1-s2.0-s0309174004001810-main

Upload: claudia-ilina

Post on 05-Apr-2018

212 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/2/2019 1-s2.0-S0309174004001810-main

    1/6

    On-line determination of fatty acid composition in intramuscularfat of Iberian pork loin by NIRs with a remote reflectance fibre

    optic probe

    I. Gonzalez-Martn *, C. Gonzalez-Perez, N. Alvarez-Garca, J.M. Gonzalez-Cabrera

    Departamento de Qumica Analtica, Nutricion y Bromatologa, Facultad de CC, University of Salamanca, Qumicas,

    C/Plaza de la Merced s/n, 37008 Salamanca, Spain

    Received 8 December 2003; received in revised form 24 June 2004; accepted 12 July 2004

    Abstract

    A near infrared spectrometer equipped with a standard 210/210 bundle remote reflectance fibre-optic probe, with a 5 5 cm

    quartz window type, was used for the determination of fatty acids in the Longissimus dorsimuscle of Iberian breed swine. The fatty

    acids C14:0, C16:0, C16:1, C17:0, C17:1, C18:0, C18:1, C18:2, C18:3, Rpolyunsaturated, Rmonounsaturated and Rsaturated were

    determined in samples of intramuscular fat from Iberian breed swine by direct application of the fibre-optic probe onto the loin

    sample, with no treatment or manipulation of the sample.

    The regression method employed was modified partial least squares. The calibration results using the fibre-optic probe for 74 loin

    samples had multiple correlation coefficients (RSQ) for C14:0, C16:0, C16:1, C17:0, C17:1, C18:0, C18:1, C18:2, C18:3, Rpolyun-

    saturated, Rmonounsaturated and Rsaturated acid of 0.785, 0.798, 0.788, 0.825, 0.762, 0.765, 0.696, 0.859, 0.878, 0.807, 0.943,

    0.858, respectively, and standard errors of prediction corrected for the same fatty acids (%) of 0.08, 0.63, 0.26, 0.02, 0.02, 0.51,

    0,77, 0.64, 0.05, 1.06, 0.34, 0.70, respectively.

    The robustness of the method was checked by applying the fibre-optic probe to unknown samples of Iberian breed pork loin in a

    slaughterhouse, using 15 samples for the external validation.

    2004 Elsevier Ltd. All rights reserved.

    Keywords: Iberian-pork loin; Fatty acid; Intramuscular fat; NIR; Fibre-optic probe; Determination

    1. Introduction

    Pork products from the Iberian breed of swine are

    widely accepted by consumers owing to the special char-acteristics of the swine from which they are obtained. In

    the present work, products from the Iberian swine, or

    the same crossed with animals of the Duroc breed, fed

    for two months on a range regimen, mainly acorns

    and grass (montanera swine) or over one month on a

    montanera regimen followed by another month with

    commercial swine feed (recebo swine) or on an intensive

    regimen with commercial feed (feed swine) (Cabeza de

    Vaca, Esparrago, Fallola, & Vazquez, 1992; MAPA,

    1984), were studied.The content of intramuscular lipids and the nature of

    the fatty acids of the Longissimus dorsimuscle from Ibe-

    rian swine at the time of sacrifice depend on the genetic

    origin of the animals, their age, and their feeding regi-

    men (such as the montanera diet). However, the determi-

    nation of fatty acids in the muscle of Iberian breed swine

    is a long and tedious task, involving extraction of total

    lipids (Folch, Lees, & Stanley, 1957), and determination

    of fatty acid methyl esters by gas chromatography.

    0309-1740/$ - see front matter 2004 Elsevier Ltd. All rights reserved.

    doi:10.1016/j.meatsci.2004.07.003

    * Corresponding author. Tel./fax: +34 23 29448.

    E-mail address: [email protected] (I. Gonzalez-Martn).

    www.elsevier.com/locate/meatsci

    Meat Science 69 (2005) 243248

    MEATSCIENCE

    mailto:[email protected]:[email protected]
  • 8/2/2019 1-s2.0-S0309174004001810-main

