A

b

i

l

h

a

s

o

s

a

#

0

d

Description and interpretation of the bracts epidermis of Gramineae

(Poaceae) with rotated image with maximum average power

spectrum (RIMAPS) technique

Eduardo A. Favret a,*, Nestor O. Fuentes b, Ana M. Molina c, Lorena M. Setten d

a Instituto de Suelos, Centro de Investigaciones de Recursos Naturales (CIRN), Instituto Nacional de Tecnologıa Agropecuaria (INTA),

De Las Cabanas y De Los Reseros s/n, 1686 Hurlingham, Buenos Aires, Argentinab Comision Nacional de Energıa Atomica, GAEN – GIDAT – UA Materiales, Av. General Paz 1499, B1650KNA San Martın, Buenos Aires, Argentina

c Instituto de Recursos Biologicos, CIRN, INTA Castelar, De Las Cabanas y De Los Reseros s/n, 1686 Hurlingham, Buenos Aires, Argentinad Centro Milstein - Fundacion Pablo Cassara - Saladillo 2468 C1440FFX, Ciudad Autonoma de Buenos Aires, Argentina

Received 17 July 2007; received in revised form 14 October 2007; accepted 15 October 2007

bstract

During the last few years, RIMAPS technique has been used to characterize the micro-relief of metallic surfaces and recently also applied to

iological surfaces. RIMAPS is an image analysis technique which uses the rotation of an image and calculates its average power spectrum. Here, it

s presented as a tool for describing the morphology of the trichodium net found in some grasses, which is developed on the epidermal cells of the

emma. Three different species of grasses (herbarium samples) are analyzed: Podagrostis aequivalvis (Trin.) Scribn. & Merr., Bromidium

ygrometricum (Nees) Nees & Meyen and Bromidium ramboi (Parodi) Rugolo. Simple schemes representing the real microstructure of the lemma

re proposed and studied. RIMAPS spectra of both the schemes and the real microstructures are compared. These results allow inferring how

imilar the proposed geometrical schemes are to the real microstructures. Each geometrical pattern could be used as a reference for classifying

ther species. Finally, this kind of analysis is used to determine the morphology of the trichodium net of Agrostis breviculmis Hitchc. As the dried

ample had shrunk and the microstructure was not clear, two kinds of morphology are proposed for the trichodium net of Agrostis L., one elliptical

nd the other rectilinear, the former being the most suitable.

2007 Elsevier Ltd. All rights reserved.

eywords: RIMAPS technique; Taxonomy; SEM; Lemma; Poaceae; Trichodium net

www.elsevier.com/locate/micron

Available online at www.sciencedirect.com

Micron 39 (2008) 985–991

K

1. Introduction

Scanning electron microscopy (SEM) is one of the most

useful tools used by botanists and taxonomists (Barthlott and

Neinhuis, 1997; Neinhuis and Barthlott, 1997; Romero Garcıa

et al., 1988; Rugolo and Molina, 1992, 1998; Molina and

Rugolo de Agrasar, 2007). The use of this microscopic

technique makes it possible to characterize biological surfaces

in a manner based only on visual observation.

Fuentes and Favret (2002) developed the rotated image with

maximum average power spectrum (RIMAPS) technique,

which allows determining the orientation and quantification of

the micro-topography of a surface. Even though this technique

* Corresponding author.

E-mail address: eafavret@cnia.inta.gov.ar (E.A. Favret).

968-4328/$ – see front matter # 2007 Elsevier Ltd. All rights reserved.

oi:10.1016/j.micron.2007.10.005

has firstly been used to characterize metallic surfaces, it has

been applied recently to the study of biological surfaces (Favret

et al., 2003, 2006; Favret and Fuentes, 2004; Favret and

Lothman, 2007). The present research work proposes RIMAPS

technique as a tool for taxonomic studies.

