Development of an image analysis system for the enumeration and sizing of aquatic bacteria

Development of an image analysis system for the enumeration and sizing of aquatic bacteria

Material Information

Development of an image analysis system for the enumeration and sizing of aquatic bacteria
David, Andrew Whitney
Place of Publication:
Tampa, Florida
University of South Florida
Publication Date:
Physical Description:
ix, 66 leaves : ill. ; 29 cm.


Subjects / Keywords:
Marine bacteria -- Measurement ( lcsh )
Dissertation, Academic -- Marine science -- Masters -- USF ( FTS )


General Note:
Thesis (M.S.)--University of South Florida, 1993. Includes bibliographical references (leaves 58-63).

Record Information

Source Institution:
University of South Florida
Holding Location:
Universtity of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
029495812 ( ALEPH )
29156806 ( OCLC )
F51-00098 ( USFLDC DOI )
f51.98 ( USFLDC Handle )

Postcard Information



This item is only available as the following downloads:

Full Text


.DEVELOPMENT OF AN IMAGE ANALYSIS SYSTEM FOR THE ENUMERATION AND SIZING OF AQUATIC BACTERIA by Andrew Whitney David A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in the Department of Marine Science in the University of South Florida May 1993 Major Professor: John H. Paul, Ph.D.


Graduate Council University of South Florida Tampa, Florida CERTIFICATE OF APPROVAL MASTER'S THESIS This is to certify that the Master's Thesis of Andrew Whitney David with a major in the Department of Marine Science has been approved by the Examining Committee on April 8, 1993 as satisfactory for the Thesis requirement for the Master of Science degree. Thesis Committee: Major John H. Paui, Ph. D. Member: Edward S. Van Vleet, Ph.D.


ACKNOWLEDGEMENTS This project was developed under the auspices of the Florida Sea Grant College Program with support from the National Oceanic and Atmosphere Administration. Office of Sea Grant, U.s. Department of Commerce, Grant NAB 6AA-D-SG068. This work is a result of research sponsored by NOAA Office of Sea Grant, Department of Commerce, under Grant NA86AA-D-SG068. This research was also supported in part by NSF Grant BSR 8605170 to John H. Paul. Several people greatly aided the development of the image analysis system and deserve recognition: John Gianmugnai of Image Technology Corporation, John Bogan of Dage MTI, and Larry Lindner of the University of South Florida. Lastly, I would like to thank my committee, John Paul, Ted Van Vleet, and Gabe Vargo for tolerating my AWOL period, Churchill Grimes and Jeff Isely for support and assistance in manuscript preparation and my wife, Kelly, and parents for fiscal and moral support. ii


TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES ABSTRACT INTRODUCTION MATERIALS AND METHODS Image Analysis System DNA Staining Slide Preparation Calibration Sample Locations RESULTS AND DISCUSSION Standard Camera Enumeration Standard Camera Measurements SIT Camera Enumerations SIT Camera Measurements Biomass Calculations CONCLUSIONS LITERATURE CITED APPENDICES APPENDIX 1 APPENDIX 2 counting and Measuring Program Program for computaion of BABI cell volume iii iv vi vii 1 11 11 18 19 20 24 26 26 38 42 46 53 57 58 64 65 66


LIST OF TABLES Table 1. Field count comparisons for Vzbrio proteolyticus Standard camera, Neuvicon phototube. 27 Table 2. Field count comparisons for Isolate #9. 29 standard camera, Neuvicon phototube. Table 3. Field count comparisons for Bayboro Harbor, 31 Prefiltration evaluation. Standard camera, Neuvicon phototube. Table 4. Mean cellml-1 comparisons for Bayboro 32 Harbor, Prefiltration evaluation. Standard camera, Neuvicon phototube. Table 5. Field count comparisons for Offshore station 34 (25241N 8250'W). Standard camera, Neuvicon phototube. Table 6. Field count comparisons for Isolate #6. 36 Standard camera, Chalnicon phototube. Table 7. Mean cellml-1 comparisons for Isolate #6. 37 Standard camera, Chalnicon phototube. Table 8. Mean cell size comparisons for Vzbrio 41 proteolyti cus and Bayboro Harbor. Standard camera, Chalnicon phototube. Table 9. Same cell comparisons for Vzbrio 42 proteolyticus (starved 5 days) and Bayboro Harbor. standard camera, Chalnicon phototube. Table 10. Field count comparisons for Vzbrio 45 pr oteo lyticus and Bayboro Harbor. SIT-66 camera. Table 11. Same cell length comparisons for bacteria from five sources. SIT-66 camera. iv 47


Table 12. Same cell measurement comparisons for Pseudomonas aeruginosa pRO2 317 SIT-66 camera. Table 13. Cell counts, sizes, and biomass values for bacteria from four sources. SIT-66 camera. v 54 55


LIST OF FIGURES Figure 1. Diagrammatic representation of dichroic 13 mirror assembly. Figure 2. Diagrammatic representation of image 15 analysis system. Figure 3. Length distribution of Alafia River 48 bacteria. Figure 4. Length distribution of crystal River 49 bacteria. Figure 5. Length distribution of Bayboro Harbor 50 bacteria. Figure 6. Length distribution of Vibrio proteolyticus 51 Figure 7. Length distribution of offshore bacteria. 52 vi


DEVELOPMENT OF AN IMAGE ANALYSIS SYSTEM FOR THE ENUMERATION AND SIZING OF AQUATIC BACTERIA by Andrew Whitney David An Abstract Of a thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in the Department of Marine Science in the University of South Florida May 1993 Major Professor: John H. Paul, Ph.D. vii


Image analysis systems, epifluorescent microscopes, and low-light video cameras have been used in a wide array of microbial investigations. This paper describes the development of a system composed of an Image Technology Model 2000 image analyzer, an Olympus BH-2 epifluorescent microscope, and standard and image intensified (Dage MTI SIT66) video cameras to enumerate and measure aquatic bacteria from a variety of marine and freshwater environments in the Southwest Florida area. Manual counts and measurements on the same fields and cells were compared with values derived from the image analysis system. Results were statistically tested for similarity. For bacterial cultures and natural populations, statistically similar cell counts were obtained with the standard camera. The SIT-66 camera enabled accurate sizing and cell counting of bacterial cultures and natural populations, when compared to manual methods. Cell densities up to 1. 14 x 107 cells ml-1 were enumerated and cell sizes from 0.18 to > 7 .00 were measured. Cell volumes were calculated from machine and manual measurements based on the shape of a prolate spheroid. Biomass estimates were made using these cel l volumes and biovolumejbiomass conversion factors taken from the literature. Biomass values ranged between 1.05 x 10-6 viii


and 2. 80 x 10-8 gC ml-1 which fall between published values for similar environments. The image analysis system developed in this project produced cell counts and measurements statistically similar to those made manually. The image analysis system measurements were achieved at a 10 fold faster rate than by manual methods, eliminated inter-operator variability, and greatly reduced operator fatigue. Abstract approved: Major John H. Paul, Ph.D. Professor Department Marine Science Date of ix


INTRODUCTION Bacterial cell counts and measurements in aquatic systems can be used to assess a wide range of water quality parameters including growth, survival, contamination, and seasonal and diel periodicity in bacterial abundance and biomass (Watson et al. 1977). Bacterial cell density is one of the most universally reported parameters in microbial studies. Nearly as common are biomass estimates based on cell densities and volumes obtained through dimensional measurements of a few representative cells. Both of these determinations require a significant amount of time to acquire, and are highly subjective due to operator experience, sample condition, and equipment sophistication. Incident light fluorescence microscopy, more commonly referred to as epifluorescence microscopy (Bradbury 1979) is a technique widely used in the examination of aquatic bacteria (Hobbie et al. 1977, Sieracki et al. 1985). This procedure involves the use of a compound microscope with a high intensity light source to view bacteria stained with a fluorochrome (Coleman 1980, Porter and Feig 1980, Paul 1982) This project aims to reduce operator induced variability as well as temporal requirements for data acquisition with digital analysis of epifluorescent microscopic images. Cell measurements will be used to generate biomass estimates which will be compared with


2 published values. Free living bacteria are normally concentrated on a filter for enumeration. Hobbie et al. (1977) and Bowden (1977) demonstrated the superiority of polycarbonate filters over cellulose filters. Among the many improvements noted are two that are particularly important to this work, uniformity of pore size and smoothness of filter surface. However as noted by Hobbie, Nuclepore polycarbonate filters (Nuclepore Corporation, Pleasanton, CA) autofluoresce under ultraviolet illumination and require counterstaining prior to use. As bacteria are nearly transparent in transmitted light, fluorochrome stains which complex with DNA are used to visualize bacteria with epifluorescence microscopy. Acridine orange ( 3, 6 bis (dimethyl amine] acridine) has long been the standard fluorochrome used in epifluorescent determinations (Francisco et al. 1973). Unfortunately acridine orange also complexes with a variety of non-bacterial materials in aquatic samples and fluoresces at the same wavelength as chlorophyll a (Caron 1983). Additionally, acridine orange is an intercalating mutagen which can cause shift mutations during DNA replication. Recently several other stains have been utilized; primulin, proflavine (3,6 Diaminoacridine hemisulfate), FITC (Fluorescein isothiocyanate), Hoechst 33258 and 33342, DAPI (4,6 Diamidino 2 phenylindole) and others. While useful in specific circumstances, the emission wavelengths of FITC and proflavine also overlap chlorophyll a


fluorescence (Davis and Sieburth 1982). 3 Pr imul in has a maximum fluorescence at acid pH (Caron 1983), a circumstance rarely found in marine environments. Paul (1982) has demonstrated the effectiveness of Hoechst 33258 and 33342 bisbenzimide dyes in staining cultured and natural populations of marine bacteria, however staining time was reported to be a minimum of one hour. Porter and Feig (1980) demonstrated the low non-specific staining characteristics and the short five minute staining time of DAPI. Coleman (1980) investigated the usage of DAPI/mithramycin mixtures for detection of bacteria attached to other marine microorganisms, primarily diatoms. Coleman noted the greater specificity of mithramycin but also the greater intensity of DAPI fluorescence. Brunk et al. (1979) concluded the DAPI/DNA complex fluoresced with 20 times the intensity of DAPI alone. Fluorescent dyes have been combined with monoclonal antibodies to detect and enumerate specific bacterial species (Hazen and Jimenez 1988, Paul 1993). Fluorescent beads have also been used to evaluate ingestion rates of microflagellates (Sieracki et al. 1987) This project involves bacterial enumeration and measurement by digital analysis of video images taken directly from microbial samples (Sieracki et al. 1985, Bjornsen 1986). In 1982, Pettipher and Rodrigues published some of the initial work linking image analysis and epifluorescence microscopy. They examined bacteria in milk and were able to measure 5.0 x


