do not appear to be severely 1.0 idepressed throughout seas that are remote -0 from sources of iron dust, as they would if the phytoplankton dominating in situ were .-I5 Fe deficient. Fe addition, however, regularly cl 0.5 enhances division rate. The increase is USUally not significant stati‘stically (Banse 199 1; de Baar et al. 1990; Cullen et al. 1992; Table 1). Some low rates in the controls are ac1 0 2v 10 12 "C -2 companied by low rates in the treatments so that, presumably, temperature or Fe were ca.13 h 16 ca.16 not the only rate-controlling factors. Fig. 3. Division rates from bioassays as dependent Overall, the stations from the Scotiaon temperature, with the “standard curve” for optimal Weddell Seas with concentrations of dis- growth (Eppley 1972), half of this maximum (broken solved Fe of 2-4 nM (de Baar et al. 1990) curve), and the relations for community division rate (thin line in Fig. 2) and bulk population division rate did better than those from the presumably line in Fig. 2) added. Estimated daylength (h) Fe-poor Ross Sea (Fig. 3; both based on (dashed shown below temperature scale. Averages for Chl- and 16-h days); the effect of Fe addition was NO,-based rates near 0°C from Table 1, above 10°C relatively small in the Scotia-Weddell Seas from Banse (199 1). Controls (open symbols) and cor(Table 1). Ironically, among the four sta- responding treatments (highest Fe additions; filled symbols) aligned. Circles based on de Baar et al. 1990 tions of Martin et al. (1990a) the highest (temperature estimated, some symbols offset for clarrates in the controls were found at the pre- ity); inverted triangles near 0°C based on Martin et al. sumably most Fe-deficient site (judging from 1990a (some symbols offset for clarity), above 10°C the location); it was at this station, though, based on Martin et al. 1989 (temperature estimated). See also 26°C value for Cullen et al. (1992) in Fig. 2. that Fe addition had the greatest absolute, as well as percentage effect (cf. Fig. 3, inverted triangles at - 0.5”C). Other evidence also suggests that the rate of in situ community division at the station was not de- Low abundance of diatoms and community division rates pressed by severe Fe limitation: Dugdale and Wilkerson (1990) here estimated a high This section reviews and discusses the occurrence and role of rare phytoplankton specific NO3 uptake rate ( VNO~). Further, I note that the observed removal of 3.7 pmol species in the offshore areas under considNO3 kg-l in the control, multiplied with a eration. Recall that a feature of the three molar C : N ratio of 6.6 and a molar C : Fe ocean regions is the low abundance of largeratio of 20,000 (from Martin et al. 1990a) celled species, although many such species suggests a consumption of dissolved Fe of occur. The predominance of small cells has 1.2 nmol kg-l. Elsewhere in the Antarctic, been established by microscopical counts Martin et al. (1990b) observed near-surface (usually of samples of a few up to - 100 ml) values about a tenth as high. Evidently, more and by size fractionation of photosynthetic Fe was available to the plankton than was rate determinations (at most a few hundred measured as dissolved, which I already notmilliliters) or chlorophyll (usually < 1 liter). ed for the Gulf of Alaska (Banse 199 1). This In experiments with contained water in the conclusion holds even if the C : Fe ratio was Gulf of Alaska, including the treatments of a few times higher than assumed above. Note the Fe enrichments, though, medium-sized finally that the average division rate of the cells tend to become abundant after - 1 week equatorial community in Fig. 2 (diamond of enclosure (Frost 199 1; see Cullen et al. at 26”C, from Cullen et al. 1992) is remark1992 for the same phenomenon in enrichably low; division rates for controls and ment experiments in the equatorial Pacific). treatments were similar. Further, in the central Gulf of Alaska, the 1894 Banse assemblage of diatoms collected in sediment traps (with a number of large-celled species prominent and with marked differences among seasons, Takahashi et al. 1990) is quite unlike that observed in routine plankton counts from the euphotic zone. Obviously, large cells are more likely to