Marine Micropaleontology 44 (2002) 163^185 www.elsevier.com/locate/marmicro Techniques for quantitative analyses of calcareous marine phytoplankton Jo«rg Bollmann a; , Mara Y. Corte¤s b , Ali T. Haidar c , Bernhard Brabec d , Anne Close e , Robert Hofmann a , So¢a Palma f , Luis Tupas g , Hans R. Thierstein a a Geological Institute, ETH Zurich, Sonneggstr. 5, 8092 Zurich, Switzerland Departemento de Geolog|¤a Marina, UABCS, Apdo. Postal 19-B, C.P. 23080 La Paz, B.C.S., Mexico c Department of Geology, American University of Beirut, P.O. Box 11-0236/26, Beirut, Lebanon d Swiss Federal Institute for Snow and Avalanche Research, Flu«elastrasse 11, 7260 Davos Dorf, Switzerland e Wrigley Institute for Environmental Studies, USC, AHF 232, Los Angeles, CA 900089-03171, USA f IPIMAR, Lab. Fitopla“ncton, Dep. Ambiente Aqua¤tico, Avenida de Bras|¤lia, 1449-006 Lisboa, Portugal Biological Oceanography Division of SOEST, University of Hawaii at Manoa, 1000 Pope Road, Honolulu, HI 96822, USA b g Received 7 February 2001; received in revised form 19 July 2001; accepted 10 August 2001 Abstract This paper discusses the techniques used to sample and analyse living marine calcareous phytoplankton. The various methods are described and tested within several research projects aimed at the determination of coccolithophore cell densities in seawater. In addition, the potential advantages and drawbacks associated with the application of light and scanning electron microscopic techniques to the quantitative analysis of coccolithophores are discussed. Several tests have been carried out in order to quantify potential errors related to: (1) homogeneity of material distribution on filter membranes; (2) use of different microscopes (scanning electron microscope versus light microscope) ; (3) use of different filter membranes (cellulose mixed-ester membranes versus polycarbonate membranes); and (4) Utermo«hl settling versus filtration method. These tests revealed that major errors in cell density calculations could result from the uneven distribution of coccolithophore specimens on a filter membrane. The error resulting from the use of a light microscope arises from its low resolution, which restricts the identification of species, especially of small coccospheres. The use of different filter membranes does not show a statistically significant difference in cell density calculations, although polycarbonate membranes can be examined much more efficiently with the scanning electron microscopy than cellulose mixed-ester membranes. The Utermo«hl method, however, gives lower cell densities consistently (several times) than the filtration method. : 2002 Elsevier Science B.V. All rights reserved. Keywords: microscopy; plankton surveys; plankton collection devices; water ¢ltration; coccolithophores 1. Introduction * Corresponding author. Tel.: +41-1-632-3684; Fax: +41-1-632-1080. E-mail address: [email protected] (J. Bollmann). In recent years, there has been an increasing interest in living calcareous marine phytoplankton 0377-8398 / 02 / $ ^ see front matter : 2002 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 7 - 8 3 9 8 ( 0 1 ) 0 0 0 4 0 - 8 MARMIC 854 3-5-02 164 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 because of their role within the global carbon cycle. In order to assess their contribution to the global carbon cycle, absolute abundance determinations (cells/l) are needed. There are two basic methods for quantifying the abundance of calcareous marine phytoplankton: (a) the Utermo«hl settling method (Utermo«hl, 1931, 1958); and (b) the ¢ltration method (McIntyre and Be¤, 1967; Okada and Honjo, 1973). Samples prepared using the Utermo«hl method can only be analysed by light microscopy (LM), whereas ¢lter preparations are suitable for both LM and scanning electron microscopy (SEM). The application of these two methods depends on the purpose of the study. The Utermo«hl method has been widely used to quantify total phytoplankton composition in water samples, whereas ¢lter preparations are mainly used to analyse calcareous marine phytoplankton. A general description of how to sample and analyse phytoplankton can be found in Sournia (1978). The use of di¡erent methods, ¢lter funnels, microscopes or ¢lter membranes may result in di¡erent cell density estimates. In this study, we have conducted several investigations to address this potential error by focussing on: (a) the homogeneity of material on ¢lter membranes; (b) the application of di¡erent microscopes (SEM versus LM); (c) the use of di¡erent ¢lter membranes (cellulose mixed-ester membranes versus polycarbonate membranes) ; and (d) the application of the Utermo«hl method versus the ¢ltration method. We also give recommendations on how to collect and analyse samples for optimal reproducibility, as well as discussing the major sources of error involved in the calculation of cell density. In addition, we provide detailed instructions on how to build a device for water sample ¢ltration. 2. Materials and methods The samples used in this study were collected at the JGOFS time series stations in Hawaii (HOT), in Bermuda (BAT) and during the EC-MAST III project CANIGO o¡ the Canary Islands. Details of all samples are given in Table 1. 2.1. Settling method (after Utermo«hl, 1931, 1958) Thirty millilitres of 20% hexamethyltetraminebu¡ered formalin (pH 7.5) were mixed in dark plastic bottles with 250 ml of seawater transferred from Niskin bottles (after Throndsen, 1978). These bottles were stored in the dark prior to analysis. For analysis, 100 ml sub-samples were introduced into a settling chamber and the phytoplankton was allowed to settle for 72 h (Margalef, 1969). Phytoplankton cells were then identi¢ed and counted by the Utermo«hl method using a Zeiss0 IM35 inverted microscope equipped with phase contrast and bright¢eld illumination (Hasle, 1978). 2.1.1. Counting Relatively large coccolithophore cells were identi¢ed and counted in alternate ¢elds of the settling chamber, at a magni¢cation of 160U. This corresponds to 50 ml of seawater analysed with a detection limit of 60 cells/l at a 95% con¢dence level. Where necessary, species identi¢cation was con¢rmed at a magni¢cation of 400U or 1000U. Relatively small species, such as Gephyrocapsa oceanica and Emiliania huxleyi, were counted and identi¢ed at a magni¢cation of 400U in 64 squares. This corresponds to 1 ml of seawater analysed with a detection limit of 3000 cells/l at a 95% con¢dence level. 