4 Costello and Monk (1984) coupled an image analysis system with a hemacytometer and enumerated yeast cells, Sacch aromyces cerev isciae, up to densities of 1. 0 x 108 cellsml-1 Singh et al. (1989) found image analysis useful as an alternative to colony forming unit (CFU) enumeration for viable cell determinations. A variety of other techniques have been reported, including photomicroscopy, electron microscopy, laser e xcitation, and flow cytometry. Paerl et al. (1973) and Bowden (1977) have investigated scanning electron microscopy (SEM) for use in bacterial enumeration and measurement. Bowden found no statistical difference between counts made by SEM and epifluoresence microscopy. However, Fuhrman (1981) reported a significant shrinkage (up to 37%) in free living marine bacteria during SEM preparation compared to epifluorescent evaluation. Paerl et al. (1973) discussed a method using ethanol, amyl acetate and liquid carbon dioxide under 72 atmospheres of pressure to dry cells without shrinkage; however, this is a very costly and time consuming process. Borsheim et al. (1990) have utilized transmission electron microscopy (TEM) for enumeration and sizing of marine bacteria and viruses although this technique required 90 minutes of ultracentrifugation at 80,000 x G. Drapeau and Laurence (1977) reported the use of a pulsed dye laser to reproducibly enumerate Escherichia coli with an electronic detection and recording system. Fry and Davies (1985) compared SEM, photomicrographs taken from an epifluorescent


5 microscope, and manual measurements with an ocular micrometer. They also found significant shrinkage with SEM (77%) and overestimation with the ocular micrometer (:S65%). These errors are based on their assumed correct value derived from image analysis of photographic slides taken through the epifluorescent microscope. This method is also very time consuming taking into account photographic development. Flow cytometry is used to electronically measure and enumerate small particles. A fine stream of fluid carries particles one at a time through a beam of light, usually a laser. As the light strikes each particle, it is reflected, the angle of reflection is determined by an array of photodetectors. The amount of reflection is proportional to the particle size (Paul 1993). In addition to size information from FALS (Forward Angle of Light Scatter), photodetectors dedicated to specific wavelengths allow detection and quantification of autofluorescing organisms. Flow cytometry examination is usually nondestructive and may be used to separate organisms in mixed assemblages. Advantages of flow cytometry include the speed of evaluation, the nondestructive nature of the evaluation, and the ability to detect and enumerate several species in a single sample through the use of different fluorochromes and photodetectors. A brief litany of the application of flow cytometry to marine microbiology follows. Flow cytometers, developed for medical applications, were used in marine microbiology by Olson et al.


(1983) and Yentsch et al. (1983). 6 Olson et al. ( 1983) assembled a simple cytometer system for quantification of chlorophyll and DNA in phytoplankton, while Yentsch et al. (1983) employed a commercial system to separate different marine microorganisms. Van Dilla et al. ( 1983) used the fluorochrome DAPI, to quantify DNA content of six bacterial species with flow cytometry. Li and Wood (1988) described a portable flow cytometer used to characterize ultraphytoplankton with regard to size, volume, and number of chloroplasts. Robertson and Button (1989) followed a rigorous protocol to improve the resolution in a commercial unit allowing accurate measurement of plastic spheres down to o. 014 and bacteria to 0.05 Sanders et al. (1990) used a dual laser flow cytometer to determine the molar percentage of guanine-cytosine in 14 bacterial species. Image analysis systems were initially developed by the coal industry forty years ago for mineshaft air quality monitoring (Bradbury 1983). As technology advanced, image analysis was developed as a research tool in the scientific disciplines. In a 1979 review, Bradbury outlines the general areas of scientific image analysis: macroscopic, optical microscopy, and electron microscopy. Optical microscopy is the most common application cited and is the method utilized in this project. Bradbury also discusses the three methods available for data collection: manual or point counting, semiautomatic, and fully automatic. Manual or point counting


7 methods require the operator to extract all information from the object images. Normally a grid is superimposed on an image and objects are counted when they lie under a grid intersection point. Cynar et al. (1985) developed a computer program for tabulating species abundance in mixed assemblages which utilized a series of "hot keys" to increment tallies for up to 4 0 species. The microscopist identified organisms manually and used the computer to track abundances. While a computer is used to analyze the data, this technique is not very far removed from totally manual determinations. Semi automatic systems also require the operator to provide image recognition while the machine produces numerical determinations. Fully automatic machines detect images as well as quantify them. The operator inputs detection parameters and monitors system performance, providing adjustments as required. Understandably, the cost of the manual systems is the lowest while fully automatic systems are the most expensive. Bradbury (1983) describes the most common problems associated with image analysis. The foremost problem is discriminatory ability. While the.human eye has little difficulty detecting subtle differences in color and contrast the same is not true of image analysis. Often image enhancement steps must be taken to accentuate these differences to aid correct machine recognition. Sieracki et al. (1989a) discusses various methods for automatically selecting threshold limits to correctly locate edges of


screen objects (segmentation). 8 Proper segmentation is critical for generating accurate size determinations. Sieracki et al. ( 1989a) found the minimum of the second derivative of the intensity gradient profile provided edge locating abilities exceeding other automated methods and equalling or exceeding manual methods depending on object shape. Advances in computer hardware and software have increased the versatility and dependability of image analysis systems while reducing the cost. Previous systems were very proprietary with individual systems offering limited flexibility. New companies are now producing components making it possible to assemble customized systems for specific applications. Maranda (1987) details the procedures followed to build a system for analyzing protein banding patterns on electrophoretic gels. New cameras with color and image intensifying circuitry have pushed image analysis far below the detection limit of the human eye. Color cameras have enabled separation of fluorescing detritus from bacterial samples and differentiation of planktonic bacteria by detection of specific wavelengths of light attributed to specific compounds (DNA, Chl a, and other pigments) (Sieracki and Viles 1990, Sieracki and Webb 1991) Hiraoka et al. (1987) described charged coupled devices, cameras that electrically detect very low light levels (down to individual photons), produce very linear intensity gradients, generate


9 'digital signals (eliminating the need for a separate digitizer), and have larger numbers of pixels than standard cameras (up to 2024 compared to 256). Commare (1988) describes cooled Charge Coupled Devices (CCDs) in systems which represent single photons as 100 pixel images after digitization. Viles and Sieracki (1992) used a cooled CCD to count and measure marine picoplankton, down to 0. 2 J.Lm, at rates >1000hr-1 with accuracy and precision exceeding that attainable with standard video cameras. Tilney and Inoue (1982) published a time series of micrographs, taken with differential interference contrast (DIC) microscopy, depicting the acrosomal reaction in sea cucumber sperm where a filament 65 nm in diameter is clearly visible. This is an object one quarter the resolution limit of light microscopy. The micrographs are the result of powerful computer digitizations involving contrast manipulations on images provided by high numerical aperture (N.A.) objectives. Inoue and Inoue (1989) demonstrated even greater advances with stereo images of intensity contours in chromosomes produced by electronic imaging cameras coupled to a video microscope. Sheppard (1987) details the physics of many of these optical imaging devices in a recent review. While these new technologies are not yet being truly mass produced, the costs are decreasing into areas affordable to more and more investigators. Cell density values are reported in nearly all microbial investigations and are clearly a very important parameter in


10 environmental evaluations. Cell measurements are required for estimations of biomass, condition, and all rates per volume. Temporal requirements for cell counting are far from insignificant and those needed for cell measurements can be quite extensive. Inter-operator variability adds error to both evaluations when they are done manually. The current project did not require detection of individual photons or measuring nanometer wide acrosomal processes. The primary goal of this research was to adapt an automatic image analysis system from Image Technology Corporation for the enumeration and measurement of aquatic bacteria. It is the desire of this project to develop and test an automated method using digital image analysis and epifluorescence microscopy for the enumeration and measurement of aquatic bacteria. It is believed results can be obtained that meet or exceed the accuracy and precision of manual determinations and that these machine derived values can be made in significantly less time. This project will be testing the null hypothesis which states: there are no statistically significant differences between bacterial counts and measurements made manually and with the Image Technology Corporation Model 2000 image analysis system. In addition bacterial volumes will be calculated and compared to published values in order to assess this system's applicability to a wider range of microbial parameters.


11 MATERIALS AND METHODS Image Analysis System Image analysis systems are normally composed of three instruments: a video camera, a microscope, and a digital processor. Two cameras were utilized in this project, a standard black and white television camera supplied with the processor and a Dage-MTI SIT 66 Silicon Intensified Target camera ( Dage-MTI, Michigan City, IN) The standard camera was originally fitted with a Videcon phototube which was replaced by a Newvicon phototube at the onset of the project. The Newvicon phototube was replaced by a Chalnicon phototube midway through the project to increase light gathering sensitivity. An Olympus BH-2 compound microscope (Olympus Optical, Atlanta, Georgia) provided magnified images of the bacteria to the video camera. This microscope was equipped for epifluorescent use with the addition of a 100W HBO mercury lamp, a B-0891 bluejUV filter set, a UV-FL100 non-fluorescing objective lens, and a dichroic mirror. The dichroic mirror allows use of ultraviolet light to illuminate the stage while preventing the dangerous wavelengths from reaching the ocular lenses.


