Examination of wooden shelves used in the ripening of a raw milk smear cheese by FTIR spectroscopy

Food Control 20 (2009) 658–663

Contents lists available at ScienceDirect

Food Control

journal homepage: www.elsevier .com/locate / foodcont

Examination of wooden shelves used in the ripening of a raw milksmear cheese by FTIR spectroscopy

N. Oulahal a,*, I. Adt a, C. Mariani a,b, A. Carnet-Pantiez a, E. Notz b, P. Degraeve a

a Université de Lyon, Université Lyon 1, Laboratoire de Recherche en Génie Industriel Alimentaire, EA n�3733, IUT A – Département Génie Biologique,technopole Alimentec, rue Henri de Boissieu, 01060 Bourg en Bresse, Franceb Actilait, Route des champs laitiers, BP 30, 74801 La Roche sur Foron, France

a r t i c l e i n f o a b s t r a c t

Article history:Received 12 November 2007Received in revised form 8 October 2008Accepted 12 October 2008

Keywords:FTIR spectroscopyWooden shelvesCheeseBiofilm

0956-7135/$ - see front matter � 2008 Elsevier Ltd. Adoi:10.1016/j.foodcont.2008.10.012

* Corresponding author. Tel.: +33 4 74 47 21 41; faE-mail address: [email protected] (N

The surface of wooden shelves used in the ripening of a raw milk smear cheese was examined by atten-uated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy. FTIR spectra of zones of woo-den shelves in contact with cheeses or not (control) were acquired. This acquisition was performed eitherdirectly after cheeses removal or after brushing with water to remove cheese compounds deposited ontowood surfaces. Their analysis revealed that the degree of similarity between the ATR-FTIR spectra ofunused and cleaned by brushing wooden shelves is higher than with the spectra of non-cleaned surfaces.The spectra of non-cleaned zones where cheeses were ripened namely differed in the 920–1180 cm�1 andin the 1485–1780 cm�1 regions. These regions can be assigned to proteins and polysaccharides and thusmight correspond to deposits of cheese rind surface as well as to the presence of microbial biofilms onwooden shelves. Similarities in these spectral regions with those of cheese rind support this hypothesis.These observations indicate that ATR-FTIR spectroscopy is a rapid and valuable analytical tool to directlyinvestigate the global chemical composition of very thin films such as microbial biofilms or cheese rinddeposits present on wooden shelves which are not accessible to standard chemical methods.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction ripening shelves colonised by living native biofilms (Mariani et al.,

Wood has a long tradition as a natural material used in foodproduction. The number of cheeses ripened on wooden shelvescan be estimated at more than 350,000 tonnes of cheese per yearin France, and consists namely in protected designation of origin(PDO) productions. Besides offering excellent mechanical resis-tance, wood has exceptional regulative water sorption propertiesthat lead to a low humidity loss of cheeses during ripening. More-over, most cheese manufacturers believe that wooden shelves fa-vour the cheese rind development and improve the organolepticqualities and typicity of cheeses. The presence of water and nutri-ents as well as the porosity of this material may favour microbialdevelopment on wooden shelves. However, little work has beencarried out to identify the microorganisms present. In a recentstudy, Mariani et al. (2007) identified micrococci–corynebacteria,yeasts and moulds as dominant microflora on the shelves used inripening the French raw milk smear cheese Reblochon de Savoieat the end of the cheese-ripening process. Since safety of theiruse has frequently been asked, the influence of the presence ofthe native multi-species biofilm present on wooden shelves onthe behaviour of two Listeria monocytogenes strains was analysed:an inhibitory effect of their growth was observed only on wooden

ll rights reserved.

x: +33 4 74 45 52 53.. Oulahal).

in press). As far as regulations are concerned, wood can be used inthe ripening of traditional cheeses under a European derogationlaw (decision 96/536/CE of July 29th, 1996).