    2/6

    NIRS methodology was used by De Pedro, Garrido,

    Bares, Casillas, and y Murrai (1992) to predict the sub-

    cutaneous fat content of oleic, linoleic, palmitic and

    stearic acids in Iberian pork. Later, Garca Olmo

    (1999), applied the method to classify pieces of fresh Ibe-

    rian ham comparing the results with those obtained with

    gas chromatography.The work of Gonzalez Martn, Gonzalez Perez,

    Hernandez Mendez, and Alvarez Garca (2002a, 2001)

    with NIRS and a fibre-optic probe used directly on the

    carcasses of Iberian swine allowed the determination

    of fat and protein and the mineralogical composition

    of pork loin. Additionally, Gonzalez Martn, Gonzalez

    Perez, Hernandez Mendez, and Alvarez Garca (2002b,

    2003) have reported on-line analysis methods of fatty

    acids (C14:0, C16:0, C18:0, C18:1, C18:2, C18:3,

    C20:1, Rpolyunsaturated, Rmonounsaturated and Rsat-

    urated) in subcutaneous fat, allowing instantaneous

    classification of Iberian pork as a function of the ani-

    mals feeding regimen.

    Here we report the rapid determination of the fatty

    acid content of the lipids of the muscle tissue of Iberian

    swine using NIRS technology with a fibre-optic probe,

    by direct application of the probe to the Longissimus

    dorsimuscle with no prior sample treatment or manipu-

    lation. This offers an on-line technique for the immedi-

    ate determination of the fatty acid composition of

    intramuscular fat in the animals.

    2. Material and methods

    2.1. Samples

    We examined 74 samples of pork loin (Longissimus

    dorsi muscle) from Iberian swine purchased from

    Ganaderos Salmantinos de Porcino Iberico S.A taken

    from the end of the loin of the animals, cut longitudi-

    nally to provide a slice measuring 8 12 2 cm. NIR

    spectra were measured by applying the fibre-optic probe

    to intact pork loin samples. The loin samples were from

    animals that had been fed on the so-called montanera,

    recebo or feed regimens. The samples were ground

    and homogenised with a Knife 1095 Mill Sample

    homogeniser from Foss Tecator, and conserved at

    18 C.

    2.2. Intramuscular fat and fatty acid analysis

    To extract total lipids from the Longissimus dorsi

    muscle, 50 g of sample previously ground and homoge-

    nised with chloroform/methanol (1/2), according to the

    method of Bligh and Dyer (1959) were used. The solvent

    was removed under vacuum in a rotary evaporator.

    Fatty acid methyl esters (FAMEs) of muscle were pre-

    pared by reaction with a solution of potassium hydrox-

    ide in methanol before analysis by gas chromatography,

    using a HewlettPackard HP-5890A gas chromato-

    graph, equipped with a flame ionisation detector

    (FID). All analyses were performed in duplicate (ISO

    Norm 5508:1990).

    2.3. NIR spectroscopy

    A Foss NIRsystems 5000, with a standard 1.5 m

    210/210 bundle fibre-optic probe, Ref. No. R6539-A,

    was used. The probe employs a remote reflectance sys-

    tem and uses a ceramic plate as reference. The win-

    dow is of quartz, with a 5 5 cm surface area,

    measuring reflectance in the IR zone close to 1100

    2000 nm.

    Measurements were carried out in reflectance mode

    between 1100 and 2000 nm. Spectra were recorded at

    intervals of 2 nm, performing 32 scans for both the ref-

    erence and samples. The average spectrum was used for

    NIR analysis. The software used was Win ISI 1.05, in-

    stalled on a HewlettPackard Pentium III computer.

    2.4. Statistical analyses

    The Mahalanobis distance (H statistic) is calculated

    from principal component analysis scores. The results

    indicate how different a sample spectrum is from the

    average sample of the set (Willians & Norris, 1987). A

    sample with an H statistic ofP3.0 standardised units

    from the mean spectrum is defined as a global Houtlier

    and is then eliminated from the calibration set.

    Calibrations were performed by modified partial leastsquares regression (MPLS). To optimise the accuracy of

    calibration, several scattering corrections and mathe-

    matical treatments were tested (standard normal variate,

    SNV; De-trending, DT; multiplicative scatter correc-

    tions, MSC; first derivative and second derivative).

    The best one was selected for each constituent on the ba-

    sis of the highest multiple correlation coefficient (RSQ)

    and the lowest standard error of calibration and cross-

    validation (SEC and SEV, respectively). Assessment of

    the calibration model was performed by cross-valida-

    tion. In this method, the set of calibration samples is di-

    vided in groups, using one of them to check the results

    (prediction) and the remaining to construct the calibra-

    tion model. The model is repeated as many times as

    there are groups in such a way that all pass through

    the calibration set and the prediction set. The same

    process was followed for the ground and intact samples.