12 Dichroic mirrors selectively reflect or transmit light utilizing the principal of Stokes Law (Figure 1). This-law describes the wavelength shift encountered in fluorescence. The mercury burner emits ultraviolet radiation in a variety of wavelengths, this light passes through an excitation filter, which removes most wavelengths above 420 nm, before entering the barrel of the microscope and striking the dichroic mirror. The mirror rests perpendicular to the light path and is canted 45 degrees below the vertical. Wavelengths below 420 nm are reflected downward through the objective and onto the sample slide, wavelengths above 420 nm (not removed by the excitation filter) pass through the mirror and strike the front of the barrel. DAPI, the fluorochrome used in this research, has an excitation maximum of 340 nm and an emission maximum of 488 nm (Brunk et al. 1979) The fluorescence emitted by the DAPI/DNA complex passes up through the objective lens and again strikes the dichroic mirror. Again any light below 420 nm is reflected by the dichroic mirror and is directed back towards the mercury lamp. The emitted light above 420 nm passes through the dichroic mirror and is viewed through the observation tubes. A light path selector knob directs the image to the ocular lenses and/or the camera tube. This selector has three positions; 100% of the light to the ocular lenses, 100% to the camera tube, or 20% to the oculars and 80% to the camera. With the low sensitivity of the standard camera, 100% of the


OPTICS OF EPIFLUORESCENCE MICROSCOPY Hg Lamp ,. ........................................... < 420 nm > 420 nm Excitation Filter IIIUIIII I IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII Sample Objective Lens Barrier Fil ter Dichroic Mirror Objective Lens Figure 1. Diagrammatic representation o f dichroic mirro r assembly. ...... w


14 light was directed to the camera in order to capture an image. The increased sensitivity of the SIT camera allowed utilization of the 20/80 split position. This allowed simultaneous viewing by the operator and the camera affording an additional time savings over the standard camera. With the standard camera, after an image was acquired and stored with the image analysis system, the light path selector was moved to the 100% ocular position and the field counted or measured manually. The time require for this operation allowed greater fading of the DAPI/DNA complex. The adhesives used in standard objective lenses fluoresce under ultraviolet light, therefore a special objective manufactured with non-fluorescing glue is used in epifluorescence microscopy. The objective utilized in this project was an Olympus UV-FL100 which provided lOOX magnification. Figure 2 illustrates the image analysis system (David and Paul 1989). The image processor used in this work was an Image Technology Corporation Model 2000 (Image Technology Corporation, Deer Park, NY). An Apple IIe microcomputer (Apple Computer, Incorporated, Cupertino, CA) provided the user interface and system controller. A frame grabber in the processor stored images for analysis and digitization. Two monitors were included in the system, one displayed current system parameters, data, and operational menus, while the other displayed the digitized image. Two additional


Image Analysis System System Controller D Video Camera ........... ... 0 Trackball I _ ='Y l ................ t Printer Epifluorescence Microscope Image Monitor 11111111111111111 lllllllll lllllll l Image Processor Video Signal Image Enhancement ----------Data Signa l Figure 2. Diagrammatic representation of image image analysis system I-' l11


16 peripheral devices completed the system, a trackball used in calibration and a printer. Image Technology Corporation also provided the software necessary to operate the Model 2000 system. There are two broad categories or modes of determinations, field specific and feature specific. In field specific mode, all objects in a field are treated collectively, only two values are reported, total objects detected and total area covered. Feature specific mode, on the other hand, deals individually with each detected screen object and evaluated them for size and position. For direct count determinations, field specific mode was utilized, while feature specific mode was used for cell size, volume and biomass estimates. System status was displayed on the Apple IIe monitor. The screen was divided into three windows, system parameters, operational menus, and data. The system parameters included information on the size, shape and location of the enhancement window and the digi tization level. The operational menus contained all of the measurement functions as well as utilities used to manipulate the digitized images. The data window contained the most recently acquired measurements, displaying X and Y coordinates, area, perimeter, height, width, and length in the feature specific mode and number and area in field specific mode. The Model 2000 scans the digitized image in four planes; horizontal, vertical, and both diagonals. The height of an object was reported as the


17 maximum vertical distance, the breadth as the maximum horizontal and the width and length as the minimum and maximum distances respectively regardless of orientation. Because the bacteria could not be arranged in a perfect horizontal/ vertical/diagonal orientation, the width and length values were used in volume/biomass calculations. The enhancement window was the portion of the camera image which was digitized. Several shapes could be chosen for the window: a circle, square or rectangle. The position and size of the window could be adjusted as required. In this work, the square window was used to facilitate the duplicate counts made manually and with the Model 2000. Manual counts were made using an ocular grid of 100 blocks (10 X 10). Due to the rapid fading rates of fluorochrome stained bacteria, a 4 X 4 or 5 X 5 section of the grid was counted. The window was superimposed over this portion of the field. The center section of the field was examined to avoid distortion due to the curvature of the UV-FL100 objective lens (Sieracki et al. 1985). The Model 2000 distinguished 256 grey levels. Grey or enhancement levels are based on the intensity of screen objects and were the demarcation above which pixels were digitized, below which they were not. This level was manually adjusted by the operator as specific conditions warranted. Increasing the digitization level allowed the operator to exclude weakly fluorescing non-bacterial objects and prevent


18 scintillation. Scintillation was a phenomenon that resulted in the undesired digitization of numerous pixels clustered around a brightly fluorescing screen object. However, if the grey level was set too high, small bacteria were occasionally missed and large bacteria were not completely digitized. There was considerable variability between samples and the grey level was adjusted for each sample. The range of grey levels utilized in this project was between 235 and 255 unless otherwise noted. DNA Staining DAPI (Sigma Chemical Company, st. Louis, MO) was chosen as the fluorochrome stain in this project. Several other stains described above were not chosen for a variety of reasons. Primulin performs best under acid conditions, FITC and proflavine fluorescence maxima overlap with chlorophyll a, acridine orange's intercalating properties are a safety hazard and Hoechst 33258 and 33342 do not stain as brightly as DAPI (Francisco et al. 1973, Davis and Sieburth 1982, Paul 1982, Caron 1983). DAPI has a fluorescence maximum over a wide range of pH (5-10), does not overlap chlorophyll a, and often yields the brightest fluorescent images (Brunk et al. 1979). A 1. o x 10-3 M stock solution of DAPI was prepared by hydrating the powdered stain with deionized water. Fresh stain was filtered through a 0.2 pm polycarbonate filter and


19 1. 0 ml aliquots were stored in the dark in 1. 5 ml microcentrifuge tubes at -20 c. Immediately prior to use, the frozen aliquots were thawed and refiltered through a 0.2 polycarbonate membrane. Freshwater samples were stained at a final concentration of 1.0 x 105 M and marine samples at 2.0 x 105 M unless otherwise noted. stain was added to sample tubes with a 25-100 Hamilton syringe. The DAPI was allowed to bind for 10 minutes before the samples were filtered for examination. While some photodegradation was unavoidable, this problem was minimized by wrapping the sample tubes in aluminum foil and storing them in the dark during staining. As previously noted, Nuclepore filters fluoresce under ultraviolet light. To eliminate this problem the filters were counterstained with irgalen black (Paul 1982). Irgalen black powder was diluted to 0.2% (wtjvol) in 2.0% (voljvol) acetic acid. 0.2 Nuclepore filters (25 mm diameter) were immersed in this solution. While thirty minutes was the minimum time required for counterstaining, the filters could be left in the irgalen black solution for several weeks with no deleterious effects, provided they were stored at 4.0 C to prevent bacterial contamination. Before use, the excess counterstain was removed by two rinses in deionized water. Slide Preparation Once the bacterial samples were stained they were


20 concentrated on counterstained 0.2 Nuclepore filters. The filters were placed on a 25 mm glass and stainless steel filtration tower (Paul 1982). One to five ml samples were filtered under vacuum (125-175 mm Hg) and placed on 76 mm x 25 mm x 1 mm glass slides. These volumes produced the most accurate counts in a filter volume comparison by Jones and Simon (1975). One drop of low fluorescence silicon immersion oil (nd = 1. 404) was spread over the slide to provide a surface conducive to filter adhesion. This technique was later discarded in favor of the breath fogging technique of Hobbie et al. ( 1977) where condensation from the breath provides a hydrophilic surface on the slide improving adhesion of the wet filters. The breath fogging technique provided an optically flatter slide, the benefits of which are described below. In either case a drop of immersion oil was applied to the top of the filter and a No. cover slip added. Slides were pressed under a 2.0 kg weight in the dark for at least two hours to insure optically flat viewing areas. Calibration A prerequisite for the measurement aspect of the project was calibration of the image analysis system. Uniform fluorescent microspheres were tested initially in a technique reported by Bjornsen (1986). After several unsuccessful attempts, this procedure was discarded in favor of a method


21 employing a stage micrometer and interactive features of the image analysis system. The same stage micrometer was also used to calibrate the ocular micrometer used for comparison measurements. Latex spheres containing a highly uv-excitable fluorescent component were acquired in five diameters between 0. 2 pm and 1.1 pm (Polysciences Inc., Warrington, PA). Suspensions of spheres were made with deionized water and five 1: 10 serial dilutions performed. Slides were prepared by the method described above for the preparation of bacterial slides, namely filtration onto counterstained polycarbonate filters and placement on glass slides. Slides were made and examined at all dilutions. The smaller-sized spheres proved inadequate due to excessive clumping while the largest size spheres were neither sufficiently round nor uniform in diameter. Spheres of 0.60 pm and 0.77 pm were acceptably dispersed and uniform for use in this calibration. The dilution most resembling the cell density most commonly encountered with natural bacterial populations, 4. 0 x 105 cellsml"1 was utilized for size calibration. This density averaged 12.3 spheres per field. The spheres fluoresced at a significantly greater intensity than bacteria stained with DAPI and scintillation resulted in erroneous results at the grey level threshold level of 235, normally used with bacteria. At a level of 100115 the spheres no longer scintillated and the size of