Microbial ecology of wooden shelves depends on physico-chemical parameters such as temperature, pH, salt concentrationand water activity and also on the presence of nutrients whichcan be released from the ripening cheeses to the surface of woo-den shelves. Since the rind of cheeses is in direct contact withwooden shelves during their ripening, it is not surprising thatthe microbial groups identified by Mariani et al. (2007) arestrongly correlated with those of cheese rinds. However, besideshaving determined that temperature, pH, salt concentration, andwater activity values varied from 13 to 14 �C, 7.1 to 8.3, 0.3% to0.4% (wt/wt), and from 0.94 to 0.97, respectively, other chemicalanalytical techniques (such as the Kjeldahl method to assay nitro-gen) appeared long-lasting and irrelevant to assay compoundspresent in low amounts and as thin films on the surface of wood.As an analytical tool, the attenuated total reflection (ATR) Fouriertransform infrared (FTIR) spectroscopy offers the possibility to di-rectly investigate the chemical composition of very thin films(Schmitt & Flemming, 1998). This technique was thus employedsuccessfully to determine the efficiency of different cleaningagents to remove proteins fouling the surface of commercialultrafiltration membranes (Zhu & Nystrom, 1998) by placing di-rectly the surfaces of virgin, fouled, and cleaned membranes onto

N. Oulahal et al. / Food Control 20 (2009) 658–663 659

the internal reflection element in the ATR-FTIR spectrometer. An-other important feature of the ATR-FTIR technique is that it al-lows for depth profiling (Chan & Chen, 2004). Since thepenetration depth depends on the angle of the incident IR beam,adjusting the angle of the incident IR beam allows for analysis ofdepths from about 1 to 5 lm. Although, whereas it has been pre-sented as a promising technique to study microbial biofilms in arelatively undisturbed state, to our knowledge, ATR-FTIR spectros-copy has never been employed to investigate the surface ofcheese-ripening wooden shelves. Unused ripening woodenshelves, wooden shelves collected immediately after ripeningand analysed either directly or following brushing with a solutionwere analysed by ATR-FTIR spectroscopy. Comparison of the FTIRspectra thus obtained with the spectra of cheese rinds or micro-bial biofilms allowed to check the validity of ATR-FTIR spectros-copy to analyse the biochemical composition of the depositsand/or the microbial biofilms present on cheese-ripening woodenshelves. Finally the ATR-FTIR spectra will be discussed and com-pared with the results obtained by microbiological, microscopicaland chemical analysis of the surface of wooden shelves.

Fig. 1. Sampling out of cheese

2. Materials and methods

2.1. Cheese-ripening wooden shelves

PDO ‘‘Reblochon de Savoie” is a raw milk smear cheese, pro-duced in the French North Alps. PDO cheeses were laid on woodenshelves in two ripening rooms. The shelves were cut lengthwise inspruce wood (Picea abies). They had been used in cheese ripeningfrom 6 months to 14 years and their aspects were different accord-ing to their age. They bore 13 visible marks on each side corre-sponding to the 13 contact zones of cheeses, more visible on theolder shelves. After contact with the cheeses, the shelves werecleaned with a brushing machine (brushing three times for 3 s withcold water) and subsequently placed in an air-drying system (stateafter cleaning and drying). They were then randomly chosen forthe different steps of the ripening, in order to homogenize biofilmspresent on wooden shelves.

The state before cleaning represented the most representativebiofilm with the dominant microflora, and the state following dryingafter cleaning represented the step where the risk of contamination

ripening wooden shelves.

660 N. Oulahal et al. / Food Control 20 (2009) 658–663

by pathogenic microorganisms was the most important, since possi-bly contaminated cheeses were laid down at this state.

In the laboratory, the shelves were then sawn into wooden sam-ples at the different states (before and after cleaning (followed bydrying)).

2.2. Removal of native biofilm and/or cheese rind deposits from shelvesby brushing

Biofilms and/or cheese rind deposits were removed on marksfor each shelf by using a brushing treatment. A surface of 10 cm2

(4 cm � 2.5 cm) was delimited with adhesive tape disinfected with70% (v/v) ethanol. After having deposited on the delimited surface1 mL of Ringer’s solution (1/4 dilution) containing 2 g Tween 80detergent/L (Merck, Lyon, France) and 3 g agar/L (Biokar, Beauvais,France), the surface was brushed 10 times horizontally and 10times vertically with a sterile toothbrush. The resulting suspensionwas recovered with sterile sponges (3 cm � 3 cm, Super Twill) andtransferred aseptically into individual sterile stomacher bag, mixedwith 30 mL of Ringer and homogenized for 2 min.

2.3. Reblochon rind

Reblochon de Savoie cheese rind was recovered from the sur-face in contact with the cheese-ripening wooden shelves with a ra-zor blade and directly laid down on the zinc selenide crystalinternal reflection element disposed in the FTIR spectrometer forspectral data acquisition (Fig. 1).