be present in settling material than small ones, but the point here is that these large species are rare in the surface layers. Finally, in a central Pacific section from 33”N to 41’S, Semina (1962) compared l-liter (!) bottle samples with hauls by nets of 180-pm mesh size made on the same stations. At the three stations near the equator, 2, 13, and 4 species of the totals of 29, 82, and 46 species, respectively, were common to both gears; at the two southernmost stations, no species were common among the totals of 26 and 12, respectively. The total specimen numbers of the rare forms collected by the (coarse!) net in the upper 100 m were rnostly 5 1 liter-l, the maximum at the equator being 6 liter-l (Semina 1963). Are many of these uncommon species rare because of Fe deficiency, which is apt to be felt more strongly by large than small cells (Hudson and Morel 1990), or because of grazing by large zooplankton? Would they remain rare after Fe fertilization (see Cullen 199 1; Frost 199 l)? Has, instead, the rarity of large cells (as contrasted with mediumsized forms) in the available Fe enrichments to do with too small inoculum sizes, or with Fe- and size-dependent growth rates, or is it a fluke because of too rare experiments? Varying plankton composition may also underlie other observations made during Fe enrichment work. The community division rates in controls and treatments at former weather station “Papa” in the central Gulf of Alaska by Martin et al. ( 1989; see Banse 199 1) were low relative to expectation from temperature (Fig. 3, inverted triangle at 12”C), suggesting appreciable limitation by other than macronutrients. At and near the same site, the average community division rate (derived from Chl change, Welschmeyer et al. 1991) of 0.36 div. d-l in 109 in situ incubations of 24 (48) h, through much of the euphotic zone during several summers, was even lower than in Martin’s experiments (in the latter, the rate was usually determined for periods starting after days 1 or 2, see Banse 199 1). At “Papa,” an important cause for a low rate of commun.ity division may be the low abundance of diatoms: the division rates from the increase in the diatom pigment fucoxanthin are much larger than the community division rates in the same samples (Strom and Welschmeyer 199 1; Welschmeyer et al. 199 1). It appears that the rapid intrinsic growth rate of diatoms, coupled in the bottles with reduction of grazing pressure on these usually larger cells (demonstrated by Strom and Welschmeyer 199 1; see also Buma et al. 199 l), tends to lead to their dominance in the treatment bottles after a few days. Thus, the Fe addition prevents early termination of biomass increase (Banse 199 1, and below) and may affect division rate differentially among species. Elsewhere, the small in situ biomass of diatoms may be a reason for the low division rates of the equatorial community (diamond in Fig. 2 at 26°C; controls and treatments were indistinguishable). A similar depression of community division rate may be expected for many sites in the Antarctic where, apart from blooms in the marginal ice zone, small cells other than diatoms can prevail. In fact, the community division rates at Sta. 145 and 159 (Table 1; at the same offshore location) are lower than the net population growth rates of most individual diatom species (from Buma et al. 199 1, table 3) in the same containers. Buma et al. (1991) already noted differences in grazing pressure on diatoms and other phytoplankton on the basis of independent evidence. For none of these SiO,-rich offshore areas is it clear why the abundance of diatoms is low to begin with, although, as stated, many species occur in low abundance. Discussion This paper shows that the division rates of phytoplankton in Fe-depleted offshore waters are not low, but most are about as high as observed in coastal waters where Chl concentrations can be large. For both regions, it is not known whether the high division rates are Fe saturated, but. a comparison of Figs. 2 and 3 with Eppley’s (1972) maximal rates demonstrates that even the Phytoplankton division rates 1895 the low Si : N ratios of surface water, which low ones are not as depressed as they would be if Fe were really