22 digitized images corresponded well to that of the ocular images. The Model 2000 normally reports detected images as the number of pixels digitized. If the size of the digitized subject is known, a calibration feature can be programmed into the system, resulting in linear values replacing pixels as the reported measurement. This was attempted using the 0.60 and 0.77 spheres, whereupon a flaw with this calibration method quickly manifested itself. By raising or lowering the focal plane, the area of digitized pixels could be considerably altered, with a concomitant variation in the pixels micron-1 value. Also numerous single pixel images resulting from the scintillation of the spheres were included in the reported data. A series of commands was inserted in the measure program to erode the perimeter of all digitized images by one pixel and then dilate all remaining digitized images by one pixel. This resulted in little effect on digitized bacterial images, however the single pixel "noise" was removed (See Appendix 1 for program). Due to the large difference between digitization levels for bacteria and the spheres as well as the potential for error incurred when determining the correct focal plane, and therefore the pixelsmicron-1 value, the spheres were discarded as a calibration tool. However, several software manipulations (described below) were developed that proved useful in cell measurements. The next attempt at calibration included the use of


anbther interactive feature of the Model 2000. 23 A stage micrometer (American Optical, Southbridge, MA) with smallest divisions = 10"5 m was placed on the stage and the scale brought into focus. The analysis system projected the micrometer image on the video monitor and the trackball was used to draw a line between micrometer divisions. The system counted and reported the number of pixels in the line. Lines between 10 and 90 were drawn in horizontal and vertical planes. Pixelsmicron1 values were determined. The horizontal value was 5.655 and the vertical value was 5. 655 pixels There was no statistical difference between these values (95% confidence level). This illustrates one of the advantages of the ITC Model 2000 over other image analysis systems which use rectangular pixels. Rectangular pixels create difficulties in calibration as one value is needed for the horizontal plane, one for the vertical, and hundreds for the many diagonals (Bradbury 1979). The reciprocal of the 5. 655 pixels value is o .177 pixel"1 this value is input into the system controller resulting in subsequent measurements reported in microns rather than pixels. Between the camera and the light path selector knob, there is an adjustment to slightly increase or decrease the magnification of the image by moving a lens in the camera tube. The value of 0.177 was acquired at the lowest magnification setting. This value would decrease if the


24 magnification were increased and thus would improve resolving power. There was a slight decrease in image quality at the higher magnification but the improvement in resolving power more than compensated for it. At the highest parfocal magnification, the stage micrometer calibration was repeated and the resulting ,urn pixel-1 value was o .119. The determination of this value was repeated anytime there was an adjustment in the optical path, i.e. camera change, phototube replacement, etc. Sample Locations The ultimate goal of this project was to count and size natural populations of aquatic bacteria, however the environmental samples containing these organisms also included undesired non-bacterial materials (sediments, algae, etc.) which complicated development of the technique. Therefore, the initial work was conducted using pure cultures of laboratory grown organisms. In addition to purity, the cultured samples offered the advantages of large cells of relatively homogeneous size and shape. The two cultured organisms studied were Vibrio proteo/yticus (Jeffrey and Paul 1986), an estuarine organism (obtained from John Paul, University of South Florida, St. Petersburg, FL), and Pseud omo nas aemgino sa a pathogenic soil organism containing the pR0-2317 plasmid (obtained from Stephen Cuskey, us EPA,


25 Gulf Breeze, FL). Two unidentified marine isolates, #6 and #9, (obtained from J. Paul) were also studied. Environmental samples were taken from five areas in West Central Florida and the adjacent Gulf of Mexico: the Alafia River, a eutrophic river heavily influenced by agricultural runoff and phosphate mining (eight sample locations between headwaters and river mouth), Bayboro Harbor, a eutrophic embayment of Tampa Bay adjacent to the University of South Florida's Marine Science Department, Crystal River, an oligotrophic river (four sample locations between headwaters and river mouth), an oligotrophic offshore station in the Gulf of Mexico approximately 80 km northeast of the Dry Tortugas (25o241N, 82 o50'W), and from the coral surface microlayer (CSM) of coral heads 100 m west of Key in the Dry Tortugas (24oJ71N, 82o551W). Samples from Bayboro Harbor were collected from the surface by 5.6 l bucket, from the surface of the Alafia and Crystal Rivers by acid-washed 20 l plastic carboys, and from 10 m in the Gulf of Mexico by 8 l Niskin bottle. The CSM samples were collected with 60 ml syringes from the coral heads by divers. Triplicate 20 ml subsamples were taken as rapidly as possible and fixed with 0.2 filtered formalin (final formaldehyde concentration 1.85% voljvol).


26 RESULTS AND DISCUSSION Standard Camera Enumeration The simplest task demanded of the Model 2000 was counting of bacterial cells. Slides were prepared in the manner described above and counts were made on the same fields manually and by image analysis. These counts were then statistically tested for similarity by t-test. The first investigations were made on pure cultures of Vibrio proteolyticus. Overnight cultures of this bacteria were grown, fixed, and stained as described above. Slides were prepared and evaluated three ways: manually through the oculars, manually off the image monitor, and with the image analysis system. Ten fields were evaluated from each of two slides. Mean count field"1 values were multiplied by a conversion factor to generate cells ml"1 values. This conversion factor took into account dilution rate, volume filtered, and percentage of filter area counted. The field counts determined by the three methods are listed in Table 1 Multivariate ANOVA with repeated measures of field counts revealed no significant difference between the three methods (P = 0.16) (Statistical Analysis Software, SAS Institute, Cary, NC) However, field counts generated manually from the


Screen 20 17 25 14 20 20 19 27 28 21 Table 1. Field count comparisons for Vibrio proteolyticus. Standard camera, Newvicon phototube Slide 1 Slide 2 Scope IAS Screen Scope 17 24 31 24 14 18 27 23 23 13 9 9 12 17 25 21 18 28 22 19 17 28 17 16 16 15 21 18 23 29 20 16 25 22 19 17 20 17 29 25 All values in cells field-1 Multivariate ANOVA of Repeated Measures IAS 19 23 20 15 18 26 31 26 13 22 No Significant Difference Between Methods by Observation p = 0.1569 Paired Sample t Tests Methods Slide 1 Slide 2 Manual/IAS 5.52 4.45 Critical t Value t0.05(2)9 = 2.262 27


28 monitor screen are not used in standard bacterial enumeration methods, therefore this method was dropped from the analysis. A paired sample t test of the two more standard methods, manual and image analysis, found them to be statistically different at the 95% confidence level (Zar, 1974). The paired sample t test was used to evaluate similarity between manually and machine generated values in the subsequent determinations of this project. The paired sample t test is a powerful and appropriate statistic to use in this situation. Because this project compared individual field counts and cell measurements, the paired sample t test was superior to the standard Student's t test designed to evaluate similarity between group means (Zar 1974). The digitization level of the analysis system was adjusted to improve image detection and another sample of Isolate #9 was examined. Table 2 lists the comparative field counts and the paired sample t test values. The discrepancies between manual and machine counts were less than in the previous examinations, however the two methods were still not similar at the 95% confidence level. The system would be required to enumerate natural populations in addition to cultures, therefore the next test was conducted on an environmental sample from Bayboro harbor. Previous examinations of Bayboro Harbor samples revealed large numbers of detrital particles, autotrophs, and larger flora and fauna. This high concentration of non-microbial


Table 2. Field count comparisons for Isolate #9 Standard camera, Newvicon phototube Isolate #9 Isolate #9 ManjiAS Man/IAS 18/20 19/18 22/21 27/26 14/14 27/27 18/22 13/15 21/24 22/25 24/25 19/21 29/31 15/18 21/25 21/23 18/21 28/33 18/22 18/19 t value t value 4.00 2.67 All values in cells field-1 Critical t value = 2.262 29


30 constituents led us to investigate prefiltration as a method of reducing the non-bacterial population. Three fractions were prepared, 3. 0 f..Lm prefiltration, 1. o f..Lm prefiltration, and unfiltered. Nuclepore polycarbonate filters were used for prefiltration. All subsamples were fixed with 0.2 J.Lm filtered formalin (final concentration 5% voljvol) and stained with 5.0 f..LM DAPI. Three slides of each of the six subsamples were prepared by filtering 1. 0 ml onto a counterstained 0. 2 f..Lm Nuclepore filter. As in the initial investigations, a representative field was brought into focus and the image frozen with the image analyzer. Prior to digitization and automatic counting, the field was counted manually through the ocular lenses. Twenty fields were evaluated for each slide. The field comparisons are presented in Table 3. A paired sample t test reveals statistically different manual and machine counts for any of the fractions at the 95% confidence level . The unfiltered samples were, however, more closely related to one another, having lower t values, than the 1.0 and 3.0 f..Lm filtered samples (median t for unfiltered= 3.94, 3.0 J.Lm = 8.16, and 1.0 f..Lm = 13.71). It was concluded the filters caused disintegration of some bacteria or autofluorescing detrital particles. Some of these small particles were misidentified as small bacteria by the operator, but were too faint to be detected by the image analysis system, and thus imparted error that was not present in unfiltered samples. When mean cellml"1 values were


31 Table 3. Field count comparisons for Bayboro Harbor Prefiltration evaluation. Standard camera, Newvicon phototube no prefilt 3.0 f.Lm filter 1. o f.Lm filter Mn/IAS Mn/IAS Mn/IAS Mn/IAS Mn/IAS Mn/IAS 21/18 19/18 11/9 22/10 9/3 13/9 16/13 18/17 15/11 10/8 11/7 11/6 14/10 17/15 14/10 18/15 12/9 9/5 22/19 17/16 19/19 10/6 10/7 13/7 20/17 14/15 16/15 10/6 10/6 14/11 17/18 18/18 12/8 10/9 12/9 12/9 14/14 12/11 15/11 17/15 11/8 13/11 24/20 15/12 13/11 13/9 10/9 9/6 23/20 18/17 12/8 10/7 15/9 12/7 15/16 16/15 11/9 10/9 13/7 17/14 20/21 15/15 11/9 11/7 14/10 15/9 15/12 20/16 12/10 10/8 11/7 10/6 23/19 20/22 13/9 9/7 14/8 13/8 15/14 14/13 12/10 12/9 14/12 14/10 12/9 19/16 14/11 11/10 22/19 13/7 20/17 17/14 14/11 10/7 17/13 10/7 20/21 12/10 10/6 12/9 15/9 14/9 14/12 16/14 14/10 11/6 16/10 16/13 20/21 13/15 8/6 12/10 12/7 11/7 15/13 13/13 11/8 10/8 15/14 14/11 t t t t t t value value value value value value 4.96 2.92 10.42 5.89 10.83 16.59 All values cells field"1 Critical t value t0.05(2)19 = 2 09 3