2.2. Filtration Up to 10 l of seawater was ¢ltered on a 47-mmdiameter cellulose mixed-ester membrane ¢lter (Millipore0 ) or polycarbonate membrane ¢lter (Nucleopore0 ) using Nalgene0 or Gelman0 1119 inline ¢lter gaskets (for details of samples see Table 1). Each ¢lter membrane was rinsed with NH4 -bu¡ered distilled water (pH 8.5) immediately after ¢ltration in order to remove all traces of sea salt. All membranes were stored in plastic Petri dishes and dried in an oven at 40^60‡C for several hours. The seawater was ¢ltered with an accuracy of P 0.1 l. A sketch of the ¢ltration apparatus used for these studies is shown in Fig. 1. For light microscope analyses a piece of ¢lter membrane cut along its radius was mounted onto MARMIC 854 3-5-02 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 165 Table 1 Details of all samples used in this study Sample Filter type Filtration device Test material distribution HS715-25-1 M 1 HS715-25-2 M 1 HS715-25-3 M 1 HS715-25-4 M 1 HOT52-2-15-17 M 2 M 2 M 2 HOT53-2-19-6 M 2 M 2 M 2 HOT52-2-15-12 M 2 M 2 M 2 HOT53-2-19-9 M 2 M 2 M 2 Test SEM versus LM HS717 M 1 M 1 HS745 M 1 M 1 HOT57-75 M 2 M 2 HOT52 M 2 M 2 Test MF versus PC HOT61-50 N 1 M 1 HOT61-75 N 1 M 1 HOT61-100 M 1 N 1 Test ¢ltration versus Utermo«hl M37-31-31-0m N 3 M37-31-31-10m N 3 M37-31-31-25m N 3 M37-31-31-50m N 3 M37-31-31-75m N 3 M37-31-31-100m N 3 M37-31-31-150m N 3 Method Amount of water ¢ltered (l) Amount of water analysed (ml) Number of specimens Cells/l Number of species Worker LM LM LM LM LM LM LM LM LM LM LM LM LM LM LM LM 4 4 4 4 8.6 8.6 8.6 8.7 8.7 8.7 8.7 8.7 8.7 8.7 8.7 8.7 3.1 3.1 3.1 3.1 15.9 15.9 15.9 16.1 16.1 16.1 16.1 16.1 16.1 16.1 16.1 16.1 320 376 343 448 241 238 290 299 309 248 81 66 81 35 48 44 101 780 119 592 109 095 142 492 151 73 149 85 182 58 18 522 19 142 15 363 5 018 4 089 5 018 2 168 2 973 2 726 ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ A A A A B B B B B B B B B B B B LM SEM LM SEM LM SEM LM SEM 3.5 3.5 4 4 8.5 8.5 8.6 8.6 3.1 6.7 11.8 11.4 7.7 12.4 15.6 13.5 306 369 302 366 303 629 305 306 99 504 55 312 25 615 32 113 39 357 50 537 19 519 22 689 5 3 16 36 13 27 18 20 A B A B A B A B SEM SEM SEM SEM SEM SEM 8.8 8.8 8.8 8.8 8.8 8.8 25.8 25.8 25.8 25.8 12.9 12.9 1062 954 946 1008 535 522 41 185 36 997 36 687 39 091 41 602 40 591 ^ ^ ^ ^ ^ ^ B B B B B B SEM Utermo«hl SEM Utermo«hl SEM Utermo«hl SEM Utermo«hl SEM Utermo«hl SEM Utermo«hl SEM Utermo«hl 4 6.8 51.0 6.2 51.0 6.2 51.0 3.4 51.0 10.2 51.0 6.8 51.0 8.0 51.0 399 33 777 3 933 35 428 24 1147 34 467 18 91 6 58 676 14 380 125 323 3 000 150 484 10 500 125 882 6 360 112 451 7 540 68 676 5 260 11 375 2 070 15 12 26 2 33 9 21 10 30 6 17 5 7 3 C D C D C D C D C D C D C D 4 4 4 4 4 4 HS samples were collected at the JGOFS time station Bermuda, 32‡10PN, 64‡30PW and HOT samples were collected at the JGOFS time station ALOHA, 22‡45PN, 158‡0PW. M37 samples were collected on R/V Meteor at the location 28‡51PN, 13‡56PW. M = Millipore0 MF cellulose mixed ester membrane; N = Nucleopore0 polycarbonate membrane. Filtration device: 1 = Gelman0 1119, 2 = Nalgene0 330^4000, 3 = Gelman0 1119 modi¢ed as described in Appendix A.2.1; LM = light microscope; SEM = scanning electron microscope. Worker: A, B, C, D indicate the scientist who has analysed the sample. MARMIC 854 3-5-02 166 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 Fig. 1. Filtration apparatus used at JGOFS time series station Bermuda, Hawaii and during the CANIGO project. (A) 10-l carboys. (B) 47-mm-diameter inline ¢lter holders. (C) Stopcock valve. (D) Vacuum pump. (E) Large water trapping steel tank. a glass slide using Canada Balsam and ¢xed beneath a cover slip. A similar sized piece of ¢lter membrane was mounted onto an aluminium stub using carbon tape and coated with 15 nm of gold for subsequent analysis in the SEM. 2.2.1. Counting Cell counts were carried out with a Zeiss0 polarising light microscope using 63U or 100U objectives and a 1.25U Optovar lens. The area represented by one ¢eld of view is 0.0389 mm2 and 0.0154 mm2 , respectively, and the total observed area (52 ¢elds of view along the radial cut) is 2 mm2 and 0.8 mm2 (see also Haidar and Thierstein, 2001). This corresponds to 7.86 and 3.13 ml of seawater analysed, if 4 l were ¢ltered using the Gelman0 1119 inline ¢lter gasket. SEM counts of the HOT and the BAT samples were performed with a Hitachi0 S2300 SEM equipped with a computer controlled stage at a magni¢cation of 1500U and 3000U. Calculation of the observed ¢ltration area relies on the positioning accuracy of the stage and the accuracy with which the area of one ¢eld of view can be calculated. The area of observation is the sum of the area of all single ¢elds of view. This is easily estimated and controlled with LM, but it is more di⁄cult to calculate and control in the SEM. In both cases, it is important that single ¢elds of view do not overlap to prevent double counting of specimens. Using a computer-driven stage makes it possible to control the size of the total observed area, to prevent overlapping and, therefore, avoiding the double counting of an area. The following set-up was used for the HOT samples (Fig. 2). A total area of 3.88 mm2 was analysed along a transect from the centre to the edge of each ¢lter membrane. This corresponds to 26 ml of seawater analysed. Forty equidistant ¢elds of observation, each 0.097 Wm2 in size, were analysed manually along this transect. Each ¢eld of observation consisted of 81 single monitor screens, each 1200 Wm2 in size (as calibrated with a standard metal grid, 10 P 0.2 Wm), at a magni¢cation of 3000U or 20 single monitor screens, each 4800 Wm2 in size (total observed area of 7.7 mm2 ), captured at a magni¢cation of 1500U (Fig. 2, see also Corte¤s et al., 2001). This corresponds to 52 ml of the seawater sample. One ¢eld of observation at 3000U magni¢cation is equivalent to approximately two ¢elds of view using a light microscope with a 63U objective and an 8U ocular. Coccolithophore counting of the CANIGO samples was carried out with a computer-controlled stage and a fully automated image-capturing system. A remotely controlled Philips0 XL 30 scanning electron microscope in combination with the imaging software analySIS0 3.0 was used to capture and store all images collected along a prede¢ned transect (Fig. 2). The system, which captures about 1000 images/h, was used to acquire MARMIC 854 3-5-02 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 167 A 95% con¢dence interval was calculated by assuming a Poisson distribution and applying the equation : AC L;U CDL;U ¼ ð2Þ av where : CDL;U = lower, upper con¢dence limit of the cell density based on the Poisson distribution and CL;U = lower, upper con¢dence limit for a Poisson distribution. qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 CL ¼ ðN þ 0:982 Þ30:98 ð3Þ Fig. 2. Schematic representation of the counting method used in this study (modi¢ed after Corte¤s et al., 2001). (A) A sector of the ¢lter is mounted on an aluminium stub. The count direction is from U1 to U2. (B) Forty ¢elds of observation were analysed. Each ¢eld consisted of 9U9 monitor screens at a magni¢cation of U3000 or 5U4 screens at U1500. (C) The area of a single screen is 30U40 Wm or 60U80 Wm, respectively. images from 135 ¢lter samples and enables nearly optimal utilisation of the SEM. Approximately 2100 images (total area of 2.5 mm2 ) were captured from each sample at a magni¢cation of 2000U and all images were stored on CD-ROM. Coccolithophore cell densities were subsequently determined by manual examination of single images using a PC. The amount of analysed seawater varied between 5 and 20 ml. CU ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 ðN þ 0:982 Þ þ 0:98Þ ð4Þ The cell density and the lower and upper con¢dence limit of each species was calculated by replacing N by the number of individuals of each species in Eqs. 1^4. 