Some grasses (Poaceae) are characterized in the lemma by

the presence of transverse thickening bars, called trichodium

net (Bjorkman, 1960; Romero Garcıa et al., 1988; Rugolo and

Molina, 1992, 1998; Rugolo de Agrasar and Molina, 2002;

Molina and Rugolo de Agrasar, 2007). The exterior cell wall of

the epidermis of the lemma is very thin, but in some species a

kind of reinforcement occurs. This reinforcement imbricates

with those of the neighboring cells and form the net known as

trichodium net.

The epidermis of the lemma of different species of grasses

was analyzed with the SEM. Simple geometrical schemes

representing the real microstructure of the lemma were

Fig. 1. SEM image of the Podagrostis aequivalvis lemma.

Fig. 3. SEM image of the Bromidium ramboi lemma.

Fig. 2. SEM image of the Bromidium hygrometricum lemma.

E.A. Favret et al. / Micron 39 (2008) 985–991986

proposed. Then, RIMAPS spectra of those schemes and the real

microstructures were compared. From this comparison it was

possible to infer how similar the geometrical schemes were to

the real microstructures. If this was the case, each correspond-

ing scheme could be used as a reference for classifying other

species.

2. Materials and methods

The abaxial epidermis of the lemma (middle section) was

observed in the following species of Poaceae: Podagrostis

aequivalvis (Trin.) Scribn & Merr; Bromidium hygrometricum

(Nees) Nees & Meyen; Bromidium ramboi (Parodi) Rugolo and

Agrostis breviculmis Hitchc.

The observations were carried out on herbarium material.

The material was previously cleaned with alcohol 908 or xilol

and then it was coated with gold–palladium. The morphology

of the epidermis of the lemma was analyzed with a SEM Philips

500. Simple patterns or schemes of the epidermis based on the

micrographs were done for comparison purposes. Only the wall

or boundaries of the cells and trichodium nets were drawn on

the scheme. The Power Point program (Microsoft Office) was

used for drawing the patterns.

2.1. RIMAPS technique

RIMAPS is a new imaging characterization technique

independent of the class of microscopy and of conditions used

for observation as long as they remain constant (Fuentes and

Favret, 2002; Favret et al., 2003, 2006; Favret and Fuentes,

2004; Favret and Lothman, 2007).

This technique involves rotating the image using available

algorithms of commercial software and computing one space-

variable integral of the two-dimensional Fourier transform

for each value of the other variable (Elliot and Rao, 1982;

Gorcester et al., 1989; Mc Donough and Whalen, 1995;

Gonzalez-Velasco, 1996; Felinger, 1998; Ozaktas et al.,

1999; Takeda, 2000; Jeffrey, 2001; Smith, 2002; Palm, 2003;

Palmblad and Bergquist, 2003). The commercial software

used was a demo version of Mountains Map, which could be

run in a computer with Intel Pentium IV processors and

512 MB RAM.

The mathematical procedure is essentially the rotation of the

digitised image I(x, y) a certain angle a, and the calculation of

the x-step of the two-dimensional Fourier transform for each y-

line of the new image Ia(x, y) obtained after rotation:

Iaðnx; yÞ ¼Z 1�1

Iaðx; yÞ expð�inxxÞ dx: (1)

Results from Eq. (1) are averaged over all the y-coordinate of

the image and an average power spectrum (APS) is obtained for

each angular position:

AaðnxÞ ¼ jhIaðnx; yÞi�hIaðnx; yÞij; (2)

where hi means average value and with Iaðnx; yÞh i� being the

complex conjugate of Iaðnx; yÞh i. If the set of maximum values,

Fig. 4. (a) Scheme of the epidermis surface seen in Fig. 1. (b) Scheme of the epidermis surface, including prickles, seen in Fig. 1.

Fig. 5. Scheme of the epidermis surface seen in Fig. 2.

E.A. Favret et al. / Micron 39 (2008) 985–991 987

in arbitrary units (a.u.), of all average power spectra (MAPS):

MðaÞ ¼ max AaðnxÞ; (3)

is plotted as a function of the angle of rotation (typically steps

of 18) of original image (RI), valuable information can be

obtained from the surface pattern under study (Fuentes and

Favret, 2002). The peaks appearing in the resulting plots

indicate surface pattern orientation.