32 calculated for each slide (Table 4), none of the means were similar at the 95% confidence level. Table 4. Mean cellml-1 comparisons for Bayboro Harbor Prefiltration evaluation Standard camera, Newvicon phototube Fraction Manual IAS cells rnl1 S.D. cells rnr1 + S.D. Unfiltered 1. 65 X 106 0.13 1.51 X 106 0.06 3.0 J.Lm filt 1. 20 X 106 0.07 9 .09 X 105 + 0.89 1.0 J.Lm filt 1.25 X 106 0.03 8.59 X 105 0 .38 Critical t value t0.05(19) = 2. 093 t value 2.140 4.700 6.450 cells rnl1 values calculated from 20 fields per fraction Prefiltration was investigated as a means to remove non-bacterial particles from mixed assemblages. However, in addition to increasing preparation time, prefiltration decreased the agreement between manual and machine generated counts. Prefiltration was discontinued at this point in the project. Samples obtained during an oceanographic research cruise were examined next. These samples carne from two sources: an oligotrophic surface station approximately 80 krn northeast of the Dry Tortugas (25'N, 8250'W) and the coral surface microlayer (CSM) from a patch reef 100 m west of Loggerhead Key (2437'N, 8255'W) in the Dry Tortugas. The CSM is a nutrient-rich environment consisting of mucus exuded by coral polyps during a self-cleaning process. In addition to large


33 numbers of bacteria, the CSM contains numerous zooxanthellae expelled by the polyps. When illuminated with the ultraviolet light used to excite the DAPI-DNA complex of stained bacteria, these zooxanthellae autofluoresced intensely. This extraneous light overwhelmed the weakly fluorescing bacteria and prevented their detection with the image analysis system. Manual direct counts of CSM samples were also very difficult to generate accurately due to bacteria adhering to the zooxanthellae. Bacteria adhering to the undersides of zooxanthellae were not visible and thus uncounted. Owing to the impossibility of acquiring machine generated counts, no manual/machine comparison counts were made with CSM samples, and there was such little confidence in the manual counts that the values are not reported here. The slides from the oligotrophic surface station revealed another deficiency in the image analysis system as it was originally configured. The bacteria in this sample were very small and produced extremely weak fluorescence during ultraviolet excitation. While visible to the human eye, many cells were either completely undetected by the image analyzer or faded below the detection threshold too rapidly to be routinely counted. The results from this sample are displayed in Table 5. There was no similarity between field counts made manually and with the image analysis system. These observations led to the first of several attempts to improve the detection threshold, replacement of the video


Table 5. Field count comparisons for Offshore station (25'N, 82'W) Standard camera, Newv icon phototube Slide 1 Slide 2 ManjiAS Man/IAS 24/8 31/14 23/4 24/7 32/8 28/8 35/10 18/5 17/4 27/6 24/6 19/4 16/6 28/5 28/5 28/13 25/5 30/9 20/6 27/6 19/4 20/7 28/10 24/6 17/4 23/6 23/6 24/11 29/5 15/4 31/7 25/9 28/12 27/6 19/6 29/12 21/5 31/8 21/6 19/3 t value t value 17.87 22.04 Critical t Value to. osc2>19 = 2.093 34


35 camera phototube. The above work was conducted with a Newvicon phototube. This was replaced with a Chalnicon phototube. As in all other cases when a change was made in the light path, the pixel J.Lm-1 value was recalculated. Slides from a pure culture, isolate #6, were counted with the Chalnicon phototube. The results from comparison counts of four slides are displayed in Table 6. Paired sample t tests revealed statistical similarity at the 95% confidence level for all slides in this sample. Additionally, group means, expressed as cells ml-1 were similar at the 95% confidence level (Table 7). Examination of the individual counts in Table 6 indicates no systematic error (i.e. discrepancies between manual and machine counts were relatively evenly divided between high manual/low machine and low manual/high machine observations). With the original Newvicon phototube discrepancies were usually high manual/low machine. The new Chalnicon phototube had superior light gathering ability. Routine examination of Vibrio proteolyricus cultures in log phase growth revealed large clumps of dividing cells. These clumps were difficult to count with the image analysis system. When cells make contact with each other, the Model 2000 cannot differentiate between individual cells. This inability to differentiate individual cells in a group of clumped cells results in the generation of erroneously low cellml-1 values and overestimates of cell volumes. To ameliorate clumping,


Table 6. Field count comparisons for Isolate #6 Standard camera, Chalnicon phototube Slide A Slide B Slide c Slide D Man /IAS Man /IAS Man /IAS ManjiAS 21/21 16/17 11/11 13/12 22/20 16/17 11/15 15/15 18/20 21/22 12/10 14/13 22/22 24/25 11/11 17/16 24/26 20/21 11/9 18/20 18/19 20/ 1 8 12/13 15/15 19/20 19/19 15/16 15/13 21/21 17/15 15/16 12/12 23/27 18/19 15/16 13/13 37/35 18/1 8 12/12 14/15 20/21 15/15 13/11 15/15 24/25 22/20 11/11 14/14 20/22 16/16 11/11 12/11 19/18 17/17 13/13 12/12 21/21 19/17 11/12 11/11 20/21 17/16 12/12 14/13 22/20 16/16 13/13 16/15 22/22 17/16 11/11 13/13 24/25 18/17 12/11 14/13 19/21 17/15 11/11 11/11 t value t value t value t value 1.60 1. 33 0.34 1. 58 All values in cells field-1 Critical t value t0.05(2)1 9 = 2.093 36


37 Table 7. Mean cellml-1 comparisons for Isolate #6 Standard camera, Chalnicon phototube Sample Manual IAS t Value Cells mr1 S.D. Cells ml-1 + S.D. Slide A 5.02 X 108 0.93 5.15 X 108 + 0.87 0.44 Slide B 4.18 X 108 0.53 4.10 X 108 0.60 0.45 Slide c 2.80 X 108 0.32 2.82 X 108 + 0.47 0.18 Slide D 3.20 X 108 0.43 3.13 X 108 0 .49 0 .47 Critical t Value = 2.093 a surfactant was applied to log phase culture samples. 20 pl of 5% Triton X-100 was added to 2.0 ml of bacterial culture (5 x 10-5 M final concentration) in an attempt to disassociate the cells. No noticeable effect was detected. Triton concentration was quintupled to 2. 5 x 10-4 M final concentration, again with no improvement. Similarly, agitation of cells suspended in 2.5 x 10-4 M Triton X-100 on a Vortex-Genie test tube mixer did not reduce cell clumping. Sonication separated clumped cells but also disrupted numerous cell membranes. As an alternative method for counting pure cultures, nutrient deprivation of cultures was examined next. Inocula of rapidly growing cultures were added to isotonic solutions of nutrient free media (artificial seawater). These cultures were incubated at optimum growth temperatures and examined after 36, 108, 204 and 360 h. Clumping continued to adversely affect machine counts at the 36, 108, and 204 hour points. However, at 360 h the cells were monodispersed and


38 easily counted with the image analysis system. Standard. Camera Measurements The 3 60 hour starved cells were also very uniform in size and shape. This starvation procedure was repeated in pure culture evaluations during the biomass estimation aspect of this project described below. It was possible to measure pure cultures that were starved for less than 15 days, but more time was required to locate single cells. At 15 days, all cells were uniformly dispersed. An increase in cell length was seen after the starvation period, similar to the bottle effects reported by Ferguson et al. (1984). The Model 2000 reported cell measurements in four axes, (height, width, length and breadth) but did not provide volume information. From the linear dimensions, area was easily computed with a simple calculation. Cell volume was initially determined with the formula of a sphere, V = where V is volume and r is radius. Although coccoid bacteria are spherical, the majority of bacteria encountered in the marine environment are rod shaped, therefore, it was decided to compute cell volume with the formula for a capsule, a cylinder with hemispherical ends. This formula, reported by Krambeck et al. (1981), is Volume= W/3) where W =cell width and L = cell length. Van Wambeke (1988) evaluated four shapes (circle, oval, rectangle, and capsule) for accuracy in


39 volume determinations of marine planktonic bacteria. Van Wambeke found the capsule to provide the most accurate volume values. A separate computer program was written in BASIC to compute cell volume with this formula when given cell widths and lengths (See Appendix 2) Sieracki et al. ( 1989b) developed an algorithm to estimate cell volume from a twodimensional image. This method assumed the two-dimensional digitized image represented the maximum diameter of a round or cylindrical object. Summation of the volumes of successive one pixel wide disks provided total cell biovolume estimates. Accurate cell counts could be generated as long as at least part of each cell was digitized. Accurate cell measurements required that each cell be digitized in its entirety. A visual comparison of digitized images and camera images estimated that approximately 50% of screen objects were completely digitized. Increasing the concentration of DAPI (up to four fold that normally used) andjor the staining time (to thirty minutes rather than ten) did not increase the intensity of DAPI/DNA fluorescence. In the counting phase of the project, over 90% of digitized were acceptable, however, for the measuring phase this value decreased to 1015% Even though half of the individual cells were correctly digitized in any one field, it was difficult to find a field where all of the cells were properly digitized. Decreasing the cell density of the sample reduced the number of bacteria per field, thus increasing the likelihood of a completely