scarce. Instead, as stated were mentioned earlier, suggest that effects by Banse (1991), the bioassays from sub- like new CO, fixation may not be predictable everywhere based simply on consumparctic and subantarctic waters suggest that tion of all excess N. Also, since SiO, has in situ grazing prevents the doubling or quabeen depleted first, I wonder whether many drupling of Chl concentrations over ambient levels, which is normally observed in large cells other than diatoms, or only small flagellates, can be expected; small cells are the controls; enough Fe is available from apt to enter microbial food webs that do not the outset, or is supplied during the interval lead to high vertical fluxes of organic matter. of l-2 cell divisions, to permit a material Moreover, in surface waters where SiO, exincrease of Chl in the controls. Fe addition in the treatments, however, results in large haustion occurs first, one consequence might blooms and NO, exhaustion. At the same be that diatoms settling out at the end of blooms would consume SiO, below the eutime, the phytoplankton species composiphotic zone and the nitracline and carry it tion in the treatments tends to change markedly. Basically the same conclusion has been to great depths, which might be a regular feature in the Barents Sea (Rey and Skjoldal drawn for the waters of the eastern equa1987). If so, the Si04 supply to the surface torial Pacific (Cullen 199 1). in subsequent years would be affected unThe hypothesis of Fe limitation of phytoplankton growth in large parts of the sea favorably. Lessons for bioassays-The Fe enrichis currently being discussed on the basis of ment experiments published so far did not, an experimental data set, which is unusually small for this major issue and which has not by their design, measure rates of cell dividivision, because answered several questions. On the as- sion but of community were always enclosed sumption that more bottle experiments on small zooplankton with phytoplankton; the bioassays dealt with community and especially population division rates are called for prior to field ma- one equation with two unknowns (Eq. 2). nipulation, this discussion emphasizes les- Further, to derive instead the instantaneous growth rate of phytoplankton and, hence, sons from algal cultures and published division rate from the cumulative NO3 upbioassays. take can serve only as a substitute for direct Lessonsfor the fild from algal cultures Figure 2 extends to field data the great role measurement because the calculation is of temperature summarized by Eppley heavily affected by the quality of the initial (1972) for division rates in algal cultures. In NO3 determinations (Banse 199 1). Other consequence, when comparing the three criticism, which can be advanced against major ocean regions and thinking of field the available assays, concerns the exclusion manipulations with Fe, algal growth proof large zooplankton from the containers cesses must be expected to proceed 2-3 times because of low abundance. This omission faster in equatorial waters than in the sub- will shift the competitive advantage in the arctic Pacific during summer and 5-6 times bottles toward the larger algae directly; faster than near an ice edge. In contrast, moreover, as noted by Buma et al. (199 1) physical processes like advection and eddy for Antarctic plankton, it will have the same diffusion will be largely independent of temeffect indirectly by reducing grazing presperature. Because of small seasonal change, sure on the small animals that prey on the temperature in each of the three regions will initially dominant small phytoplankton. matter little for division rates, although it Again, more than one unknown is present may affect changes of assemblage compoin the equation to be solved by the enrichsition and, thus, indirectly influence diviment experiments. I believe, however, that sion rates (e.g. via cell size). even so, several outstanding questions in Further, in respect to large-scale experithe Fe controversy can be addressed by difments with Fe additions in the Southern ferently designed bottle assays. Ocean, the high half-saturation constants for Proximally at issue for understanding the SiO, uptake by some dominant