ARTICLE IN PRESS

Ultramicroscopy 107 (2007) 669–673

www.elsevier.com/locate/ultramic

A new method based on image analysis for determining cyanobacterial

biomass by CLSM in stratified benthic sediments
A. Solé, J. Mas, I. Esteve
Department of Genetics and Microbiology, Autonomous University of Barcelona, Bellaterra, 08193 Barcelona, Spain
Received 27 July 2006; received in revised form 27 November 2006; accepted 9 January 2007

Abstract

Cyanobacteria are the dominant primary producers in microbial mats, which are stratified benthic microbial ecosystems found in
coastal environments. Some cyanobacteria form long filaments, which make difficult to apply classical methods to estimate their biomass
because they establish strong interactions with detritic particles. In a previous study, we described a method for determining
cyanobacterial biomass by means of confocal laser scanning microscopy (CLSM). However, the manual method used, based on summa
projection images, was difficult to apply when analyzing a large number of samples.
In this paper, we described a new automated method, based on stacks and applying the plugin voxel counter in the ImageJ analysis
system, more adequate for obtaining biomass data quickly from a large number of CLSM images.
r 2007 Elsevier B.V. All rights reserved.

Keywords: Confocal laser scanning microscopy (CLSM); Image analysis; Cyanobacteria; Biomass; Microbial mats

1. Introduction photosynthetic microorganisms emit naturally autofluor-

Cyanobacteria are the most abundant photosynthetic
microorganisms colonizing a wide range of different
habitats. Among these, microbial mats (benthic ecosystems
stratified in millimeter-thick colored layers), have been
studied by different techniques [1–6]. Cyanobacteria are
located in the upper part of microbial mats, where light is
available [7–10]. These microorganisms in general and
particularly the filamentous type belonging to the order
Oscillatoriales, form a complex network throughout the
mat where they interact strongly with the detritus material
which surrounds them. Moreover, some of them produce
exopolysaccharides [11–13]. Due to this, the upper part of
the microbial mat, which coincides with the green
pigmented layer where the cyanobacteria live, has a very
compact texture. Among the microscopical techniques used
for identifying microorganisms in this type of environment,
fluorescence microscopy has been applied repeatedly since
Corresponding author. Tel.: +34 93 5813255; fax: +34 93 5812387.
E-mail address: antoni.sole@uab.es (A. Solé).
escence [14–16]. However, the density of the filamentous
cyanobacteria, which can be seen in each sample, makes
it difficult to correctly identify them and determine the
biomass that these microorganisms represent in the
natural environment. Despite this, confocal microscopy
scanning microscopy (CLSM) provides high-resolution
images which have been used to determine the diversity
of the cyanobacteria in the microbial mats from the
Ebro delta [17,18], and from other places around Europe
[6,8,9].
In 2001, we published a paper in which we applied a
manual method using images obtained with CLSM for
identifying and quantifying (biomass) the cyanobacteria in
Ebro delta microbial mats [3]. This manual method is
far more efficient than estimating obtained biomass
through pigment extraction. It also improved high-resolu-
tion microscope techniques (SEM and TEM) by not
requiring the use of lengthy protocols in sample prepara-
tion [1,19].
As regards the use of molecular techniques for identifi-
cation, many cyanobacteria (and particularly Microcoleus