40 digitized object in the field. The mean number of bacteria per field was 14.3 for counting, but was reduced to 1.5 for measuring. Because the DAPI/DNA fluorescence began to fade immediately after exposure to ultraviolet light, the image had to be grabbed very rapidly after it had been brought into the field of view. This rapid slide manipulation often did not allow adequate time to properly focus the microscope, and thus many images were discarded due to an incorrect focal plane. The difficulty in obtaining accurate images dramatically increased the time required to acquire data. The speed of data acquisition was a major factor in developing automated counting and measuring systems. Other investigators have also investigated methods for correct edge determination. Sieracki et al. (1989a) evaluated several mathematical techniques to most accurately determine cell edges. Best results were attained by using the minimum of the second derivative of the intensity gradient encompassing the cell edge. Eight day starved cells of Vibrio proteolyticus were measured by machine and manually with an ocular micrometer (calibration described above) Length and width measurements were generated and fed into a separate computer for calculation of volume by the above described BABI formula for capsules. In addition to the pure culture, manual and machine measurements were compared from cells in Bayboro Harbor. Due to the difficulty in acquiring acceptable images with the standard camera these were not same cell comparisons and thus the less I


41 _powerful pooled sample t-test was applied to the results. The results showed similarity at the 95% confidence level for all coml'_arisons except length in V. proteolyticus (Table 8) Table 8. Mean cell size comparison for Vibrio proteolyticus and Bayboro Harbor Standard camera, Chalnicon phototube Measurement N Manual IAS Man/IAS Mean S.D. Mean + S.D. Length V.p. 25/178 1.62 0.50 1.13 1.21 Breadth V.p. 25/178 0.66 0.29 0 70 0.70 Volume V.p. 25/178 0.49 0.36 0.48 0.64 Length BBH 25/186 1.00 0.40 0.62 1.21 Breadth BBH 25/186 0.56 0.24 0.40 0.90 Volume BBH 25/186 0.28 0.35 0.18 0.31 Critical t Value t0.05(2)2 = 1.972 Length and Breadth values in Volume values in V.p. Vibrio proteolyticus BBH Bayboro Harbor t 2 05 1.65 0.28 0.08 0.88 1.49 However, these similarities are quite likely statistical anomalies caused by the high standard deviations about the means due to comparison of group means rather than same cell comparisons (seven of the twelve values have standard deviations equal to or exceeding the mean). With considerable effort, a same cell comparison of 15 day starved V. proteolyticus and Bayboro Harbor cells was made. Table 9 displays the results of these comparisons, only the breadth and volume


Table 9. Same cell comparison for Vibrio proteolyticus (Starved 5 days) and Bayboro Harbor Standard camera, Chalnicon phototube Measurement N Manual IAS Mean S.D. Mean + S.D. Length V.p 50 1.52 0.38 1.44 0.45 Breadth V.p. 50 0.96 0.23 0.94 0.18 Volume V.p. 50 0.99 0.84 0.86 0.50 Length BBH 50 0.92 0.29 0.80 + 0.24 Breadth BBH 50 0.72 0.27 0.57 + 0.22 Volume BBH 50 0.36 0 .33 0.20 + 0.19 Critical t Value t0.05(2)49 = 2. 009 Length and Breadth values in Volume values in V.p Vibrio pro teolyti c us BBH BayboroHarbor of V. prote o lyti c us passed this more sensitive test. 42 t Value 2.05 1.07 0.05 5.36 5.66 7.33 Because nothing could be done to increase fluorescent e missions, an effort was made to increase light detection. We therefore purchased a Model 66 silicon intensified target (SIT) c amera from D age-MTI (Michigan City, IN). SIT Camera Enumerations Upon initial installation of the SIT-66 it became apparent that there were several adjustments that had to be made to make the camera compatible with the Model 2000 analysis system. Images produced by the SIT-66 camera were as


43 they appear to the eye, that is, light objects were light and dark objects were dark. Unfortunately, the Model 2000 analysis system required an inverted video signal, resembling a photographic negative, where light objects were dark and dark objects were light. The horizontal and vertical video synchronization signals, required to coordinate the frame generation rates of the camera and image analyzer, were also incompatible. Several modifications were made to the SIT-66 camera to accommodate the image analysis system's video requirements without success. The SIT-66 had the two sync signals combined with the video signal in a single composite video output. The principal difficulty was separating the horizontal and vertical video synchronization from the video signal so the latter could be inverted. An external synchronization separator was constructed by Mr. Larry Lindner in the USF Marine Electronics workshop and inserted between the SIT-66 camera and the analysis system. Separating the composite video signal into its component parts allowed inversion of the video signal but a frame rate synchronization problem persisted. The sync separator also injected electronic noise into the system which resulted in a grainy video image. Eventually the sync rate problem was corrected through hardware manipulation on the gen-lock board of the video camera and video inversion was accomplished through software operations of the image analysis system. It was then possible to begin investigations with the SIT-66 camera.


44 The SIT-66 camera had much greater light gathering ability than the conventional camera that it replaced. Unfortunately, the electronic intensifier that allowed this light gathering ability also introduced a significant amount of electronic noise into the video signal (Reynolds and Taylor 1980), even without the above mentioned sync separator. This noise resulted in a grainy appearance in the generated image. A control panel supplied with the SIT-66 camera allowed manipulation of the gain and kV levels of the new camera. Adjustments of these controls (decrease gain, increase kV) reduced the majority of the noise, further improvements were achieved by decreasing the digitization level of the Model 2000 to 225 from 250 used with the standard camera. The end result of these manipulations was an image exhibiting only slightly more noise than the standard camera but having far greater light gathering ability. The benefits far outweighed the detriments. The SIT-66 camera was initially evaluated with cell counts. In determinations of pure cultures and environmental samples, the SIT-66 far surpassed the camera. Not only were manual and machine counts statistically similar at the 95% confidence level, they were nearly identical (Table 10). Of the 80 fields counted, only three had any discrepancy between manual and machine generated counts (the largest difference was two cells), and unlike the standard cameraderived counts the higher counts were produced by the image


-Table 10. Field count comparison for Vibrio proteolyticus and Bayboro Harbor SIT-66 Camera Vibrio proteolyticus Bayboro Harbor ManjMac ManjMac ManjMac Man/Mac 23/23 17/17 21/21 18/18 18/18 19/19 23/23 12/12 17/17 16/16 19/19 17/17 29/29 25/25 21/20 21/21 21/21 20/20 27/27 23/23 21/21 24/24 23/23 21/21 19/19 21/21 24/24 18/18 17/17 18/18 13/13 20/20 23/23 23/23 19/19 28/28 18/18 23/23 22/22 21/21 11/11 21/21 16/16 19/19 16/16 16/16 19/19 24/22 22/22 9/9 19/19 16/16 19/19 25/25 20/19 20/20 24/24 21/21 23/23 21/21 19/19 17/17 17/17 18/18 30/30 14/14 24/24 27/27 21/21 20/20 19/19 22/22 18/18 23/23 18/18 19/19 20/20 19/19 23/23 25/25 t value t value t value t value o.oo 0.00 1.43 1. 43 All values are cells field-1 Critical t value t0.05(2)19 = 2 0 093 45


46 analysis system. Frozen images were only discarded due to focusing problems caused by large depths of field on slides that were not optically "flat". The speed of counts far surpassed that achievable with manual determinations. The exceedingly bright images generated by the SIT-66 camera, and the significant similarity between manual and machine counts led us to terminate the counting phase of the project and conclude that the analysis system was able to consistently produce cell counts equal to or exceeding those made manually. SIT Camera Measurements With the counting phase successfully completed, attention focused on the cell measurement aspect of the project. A same cell comparison of cell lengths measured manually and with the image analysis system was performed to test the SIT-66 camera. As in the counting work, the SIT-66 provided high quality images that were rarely discarded for digitization deficiencies. Measurements were made on bacteria from a wide variety of sources: cultures of V. proteolyticus, and natural populations from Bayboro Harbor, Offshore Gulf of Mexico, Alafia River, and Crystal River. 50 cells from each source were measured along the longest axis. The results are listed in Table 11. All comparisons were statistically similar at the 95% confidence level. Not only were all measurements similar, but they were obtained with relative ease, only


Table 11. Same cell length comparison for bacteria from five sources. SIT-66 Camera. Measurement N Manual Machine Mean S.D. Mean + S.D. Length V.p. 50 0.96 0.27 0.97 0 .26 Length BBH 50 0.72 + 0.26 0.73 + 0.25 Length Off 50 0.87 0.45 0.87 0.44 Length Ala 50 0.77 + 0.29 0.78 + 0.28 Length CR 50 0.94 0.25 0.95 0 .25 Critical t Value t0. 05(2)49 = 2.009 Length values in V.p. Vibrio proteolyticus BBH Bayboro Harbor Off Offshore Ala Alafia River CR Crystal River 47 t Value 0.89 0.68 0.75 0.97 0.98 rarely would the digitization of a: field be unacceptable. Properly digitized fields were encountered over 90% of the time. A field was properly digitized when all bacteria were detected and the digitized pixels superimposed over them were of similar size and shape. Another method of evaluating the accuracy of cell measurements is examination of their range and distribution. Figures 3 -7 illustrate the distribution of 50 cells from each of five environments: cultured V. proteolyticus, and natural populations from Bayboro Harbor, Alafia River, Crystal River, and the offshore station. The distributions are categorized into 0 5 size ranges and values are presented for manual


50 45 40 35 30 <1> ..0 25 E :J z 20 15 10 5 0 Freshwater Organisms Alafia River 0.00-0.50 Method Ill Manual Cell Length (urn) 0.51-1 00 1.01-1.50 1.51-2.00 Figure 3. Length distribution of Alafia River bacteria. ())


Freshwater Organisms 40 35 30 25 ..... 20 E ::J z 15 10 5 0 ...._I ----,=..,.___-Crystal River 0 00-0 .50 0.51-1 00 1 .011 .50 1.51-2 .00 Cell Length (urn) Method Figure 4 Length distribution o f C rystal River bacteria. """ \0


L-Q) ..a E ::l z 50 45 40 35 30 25 20 15 10 5 0 Marine Organisms Bayboro Harbor 0.00-0.50 0.51-1.00 1.01-1 .50 1.51-2.00 Cell Length (um) Method Figure 5. Length distribution of Bayboro Harbor bacteria. U1 0