2.2.3. Detection limits The detection limit of a species is ¢nding at least one cell of a species in a de¢ned volume of water, and depends on both the total amount of seawater analysed and the total cell density. It can be calculated from a Poisson distribution as follows: P6X ¼ 0s ¼ expð3 M Þ ð5Þ where 2.2.2. Cell density calculations A detailed description of the number of cells to count and the type of distribution which can be expected (e.g. Poisson or Normal distribution), was presented by Venrick (1978a, 1978b and references therein). The number of coccolithophore cells in 1 l of water was calculated using the following equation: AN ð1Þ CD ¼ av where CD = cell density (cells/l water) ; A = ¢ltration area ; N = total number of cells counted ; a = analysed area ; and v = volume of water ¢ltered (l). M ¼ xy where x = number of cells/l and y = amount of sample analysed. Thus, the probability at any given M of ¢nding one or more cells is: 13P6 M s0 ð6Þ For example, if in a given amount of sample one or more cells should be detected at 95% probability level. 13expð3 M Þs0:95 MARMIC 854 3-5-02 ð7Þ 168 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 3. Results 3.1. Material distribution on a ¢lter membrane Fig. 3. Detection limits (DTL) for samples with 10^10 000 cells/l and an analysed volume of 1^100 ml. The detection limit is about 3000 cells/l at a 95% probability level if 0.001 l was analysed or 30 cells if 0.1 l was analysed. From this follows for M (see Fig. 3): M s3lnð0:05Þ ð8Þ and similarly for other probabilities 99% : M s3lnð0:01Þ 99:9% : M s3lnð00:01Þ ð9Þ ð10Þ The detection limit is about 3000 cells/l at a 95% probability level if a volume of 0.001 l is analysed, or 30 cells if 0.1 l is analysed. Most ¢ltration devices produce an uneven distribution of material. Although these patterns are sometimes obvious to naked eye, they are often overlooked because of the thin ¢lm of material on the membrane. In order to assess the distribution of material on membranes, we put membranes collected with di¡erent ¢ltration devices on a light table and applied digital contrast enhancement to the captured images (Table 2, Fig. 4). The observed pattern of uneven material distribution can be attributed to: (a) the structure of the support sieve (Fig. 4AI, AII, BI, BII); (b) air locking of parts of the ¢lter membrane (Fig. 4AIII, BII, BIII) ; (c) improper ¢lter handling (Fig. 4BIII); and (d) di¡erent £ow velocities of the ¢ltered water in a ¢lter funnel or gasket (Fig. 5). Air locking occurs in every low vacuum ¢ltration system as the attached vacuum ( 6 200 mm Hg) is too weak to remove air bubbles trapped above the ¢lter membrane and below or inside the ¢lter support (Fig. 4BII) and membrane. The former occurs especially in ¢lter gaskets without an evacuation valve (Fig. 4AIII), while the latter occurs in every type of ¢ltration device and especially if the ¢lter membrane was not properly moisturised. These air bubbles clog up parts of Fig. 4. Support sieves and related material distribution of several ¢ltration devices tested. (AI) Nalgene0 330^4000 inline ¢lter support (same type as Millipore Swinnex0 47-mm-diameter inline ¢lter support. (AII) Millipore0 cellulose mixed-ester membrane ¢ltered with Nalgene0 330^4000 47-mm-diameter inline ¢lter gasket. The membrane shows dark dots with an increased amount of material. In addition, the lower part of the membrane was clogged by a large air bubble above the membrane. (AIII) Millipore0 cellulose mixed-ester membrane ¢ltered with Nalgene0 330^4000 47-mm-diameter inline ¢lter. The binarised image shows that large areas (white) were clogged up by air bubbles above the membrane. (BI) Gelman0 1119 inline ¢lter plastic support. (BII) Nucleopore0 polycarbonate membrane used with Gelman0 1119 inline ¢lter gasket and a plastic support. The membrane re£ects the gutter structure of the plastic ¢lter support and small areas of the membrane were clogged up by air bubbles inside the support sieve. (BIII) Nucleopore0 polycarbonate membrane used in a Gelman0 1119 inline ¢lter gasket and a plastic ¢lter support. The material distribution was disturbed because the gasket was accidentally ¢lled with water while opening it. (CI) Gelman0 1119 inline ¢lter metal grid support. (CII) Nucleopore0 polycarbonate membrane used in a modi¢ed Gelman0 1119 inline ¢lter gasket and metal grid support combined with a Millipore0 cellulose mixed-ester membrane for improved drainage. The membrane shows a homogenous material distribution; however, the distribution was disturbed partly by a large gelatinous object. (CIII) Nucleopore0 polycarbonate membrane used in a modi¢ed Gelman0 1119 inline ¢lter gasket equipped with metal grid support combined with a Millipore0 cellulose mixed-ester membrane for improved drainage. The membrane shows homogenous material distribution. Note: for all images, the contrast was enhanced by using a light table and digital contrast enhancement. MARMIC 854 3-5-02 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 the ¢ltration area and therefore, the ¢ltration area is reduced, leading to potentially erroneous cell density determinations (Fig. 4AIII,BII). A second cause of uneven distribution on ¢lter membranes is the texture/structure of the support sieve (Fig. 4AI,BI,CI). Rings, dots (Fig. 4AII) or gutter structures (Fig. 4BII) are occasionally observed patterns in the resulting ¢lter materials (Table 2). 169 The last cause of uneven material distribution is related to variation in the £ow velocity of ¢ltered water within a ¢lter funnel or inline ¢lter gasket (Fig. 5). The number of cells per ¢eld along radial transects of membranes used in two di¡erent inline ¢lter gaskets (47 mm diameter, Gelman0 1119 and Nalgene0 330^4000), occasionally show a signi¢cant decrease from the edge to the centre (Fig. 5). This is an indication that the £ow MARMIC 854 3-5-02 170 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 Fig. 5. Variation of cell counts per ¢eld of view in three radial transects of one ¢lter for di¡erent cell densities. In the top row, a sketch shows how the analyses were carried out. Cell densities decrease from top to bottom of the graph. The solid line smoothes the individual analyses and represents the locally weighted regression scatter plot smoothing with a 20% span. All analyses were done with a light microscope. Data are from Corte¤s (1998) and Haidar (1997). velocity of ¢ltered water varies from the inner to the outer part of the ¢lter gasket. This pattern seems to become more evident with increasing total cell density (Fig. 5). The calculations of the cell densities with a 95% con¢dence interval for each analysed transect show no statistically significant di¡erence between transects of one ¢lter for the samples with low and middle cell densities, if a MARMIC 854 3-5-02 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 171 Table 2 List of all ¢ltration devices tested Brand Filter Type Support sieve Order number Nalgene0 Gelman0 Gelman0 Schleicher and Schuell0 Millipore0 Millipore0 Millipore0 Millipore0 Bioblock0 47 mm 47 mm Modi¢ed 47 mm FP 050/1 Swinnex 25 mm Swinnex 47 mm Pyrex 47 mm Pyrex 47 mm 47 mm Inline Inline Inline Inline Inline Inline Funnel Funnel Funnel Radial plastic Plastic grid Metal grid Radial plastic Radial plastic Radial plastic Fritted glass Steel grid Fritted steel 330^4000 1119 1119 461 300 SX0002500 SX0004700 XX1004700 XX1004730 C17303 complete transect was counted. The mean values vary from P 11 to 16% of each sample with respect to the mean (Fig. 6). Cell density calculations from the ¢lter with high cell density, however, exhibit statistically signi¢cant di¡erences between transects 715-25-1 and 715-25-4 (Fig. 6) which is probably the result of a clogged membrane. 