In the present paper we focus our study on the shape of the

curves and the relative position of peaks, but not specifically on

the intensity and shape of them. Images from the ideal regular

patterns and the experimental patterns were analyzed using this

technique. As the RIMAPS spectra show symmetry after a full

rotation of 3608, only the first 1808 are presented in the graphs.

The apical direction corresponds, in the graphs, to the angular

position of 908.

3. Results

Figs. 1–3 shows the microstructure of the lemma of P.

aequivalvis, B. hygrometricum and B. ramboi. Figs. 4–6

represent the respective simplified schemes of the epidermis

surface of the lemma. These schemes only consider the

boundaries of the epidermis cells and trichodium net seen in

Figs. 1–3 and not the prickles, except in Fig. 4b where they are

taken into account. They correspond to the studied species: P.

aequivalvis (Fig. 4a and b), B. hygrometricum (Fig. 5) and B.

ramboi (Fig. 6). As the number of prickles seen in Fig. 2 is very

low, they have not been considered in the corresponding

scheme design (Fig. 5).

Fig. 6. Scheme of the epidermis surface seen in Fig. 3.

Fig. 8. RIMAPS spectra of Figs. 2 (black line) and 5 (gray line).

Fig. 9. RIMAPS spectra of Figs. 3 (black line) and 6 (gray line).

E.A. Favret et al. / Micron 39 (2008) 985–991988

RIMAPS spectra (Figs. 7–9) show differences among the

species. The sinusoidal (waviness) morphology of the cell walls

is distinguished as a double peak at both sides of the apical

direction, that is, around 908. The distance between both

doublet peaks increases as the waviness becomes notorious.

The trichodium net, which has a structure similar to ellipses

with a pronounced eccentricity and with its biggest semi-axe

perpendicular to the apical direction, can be detected in

RIMAPS spectrum as a maximum value around 08 and 1808.The P. aequivalvis structure is similar to the scheme of

Fig. 4a, without considering the prickles. Fig. 4b is a better

representation of the microstructure, as it can be appreciated in

Fig. 7. RIMAPS spectra of Figs. 1 (black line), 4a (gray line) and 4b (light gray

line).

Fig. 10. Smoothed RIMAPS spectra of Figs. 1 (black line) and 2 (gray line) to

emphasize the main maxima. (1) Maxima due to the cell wall direction (the

doublet is caused by the waviness of the cell wall); (2) maxima due to the

alignment direction of the different trichodium net feature neighbors; (3)

maxima due to the long side direction of the trichodium net unit; (4) maxima

due to the alignment of the prickles.

Fig. 11. A poor SEM image of Agrostis breviculmis lemma. Fig. 13. RIMAPS spectra of Figs. 5 (light gray line), 11 (black line) and 12

(gray line).

E.A. Favret et al. / Micron 39 (2008) 985–991 989

the RIMAPS spectrum. Maximum values obtained in 1058 and

1658 make the RIMAPS spectrum more similar to the real one.

The B. hygrometricum structure looks like the scheme seen

in Fig. 5. Maxima observed between 308 and 408, 1408 and 1508are possibly due to the prickles, even though the density of them

is very low.

As no boundary walls between cells are seen in Fig. 3, the B.

ramboi structure is similar to the scheme seen in Fig. 6. The

main direction of the trichodium net appears in 08 and 1808.However, around 908 (apical direction) the maximum value of

the real structure is higher than the one observed in the scheme.

Fig. 12. Scheme of the epidermis surface seen in Fig. 11, considering a

rectilinear trichodium net instead of an elliptical one, as seen in Fig. 5.

A possible explanation to this behavior is the difference between

the density of the trichodium as well as the misalignment of the

real trichodium net.

Summarizing, if boundary walls between cells with

waviness morphology and transverse thickening bars (tricho-

dium net) are present, the RIMAPS spectrum varies around a

constant value along the ordinate axe.