diatoms and high-nutrient, low-chlorophyll paradox is 1896 Banse how small or large p may be in situ, whether toms?) flourish in assays because of removal large cells (or diatoms as a whole) are usu- of Fe deficiency” vs. “they do so because of ally Fe limited while small cells are not, and absence of suitable grazers,” can be examhow even a small p is balanced by grazing ined from bottle experiments by the process and physical processes so that the local time of elimination. If the former statement does change is close to zero. The biological issues not hold, the latter can well be true. Cumneed to be seen in the framework of Eq. 3, bersome as such species-specific approaches with p and m varying among species. Among will be for the individual researcher, much the questions that can be addressed with less manpower and funds will be required small containers on shipboard are: where, than for a manipulation of an area in the when, and to what degree does Fe addition open sea where grazers are present in notenhance the division rate of the dom.inant controllable numbers and kinds. The bottle phytoplankton species vs. the stimulation experiments can be done first and soon, but, of rare, larger species that might require most importantly, they will yield less amhigher concentrations of Fe? what effect biguous results than the field manipulation does the reduction of grazing by larger an- because of better definition of conditions imals, which are normally excluded from before and during the incubations, includbottles, have on the enhanced population ing rigorous control experiments and repgrowth of small predators, which depress lication, and the absence of eddy diffusion the abundance of their prey (i.e. of the phyand advection. toplankton dominant in situ), and on the population growth of the larger (rarer) phyReferences toplankton species? The enhanced NC& upAHLGREN, G. 1987. Temperature functions in biology take upon Fe addition in short assays (e.g. and their application to algal growth constants. Morel et al. 199 1) suggests effects on phyOikos 49: 177-190. toplankton physiology; the chlorophyll in- ANTIA, N. J., C. D. MCALLISTER, T. R. PARSONS, K. STEPHENS,AND J.D.H. STRICKLAND. 1963. Furcrease in the controls of the published assays ther measurements of primary production using a indicates grazing effects. large-volume plastic sphere. Limnol. Oceanogr. 8: These issues can be tackled by species166-183. BANSE, K. 1982. Cell volumes, maximal growth rates specific work with small containers, either of unicellular algae and ciliates, and the role of directly or (for large zooplankton) by the ciliates in the marine pelagial. Limnol. Oceanogr. process of elimination. For example, we can 27: 1059-107 1. determine the uptake of a compound or ra- --. 199 1. Iron availability, nitrate uptake, and dioisotope into individual cells (cf. overexportable new production in the subarctic Pacific. J. Geophys. Res. 96: 741-748. view by Furnas 1990) or species-specific cell BIENFANG, P. K., AND M. TAKAHASHI. 1983. Ultradivision rates (e.g. frequency-of-dividing plankton growth rates in a subtropical ecosystem. cells, FDC). Although measurements of upMar. Biol. 76: 213-218. take rates will require incubations and be BRAND, L. E., AND R. R. L. GUILLARD. 1981. The effects of continuous light and light intensity on subject to the uncontrollable death of cells the reproduction rates of twenty-two species of in the containers, the incubations using isomarine phytoplankton. J. Exp. Mar. Biol. Ecol. 50: topes do not have to extend for a week and 119-132. therefore eliminate the errors arising from -,W.G.SUNDA,AND R.R.L. GUILLARD. 1983. Limitation of marine phytoplankton reproductive species succession in controls, which differs rates by zinc, manganese, and iron. Limnol. from that in treatments. Brief incubation Oceanogr. 28: 1182-l 198. holds also for FDC; this method has the BUMA, A. G. J., H. J. W. DE BAAR, R. F. NOLTING, additional advantage that cell division rates AND A. J.VAN BENNEKOM. 1991. Metalenrichin the controls can be checked against those ment experiments in the Weddell-Scotia Seas: Effects of iron and manganese on various plankton prevailing in the sea at the same time and communities. Limnol. Oceanogr. 