2.4. FTIR spectroscopy and spectral analysis by curve-fitting

All samples (unused wooden shelves, cheese-ripening woodenshelves, cheese-ripening wooden shelves cleaned and cheese rinds)were analysed by ATR-FTIR spectroscopy. FTIR spectra were re-corded between 4000 and 800 cm�1 using a Bomem FLTA 200–100 (ABB process, Montluel, France) spectrometer. Spectra weremeasured at a resolution of 8 cm�1 and 20 scans per sample wererecorded. After acquisition, spectra were pre-processed (rubber-band correction of baseline, vectorial normalisation, and secondderivative) using OPUS NT 3.1 software (Bruker optics, Wissem-bourg, France) and a curve-fitting analysis was performed.

0

0.01

0.02

0.03

0.04

0.05

2820 2870

0

0.02

0.04

0.06

0.08

0.1

0.12

900 1100 1200

1035

1017

1054

1159

1262

1315

1104

wavenumb

norm

alis

ed a

bsor

banc

e un

its

01000 1200 1300

1035

1017

1054

1159

1262

1315

1104

Fig. 2. Normalized FTIR spectra of unused wooden shelves (dotted grey line), ripening shrind (grey line).

Curve-fitting is a mathematical procedure which can be appliedon FTIR spectra in order to solve overlapping due to the presence ofvarious biological materials. Effectively, when the bandwidth isgreater than adjacent peak-to-peak separation, it is often difficultto observe detailed structure in the FTIR spectra. The curve-fittingprocedure used was based on a least-square method using Gauss-ian and Lorentzian bands (Galichet, Sockalingum, Belarbi, & Man-fait, 2001; Sockalingum et al., 1997). This method calculates atheoretical spectrum which best fits the experimental one. Theaccuracy of the fit is given by the chi-square value: the lower thevalue of v2, the better is the fit.

Prior to curve-fitting analysis, band position was found usingthe ‘‘peak search” function (OPUS NT 3.1) with a detection levelof 1% followed by calculation of second derivative spectra in orderto separate overlapping signals and to visualise the exact positionsof the maxima for each band. These positions were then used in thecurve-fitting method in order to calculate the theoretical spectrum.In this calculation, the peak position was fixed and the width athalf height of the mixed Gaussian–Lorentzian bands was restrictedto less than 30 cm�1. In these conditions, the fit spectrum gavegood v2 values in the range 10�6–10�8.

3. Results and discussion

Comparison of normalized FTIR spectra presented in Fig. 2 re-vealed differences in chemical composition of unused woodenshelves, cheese-ripening wooden shelves analysed directly afterthe end of cheese ripening or following cleaning and cheese rinds.Thus, although some similarities in spectral profiles between un-used wooden shelves and wooden shelves after cleaning and dry-ing can be observed, on one hand, and between ripening shelvesand Reblochon rind, on the other hand, it appears that samplespresent major variations all around the studied spectral windows.In order to investigate more precisely these structural differences,the four main zones of FTIR spectra – i.e., 920–1180 cm�1, 1180–1485 cm�1, 1485–1780 cm�1, and 2820–2990 cm�1 corresponding,respectively, to polysaccharide, protein, mixed, and lipid absorp-tion regions – were separately analysed using a mathematical pro-cedure: the curve-fitting method. These compounds were chosensince cheeses contain lipids and proteins while microorganismsin general (and namely microbial biofilms) contain both proteinsand polysaccharides (Marcotte, Kegelaer, Sandt, Barbeau, & Lafleur,

2920 2970

2920

2970

1600 1800

1372

1414

1509

1542

1646

1636

1246

1456

2920

297029

20

297029

20

2970

er (cm-1)

1400 1500 1700

1372

1414

1509

1542

1646

1636

1246

1456

elves (dotted black line), wooden shelves after brushing (black line) and Reblochon

N. Oulahal et al. / Food Control 20 (2009) 658–663 661

2007). A comparative analysis of most important results of curve-fitting analysis is presented below.

Comparison of the total band surface areas obtained for the fourabove-mentioned regions using curve-fitting method is presentedin Table 1. It appears that proteins and polysaccharides are thetwo major components for all studied samples. However, as itwas previously noted, band area calculations show similarities be-tween unused wooden shelves and cheese-ripening woodenshelves after cleaning and drying in contrast to ripening shelvesand Reblochon rind. Thus, we observed an inversion of the pro-teins/polysaccharides ratios (respectively, 0.5 and 2). Each of thefour spectral regions have then been analysed in more detail sepa-

Table 1Comparison of the total band surface areas (%) calculated using the curve-fitting procedurbefore cleaning, wooden shelves after cleaning and Reblochon rind.