45 40 35 30 Q5 25 ...0 20 z 15 10 5 Marine Organisms Vibrio proteolyticus 0 I f''-."\.'\. '"\'\'i] !i 0.00-0.50 0.51-1.00 1 .01-1. 50 1.51-2.00 C e ll Length (urn) Method lAS Figure 6. Length distribution of Vibno proteolyticus 01 t-'


..... <1) ..Q E :::J z 25 23 21 19 17 15 13 11 9 7 5 3 Marine Organisms Offshore 25 24' N 82 50' W 0.00-0.50 0.51-1.00 1.01-1 .50 1.51-2.00 Cell Length (um) Method Figure 7. Length distributio n of offshore bacteria. U1 IV


53 and machine generated determinations. Of the 250 cells measured, one V. proteolyticus cell and one offshore cell were classified in different size groups by each measurement method. The shapes of the distributions are similar to those reported by Fry and Davies (1985) for freshwater organisms and Maeda and Taga (1983) for marine organisms. Maeda and Taga (1983) also report narrower size ranges for mesotrophic samples compared to eutrophic and oligotrophic samples. This trend is not clearly seen here where four of the samples have similar distributions, while the offshore sample does have a broader distribution. An unrelated project evaluating production of extracellular DNA by genetically engineered microorganisms (GEMs) provided an additional source of bacteria in pure culture for study in this project. Cells from one of these organisms, Pseudomonasaeruginosa with the pR0-2317 plasmid, were measured manually and by machine for length and breadth. Cell volumes were calculated with the above described BABI formula. Results are shown in Table 12. All measurements were statistically similar at the 95% confidence interval. The volume value is consistent with that reported in cultured Escherichia coli by watson et al. ( 1977) of 2. 4 J.Lm3 cell-1 Biomass Calculations The final determinations to be generated were biomass


Table 12. Same cell comparison for Pseudomonas aeruginosa pR0-2317 SIT-66 Camera Measurement N Manual Machine Mean S.D. Mean + S.D. Length 30 7.40 + 0.52 7.43 + 0.51 Breadth 30 0.59 0.10 0.60 + 0.08 Volume 30 2.03 0.71 2.09 0.63 t Value = 2. 042 Length, Breadth values 1n Volume values in 54 t Value 0.95 1.09 0.86 values. These values were calculated by multiplying the mean cell volume by the cell density (cellsml-1 ) to derive a total cell volumemr1 and then multiplying this figure by a conversion factor. The conversion factor utilized in this project was the mean of factors reported in four marine investigations, 3 .53 x 1013 gCmr1 (Watson et al. 1977, Bratbak 1985, Bj0rnsen 1986, Lee and Fuhrman 1987). Cell lengths and breadths were measured for bacteria from four sources. : Alaf ia River, Crystal River, Bayboro Harbor, and Offshore. Volumes were calculated with the BABI formula and biomass values determined with the above conversion factor (Table 13). Cell values ranged between.0.16 for offshore and 0.26 for Alafia River and are consistent with the mean volume, 0.23 0.05, reported by others (Salonen 1977, watson et al. 1977, Fuhrman 1981, Fry and Zia 1982, Riemann 1983, Rublee et al. 1983, Bratbak 1985, Fry and Davies 1985, Sieracki et al. 1985, Bjornsen 1986, and Lewis et al. 1986). We had very high confidence in the SIT-66 camera-derived


55 Table 13. Cell counts, sizes, and biomass values for four sample locations. SIT-66 Camera. Sample Alafia R. Crystal R. Bayboro H. Offshore* N 63 55 61 52 Cellsml-1 1.14 X 107 2.95 X 105 2.43 X 106 4.95 X 105 0.02 + 0.00 + 0.03 + 0.08 Length 0.89 0.87 0.85 0.75 0.08 0.07 0.09 0.08 Breadth 0.62 0.54 0.54 0.50 0.07 0.07 0.08 0.07 Volume 0.26 0.22 0.22 0.16 0.19 0.23 0.21 0.14 gC mr1 1. 05 X 10-6 2.17 X 10-8 1. 89 X 10-7 2.80 X 10-8 Length and Breadth values in Volume values in Erode and dilate commands deleted to insure detection of small cells (single cell digitizations) measurements due to the previous cell measurement results, therefore no manual measurements were made. Biomass values ranged between 2. 80 x 10-8 gC ml-1 Offshore and 1. 05 x 10-6 counted in gCml-1 in the Alafia River, and are also consistent with the above authors. The temporal requirements for bacterial counts were improved significantly with the image analysis system and measurements were dramatically improved. The time needed to manually measure cells with the ocular micrometer was 15-20 seconds per cell, a rate of 3-4 cells minute-1 With image analysis, 30-40 cellsminute-1 was not an uncommon cell measurement rate. The temporal disparity between manual and machine counts was even greater. Typically 300 cells are


56 determining a cell density value for a bacterial sample. Using the image analysis system, over 14,000 individual cells were counted in one day, a figure corresponding to 47 samples, 3 8 more than the best day of manual determinations. In addition to the time savings, image analysis provided results unbiased by inter-operator variability and greatly reduced operator fatigue.


57 CONCLUSION Image analysis was developed in the 1950's by the coal industry to monitor mineshaft air quality (Bradbury 1983). Epifluorescence microscopy has been used for over 20 years as an enumeration technique for aquatic bacteria (Francisco et al. 1973) In 1984, Costello and Monk linked these two techniques for counting bacteria in milk. Subsequently, image analysis evaluations of epifluorescent images has been applied to a wide range of microbial studies. This paper has demonstrated the applicability of an ITC Model 2000 image analyzer coupled with an BH-2 epifluorescent microscope and Dage MTI SIT-66 video camera to accurately count and measure aquatic bacteria from a variety of cultured, freshwater, and marine environments. The results conclusively show that image analysis provides counts and measurements statistically similar to those made manually. The elimination of inter-operator bias and reduction of operator fatigue provided by image analysis justify its use in bacterial enumeration and measurements. When the temporal advantages are taken into account, image analysis far exceeds the capabilities of manually derived determinations in the applications evaluated in this project.


58 LITERATURE CITED of Appl. Bjornsen, P. K. 1986. Automatic determination bacterioplankton biomass by image analysis. Environ. Microbiol. 51:1199-1204. Borsheim, K. Y., G. Bratbak, and M. Beldal. 1990. Enumeration and biomass estimation of planktonic bacteria and viruses by transmission electron microscopy. Appl. Environ. Microbiol. 56:352-356. Bowden, w. B. 1977. Comparison of two direct techniques for enumerating aquatic bacteria. Environ. Microbiol. 33:1229-1232. count Appl. Bradbury, s. 1979. Microscopical image analysis: Problems and approaches. J. of Microsc. 115:137-150. Bradbury, s. 1983. Commercial image analysers and the characterization of microscopical images. J. of Microsc. 131:203-210. Bratbak, G. 1985. Bacterial biovolume and biomass estimations. Appl. Environ. Microbiol. 49:1488-1493. Brunk, c. F., K. c. Jones, and T. w. James. 1979. Assay for nanogram quantities of DNA in cellular homogenates. Anal. Biochem. 92:497-500. Caron, D. A. 1983. Technique for enumeration of heterotrophic and phototrophic nanoplankton, using epifluorescence microscopy, and comparison with other procedures. Appl. Environ. Microbiol. 46:491-498. Coleman, A. w. 1980. Enhanced detection of bacteria in natural environments by fluorochrome staining of DNA. Limnol. oceanogr. 25:948-951. commare, D. L. 1988. Video Microscopy: An integrated systems approach. Am. Lab. 20(12):56-65. costello, P. J. and P. R. Monk. 1985. Image analysis method for the rapid counting of Saccharomyces cerevisiae cells. Appl. Environ. Microbiol. 49:836-866.


59 Cynar, F. J., M. E. Sieracki, and J. MeN. Sieburth. 1985. Estimating plankton populations with the Apple II. Bioscience 35:652-655. David, A. W. and J. H. Paul. 1989. Enumeration and sizing of aquatic bacteria by use of a silicon-intensified target camera-linked image-analysis system. J. Microbiol. Methods 9(4):257-266. Davis, P. G. and J. MeN. Sieburth. 1982. Differentiation of phototrophic and heterotrophic nanoplankton populations by epifluorescence microscopy. Ann. Inst. Oceanogr. 58(Suppl.):249-260. Drapeau, A. J. and R. A. for the fluorescent Escherichia coli on a 11(10) :937-944. Laurence. counting membrane 1977. Pulsed dye laser of entero pathogenic filter. Water Res. Ferguson, R. L., E. N. Buckley, and A. v. Palumbo. 1984. Response of marine bacterioplankton to differential filtration and confinement. Appl. Environ. Microbiol. 47:49-55. Francisco, D. E., R. A. Mah, and A. c. Rabin. 1973. Acridine orange-epifluorescence technique for counting bacteria in natural waters. Trans. A.m. Microsc. Soc. 92:416-421. Fry, J. c. and A. R. Davies. 1985. An assessment of methods for measuring volumes of planktonic bacteria, with particular reference to television image analysis. J. Appl. Bact. 58:105-112. Fry, J. c. and T. zia. 1982. Viability of heterotrophic bacteria in freshwater. J. Gen. Microbiol. 128:2841-2850. Fuhrman, J. 1981. Influence of method on the apparent size distribution of bacterioplankton cells: Epifluorescence microscopy compared to scanning electron microscopy. Mar. Ecol. Prog. Ser. 5:103-106. Hazen, T. c. and L. Jimenez. identification of bacteria using nucleic acid probes. 1988. Enumeration and from environmental samples Microbiol. Sci. 5:340-343. Hiraoka, y., J. w. sedat, and D. A. Agard. 1987. The use of a charge-coupled device for quantitative optical microscopy of biological structures. Science 238:36-41. Hobbie, J. E., R. J. Daley, and s. Jasper. 1977. Use of