36: 1865-l 878. thus serve as a test for direct container efCULLEN, J. J. 1991. Hypotheses to explain high-nufects on phytoplankton. trient conditions in the open sea. Limnol. OceanThe appeal of determining species-speogr. 36: 1578-1599. cific cell division rates is that the alternaM. R.LEWIS,C.O.DAVIS,AND R.T. BARBER. --3 1992. Photosynthetic characteristics and estispecies (or diatives, ‘“median-large-size 1897 Phytoplankton division rates mated growth rates indicate grazing is the proximate control of primary production in the equatorial Pacific. J. Geophys. Res. 97: 639-654. DE BAAR, H. J. W., AND OTHERS. 1990. On iron limitation of the Southern Ocean: Experimental observations in the Weddell and Scotia Seas. Mar. Ecol. Prog. Ser. 65: 105-122. DOWNS,J. N., AND C. J. LORENZEN. 1985. Carbon : pheopigment ratios of zooplankton fecal pellets as an index of herbivorous feeding. Limnol. Oceanogr. 30: 1024-l 036. DUGDALE, R. C., AND F. P. WILKERSON. 1990. Iron addition experiments in the Antarctic: A reanalysis. Global Biogeochem. Cycles 4: 13-19. EPPLEY, R. W. 1972. Temperature and phytoplankton growth in the sea. Fish. Bull. 70: 1063-1085. -, AND OTHERS. 197 1. Phytoplankton growth and composition in shipboard cultures supplied with nitrate, ammonium, or urea as the nitrogen source. Limnol. Oceanogr. 16: 74 l-75 1. FIALA, M., AND L. ORIOL. 1990. Light-temperature interactions in the growth of Antarctic diatoms. Polar Biol. 10: 629-636. FROST, B. W. 199 1. The role of grazing in nutrientrich areas of the open sea. Limnol. Oceanogr. 36: 1616-1630. FURNAS, M. J. 1982. Growth rates of summer nanoplankton (< 10 pm) populations in lower Narragansett Bay, Rhode Island, USA. Mar. Biol. 70: 105-l 15. -. 1990. In situ growth rates of marine phytoplankton: Approaches to measurement, community and species growth rates. J. Plankton Res. 12: 1117-1151. 199 1. Net in situ growth rates of phytoplank-. ton in an oligotrophic, tropical shelf ecosystem. Limnol. Oceanogr. 36: 13-29. GEIDER, R. J., T. PLATT, AND J. A. RAVEN. 1986. Size dependence of growth and photosynthesis in diatoms: A synthesis. Mar. Ecol. Prog. Ser. 30: 93104. GIFFORD, D. J. 1988. Impact of grazing by microzooplankton in the Northwest Arm of Halifax Harbour, Nova Scotia. Mar. Ecol. Prog. Ser. 47: 249-258. GILSTAD, M., AND E. SAKSHAUG. 1990. Growth rates of ten diatom species from the Barents Sea at different irradiances and day lengths. Mar. Ecol. Prog. Ser. 64: 169-173. GOLDMAN, J. C., AND E. J. CARPENTER. 1974. A kinetic approach to the effect of temperature on algal growth. Limnol. Oceanogr. 19: 756-766. HUDSON, R. J. M., AND F. M. M. MOREL. 1990. Iron transport in marine phytoplankton: Kinetics of cellular and medium coordination reactions. Limnol. Oceanogr. 35: 1002-l 020. JACQUES, G. 1983. Some ecophysiological aspects of the Antarctic phytoplankton. Polar Biol. 2: 27-33. AND M. MINAS. 198 1. Production primaire dans le secteur indien de I’odan Antarctique en fin d’ete. Oceanol. Acta 4: 33-4 1. JAHNKE, J., U. H. BROCKMANN, L. ALETSEE, AND K. D. HAMMER. 1983. Phytoplankton activity in enclosed and free marine ecosystems in a southern Norwegian fjord during spring 1979. Mar. Ecol. Prog. Ser. 14: 19-28. KUIPER, J., U. H. BROCKMANN, H. VAN HET GROENEWO~D,G.HOORNSMAN,ANDK. D.~MER. 1983. Influences of bag dimensions on the development of enclosed plankton communities during POSER. Mar. Ecol. Prog. Ser. 14: 9-17. LANDRY, M. R., L. W. HAAS, AND V. L. FAGERNESS. 1984. Dynamics of microbial plankton communities: Experiments in Kaneohe Bay, Hawaii. Mar. Ecol. Prog. Ser. 16: 127-133. LANGDON, C. 1988. On the causes of interspecific differences in the growth-irradiance relationship for phytoplankton. 2. A general review. J. Plankton Res. 10: 1291-1312. LAWS, E. A. 1984. Improved estimates of phytoplankton carbon based on 14Cincorporation into chlorophyll a. J. Theor. Biol. 110: 425434. -, G. R. DITULLIO, AND D. G. REDALJE. 1987. High phytoplankton growth and production rates in the North Pacific subtropical gyre. Limnol. Oceanogr. 32: 905-9 18. -, PND OTHERS. 1984. High phytoplankton growth and production rates in oligotrophic Hawaiian coastal waters. Limnol. Oceanogr. 