Fig. 10 shows what features of the microstructure produce

the maximum values seen in RIMAPS spectrum. Only the cases

of P. aequivalvis and B. hygrometricum are analyzed.

4. Discussion and conclusions

RIMAPS technique is a useful tool for studying the lemma

surface when its trichodium net is not well resolved or the

surface has suffered certain shrinkage. As an example, let us see

the case of one sample of A. breviculmis. Fig. 11 shows the

microstructure of its lemma. It is not possible to certify whether

its trichodium net has an elliptical or rectilinear morphology. To

Fig. 14. A good and magnified SEM image of the Agrostis breviculmis lemma.

The eliptical shape of the trichodium net is observed.

Fig. 17. Scheme of the epidermis surface equivalent to the one seen in Fig. 5,

Fig. 15. RIMAPS spectra of Figs. 5 (light gray line) and 14 (black line).

E.A. Favret et al. / Micron 39 (2008) 985–991990

decide this, the RIMAPS spectra of Figs. 5 and 12 are

compared. The main difference between both schemes is the

trichodium net, one elliptical and the other rectilinear. In Fig. 13

the RIMAPS spectra of Figs. 5, 11 and 12 are seen. The real

microstructure of A. breviculmis coincides better with the

RIMAPS spectrum of Fig. 5, indicating that the trichodium net

has an elliptical morphology. For a better verification of this

conclusion, a higher quality image of A. breviculmis was

obtained (see Fig. 14) where it is clearly noticed that the

trichodium net has an elliptical shape, a fact which is again

but without the sinusoidal walls.

Fig. 16. Scheme of the epidermis surface showing the overlapping of the

trichodium net.

corroborated by its RIMAPS spectrum (see Fig. 15). However,

in this image (Fig. 14), the boundary walls between cells are not

clearly discernible. Three schemes were proposed: Fig. 5 shows

the trichodium net and the boundary walls, Fig. 16 shows only

the overlapping of the trichodium net, and Fig. 17 is similar to

Fig. 5 but without the sinusoidal walls between the cells. Fig. 18

Fig. 18. RIMAPS spectra of Figs. 5 (black line), 16 (gray line) and 17 (light

gray line).

E.A. Favret et al. / Micron 39 (2008) 985–991 991

compares the RIMAPS spectra of the three schemes, and shows

the coincidence between Fig. 5 and the microstructure of A.

breviculmis.

These geometrical schemes used as models contribute to

represent the microstructure of the lemma in a simple way. The

comparison between the RIMAPS spectra of real surfaces and

the designed patterns will enable the researcher to find the real

trichodium net morphology when the SEM images are not clear

enough.

Acknowledgements

The authors express their appreciation to Dr. Patricia

Bozzano, Mrs. Adriana Domınguez and Lic.Graciela Favret for

their help on the work.

References

Barthlott, W., Neinhuis, C., 1997. Purity of the sacred lotus, or escape from

contamination in biological surfaces. Plantarum 202, 1–8.

Bjorkman, J., 1960. Studies in Agrostis and related genera. Symb. Bot. Upsala

17 (1), 1–112.

Elliot, D., Rao, K., 1982. Fast Transforms Algorithms, Analyses, Applications.

Academic Press, New York, pp. 448.

Favret, E., Fuentes, N., 2004. RIMAPS and variogram analysis of barley leaf

surfaces. Microsc. Today.

Favret, E., Lothman, P., 2007. RIMAPS image analysis of biological and

technical non-wettable surfaces. Microsc. Anal. 21 (1), 7–9.

Favret, E., Fuentes, N., Molina, A., 2003. Identification of botanical species

using RIMAPS analysis on images from leaf surfaces. Microsc. Microanal.

9 (Suppl. 2), 1338–1339.

Favret, E., Fuentes, N., Molina, A., 2006. Quantitative study of Xanthosoma

violaceum leaf surfaces using RIMAPS and variogram techniques. Microsc.

Res. Tech. 69, 684–688.

Felinger, A., 1998. Data Analysis and Signal Processing in Chromatography.