3.2. SEM versus light microscope counts The use of di¡erent microscopes may result in di¡erent cell density values and species compositions for the same sample. This hypothesis was tested with four di¡erent samples, one with high cell densities (HS717), two with high species richness (HOT57-75, HS745), and one with a high proportion of small cells ( 6 2.5 Wm) (HOT52). Each sample was analysed with both the light microscope and a SEM. Species richness and assemblage composition determined with both microscopes are similar for samples with high cell densities or small cells (HS717 and HOT52, Table 1, Appendix A, Section A.1). However, the number of species obtained in the high diversity samples HS745 and HOT57-75 was much higher with the SEM than with the light microscope (Table 1, Appendix A, Section A.1). This is due to the occurrence of syracosphaerids, small reticulofenestrids, small gephyrocapsids and holococcolithophores, which could be identi¢ed with greater certainty using the SEM rather than with the light microscope. However, the occasionally large di¡erences in total cell density (Fig. 7) result from a combination of two e¡ects: (a) di¡erent de¢nitions of Fig. 6. Coccolithophore cell densities determined from at least three radial counting transects for the four samples shown in Fig. 5. Error bars show the 95% con¢dence interval for a Poisson distribution. MARMIC 854 3-5-02 172 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 Fig. 7. Mean cell densities for four di¡erent ¢lters each analysed with the light microscope and SEM, respectively. Error bars show the 95% con¢dence interval for a Poisson distribution. what constitutes a coccosphere ; and (b) the lower recognition rate of small cells with the light microscope. The higher cell density obtained with the SEM from sample HS717 can be explained by the ¢rst point. A re-examination of the same sample in light microscope and SEM showed that there are many single placoliths of Emiliania huxleyi lying relatively close together. But this impression varies from light microscope to SEM and it turned out that this sample was analysed using two di¡erent concepts of what constitutes a coccosphere. This may also be the reason for the di¡erent cell densities recorded for Umbellosphaera tenuis and Umbellosphaera irregularis in sample HOT57-75. The limited recognition rate of small cells with the light microscope is demonstrated by samples HS745, HOT57-75 and HOT52. A di¡erence in cell density of up to 23% (e.g. HS745) between the light microscope and the SEM counts can be explained by larger proportions of the small taxa Emiliania huxleyi, Gephyrocapsa ericsonii and Gephyrocapsa protohuxleyi which were determined with the SEM (for details see Appendix A, Section A.1). 3.3. Counts on di¡erent ¢lter membranes The most commonly used ¢lters are cellulose mixed-ester membranes (Millipore0 MF HAWP) and polycarbonate membranes (Nucleopore0 PC111109 or Millipore0 ISOPORE ATTP). Cellulose mixed-ester membranes are widely used for light microscope studies because they appear transparent in smear slides; however, they are not optimal for SEM analysis because of their irregular surface. In this respect, polycarbonate membranes are much more suited to the analysis of coccolithophores with the SEM because a £at surface is advantageous for the identi¢cation of small species and therefore much less tiring for the SEM operator. They are, however, not opti- MARMIC 854 3-5-02 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 Fig. 8. Comparison between cell density determinations using Millipore0 cellulose mixed-ester membranes and Nucleopore0 polycarbonate membranes for the same sample. In total, three di¡erent samples are shown. N = Nucleopore0 polycarbonate membranes; M = Millipore0 cellulose mixedester. Error bars show the 95% con¢dence interval for a Poisson distribution. Data from Corte¤s (1998). mal for light microscope analysis due to a slight darkening in transmitted light. The sti¡er cellulose-mixed ester membranes are much easier to handle than the polycarbonate membranes that are prone to cling and wrinkle. The identi¢cation of small cells and of holococcolithophores on the irregular-structured cellulose mixed-ester membranes is clearly more di⁄cult than on the £at polycarbonate membranes, which may result in slightly di¡erent recognition rates and cell densities in SEM studies. In our examples, however, SEM counts on both kinds of ¢lters show no statistically signi¢cant di¡erence (Fig. 8). 3.4. Filtration (SEM) versus Utermo«hl settling method Reid (1980) reported that cell density calculations of small coccospheres obtained with the inverted light microscope technique are probably not reliable. The comparison of seven sample pairs that were analysed both with the ¢ltration (SEM) and the Utermo«hl technique during the EC-MAST III project CANIGO, support this assumption (Table 1, Appendix A, Section A.1). Coccolithophore cell densities obtained with the 173 ¢ltration technique are consistently several times higher than those obtained with the Utermo«hl technique (Table 1), although the sub-samples for the di¡erent analyses were always taken from the same Niskin0 bottle. In most samples, the overall number of species determined with the Utermo«hl technique is lower than with the ¢ltration (SEM) technique. However, in those samples in which the overall number of species is comparable between the two methods, the assemblage composition is quite di¡erent (Appendix A, Section A.1, sample M37-31-31-0m). With the Utermo«hl technique relatively large species with low cell densities are still encountered because the probability of detecting these species is higher (50 ml of analysed water and therefore a detection limit of 60 cells/l, see Fig. 3) than in the SEM (6.8 ml of analysed water and therefore a detection limit of 430 cells/l, see Fig. 3). However, the recognition rate of small and delicate species is much better with the SEM than with the inverted light microscope. Therefore, the large di¡erences in cell density and species composition are likely to result from a combination of: (a) the restricted recognition rate for small species in the light microscope at a magni¢cation of 400U, as most samples analysed were dominated by small G. ericsonii and Emiliania huxleyi cells ; and (b) higher probability of encountering large species using the Utermo«hl technique because of larger volume investigated. However, it remains unexplained why some species such as Florisphaera profunda and Umbellosphaera tenuis were never encountered in the light microscope although they were found frequently in ¢lter preparations. Filtration and subsequent analyses of some Utermo«hl preparations with the SEM suggest that carbonate dissolution might take place during the extended particle settling times, however, this has to be tested in future studies. 4. Conclusion Cell densities obtained with the Utermo«hl method are consistently lower than those obtained with the ¢ltration method because of the restricted recognition rate of species in the light MARMIC 854 3-5-02 174 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 microscope. The major error in cell density calculations using the ¢ltration technique results from the uneven distribution of plankton on ¢lter membranes. This can be attributed to: (a) air locking of parts of the ¢lter membrane; (b) the structure of the support sieve; (c) improper ¢lter handling; and (d) uneven £ow velocities of the ¢ltered water across a ¢lter funnel or gasket. The bias resulting from the use of a light microscope or scanning electron microscope arises because of the restricted recognition rate of species in the light microscope, especially of small coccospheres. In contrast, the use of di¡erent ¢lter membranes does not show a statistically signi¢cant di¡erence in cell density calculations, although polycarbonate membranes can be examined much more e⁄ciently with the SEM than cellulose mixed-ester membranes. 