Spectral window Unused wood shelves Wood she

920–1180 cm�1 (polysacharides) 55.1 32.31180–1485 cm�1 (mixed) 15.9 12.41485–1780 cm�1 (proteins) 23.8 52.52820–2990 cm�1 (lipids) 5.2 2.7

Table 2Quantitative analysis by curve-fitting of the major component bands of the FTIR spectraspectral windows: 920–1180 cm�1, 1180–1485 cm�1, 1485–1780 cm�1, and 2820–2990 c

Unused woodshelves

Wood shelves beforecleaning

Wood shelves after cledrying

Polysaccharides region 920–1180 cm�1

946 – – 3.9976 – 3.8 –

989 15.6 – 27.21030 32.9 49.3 31.31057 38.5 – 25.91078 – 32.1 –1108 10.2 10.5 101155 2.7 4.2 1.6

Mixed region 1180–1485 cm�1

1206 1.7 – 6.41237 14.9 22.8 –1266 24.4 – 15.21312 15.3 9.2 3.81338 6 – 0.151373 17.4 13 16.51410 – 42.6 49.51423 11.7 – –1459 8.7 12.4 8.3

Proteins region 1485–1780 cm�1

1510 4.9 – 3.71516 – 0.6 –1544 3 18.8 3.61565 – – 4.71586 13.2 – 13.41627 32.6 56.8 –1643 16.4 14.2 68.11673 14.2 9.5 2.21692 6.4 – 4.21725 9.2 – –1742 – – 2

Lipids region 2820–2990c2851 17.7 13.3 2.22879 – 9.7 7.62896 43.6 – –2925 32.7 62.1 70.12944 – – –2961 6 14.9 20.12973 – – –

a Galichet et al. (2001), Legal, Manfait, and Theophanides (1991), Maquelin et al. (200(1997).

rately. All bands assignments were notably done with respect toliterature and FTIR spectra of samples representative of cheesecomponents and Reblochon endogenous microorganisms (unpub-lished data).

Peaks in the 2820–2990 cm�1 spectral window are generally as-signed to C–H stretching modes mainly from lipids. Results pre-sented in Tables 1 and 2 confirm that wooden shelves (unused orcleaned and dried) have some common characteristics contraryto wooden shelves before cleaning and Reblochon rind. Thus,bands at 2879 cm�1 and 2961 cm�1, which can be assigned, respec-tively, to symmetric and asymmetric –CH3 modes are quite similarfor unused and cleaned wood while the repartition of bands corre-

e applied on the FTIR spectra obtained from unused wooden shelves, wooden shelves

lves before cleaning Wood shelves after cleaning Reblochon

67.8 27.813.5 12.114.4 59.7

4.2 0.4

of unused, used and brushed wooden shelves and Reblochon rind for the followingm�1.

aning and Reblochonrind

Assignmenta

– None– Could be attribute to mannans from Geotrichum

cell wall– None

33.3 Sugars or polysaccharides– Cellulose, hemicellulose

59.5 Glycogen, nucleic acidsOH association from COOH in wood

7.1 Esters (lipids), cellulose, hémicellulose

4.3 None32.8 Nucleic acids, phospholipids

– Lignin13 None

– Cellulose15.8 Lignin, hemicellulose27 Proteins or carboxyl functions

– Lignin4.4 Proteins, lipids, lignin

– Lignin1.1 Tyrosin

17.8 Amide II (antiparallel b-sheet)– Amide II– Lignin

56.1 Amide I13.3 Amide I (antiparallel b -sheet)

8.6 Amide I (antiparallel b -sheet)– Amide I– Lipids, polysaccharides– None

6.7 ts CH2 from lipids, cellulose4.5 ts CH3 from lipids– CH from methyl groups

48.4 tas CH2 from lipids27.2 none

– tas CH3 from lipids1.2 None

2), Naumann et al. (2005), Pandey (2005), Polovka et al. (2006), Sockalingum et al.

662 N. Oulahal et al. / Food Control 20 (2009) 658–663

sponding to symmetric (2851 cm�1) and asymmetric �CH2 modes(2925 cm�1) is very heterogeneous in the different studiedsamples.