0304-3991/$ - see front matter r 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.ultramic.2007.01.007
 ARTICLE IN PRESS
670 A. Solé et al. / Ultramicroscopy 107 (2007) 669–673
sp.) produce abundant exopolysaccharides, which may
frequently cause difficulties in DNA extraction [4].
Fluorescence ‘‘in situ’’ hybridization (FISH) has been used
as a suitable method for the quantification of phylogenetic
groups in environmental samples [20], nevertheless, this
method has limitations when cell permeability is dimin-
ished. Furthermore, molecular techniques also involve long
protocols.
The manual method described was also applied to
determine the spatio-temporal dynamic of the total
biomass and the biomass distribution of each type of
cyanobacteria during one annual cycle [18]. However, this
method required that the limits of each cyanobacterium
presented in each summa projection were established
manually on screen to obtain different morphometric data.
Furthermore, different stereometric corrections were ne-
cessary to obtain the final biomass values. This laborious
manual procedure was tedious to apply, especially when it
was necessary to analyze a large number of images or
samples.
This report describes a reliable and simple method of
computer-assisted microscopy for the quantitative analysis
of confocal laser scanning microscopy images from
stratified benthic sediments in which complex populations
of cyanobacteria are present, and which at the same time
allows a larger number of samples to be analyzed quickly
and easily.
Fig. 1. Confocal summa projection images showing different levels of cyanobacteria complexity: (a) high density of Microcoleus chthonoplastes filaments;
(b) Oscillatoria sp. filaments; (c) low density of Microcoleus chthonoplastes filaments; (d) Pseudanabaena sp. filaments.
2. Materials and methods
2.1. Ebro delta microbial mats
Microbial mats were obtained from the delta of the Ebro
river, near Tarragona, in the NE of Spain (401400 N,
01400 E). Samples were taken from a site near the ‘‘Salines
de la Trinitat’’ (P8) in the Alfacs Peninsula between April
1997 and May 1998. They were collected monthly; more
samples were collected during the summer months in which
greatest microorganism growth occurred.
2.2. Confocal microscopy and image processing analysis
Stacks of images were obtained with a Leica True
Confocal Scanner TCS 4D (Leica Lasertechnik GmbH,
Heidelberg, Germany) under an excitation beam of
wavelength 568 nm, from an argon–krypton laser, and
stored on CD ROMs with a 650 Mb capacity.
In a previous study, we published a manual method for
determining cyanobacteria biomass in microbial mats by
means of CLSM optical series and the computer-aided
drafting program, AutoCAD release version 14 [3].
In this paper, we reuse the optical series or stacks of
images obtained with CLSM from samples of Ebro delta
microbial mats. In order to determine the cyanobacterial
biomass using the new computerized method described in
 ARTICLE IN PRESS
A. Solé et al. / Ultramicroscopy 107 (2007) 669–673 671

Section 3. The Voxel Counter plugin, created by Wayne
Rasband and available on-line, was applied to the ImageJ
1.33f analysis system, a free image-processing and analysis
program.
2.3. Statistical analysis
available on-line, was applied to this software. This specific
application calculates the ratio of thresholded voxels or
cyanobacterial volume (Threshold Voxel Count) to all
voxels or the total sediment volume (Total Voxel Count)
from each binary image in every stack. This value (Volume
Fraction) was multiplied by a conversion factor of

With the goal of comparing both methods (manual

versus computerized), a mixed Lineal Model [21] for
repeated measures was established, with the biomass
logarithm as the response variable, and month, depth and
their interaction and method, as the covariates. Statistical
significance was fixed at a ¼ 0.05. All statistical analyses
were performed with SPSS 13.0 for Windows.

3. Results and discussion

When sediment samples, such as microbial mats, are

analyzed with confocal laser microscopy, high-quality
images are obtained. These high contrast images show
the natural autofluorescence emitted by cyanobacteria
(foreground), against a black background (detritus and
non-fluorescent microorganisms).
Summa projection images, generated from optic series or
stacks in which all cyanobacteria are summarized in one
single image can be easy or more to analyze depending on
the type of cyanobacteria present. Among the oxygenic
phototrophic microorganisms described in the Ebro delta
microbial mats, Microcoleus sp. is the dominant cyano-
bacterium. It has filaments which are usually long and
curved and that are normally grouped together in bundles
surrounded by a exopolysaccharide matrix. The images
that this microorganism generates have a high degree of
overlapping and intercrossing filaments (Figs. 1(a) and (c)).
Oscillatoria sp. is characterized by its generally short, thick
filaments that have a rectilinear trajectory, which means
that in the images where it appears there is not as much
overlapping as in the case of Microcoleus sp. (Fig. 1(b)). As
well as these two types in the Ebro delta, other types of
filamentous and unicellular cyanobacteria are found in less
abundance. Underneath the green layer coinciding with the
anoxic zone, the images taken show very thin filaments of
varying lengths, which normally do not occupy many
microns of depth (Fig. 1(d)).
In the present study, a free image processing and analysis
program, ImageJ 1.33f, which is available on-line [22] was
used to determine the cyanobacterial biomass from mat
samples. Confocal stacks of varying complexity and
thickness were used instead of summa projection images.
Stacks of images were opened and imported as 8-bit Fig. 2. Summa projection and optical sections from the same CLSM stack
512  512 tiff image sequences (original image format) and are shown. This stack was composed by 14 optical sections in an 11 mm
transformed to binary images (black/white) using the depth. The distance between each two optical sections was 0.88 mm. (a)
threshold adjust commands in which an optimum thresh- Original confocal summa projection image from filamentous cyanobacter-
old was required, so that the least possible information ia in an Ebro delta microbial mat sample, (b) AutoCAD manual drawing
from image (a), (c), (e) and (g) the 3, 6 and 9 optical sections taken in
from the images was lost. This optimum value was depth in this stack; (d), (f) and (h) binary images, from images c, e and g,
established in each of the stacks analyzed. Later, a plugin obtained with ImageJ 1.33f software using the adjust threshold command
called Voxel Counter, created by Wayne Rasband and to obtain an optimum black and white image without losing information.
 ARTICLE IN PRESS
672 A. Solé et al. / Ultramicroscopy 107 (2007) 669–673