5. Recommendation For optimal reproducibility of coccolithophore analyses, some recommendations are given below. We strongly recommend testing the material distribution of any ¢ltration device before routine application. We suggest using inline ¢lter gaskets, as onboard a research vessel funnel systems are very di⁄cult to use due to the limited amount of seawater that can be ¢ltered (0.5^1 l) before re¢lling the funnel. Furthermore, each re¢ll may disturb the material distribution on the membrane. In addition, funnel systems are relatively large compared to inline ¢lter gaskets. This makes it di⁄cult to build a ¢ltration device that simultaneously ¢lters several samples for high resolution plankton studies (Fig. 1). Fig. 9. Construction plan of a multi-depth level ¢ltration device. All materials used are listed in Appendix A, Section A.2.6. MARMIC 854 3-5-02 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 We suggest using a modi¢ed Gelman0 inline ¢lter gasket or a similar product that has a special valve for releasing the air from the gasket. We recommend replacing the plastic support sieve that has a gutter structure, with a stainless steel support combined with a £eece (e.g. Millipore0 membrane SCWP 04700) to increase the drainage (Fig. 9). Some recommendations concerning the volume of water to be ¢ltered are given in Table 3. In the case of the clogging of a ¢lter membrane by large objects or air bubbles, we recommend calculating the e¡ective area of ¢ltration with help of a light table and digital image processing. If possible, it is advisable to count random ¢elds of view on a given membrane in order to reduce the error in cell density determinations caused by the observed trends in the inline ¢lter gaskets and the potential clogging of the membrane by large objects or air bubbles. However, it is not, as yet, possible to mount a complete 47mm-diameter membrane on a SEM stub or a glass slide. Therefore, we recommend counting a large area on a piece of ¢lter (e.g. a full transect from the centre to the edge of a ¢lter). Light microscope analyses are suitable, if e⁄cient estimates of total coccolithophore cell densities of only the most common species and species groups are required. Most common species are easily recognised with the LM. However, it should be clearly stated for all taxa what constitutes a coccosphere. If there are signi¢cant numbers of dubious cells, additional SEM analysis should be made to calculate the potential error. SEM counts are required for species diversity Table 3 Recommended ¢ltration volumes for 47-mm-diameter ¢lter Season Eutrophic areas Above seasonal thermocline Spring 1^2l Summer 1^4l Fall 1^4l Winter 1^2l Below seasonal thermocline Spring 1^2l Summer 1^4l Fall 1^4l Winter 1^2l Oligotrophic areas 2^4l 6^8l 6^8l 4^6l 175 analysis and for the determination of cell densities of certain taxa, especially small and delicately structured coccolithophores. If a computer-controlled stage is available, the time needed to analyse a sample with the SEM is comparable to that in a light microscope. We recommend the use of polycarbonate membranes (e.g. Nucleopore0 ) rather than the cellulose mixed ester membranes for SEM analysis because of their £at surface. Furthermore, polycarbonate membranes with a pore size of 0.8 Wm instead of 0.45 Wm should be used because their improved drainage. Cellulose mixed-ester membranes, however, are recommended for light microscope analysis because of their transparency. In addition, a pore size of 0.45 Wm instead of 0.8 Wm should be used as small coccospheres may be hidden in large pores if the membrane has to be analysed with the SEM. Acknowledgements We are grateful to Uwe Koy (IFM Kiel) and Andrea Spiedt for their very useful advice and their help during several cruises. Furthermore, we would like to thank Prof. Gerold Siedler, IFM Kiel, for his support during Poseidon cruise 212. We wish to thank L. Fujieki, D. Hebel, R. Johnson, D. Karl, A. Knap, A. Michaels and the technical staff of the BATS group at Bermuda Biological Station for Research and the Hawaiian time-series group for the collection of water samples. The manuscript profited from comments by Hilary Paul, Patrick Quinn and the reviews of Jeremy Young, Luc Beaufort and Dave Lazarus. This work is a contribution to the EC-MAST program CANIGO, EC contract MAS3-CT960060, and the EC-TMR program CODENET, EC contract ERB-FRMX-CT97-0113. It was funded by the Swiss Federal Office for Education (BBW 95.0355). 10l 10l 10l 10l MARMIC 854 3-5-02 176 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 Appendix A A.1. Comparison between cell counts obtained with the SEM, the light microscope and the Utermo«hl settling method Method SEM Sample HS717 N No. of species 3 Total 369 Acantoica quattrospina Acantoica sp. Algirosphaera oryza Algirosphaera quadricornu Algirosphaera robusta Algirosphaera sp. Alisphaera spatula Alisphaera unicornis Alisphaera sp. Alveosphaera bimurata Anaplosolenia brasiliensis Braarudosphaera bigelowi Calcidiscus leptoporus Calciopappus caudatus Calciopappus rigidus Calciosolenia murrayi Ceratolithus cristatus Coronosphaera sp. Cyrtosphaera aculeata Discosphaera tubifera Emiliania huxleyi 346 Florisphaera profunda Gaarderia corolla Gephyrocapsa ericsonii 22 Gephyrocapsa protohuxleyi Gephyrocapsa muellerae Gephyrocapsa oceanica Helicosphaera carteri Helicosphaera pavimentum Michaelsarsia adriaticus Michaelsarsia elegans Neosphaera coccolithomorpha Oolithotus fragilis Ophiaster hydroideus Ophiaster reductus Ophiaster sp. Polycrater galapagensis Reticulofenestra punctata Reticulofenestra sessilis Reticulofenestra sp. Rhabdosphaera clavigera Rhabdosphaera stylifera Rhadbosphaera xiphos Scyphosphaera apsteinii Syracosphaera sp. Syracosphaera anthos Syracosphaera corrugis Syracosphaera epigrosa LM % SEM SEM HS745 c/l N 55 312 5 306 % c/l N 99 504 36 368 3 2 4 1 1 93.8 51 864 6.0 LM 298 0.3 0.3 % 0.8 0.5 1.1 HOT57-75 c/l 32 113 262 N % c/l 2 0.7 27 25 615 629 6 170 1 0.3 85 2 0.7 170 16 302 % 1.0 c/l 50 537 482 N % 13 303 c/l 39 357 8 2.6 1 039 4 0.6 321 19 3.0 1 527 1 5 0.3 1.7 130 651 2 13 22 0.3 2.1 3.5 161 1 044 1 768 8 21 2.6 6.9 1 039 2 728 24 3.8 1 928 11 11 1.7 1.7 884 884 8 2.6 1 039 3 26 0.5 4.1 241 2 089 1 0.3 130 325 325 97.4 96 903 0.3 N 175 349 2 3 0.5 0.8 175 262 7 44 1.9 611 12.0 3 840 2 0.5 80 21.7 6 981 2 4 6 2 0.5 1.1 1.6 0.5 175 349 524 175 2 2 0.5 0.5 8 2.2 1 0.3 85 3 95 1.0 254 31.5 8 058 175 3 298 1 LM 325 1 6 0.3 2.0 85 509 175 175 7 2.3 594 698 1 0.3 85 3 10 5 0.8 2.7 1.4 262 873 436 1 0.3 87 MARMIC 854 3-5-02 14 1 4.6 0.3 1 187 85 9 14 1.4 2.2 723 1 125 1 9 0.3 3.0 130 1 169 16 5.3 1 357 2 0.3 161 6 2.0 779 6 1.0 482 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 Method SEM Sample HS717 N Syracosphaera haldalli Syracosphaera histrica Syracosphaera molischii 1 Syracosphaera nana Syracosphaera nodosa Syracosphaera orbiculus type Syracosphaera