60 filters for counting bacteria by fluorescence m1croscopy. Appl. Environ. Microbial. 33:1225-1228. Inoue, s. and T. Inoue. 1989. Video enhancement and image processing in light microscopy. Part 2. Digital image processing. Am. Lab. 21(4):62-70. Jeffrey, W. H. and J. H. Paul. 1986. Activity of an attached and free living sp. as measured by thymidine incorporation, p-iodonitrotetrazolium reduction, and ATP/DNA ratios. Appl. Environ. Microbial. 51:150-156. Jones, J. G. and B. M. Simon. errors in direct counts epifluorescence microscopy, method for dyeing membrane 39:317-329. 1975. An investigation of of aquatic bacteria by with reference to a new filters. J. Appl. Bact. Krambeck, c., H. J. Krambeck, and J. Overbeck. 1981. Microcomputer-assisted biomass determination of plankton bacteria on scanning electron micrographs. Appl. Environ. Microbial. 42:142-149. Lee, s. and J. A. Fuhrman. 1987. Relationships between biovolume and biomass of naturally derived marine bacterioplankton. Appl. Environ. Microbial. 53:1298-1303. Lewis, w. M., T. Frost, and D. Morris. 1986. Studies of planktonic bacteria in Lake Valencia, Venezuela. Arch. Hydrobiol. 106:289-305. Li, w. K. w. and A. M. Wood. 1988. Vertical distribution of North Atlantic ultraplankton: analysis by flow cytometry and epifluorescence microscopy. Deep Sea Res. 35:1615-1638. Maeda, M. and N. 'l'aga. 1983. Comparisons of cell size of bacteria from four marine locations. La mer 21:207-210. Maranda, B. 1987. An economical, powerful system for videoenhanced microscopy. Am. Lab. 19(4):41-52. olson, R. J., s. L. Frankel, s. w. Chisholm, and H. M. Shapiro. 1983. An inexpensive flow cytometer for the analysis of fluorescence signals in phytoplankton: chloraphyll and DNA distributions. J. Exp. Mar. Biol. Ecol. 68:129-144. Paerl, H. w. filtered and s. plankton L. and Shimp. 1973. Preparation of detritus study with scanning


61 electron microscopy. Limnol. Oceanogr. 18:802-805. Paul, J. H. 1982. Use of Hoechst dyes 33258 and 33342 for enumeration of attached and planktonic bacteria. Appl. Environ. Microbial. 33:1229-1232. Paul, J. H. 1993. The advances and limitations of methodology. In: Ford, T. ( ed. ) Aquatic Microbiology: An Ecological Approach. Cambridge, MA (in press) Pettipher, G. L. and u. M. Rodrigues. 1982. Semiautomated counting of bacteria and somatic cells in milk using epifluorescence microscopy and television image analysis. J. Appl. Bact. 53:323-329. Porter, K. G. and Y. s. Feig. 1980. The use of DAPI for identifying and counting aquatic microflora. Limnol. Oceanogr. 25:943-948. Reynolds, G. T. and o. L. Taylor. 1980. Image intensification applied to light microscopy. Bioscience 30:586-592. Riemann, B. 1983. Biomass and production of phyto-and bacterio-plankton in eutrophic Lake Tystrup. Denmark. Freshwater Biol. 13:389-398. Robertson, B. R. and o. K. Button. 1989. Characterizing aquatic bacteria according to population, cell size, and apparent DNA content by flow cytometry. Cytometry 10:7 o-76. Rublee, P. A., s. M. Merkel, and M. A. Faust. 1983. The transport of bacteria in the sediments of a temperate marsh. Estuarine Coastal Shelf Sci. 16:501-509. Salonen, K. 1977. The estimation of bacterioplankton numbers and biomass by phase contrast microscopy. Ann. Bot. Fenn. 14:25-28. sanders, c. A., o. M. Yajko, w. Hyun, R. G. Langlois, P. s. Nassos, M. J. Fu1wyler, and w. K. Hadley. 1990. Determination of guanine-plus-cytosine content of bacterial DNA by dual-laser flow cytometry. J. Gen. Microbial. 136:359-365. Sheppard, c. J. R. 1987. Scanning optical microscopy. In: Barer, R. and v. E. Cosslett (eds) Advances in Optical and Electron Microscopy, VollO. Academic Press, New York. Sieracki, M. E., L. w. Haas, o. A. Caron, and E. J.


62 1987. Effect of fixation on particle retention by Underestimation of grazing rates. Mar. Ecol. Prog. Ser. 38:251-258. Sieracki, M. E., P. w. Johnson, and J. MeN. sieburth. 1985. Detection, enumeration, and sizing of planktonic bacteria by image-analyzed epifluorescence microscopy. Appl. Environ. Microbiol. 49:799-810. Sieracki, M. E., s. E. Reichenbach, and K. L. Webb. 1989a. Evaluation of automated threshold selection methods for accurately sizing microscopic fluorescent cells by image analysis. Appl. Environ. Microbiol. 55:2762-2772. Sieracki, M. E. and c. L. Viles. 1990. Color image-analyzed fluorescence microscopy: A new tool for marine microbial ecology. Oceanography 3:30-35. Sieracki, M. E., c. L. Viles, and K. L. Webb. 1989b. Algorithm to estimate cell biovolume using image analyzed microscopy. Cytometry 10:551-557. Sieracki, M. E. and R. L. Webb. 1991. Applications of image analyzed fluorescence microscopy for characterizing planktonic bacteria and protists. In: Reid, P. c., c. M. Turley,and P. H. Burkill (eds) Protozoa and their Role in Marine Processes. Springer-Verlag Berlin, Heidelberg, New York. Singh, A. 1 B. H. Pyle, and G. A. McFeters. 1989. Rapid enumeration of viable bacteria by image analysis. J. Microbiol. Methods. 10:91-101. Tilney, L. G. and s. Inoue. 1982. Acrosomal reaction of Vo/one sperm II. The kinetics and possible mechanism of acrosomal process elongation. J. Cell. Biol. 93:820. van Dilla, M. A., R. G. Lanqlois, D. Pinkel, D. Yajko, and R. Hadley. 1983. Bacterial characterization by flow cytometry. Science 220:620-622. van Wambeke, F. 1988. Enumeration and sizing of planktonic bacteria by image-analyzed epifluorescence microscopy. Ann. Inst. PasteurjMicrobiol. 139:261-272. Viles, c. L. and M. E. Sieracki. 1992. Measurement of marine picoplankton cell size by using a cooled, chargecoupled device camera with image-analyzed fluorescence microscopy. Appl. Environ. Microbiol. 58:584-592. watson, s. w., T. J. Novitsky, H. L. Quinby, and F. w.


63 Valois. 1977. Determination of bacterial number and biomass in the marine environment. Appl. Environ. Microbiol. 33:940-946. Yentsch, c. M., P. K. Horan, K. Muirhead, Q. Dortch, E. Haugen, L. Legendre, L. s. Murphy, D. A. Phinney, s. A. Pomponi, R. w. Spinrad, M. Wood, c. s. Yentsch, and B. J. Zahuranec. 1983. Flow cytometry and cell sorting: A technique for analysis and sorting of aquatic particles. Limnol. Oceanogr. 28(6):1275-1280. Zar, J. H. 1974. Biostatistical Analysis. pp 121-127, 413. Prentice Hall, Englewood Cliffs, New Jersey.




APPENDIX 1 Counting and Measuring Program: E Enter Edit menu B Blank screen G Enter Logical Operations screen Pause (Locate and focus image on slide) SN Invert video image Escape to Logical Operations menu 0 Erode image D Dilate image Escape to Edit menu Escape to Main menu M Enter Measure menu A Choose Automatic Measure Y Proceed with Automatic Measure Escape to Main menu 65


APPENDIX 2 Program for computation of BABI cell volume 10 REM THIS IS A PROGRAM FOR ANDY DAVID 1 0 0 DIM W ( 10 0) L ( 10 0) V ( 10 0) 130 INPUT "ENTER THE NUMBER OF DATA";N 131 SM = 0 132 sx = 0 145 PRINT "ENTER BREADTH AND LENGTH AFTER THE '?'" 150 FOR I = 1 TO N 175 INPUT W(I) ,L(I) 66 200 V(I) = 0.7854 ((W(I)) "'2) ((L(I))0.333 (W(I))) 240 SM = SM + V(I) 245 SM = SM + (V(I) "' 2) 250 NEXT I 275 AV = SM + N 280 XS = SM + 2 2 8 5 D = ( ( s X -( xs I N) ) I ( N -1 ) ) 290 SD = D "' 0.5 300 PR# =PRINT "NUMBER"."BREADTH","LENGTH","VOLUME" 3 50 PRINT = PRINT "*********************************************" 360 400 420 450 4 PRINT FOR I = 1 TO N PRINT I,W(I),L(I),V(I) NEXT I 6 0 p R I N "*****************************************************" 470 PRINT "MEAN", "STANDARD DEVIATION" 480 PRINT AV,SD 500 PR# 0 550 INPUT "MORE DATA (Y,N)?";X$ 600 IF X$ = "Y" THEN 130 650 END T


Download Options

No images are available for this item.
Cite this item close


Cras ut cursus ante, a fringilla nunc. Mauris lorem nunc, cursus sit amet enim ac, vehicula vestibulum mi. Mauris viverra nisl vel enim faucibus porta. Praesent sit amet ornare diam, non finibus nulla.


Cras efficitur magna et sapien varius, luctus ullamcorper dolor convallis. Orci varius natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Fusce sit amet justo ut erat laoreet congue sed a ante.


Phasellus ornare in augue eu imperdiet. Donec malesuada sapien ante, at vehicula orci tempor molestie. Proin vitae urna elit. Pellentesque vitae nisi et diam euismod malesuada aliquet non erat.


Nunc fringilla dolor ut dictum placerat. Proin ac neque rutrum, consectetur ligula id, laoreet ligula. Nulla lorem massa, consectetur vitae consequat in, lobortis at dolor. Nunc sed leo odio.