310 fg C/mm3 to convert it to biomass [23,24]. This factor
was also applied previously in the manual method.
A summa projection drawn with the manual method
(AutoCAD) and some optical sections binarized with the
ImageJ program from the same CLSM stack are shown in
Fig. 2. In this figure, Microcoleus chthonoplastes images
have been selected as they are more abundant in the Ebro
delta microbial mats.
The spatial dynamics of cyanobacterial biomass deter-
mined from stacks (white points) and summa projections
(black points) corresponding to different periods of time
are shown in Fig. 3. The results obtained with the
computerized method show the same profile throughout
the study as those obtained manually. However, the values
of the computerized method applied to stacks are higher
(38.98 mg C/cm3 of sediment at 0.45 cm depth in April
1997) than those obtained using the manual method
(2.26 mg C/cm3 of sediment at 0.45 cm depth in April
1997), especially in layers where cyanobacteria were most
abundant (green layer). Applying the program ImageJ
achieves, in this case, much higher biomass as it provides us
with information about the entire stack. The lower values
obtained by the manual method could be due to the fact
that applying different mathematical formulae can cause
loss of biomass due to the error produced when the outlines
drawn of the bacteria do not correspond to well-defined
geometric figures, as in the case of filamentous cyanobac-
teria. At deeper layers of the mat, the filamentous
cyanobacteria are not so abundant, and the images coming
from this area overlap less, so that both methods show
similar biomass values (6.01 mg C/cm3 of sediment using
the computerized method and 4.81 mg C/cm3 of sediment
using the manual method, both results at 0.55 cm depth in
July 1997).
The statistical analysis detected significant differences
between both methods tested (po0.001), with a 95%
confidence interval [2.31, 4.14]. The results demonstrate
that the new method described here is much more accurate
than the manual method, particularly when analyzing
samples with high population density of filamentous
cyanobacteria. This is why, analyzing each image of the
same stack provides quantitative information that is more
accurate than that obtained from a bidimensional image of
the entire stack.
Furthermore, the computerized method offers other
advantages compared to the manual method. Firstly, it

0 0
b
0.2 0.2

0.4 0.4

0.6 0.6
a c
0.8 0.8 d

1 1
29/04/97 07/07/97
Depth (cm)

1.2 1.2

0 0

0.2 0.2

0.4 0.4

0.6 0.6

0.8 0.8

1 1
04/08/97 27/02/98
1.2 1.2
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Biomass (mgC/cm3 sediment)