ossa Syracosphaera pirus type Syracosphaera pulchra Syracosphaera rotula Syracosphaera type ‘I’ Syracosphaera type ‘K’ Gladiolithus £abellata Umbellosphaera irregularis Umbellosphaera tenuis Umbilicosphaera hulburtiana Umbilicosphaera sibogae Other heterococcoliths Calyptrolithophora papillifera Calyptrosphaera oblonga Corisphaera gracilis Daktylethra pirus Helladosphaera sp. Homozygosphaera arethusae Periphyllophora mirabilis Syracolithus schilleri Zygosphaera hellenica Other holococcolithophores LM LM N % c/l N % c/l N % c/l 0.3 150 1 0.3 325 1 5 0.3 1.4 87 436 14 4.6 1 187 5 1.4 436 2 0.5 175 4 1.1 349 1 5 0.3 1.4 87 436 2 131 0.5 175 35.6 11 432 1 109 0.3 85 36.1 9 245 1 3 1 2 5 0.3 0.8 0.3 0.5 1.4 87 262 87 175 436 17 5.6 5 2 1.4 0.5 436 175 1 0.3 87 2 2 SEM HOT 52 No. of species Total Acantoica quattrospina Acantoica sp. Algirosphaera oryza Algirosphaera quadricornu Algirosphaera robusta Algirosphaera sp. Alisphaera spatula Alisphaera unicornis Alisphaera sp. Alveosphaera bimurata Anaplosolenia brasiliensis Braarudosphaera bigelowi Calcidiscus leptoporus Calciopappus caudatus Calciopappus rigidus Calciosolenia murrayi Ceratolithus cristatus Coronosphaera sp. Cyrtosphaera aculeata Discosphaera tubifera Emiliania huxleyi 20 306 0.7 0.7 650 650 LM % 11 SEM N % 17 22 689 305 c/l N 19 519 15 400 % c/l N 58 707 12 33 % 6.1 3.0 % c/l 1 0.2 80 3 4 8 7 8 6 0.5 0.6 1.3 1.1 1.3 1.0 241 321 643 562 643 482 7 1.1 562 248 163 39.4 19 926 25.9 13 096 1 0.2 80 1 0.2 80 933 Utermo«hl 2 1 0.3 3.6 N 1 442 N % c/l 1 0.3 130 185 11 61.1 24 030 3.6 1 429 30 9.9 3 897 8 2.6 1 039 SEM Utermo«hl M37-31-31-10m M3-31-31-10m c/l N N 14 380 26 777 2 125 925 3 2 324 M37-31-31-0m c/l LM HOT57-75 c/l Sample 1 SEM HS745 % Method N SEM 177 % c/l % c/l 3 000 40 20 74 1 0.3 74 1 6 0.3 2.0 74 445 3 1.0 222 1 95 0.3 74 31.0 7 044 3 3 9 173 1.0 1.0 192 192 3.0 576 56.7 1 1071 3 0.8 440 4 106 1.0 587 26.5 15 557 MARMIC 854 3-5-02 4 12.1 80 1 3.0 4 10 12.1 80 30.3 10 000 4 1.0 648 1 253 162 33.0 41 003 20 2 66.7 2 000 178 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 Method SEM Sample HOT 52 N Florisphaera profunda 3 Gaarderia corolla Gephyrocapsa ericsonii 101 Gephyrocapsa protohuxleyi Gephyrocapsa muellerae Gephyrocapsa oceanica 34 Helicosphaera carteri 1 Helicosphaera pavimentum Michaelsarsia adriaticus 21 Michaelsarsia elegans Neosphaera coccolithomorpha Oolithotus fragilis Ophiaster hydroideus Ophiaster reductus Ophiaster sp. Polycrater galapagensis Reticulofenestra punctata Reticulofenestra sessilis Reticulofenestra sp. Rhabdosphaera clavigera 4 Rhabdosphaera stylifera Rhadbosphaera xiphos Scyphosphaera apsteinii Syracosphaera sp. 1 Syracosphaera anthos Syracosphaera corrugis Syracosphaera epigrosa Syracosphaera haldalli Syracosphaera histrica Syracosphaera molischii Syracosphaera nana Syracosphaera nodosa Syracosphaera orbiculus type Syracosphaera ossa Syracosphaera pirus type Syracosphaera pulchra 1 Syracosphaera rotula 2 Syracosphaera type ‘I’ Syracosphaera type ‘K’ Gladiolithus £abellata Umbellosphaera irregularis 18 Umbellosphaera tenuis 10 Umbilicosphaera hulburtiana 1 Umbilicosphaera sibogae 1 Other heterococcoliths Calyptrolithophora papillifera Calyptrosphaera oblonga Corisphaera gracilis Daktylethra pirus Helladosphaera sp. Homozygosphaera arethusae Periphyllophora mirabilis Syracolithus schilleri Zygosphaera hellenica Other holococcolithophores LM SEM Utermo«hl M37-31-31-0m % c/l N % c/l N % c/l 1.0 222 1 0.3 64 2 0.5 294 33.0 7 489 1 0.3 64 243 60.8 35 665 11.1 2 521 0.3 74 45 1 14.8 2 880 0.3 64 5 5 1.3 1.3 6.9 4 2 1 1.3 0.7 0.3 256 128 64 1 0.3 64 1.3 1 557 297 9 3.0 576 11 3.6 704 734 734 2 2 0.5 0.5 294 294 3 0.8 440 N 74 3 0.3 0.7 5.9 3.3 0.3 0.3 74 148 1 335 741 74 74 1 1 3 12 0.3 1.0 3.9 64 192 768 23 7.5 1 472 0.8 0.3 c/l 4 1 12.1 4 000 3.0 20 1 3.0 20 Utermo«hl M3-31-31-10m N % c/l N % 5 1.0 810 360 46.0 58 343 1 33.3 1 000 5 6 1.0 1.0 1 3.0 3.0 20 324 2 324 4.0 1.0 5 348 162 810 1 11 1.0 162 17 83 14 2.0 2269 10 1 1.0 1621 162 6 4 1.0 1.0 972 648 9 1.0 1 459 9 1.0 1459 4 2 1.0 648 324 20 440 147 1 0.3 147 8 1 2.0 0.3 1 174 147 11 2.8 1 614 1 2 3.0 6.1 20 40 486 3 2 0.7 810 972 2 33 1 5 3 1 0.3 % SEM M37-31-31-10m 128 MARMIC 854 3-5-02 11 486 1.0 1 783 c/l J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 Method SEM Sample M37-31-31-25m No. of species Total Acantoica quattrospina Acantoica sp. Algirosphaera oryza Algirosphaera quadricornu Algirosphaera robusta Algirosphaera sp. Alisphaera spatula Alisphaera unicornis Alisphaera sp. Alveosphaera bimurata Anaplosolenia brasiliensis Braarudosphaera bigelowi Calcidiscus leptoporus Calciopappus caudatus Calciopappus rigidus Calciosolenia murrayi Ceratolithus cristatus Coronosphaera sp. Cyrtosphaera aculeata Discosphaera tubifera Emiliania huxleyi Florisphaera profunda Gaarderia corolla Gephyrocapsa ericsonii Gephyrocapsa protohuxleyi Gephyrocapsa muellerae Gephyrocapsa oceanica Helicosphaera carteri Helicosphaera pavimentum Michaelsarsia adriaticus Michaelsarsia elegans Neosphaera coccolithomorpha Oolithotus fragilis Ophiaster hydroideus Ophiaster reductus Ophiaster sp. Polycrater galapagensis Reticulofenestra punctata Reticulofenestra sessilis Reticulofenestra sp. Rhabdosphaera clavigera Rhabdosphaera stylifera Rhadbosphaera xiphos Scyphosphaera apsteinii Syracosphaera sp. Syracosphaera anthos Syracosphaera corrugis Syracosphaera epigrosa Syracosphaera haldalli Syracosphaera histrica Syracosphaera molischii Syracosphaera nana Syracosphaera nodosa Syracosphaera orbiculus type Syracosphaera ossa Syracosphaera pirus type Syracosphaera pulchra 33 933 N 5 Utermo«hl % c/l Utermo«hl SEM Utermo«hl M37-31-31-75m M3-31-31-75m c/l N N 6 360 30 1 147 M37-31-31-50m N % 9 151 207 35 1.0 SEM c/l N 10 500 21 428 810 5 14.3 100 4 4 % c/l N % 10 125 634 24 1.0 1.0 1 174 1 174 1 4.2 179 % c/l 162 7 1 162 1 98 196 3 486 8 249 1 1.0 1 297 27.0 40 354 162 477 51.0 77 305 2 17 1 2.0 3 6 10 2 29 2 10 9 4.0 2.0 3.0 1.0 1.0 4 4 10 1 1 1 1 881 2 105 2 587 25.0 30 821 587 239 56.0 70 155 2.9 20 11 3.0 3 229 10 2.0 2 935 7 540 1 4.2 7 29.2 140 2 2 8.3 40 17 1 4.2 20 5 3 4 12.5 60 16.7 4 000 2 1 8.3 4.2 1.0 688 20 2000 20 1.0 14 41.2 280 9 6 26.5 180 17.6 6 000 1 1 2.9 2.9 1 000 20 3 8.8 60 1 670 491 10 381 5 9 615 1.0 982 33.0 37 422 491 1.0 884 54.0 60 406 3 7 1.0 2 295 688 196 324 1 3 4 14 2 17.1 120 28.6 10 000 1.0 324 2 755 162 1 36 22.9 160 c/l 20 1 8 % 6 112 660 34 5 834 16 162 5 491 2 4 8 196 393 786 648 2 269 324 1 2.9 20 1 2.9 20 4 700 324 1 621 1 459 648 648 1.0 1 621 162 162 162 162 3 1.0 881 12 3.0 3 522 2 5.7 40 1.0 1761 8 1.0 587 5 1 2 587 294 1 294 2 587 2 1.0 1.0 9 2 6 2 2 1 6 1.0 MARMIC 854 3-5-02 2 8.3 40 1.0 1 572 884 196 589 196 786 491 98 196 180 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 Method SEM Sample M37-31-31-25m Syracosphaera rotula Syracosphaera type ‘I’ Syracosphaera type ‘K’ Gladiolithus £abellata Umbellosphaera irregularis Umbellosphaera tenuis Umbilicosphaera hulburtiana Umbilicosphaera sibogae Other heterococcoliths Calyptrolithophora papillifera Calyptrosphaera oblonga Corisphaera gracilis Daktylethra pirus Helladosphaera sp. Homozygosphaera arethusae Periphyllophora mirabilis Syracolithus schilleri Zygosphaera hellenica Other holococcolithophores Utermo«hl N % c/l 14 2.0 2 269 1 5 1 SEM Utermo«hl M37-31-31-50m N % c/l N % c/l 6 1.0 1761 3 1 1.0 881 294 3 1.0 881 6 1.0 1761 N % c/l SEM Utermo«hl M37-31-31-75m M3-31-31-75m