Except for the unused wooden shelves, the major band is the2925 cm�1 one; it also appears that the brushing treatment pro-cess had no effect on this band. Furthermore, this peak is around2 folds higher for ripening wooden shelves (before or after clean-ing) than for unused ones. Concerning the unused woodenshelves, the main band (43.6%) is at 2896 cm�1 (correspondingto C–H vibrations from methyl groups) and was not observedfor the other samples.

Curve-fitting analysis of the 1485–1780 cm�1 spectral window,called protein absorbing region, showed that bands can be attributedto amide I and amide II groups from proteins vibrations and also toa very characteristic wood component: the lignin, which is a com-plex polymer of aromatic structures (phenols) specific of plantmaterials. More precisely, two bands likely corresponding to thearomatic skeletal vibrations, one at 1586 cm�1 and the other oneat 1510 cm�1 (Naumann, Navarro-Gonzalez, Peddireddi, Kues, &Polle, 2005) were assigned to lignin. These two bands representaround 17–18% of proteins regions for unused or cleaned woodenshelves spectra. Analysis of results from samples revealed mainlybands of amide I groups (1627, 1643, and 1673 cm�1). However,repartition of those bands was different: the 1627 cm�1 bandwas not found in brushed wooden shelves. Other peaks in thisstudied range corresponded to amide II groups (1544 and1565 cm�1) and tyrosine (1516 cm�1) and were quite similar be-tween wooden shelves and Reblochon rind. After brushing, woo-den shelves spectra presented intermediate features betweenthose of control unused wooden shelves and those of Reblochonde Savoie cheese rinds. Furthermore, FTIR control spectra fromcheese extracts revealed that the bands at 1544 cm�1 and1627 cm�1 could be assigned to cheese whey (data not shown):those data are in accordance with literature (Chen & Irudayaraj,1998; Van Der Ven et al., 2002). The band at 1627 cm�1 was com-mon to cheese whey spectrum and to wooden shelves with Reblo-chon cheese traces and thus could correspond to a specific proteinfrom cheese. It thus indicates that contact between woodenshelves and Reblochon cheese can interfere with all specific woodsignals.

Analysis by curve-fitting of the mixed (1180–1485 cm�1) absorb-ing region which is not specific of a macromolecules group revealedpeaks which can be assigned to proteins (1439 and 1410 cm�1),nucleic acids or lipids (1237 and 1459 cm�1) and polysaccharidicmolecules (1338, 1373, and 1410 cm�1). However, most of bandscorresponded to specific components of wood such as lignin(1266, 1423, and 1459 cm�1), hemicellulose and cellulose (1338and 1373 cm�1). Cellulose and hemicellulose are the main polysac-charidic constituents of wood and represent around 50% and 20% ofPicea species parietal chemical structure, respectively (Alen, Koti-lainen, & Zaman, 2002; Wikberg & Maunu, 2004; Yildiz, Gezer, &Yildiz, 2006). As those components are specific for plant forms,those bands were exclusively found on wooden shelves. Con-versely, the nucleic acid/phospholipids peak (1237 cm�1) appearedmore characteristic of Reblochon cheese materials (x 1.5–2.2 foldsincrease of height between unused wooden shelves and woodenshelves with Reblochon traces); this band totally disappeared afterbrushing. Thus, analysis of this region confirmed that, as for theprotein region, wooden shelves with Reblochon fingerprints pres-ent common characteristics with Reblochon rind and that woodafter brushing presents a mixed profile between unused and usedwood.

Bands in the spectral window 920–1180 cm�1 are likely con-nected to polysaccharides vibrations, some peaks can also beattributed to phospholipids and nucleic acids (Sockalingum et al.,1997). Other signals were characteristic of wood (used or unused)

materials or Reblochon cheese rind. Bands at 989 and 1057 cm�1

were found only for unused wood or wood after brushing samples.Contrary to those bands, a peak at 1078 cm�1, which can be attrib-uted to DNA and glycogen from microorganisms, was only presentin Reblochon rind or in Reblochon traces on wood. So, this band is aspecific marker of Reblochon while bands at 1057, 1266, 1338, and1510 cm�1 are specific of wood components (mainly cellulose,hemicellulose and lignin).