Fig. 3. Spatial and temporary cyanobacteria total biomass during four different months from the annual cycle studied. The data from the two methods
based on summa projections (AutoCAD) and on stacks (ImageJ) are shown. Results are expressed in mg C/cm3 of sediment. The depth of the pigmented
layers is expressed in cm. Green layer ; red layer ; summa projections results K; stack results J. Letters a, b, c, and d (indicated by arrows) show where
images from Fig. 1 are obtained.
 ARTICLE IN PRESS
A. Solé et al. / Ultramicroscopy 107 (2007) 669–673
allows all the image sequences to be processed at the same
time, which is much more efficient than if they had to be
analyzed one by one. Secondly, it uses an image format in
pixels in which different specific computer applications
(plugins) can be applied using the ImageJ analysis system.
The plugin Voxel Counter makes this program appropriate
for obtaining diverse information at microscale level to
evaluate cyanobacterial biomass variations in time and space.
4. Conclusion
The method described is the most appropriate: (a) when
the samples to be analyzed generate images with a large
amount of intercrossing and overlapping filaments, (b) for
rapidly obtaining biomass data from a large number of
images, and (c) when it is necessary to work with a large
number of samples.
Acknowledgment
This work was supported by Spanish Grant DGICYT
BOS2001-2033 to I.E.
References
[1] I. Esteve, D. Ceballos, M. Martı́nez-Alonso, N. Gaju, R. Guerrero,
Microbial mats: structure, development and environmental signifi-
cance, in: L. Stal, P. Caumette (Eds.), NATO ASI Series G:
Ecological Sciences, Springer, Heidelberg, 1994, pp. 4165–4420.
[2] G. Muyzer, Curr. Opin. Microbiol. 2 (1999) 317.
[3] A. Solé, N. Gaju, S. Méndez-Álvarez, I. Esteve, J. Microsc. 204
(2001) 258.
[4] F. Garcı́a-Pichel, A. López-Cortés, U. Nübel, Appl. Environ.
Microbiol. 67 (2001) 1902.
673
[5] S. Ramı́rez-Moreno, S. Méndez-Álvarez, M. Martı́nez-Alonso,
I. Esteve, N. Gaju, Curr. Microbiol. 48 (2004) 285.
[6] E. Diestra, A. Solé, I. Esteve, Ophelia 58 (2004) 151.
[7] C. Demergasso, G. Chong, P. Galleguillos, L. Escudero,
M. Martı́nez-Alonso, I. Esteve, Rev. Chil. Hist. Nat. 76 (2003) 485.
[8] A. Wieland, M. Kühl, L. McGowan, A. Solé, E. Diestra, I. Esteve,
T. Garcı́a de Oteyza, J.O. Grimalt, R. Duran, A. Fourc- ans,
P. Caumette, R. Herbert, Microbiol. Ecol. 46 (2003) 371.
[9] A. Fourc- ans, T. Garcia de Oteyza, A. Wieland, A. Solé, D. Diestra,
J. van Bleijswijk, J.O. Grimalt, M. Kühl, I. Esteve, G. Muyzer,
P. Caumette, R. Duran, FEMS Microbial. Ecol. 51 (2004) 55.
[10] M. Martı́nez-Alonso, J. Mir, P. Caumette, N. Gaju, R. Guerrero,
I. Esteve, Int. Microbiol. 7 (2004) 19.
[11] G. Mazor, G.J. Kidron, A. Vonshak, A. Abeliovich, FEMS
Microbiol. Ecol. 21 (1996) 121.
[12] R. de Philippis, M.C. Margheri, R. Materassi, M. Vincenzini, Appl.
Environ. Microbiol. 64 (1998) 1130.
[13] B.S. McSwain, R.L. Irvine, M. Hausner, P.A. Wilderer, Appl.
Environ. Microbiol. 71 (2) (2005) 1051.
[14] M. Schallenberg, J. Kalff, J.B. Rasumssen, Appl. Environ. Microbiol.
55 (1989) 1214.
[15] L. Raymond, J.R. Kepner, J.R. Pratt, Microbiol. Rev. 58 (1994)
603.
[16] T. Kuwae, Y. Hosokawa, Appl. Environ. Microbiol. 65 (1999)
3407.
[17] A. Solé, N. Gaju, R. Guerrero, I. Esteve, Eur. Microsc. Anal. 52
(1998) 13.
[18] A. Solé, N. Gaju, I. Esteve, Scanning 25 (2003) 1.
[19] J. Mir, M. Martı́nez-Alonso, I. Esteve, R. Guerrero, FEMS
Microbiol. Ecol. 86 (1991) 59.
[20] J. Detmers, H. Strauss, U. Schulte, A. Bergmann, K. Knittel,
J. Kuever, Microbiol. Ecol. 47 (2004) 236.
[21] M.J. Crowder, D.J. Hand, Analysis of Repeated Measures, Chapman
& Hall, New York, 1990.
[22] W.S. Rasband, ImageJ, US National Institutes of Health, Bethesda,
MD, USA /http://rsb.info.nih.gov/ij/S, 1997–2006.
[23] J.C. Fry, Methods Microbiol. 22 (1990) 411.
[24] J. Bloem, M. Veninga, J. Shepherd, Appl. Environ. Microbiol. 61
(1995) 926.