N N % c/l % c/l 162 1.0 810 162 1 2.9 20 486 Method SEM Sample M37-31-31-100m c/l N No. of species Total Acantoica quattrospina Acantoica sp. Algirosphaera oryza Algirosphaera quadricornu Algirosphaera robusta Algirosphaera sp. Alisphaera spatula Alisphaera unicornis Alisphaera sp. Alveosphaera bimurata Anaplosolenia brasiliensis Braarudosphaera bigelowi Calcidiscus leptoporus Calciopappus caudatus Calciopappus rigidus Calciosolenia murrayi Ceratolithus cristatus Coronosphaera sp. Cyrtosphaera aculeata Discosphaera tubifera Emiliania huxleyi Florisphaera profunda Gaarderia corolla Gephyrocapsa ericsonii Gephyrocapsa protohuxleyi Gephyrocapsa muellerae Gephyrocapsa oceanica Helicosphaera carteri Helicosphaera pavimentum Michaelsarsia adriaticus Michaelsarsia elegans 17 467 68 541 5 18 1 1 147 147 10 55.6 200 2 294 1 3 5.6 16.7 20 3 000 N Utermo«hl % 1 2 1 98 196 98 2 196 2 196 SEM Utermo«hl M37-31-31-150m 5 145 10 1.0 31.0 2.0 734 21 281 1 468 246 52.0 36 105 7 7 1 1.0 1.0 1 027 1 027 147 2 % 11.1 c/l N 5 260 7 91 2 000 MARMIC 854 3-5-02 % c/l N 11 448 3 6 4 62 4.0 68.0 503 7 800 12 13.0 1 510 % c/l 2080 3 50.0 60 2 33.3 2 000 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 Method SEM Sample M37-31-31-100m Neosphaera coccolithomorpha Oolithotus fragilis Ophiaster hydroideus Ophiaster reductus Ophiaster sp. Polycrater galapagensis Reticulofenestra punctata Reticulofenestra sessilis Reticulofenestra sp. Rhabdosphaera clavigera Rhabdosphaera stylifera Rhadbosphaera xiphos Scyphosphaera apsteinii Syracosphaera sp. Syracosphaera anthos Syracosphaera corrugis Syracosphaera epigrosa Syracosphaera haldalli Syracosphaera histrica Syracosphaera molischii Syracosphaera nana Syracosphaera nodosa Syracosphaera orbiculus type Syracosphaera ossa Syracosphaera pirus type Syracosphaera pulchra Syracosphaera rotula Syracosphaera type ‘I’ Syracosphaera type ‘K’ Gladiolithus £abellata Umbellosphaera irregularis Umbellosphaera tenuis Umbilicosphaera hulburtiana Umbilicosphaera sibogae Other heterococcoliths Calyptrolithophora papillifera Calyptrosphaera oblonga Corisphaera gracilis Daktylethra pirus Helladosphaera sp. Homozygosphaera arethusae Periphyllophora mirabilis Syracolithus schilleri Zygosphaera hellenica Other holococcolithophores Utermo«hl % c/l 16 3.0 2 348 4 1 1.0 587 147 1.0 2 Utermo«hl M37-31-31-150m N 3 SEM 181 N % c/l 2 11.1 40 N % c/l 3 3.0 377 1 1.0 126 2 2.0 252 N % c/l 1 16.7 20 440 294 5 1.0 734 1 1.0 126 1 10 2.0 147 1 468 6 7.0 755 N = number of specimens counted; % = relative proportion of species; c/l = cells/l; SEM = scanning electron microscope; LM = light microscope; Utermo«hl = Utermo«hl settling method. MARMIC 854 3-5-02 182 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 A.2. Description of ¢ltration equipment and procedure The following descriptions provide assistance in: (a) the construction of a ¢ltration device (Appendix A, Section A.2.6 Fig. 10) and (b) the sampling of coccolithophores. The ¢ltration system and the sampling strategy described below, were developed within the sampling campaigns at the JGOFS stations near Bermuda and Hawaii and during several cruises within the EC MAST-III project CANIGO. A.2.1. Modi¢cation of a Gelman0 ¢lter holder (catalogue number 1119, see Fig. 10) We have modi¢ed the inline ¢lter holder in order to: (a) obtain a more even distribution of material ; (b) increase the ¢ltration e⁄ciency; and (c) improve the handling of Nucleopore0 PC111109 membrane ¢lters. (1) Replace the plastic support sieves (K), which produce a cross on the membrane ¢lter, with the stainless steel support sieve (J), which allows for a homogeneous distribution. (2) Screw the two hose connectors (A) into the lower (L) and the upper ¢lter holder (E). (3) Replace the closing cap (B) from the evacuation valve with a 5-cm piece of silicon tube (D). (4) Mount one part of the PVC quick connector at the end of the 5-cm piece of the silicon tube (D). (5) Fasten a tube clip (C) at the 5-cm piece of the silicon tube and the hose connector. A.2.2. Modi¢cation of carboys (SEMADENI0 order number 1531, see Fig. 11) We have modi¢ed the ‘normal’ carboys because they are more robust onboard than for example Nalgene0 carboys, which have a protruding spigot/outlet at the bottom that can be easily broken o¡. (1) Remove the evacuation valve/lid (B) and drill a hole into the plastic lid. (2) Press the female part of a quick connector from the inner side of the lid into the hole. (3) Connect 47 cm of a hard PVC tube (R) of 6 mm diameter to the barbed part of the tubing connector and mount a check valve (Q) at the end of the tube. (4) Put the tube into the carboy and fasten the lid. If the opening of the evacuation valve is too small for the check valve, introduce the tube with- Fig. 10. Modi¢cation of a Gelman0 inline ¢lter gasket. A, hose connector; B, evacuation valve and closing cap; C, tube clip; D, silicon tube; E, ¢lter holder top; F, red sealing ring; G, white Te£on ring; H, Isopore0 /Nucleopore0 polycarbonate membrane; I, millipore0 cellulose mixed-ester membrane; J, support sieve stainless steel; K, plastic support sieve; L, ¢lter holder base. MARMIC 854 3-5-02 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 183 Fig. 11. Schematic of a single ¢ltration line. A, carboy; B, modi¢ed outlet with female quick connector (former evacuation valve); C, male quick connector; D, PVC tube, 7 mm diameter; E, female quick connector; F, silicon tube; G, tube clip; H, evacuation valve; I, ¢lter holder; J, PVC tube, 5 cm long, 7 mm diameter; K, male quick connector; L, PVC tube, 8 mm diameter; M, vacuum pump; N, stopcock valve; O, metal sheets for gasket holding; P, outlet; Q, check valve; R, PUR tube; S, ¢lling inlet. out the check valve and get the tube with the help of a wire loop through the large ¢lling inlet (S) and connect the check valve to the tube. Note: one disadvantage of this set-up is that it is necessary to ¢ll the inner tube and the connection tube between carboy and ¢lter holder with water before ¢ltration can start. This is necessary because air trapped inside the tubes and the ¢lter gasket can prevent the gravity £ow of water. A.2.3. Mounting of a membrane ¢lter (see Fig. 10) (1) Put a Millipore0 cellulose mixed-ester membrane (I), pore size of 8 Wm, directly on the stain- less support sieve (J) and moisten this membrane with distilled water to get a £at surface. Make sure that the whole membrane is wet and that there are no white areas with air bubbles. This membrane prevents formation of pressure rings on the ¢lter as well as air bubbles and it supports the drainage. (2) Place a 0.8-Wm Isopore0 /Nucleopore0 polycarbonate membrane (H) on top of membrane (I). Make sure that there are no air bubbles between these two membranes (this membrane will ¢lter the coccolithophores). In order to hold the membrane in position we recommend attaching a low vacuum under the membrane ¢lter (connection to lower hose adapter A). MARMIC 854 3-5-02 184 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 (3) Put the white Te£on ring (G) on top of the 0.8-Wm Isopore0 /Nucleopore0 polycarbonate membrane (H). This prevents wrinkling of the membrane during closing of the ¢lter holder. (4) Put the red sealing ring (F) on the top of the white Te£on ring (G). (5) Close the ¢lter holder (E+L) carefully and check for membrane wrinkling. The ¢lter holder is now ready for use. A.2.4. Filtration