29: 116 l1169. LONGHURST, A. R. 1991. Role of the marine biosphere in the global carbon cycle. Limnol. Oceanogr. 36: 1507-l 526. MCALLISTER, C. D., T. R. PARSONS, K. STEPHENS, AND J. D. H. STRICKLAND. 1961. Measurements of primary production in coastal sea water using a large-volume plastic sphere. Limnol. Oceanogr. 6: 237-258. MARTIN, J. H., S. E. FITZWATER? AND R. M. GORDON. 1990a. Iron deficiency limits phytoplankton growth in Antarctic waters. Global Biogeochem. Cycles 4: 5-12. -, R. M. GORDON, AND S. E. FITZWATER. 1990b. Iron in Antarctic waters. Nature 345: 156-l 58. , AND W. W. BROENKOW. 1989. --vVERTEX: Phytoplankton/iron studies in the Gulf of Alaska. Deep-Sea Res. 36: 649-680. MOREL, F. M. M., R. J. M. HUDSON, AND N. M. PRICE. 199 1. Limitation of productivity by trace metals in the sea. Limnol. Oceanogr. 36: 1742-1755. PARANJAPE, M. A. 1987. Grazing by microzooplankton in the eastern Canadian arctic in summer 1983. Mar. Ecol. Prog. Ser. 40: 239-246. REY, F., AND H. R. SKJOLDAL. 1987. Consumption of silicic acid below the euphotic zone by sedimenting diatom blooms in the Barents Sea. Mar. Ecol. Prog. Ser. 36: 307-3 12. RICH, H. W., AND F. M. M. MOREL. 1990. Availability of well-defined iron colloids to the marine diatom Thalassiosira weissflogii. Limnol. Oceanogr. 35: 652-662. RIEMANN, B., T. G. NIELSEN, S. J. HORSTED, P. K. BJORNSEN, AND J. POCK-STEEN. 1988. Regulation of phytoplankton biomass in estuarine enclosures. Mar. Ecol. Prog. Ser. 48: 205-2 15. RIVKIN, R. B., AND M. PUTT. 1987. Photosynthesis and cell division by Antarctic microalgae: Comparison of benthic, planktonic and ice algae. J. Phycol. 23: 223-229. 1898 SAKSHAUG, E., AND 0. HOLM-HANSEN. Banse 1986. Photoadaptation in Antarctic phytoplankton: Variations in growth rate, chemical composition and P versus I curves. J. Plankton Res. 8: 459-473. SCHLESINGER,D.A.,L. A. MOLOT, AND B.J. SHUTER. 198 1. Specific growth rates of freshwater algae in relation to cell size and light intensity. Can. J. Fish. Aquat. Sci. 38: 1052-1058. SEMINA, G. I. 1962. Phytoplankton from the central Pacific collected along the meridian 174”W. 1. Methods and taxonomy [in Russian]. Tr. Inst. Okeanol. 58: 3-26. -” 1963. The phytoplankton along the meridian 174’W in the central Pacific. 2. Horizontal distribution and abundance [in Russian]. Tr. Inst. Okeanol. 71: 5-2 1. SOMMER, U. 1986. Nitrate- and silicate-competition among antarctic phytoplankton. Mar. Biol. 91: 345351. -. 1989. Maximal growth rates of Antarctic phytoplankton: Only weak dependence on ccl1size. Limnol. Oceanogr. 34: 1109-l 112. -, AND H.-H. STABEL. 1986. Near surface nutrient and phytoplankton distribution in the Drake Passageduring early December. Polar Biol. 6: 107110. SPIES, A. 1987. Growth rates of Antarctic marine phytoplankton in the Wcddell Sea. Mar. Ecol. Prog. Ser. 41: 267-274. STROM,~. L., AND N. A. WELSCHMEYER. 1991. Pigment-specific rates of phytoplankton growth and microzooplankton grazing in the open subarctic Pacific Ocean. Limnol. Oceanogr. 36: 50-63. SUNDA, W. G., D. G. SWIFT, AND S. A. HIJNTSMAN. 199 1. Low iron requirement for growth in oceanic phytoplankton. Nature 351: 55-57. TAKAHASHI, K., J. D. BILLINGS, AND J. K. MORGAN. 1990. Oceanic province: Assessment from the time series diatom fluxes in the northeastern Pacific. Limnol. Oceanogr. 35: 154-l 65. TAYLOR,G. T., AND P. R. HABERSTROH. 1988. Microzooplankton grazing and planktonic production in the Bransfield Strait observed during the RACER program. Antarctic J. U.S. 23(5): 126128. WELSCHMEYER, N., R. GOERICKE, S. STROM, AND W. PETERSON. 199 1. Phytoplankton growth and herbivory in the subarctic Pacific: A chemotaxonomic analysis. Limnol. Oceanogr. 36: 163 l-l 649. --, AND C. J. L~RENZEN. 1984. Carbon-14 labeling of phytoplankton carbon and chlorophyll a carbon: Determination of specific growth rate. Limnol. Oceanogr. 29: 135-145. WILKERSON, F. P., AND R. C. DUGDALE. 1987. The use of large shipboard barrels and drifters to study the effects of coastal upwelling on phytoplankton dynamics. Limnol. Oceanogr. 32: 368-382. ZENTARA, S.-J., AND D. KAMYKOWSKI. 1977. Latitudinal relationships among temperature and selected plant nutrients along the west coast of North and South America. J. Mar. Res. 35: 321-337.
© Copyright 2026 Paperzz