Elsevier Science, New York, pp. 19–40; 167–173; 241–247; 279–286;

385–406.

Fuentes, N., Favret, E., 2002. A new surface characterization technique:

RIMAPS (rotated image with maximum average power spectrum). J.

Microsc. 206, 72–83.

Gonzalez-Velasco, E., 1996. Fourier Analysis and Boundary Value Problems.

Academic Press, New York, p. 551.

Gorcester, J., Millhauser, G.L., Freed, J.H., 1989. Two-dimensional and Fourier

transform EPR. In: Hoff, A.J. (Ed.), Advanced EPR. Applications in

Biology and Biochemistry. Elsevier Science, Amsterdam, pp. 105–125.

Jeffrey, A., 2001. Advanced Engineering Mathematics ISE International Edi-

tion. Academic Press, New York, pp. 545–622.

Mc Donough, R., Whalen, A., 1995. Detection of Signals in Noise, second ed.

Academic Press, New York, pp. 24–25; 52–54; 96–101.

Molina, A.M., Rugolo de Agrasar, Z.E., 2007 (inedito). Estudio Taxonomico de

las especies americanas del genero Podagrostis (Griseb.) Scribn. & Merr.

(Poaceae: Agrostideae).

Neinhuis, C., Barthlott, W., 1997. Characterisation and distribution of water-

repellent, self-cleaning plant surfaces. Ann. Bot. 79, 667–677.

Ozaktas, H.M., Alper Kutay, M., Mendlovic, D., 1999. Introduction to the

fractional Fourier transform and its applications, in: Kazan, B., Mulvey, T.,

(Serial Eds.) Hawkes, P., Editor in Chief, Advances in Imaging and Electron

Physics, vol. 106. Academic Press, New York, pp. 239–286.

Palm, A., 2003. Capillary isoelectric focusing developments in protein analysis.

In: Marko-Varga, G., Oroszlan, P. (Eds.), Emerging Technologies in Protein

and Genomic Material Analysis. Journal of Chromatography Library, vol.

68. Elsevier Science, New York, pp. 118–122; 209–210.

Palmblad, M., Bergquist, J., 2003. Identification and characterization of pep-

tides and proteins using Fourier transform ion cyclotron resonance mass

spectrometry. In: Marko-Varga, G., Oroszlan, P. (Eds.), Emerging Tech-

nologies in Protein and Genomic Material Analysis. Journal of Chromato-

graphy Library, vol. 68. Elsevier Science, New York, pp. 209–210.

Romero Garcıa, A.T., Blanca Lopez, G., Morales Torres, 1988. Revision del

genero Agrostis (Poaceae) en la Penınsula Iberica. Ruizia 7, 1–160.

Rugolo, Z.E., Molina, A.M., 1992. Las especies del genero Agrostis (Grami-

neae: Agrosteae) de la Argentina. Parodiana 7 (1–2), 179–255.

Rugolo, Z.E., Molina, A.M., 1998. Las especies del genero Agrostis L.

(Gramineae: Agrostideae) de Chile, Gayana. Botany 54, 91–156.

Rugolo de Agrasar, Z.E., Molina, A.M., 2002. El genero Lachnagrostis

(Gramineae:Agrostideae) en America del Sur: 20–32. In: En Freire-Fierro,

A., Nelly, D.A. (Eds.), La Botanica en el Nuevo Milenio, Memorias del III

Congreso Ecuatoriano de Botanica. Publicaciones de la Fundacion Ecua-

toriana para la Investigacion y el Desarrollo de la Botanica FUNBOTA-

NICA 4. Quito, 260 pp.

Smith, S., 2002. Digital Signal Processing: A Practical Guide for Engineers and

Scientists. Newnes, New York, 141–260, 311–318, 567–580.

Takeda, M., 2000. Fourier transform profilometry. In: Rastogi, P.K., Inaudi, D.

(Eds.), Trends in Optical Non-Destructive Testing and Inspection. Elsevier

Science, New York, pp. 576–600.