procedure (see also Fig. 11) The sampling and ¢ltration procedures can be summarised as follows: (1) Draw o¡ the desired amount of water from the Niskin0 bottles, from di¡erent depth levels, into a carboy (A). (2) Connect tube (L) to the ¢lter holder and close valve (N). Be sure to close valve (N) after connecting the ¢lter holder because the membrane might be damaged by air pressure. (3) Connect tube (D) to the ¢lter holder and ¢ll both with distilled water. Note: this step is not necessary if carboys with a spigot/outlet at the bottom of a carboy are used. (4) Connect this tube (D) to the carboy (A, B). (5) Connect a silicon tube to the silicon tube (F) that is attached to the evacuation valve (H) (use the quick connector E) and put the end of the tube in a beaker. (6) Open the tube clip (G). The water will now start £owing (gravity £ow). Make sure that large air bubbles migrate out of the ¢lter gasket through the evacuation valve. Then close the tube clip. It is best to tilt the ¢lter holder a little so that the evacuation valve is slightly elevated. (7) Make sure that the vacuum is not higher than 200 mm Hg (otherwise cells may be deformed) and then open valve (N). (8) If large air bubbles are observed in the ¢lter holders during the ¢ltration, close valve (N) (disconnect the vacuum) and repeat the procedure as described in steps 5 and 6. (9) After all water has been ¢ltered, close the valve (N), disconnect tubes (D) at the carboy. Fill tube (D) with bu¡ered distilled water (NH4 , pH 8.5) until the gasket is £ooded. If needed, open tube clip (G) carefully to release the air, and open valve (N) again. (10) After rinsing the ¢lter with distilled water under the vacuum, remove tube (D) and open the ¢lter holder carefully. Make sure that there is no water left in the ¢lter holder, otherwise some material will bloat and change the distribution. Drops of water on the white Te£on ring should be carefully removed with a tissue before removing the membrane. (11) Close the valve (N) and remove the ¢lter membrane and store it in a labelled petri dish on top of the absorber pads. You may have to exchange absorber pads several times because the polycarbonate membrane tends to roll up. (12) Dry the membrane immediately in an oven at 40^60‡C for several hours. It is very important to remove the membrane as soon as possible after ¢ltration has ¢nished, otherwise any coccolithophores may begin to dissolve. A.2.5. Sample preparation We strongly recommend using Canada Balsam for light microscope preparations because of the improved durability of the slides. Canada Balsam preparations of the Ehrenberg collection are after 150 yr still usable (Dave Lazarus, personal communication, 2001). The examination of old SEM stubs (older than 1 yr) that were prepared using conductive carbon tape revealed decomposition of the gold coating. These samples could not be recovered. We suspect that a chemical reaction takes place between the conductive glue, the ¢lter membrane and the gold. Therefore, we currently mount ¢lter samples with the help of alcohol and liquid conductive silver directly onto the aluminium stub (or glass). We put a drop of alcohol on the stub and, onto the drop, place the piece of ¢lter. The ¢lter membrane clings tightly to the aluminium stub as a result of the strong capillary adhesion. In order to mount the ¢lter permanently, we add a drop of conductive silver to the corners of the ¢lter and wait for the alcohol to evaporate. Subsequently, the stub is sputtered with gold. Ongoing tests have shown that these preparations last at least 3 yr. MARMIC 854 3-5-02 J. Bollmann et al. / Marine Micropaleontology 44 (2002) 163^185 185 A.2.6. List of materials used to build a ¢ltration device (including supplier information) Item Quantity Brand Order number Te£on ring 47U37U1 mm Dual head membrane vacuum pump Inline ¢lter gasket Photo etched steel support sieve Tube clip 60 mm long Carboy 9 l Aluminium sheets, 1.5 and 2 mm Stopcock valve, 7 mm diameter Carboy 10 l PVC tubes, 7 mm diameter T-tube connector, 7 mm diameterU10 mm PVC Tubes, 8 mm diameterU12 mm PUR tubes, 8 mm diameterU10 mm Check or no return valve, 6 mm diameter Quick disconnectors, 6 mm diameter Quick disconnectors, 7 mm diameter Vacuum carboy 10 l Filling/venting closure Press nuts, m 6 Stainless steel tank MF-membrane ¢lter, 8 Wm pore size ISOPORE membrane ¢lter, 47 mm diameter, 0.8 Wm Polycarbonate membrane, 47 mm diameter, 0.8 Wm 12 1 12 12 12 12 2 m2 14 12 12 m 12 10 m 3m 12 12 12 1 1 20 1 ADIPAC0 Cole-Parmer0 Gelman0 Gelman0 Merck0 Nalgene0 NN Semadeni0 Semadeni0 Semadeni0 Semadeni0 Semadeni0 Semadeni0 Semadeni0 /Nalgene0 Semadeni0 /Nalgene0 Semadeni0 /Nalgene0 Semadeni0 /Nalgene0 Semadeni0 /Nalgene0 Tubtara0 NN Millipore0 Millipore0 Nucleopore0 Special order E-07061-00 1119 72970 3802002 2320-0020 Special order 03097 1531 1345 0485 1348 2591 0523/6120-0010 00517/6150-0010 00518/6150-0010 00970/2210 01409/2162 References Corte¤s, M.Y., 1998. Coccolithophores at the time-series station ALOHA, Hawaii: population dynamics and ecology. Ph.D. Thesis, University of Zurich, Zurich, unpublished, 176 pp. Corte¤s, M.Y., Bollmann, J., Thierstein, H.R., 2001. Coccolithophore ecology at the HOT station ALOHA, Hawaii. Deep-Sea Res. II 48, 1957^1981. Haidar, T.A., 1997. Calcareous phytoplankton dynamics at Bermuda (N. Atantic). Ph.D. Thesis, No. 12084, ETH Zurich, unpublished, 168 pp. Haidar, A.T., Thierstein, H.R., 2001. Coccolithophore dynamics o¡ Bermuda (N. Atlantic). Deep-Sea Res. II 48, 1925^ 1956. Hasle, G.R., 1978. The inverted microscope method. In: Sournia, A. (Ed.), Phytoplankton Manual. Monographs on Oceanic Methodology, UNESCO, Vol. 6. Paris, pp. 88^96. Margalef, R., 1969. Counting. In: Vollenweider, R.A., Talling, J.F., Westlake, D.F. (Eds.), A Manual on Methods for Measuring Primary Production in Aquatic Environments including a Chapter on Bacteria. International Biological Programme (IBP Handbook 12). Blackwell Scient¢c, Oxford, pp. 7^14. McIntyre, A., Be¤, A.W.H., 1967. Modern coccolithophoridae of the Atlantic Ocean-I. placoliths and cyrtoliths. Deep-Sea Res. 14, 561^597. special order SCWP04700 ATTP 04700 111109 Okada, H., Honjo, S., 1973. The distribution of oceanic cocolithophorids in the Paci¢c. Deep-Sea Res. 20, 55^374. Reid, F.M.H., 1980. Coccolithophorids of the North Paci¢c Central Gyre with notes on their vertical and seasonal distribution. Micropaleontology 26, 151^176. Sournia, A., 1978. Phytoplankton Manual. Monographs on Oceanographic Methodology, UNESCO, Vol. 6. Paris, 337 pp. Throndsen, J., 1978. Preservation and storage. In: Sournia, A. (Ed.), Phytoplankton Manual. Monographs on Oceanographic Methodology, UNESCO, Vol. 6. Paris, pp. 69^ 74. Utermo«hl, H., 1931. Neue Wege der quantitativen Erfassung des Planktons (mit besonderer Beru«cksichtigung des Ultraplanktons). Verh. Int. Ver. Theor. Angew. Limnol. 5, 567^ 596. Utermo«hl, H., 1958. Zur Vervollkommnung der quantitativen Phytoplankton-Methodik. Mitt. Int. Ver. Theor. Angew. Limnol. 9, 1^38. Venrick, E.L., 1978a. The implication of sub-sampling. In: Sournia, A. (Ed.), Phytoplankton Manual. Monographs on Oceanographic Methodology, UNESCO, Vol. 6. Paris, pp. 75^87. Venrick, E.L., 1978b. How many cells to count? In: Sournia, A. (Ed.), Phytoplankton Manual. Monographs on Oceanographic Methodology, UNESCO, Vol. 6. Paris, pp. 167^180. MARMIC 854 3-5-02
© Copyright 2026 Paperzz