Curve-fitting analysis results reveal interactions betweencheese and wooden shelves during the ripening process. First,deposits of cheese components – clearly pointed out by the pres-ence of bands at 1544 cm�1 and 1627 cm�1, corresponding tocheese whey, on used wooden shelves – were probably respon-sible for modifications of specific wood signals on used shelves.Reblochon traces could notably explain that the main band onwood (at 2896 cm�1) was not found after cheese contact withshelves and the 2 folds increase of the 2925 cm�1 peak heighton used wooden shelves (before and after brushing). In the sameway, the 1643 cm�1 peak was 4.15 folds higher for woodenshelves after ripening than for unused ones. Secondly, it alsoseemed that the cleaning process by brushing was soft and didnot remove all the materials from wooden shelves. Indeed, afterbrushing, shelves spectra presented intermediate features be-tween those of control wooden shelves and of Reblochon rind.This could result from the presence of compounds from Reblo-chon rind like fatty acids and microorganisms which wouldnot be removed by the step of cleaning. In fact, Reblochon rindis mainly composed by microorganisms and more precisely byGeotrichum candidum, a yeast-like fungus, G. candidum is usedas a starter in the dairy industry and colonies nearly all sur-face-ripened cheeses during the early stages of ripening (Berger,Khan, Molimard, Martin, & Spinnler, 1999). It starts to grow onthe surface of the rind of Camenbert, Pont l’Eveque, Munster,Limburger and Livarot cheeses as well as on fresh goats’ andewe’s milk cheeses at the beginning of the cheese-ripening pro-cess, and on semi-hard cheeses such as St Nectaire and Reblo-chon (Guéguen, 1984). In a previous work, we pointed out thepresence of biofilms (i.e., microbes associated with a surface,typically developing into a densely packed community of cellsinterconnected by a biopolymeric matrix) from rind cheese onwooden shelves. The dominant microflora was composed bymicrococci–corynebacteria, yeasts and moulds (Mariani et al.,2007). FTIR signal of rind cheese is thus a complex spectrumrepresentative of cheese matrix and cheese surface microorgan-isms and, in this context, band at 1078 cm�1 is probably charac-teristic of Reblochon cheese endogenous flora, since suchmicroorganisms were the most likely source of glycogen or nu-cleic acids in cheese rind. The presence of this band on woodenshelves thus indicates that some microorganisms, which com-posed the biofilm, were not removed by the cleaning and dryingprocedures. This conclusion is in accordance with SEM observa-tions of the surface of wooden shelves at the end of the ripeningprocess and after cleaning which showed presence of bacteriaand yeasts on wood fibres and in the cracks (data not shown).The prevalence of bands characteristic of microorganisms fromsuch biofilms could also explain the inversion of the protein/polysaccharide ratio (Table 2). Thus, contribution of proteinsfrom microorganisms’ cell wall could be responsible for the in-crease of protein bands in cheese-ripening shelves and Reblo-chon rind. However, it appears that wood composition wasaffected by brushing, because we observed that the intensity ofbands from cellulose and hemicellulose on wood materials de-creased after brushing. For example, the contribution of the1057 cm�1 band (which was assigned to cellulose and hemicellu-lose) was reduced between unused wood (38.5%) and wood aftercleaning (25.9%).

N. Oulahal et al. / Food Control 20 (2009) 658–663 663

4. Conclusion

The common practice in biofilm investigation is the destructiveanalysis of samples on which biofilms developed. Furthermore,very thin films and surface coatings, which are not accessible tostandard chemical methods, can be examined. The compositionof surface coatings like biomass or other surface contaminantscan be detected. FTIR studies data support the results obtainedfrom other tests. These different measurement techniques demon-strate that ATR-FTIR spectroscopy is suitable for biofilm and sur-face analysis and can be applied in many different ways and forwood analysis, an important material for the production of tradi-tional foods.

Acknowledgments

This work was financially supported by the French ‘‘Associationde Coordination Technique pour l’Industrie Agro-alimentaire” (AC-TIA). We are very grateful to the SARL Joseph Paccard for providingthe shelves and to our partners (Syndicat Interprofessionnel deReblochon, Arilait Recherches, Comité National des Appellationsd’Origines Laitières (CNAOL), Entremont-Alliance, and Aérial) fortheir financial and scientific participation.

References

Alen, R., Kotilainen, R., & Zaman, A. (2002). Thermochemical behavior of Norwayspruce (Picea abies) at 180–225 degrees C. Wood Science and Technology, 36,163–171.

Berger, C., Khan, J. A., Molimard, P., Martin, N., & Spinnler, H. E. (1999). Production ofsulfur flavors by ten strains of Geotrichum candidum. Applied and EnvironmentalMicrobiology, 65, 5510–5514.

Chan, R., & Chen, V. (2004). Characterization of protein fouling on membranes:Opportunities and challenges. Journal of Membrane Science, 242, 169–188.

Chen, M. X., & Irudayaraj, J. (1998). Sampling technique for cheese analysis by FTIRspectroscopy. Journal of Food Science, 63, 96–99.

Galichet, A., Sockalingum, G. D., Belarbi, A., & Manfait, M. (2001). FTIR spectroscopicanalysis of Saccharomyces cerevisiae cell walls: Study of an anomalous strain

exhibiting a pink-colored cell phenotype. FEMS Microbiology Letters, 197,179–186.

Guéguen, M. (1984). Contribution to knowledge of Geotrichum candidum and inparticular to its variation. Consequences for dairy technology. Ph.D. Thesis,Université de Caen Basse-Normandie, Caen, France.

Legal, J. M., Manfait, M., & Theophanides, T. (1991). Applications of FTIRspectroscopy in structural studies of cells and bacteria. Journal of MolecularStructure, 242, 397–407.

Maquelin, K., Kirschner, C., Choo-Smith, L. P., Van Den Braak, N., Endtz, H. P.,Naumann, D., et al. (2002). Identification of medically relevant microorganismsby vibrational spectroscopy. Journal of Microbiological Methods, 51, 255–271.

Marcotte, L., Kegelaer, G., Sandt, C., Barbeau, J., & Lafleur, M. (2007). An alternativeinfrared spectroscopy assay for the quantification of polysaccharides inbacterial samples. Analytical Biochemistry, 361, 7–14.

Mariani, C., Briandet, R., Chamba, J. F., Notz, E., Carnet-Pantiez, A., Eyoug, R. N., et al.(2007). Biofilm ecology of wooden shelves used in ripening the French raw milksmear cheese Reblochon de Savoie. Journal of Dairy Science, 90, 1653–1661.

Mariani, C., Oulahal, N., Chamba, J. F., Dubois-Brissonnet, F., Notz, E., & Briandet, R.(in press). Inhibition of Listeria monocytogenes by biofilms present on woodenshelves used for cheese ripening. International Journal of Food Microbiology.

Naumann, A., Navarro-Gonzalez, M., Peddireddi, S., Kues, U., & Polle, A. (2005).Fourier transform infrared microscopy and imaging: Detection of fungi in wood.Fungal Genetics and Biology, 42, 829–835.

Pandey, K. K. (2005). Study of the effect of photo-irradiation on the surfacechemistry of wood. Polymer Degradation and Stability, 90, 9–20.

Polovka, M., Polovkova, J., Vizarova, K., Kirschnerova, S., Bielikova, L., & Vrska, M.(2006). The application of FTIR spectroscopy on characterization of papersamples, modified by bookkeeper process. Vibrational Spectroscopy, 41,112–117.

Schmitt, J., & Flemming, H. C. (1998). FTIR-spectroscopy in microbial and materialanalysis. International Biodeterioration & Biodegradation, 41, 1–11.

Sockalingum, G. D., Bouhedja, W., Pina, P., Allouch, P., Mandray, C., Labia, R., et al.(1997). ATR-FTIR spectroscopic investigation of imipenem-susceptible and –resistant Pseudomonas aeruginosa isogenic strains. Biochemical and BiophysicalResearch Communications, 232, 240–246.

Van Der Ven, C., Muresan, S., Gruppen, H., De Bont, D. B. A., Merck, K. B., & Voragen,A. G. J. (2002). FTIR spectra of whey and casein hydrolysates in relation to theirfunctional properties. Journal of Agricultural and Food Chemistry, 50, 6943–6950.

Wikberg, H., & Maunu, S. L. (2004). Characterisation of thermally modified hard-and softwoods by C-13 CPMAS NMR. Carbohydrate Polymers, 58, 461–466.

Yildiz, S., Gezer, E. D., & Yildiz, U. C. (2006). Mechanical and chemical behavior ofspruce wood modified by heat. Building and Environment, 41, 1762–1766.

Zhu, H. H., & Nystrom, M. (1998). Cleaning results characterized by flux, streamingpotential and FTIR measurements. Colloids and Surfaces A – Physicochemical andEngineering Aspects, 138, 309–321.