    Samples from the validation set were then analysed

    with these equations, which gave a standard error of

    prediction corrected (SEPC) and bias (mean of residu-

    als, defined as the difference between the laboratory va-

    lue and the value predicted by the equation) for each

    constituent. In this step, samples with high residual val-

    ues were eliminated, using the T> 2.5 criterion.

    244 I. Gonzalez-Martn et al. / Meat Science 69 (2005) 243248

  • 8/2/2019 1-s2.0-S0309174004001810-main

    3/6

    After the calibration equations for the fibre-optic

    probe had been obtained, they were subjected to exter-

    nal validation by application to a set not involved in

    the calibration process (15 samples) checking the func-

    tioning of the fibre-optic probe for instantaneous analy-

    sis at production level at the slaughterhouse of the

    Ganaderos Salmantinos de Porcino Iberica S.A Com-pany in Salamanca (Spain). The spectra were recorded

    by direct application of the probe on intact pieces of Ibe-

    rian pork loin. Spectra were recorded in triplicate and

    the spectral average was taken. The calibration equa-

    tions obtained in the development of the procedure were

    applied and the predicted values were compared with the

    laboratory results obtained later.

    3. Results and discussion

    3.1. Chemical analyses and spectral information

    The chemical compositions of the pork loin samples

    used for calibration are shown in Table 1, 72 samples

    were used. The results obtained for the content of the

    different fatty acids revealed a broad range of variabil-

    ity, which is important when searching for calibrations

    equations to be used later in the prediction. The causes

    of such high variability must be sought in the feeding

    regimen and genetics of the animals (the Iberian swine

    used in this work were Iberian swine of the subgenus

    Sus mediterraneus crossed with Duroc-Jersey, with

    Iberian genetics varying between 50% and 75%).

    Fig. 1 shows the spectra of a sample of loin ob-tained with the fibre-optic probe by direct application

    on the sample together with one of the mathematical

    treatments that proved optimum for the calibrations

    of the fatty acid C18:1 in the intramuscular fat of

    the loin, using second derivative and DT. The best

    of the different mathematical treatments for each fatty

    acid was applied (Table 2). The effects of dispersion

    Table 1

    Statistical overview of chemical analysis (all units in %) N= 74

    Fatty acid Minimum Maximum Mean SD

    12:0 0.05 0.10 0.07 0.01

    14:0 0.95 1.70 1.30 0.18

    16:0 20.59 27.13 23.64 1.58

    16:1 2.59 5.51 3.73 0.61

    17:0 0.09 0.31 0.16 0.0517:1 0.13 0.46 0.21 0.05

    18:0 8.94 13.71 11.09 1.14

    18:1 49.06 58.30 52.83 1.74

    18:2 2.46 9.23 5.42 1.81

    18:3 0.14 0.82 0.36 0.17

    20:0 0.10 0.32 0.19 0.05

    20:1 0.76 1.55 0.99 0.13

    SFA 31.03 41.47 36.45 2.62

    MUFA 54.39 64.76 57.76 1.78

    PUFA 2.62 9.91 5.79 1.97

    Fig. 1. (a) NIR spectrum of the sample of loin measured with the fibre-

    optic probe. (b) Corrected spectrum using second derivative and DT

    for the fatty acid C18:2.

    Table 2

    Calibration statistical descriptors for the NIR determination of fatty acids in intramuscular fat from loin

    Fatty acid N Mathematical treatment RSQ SEC SECV No. of principal components Probability explained (%)

    12:0 69 None/2a derivada 0.622 0.007 0.008 7 99.46

    14:0 70 None2a derivada 0.785 0.080 0.137 7 99.51

    16:0 69 DT/2a derivada 0.798 0.662 1.051 7 99.54

    16:1 71 DT/2a derivada 0.788 0.272 0.472 9 99.28

    17:0 68 None/2a derivada 0.825 0.017 0.026 9 99.30

    17:1 72 DT/2a derivada 0.762 0.019 0.031 7 99.54

    18:0 73 Standar MSC/2a der 0.765 0.538 0.934 8 99.52

    18:1 67 DT/2a derivada 0.696 0.816 1.094 7 99.46

    18:2 71 DT/2a derivada 0.859 0.681 1.146 9 99.28

    18:3 67 DT/2a derivada 0.878 0.054 0.099 7 99.36

    20:0 71 SNV/2a derivada 0.458 0.035 0.043 9 99.30

    20:1 71 DT/2a derivada 0639 0.070 0.098 8 99.52

    SFA 70 Estandar MSC /2a deiv 0.807 1.116 1.763 8 99.52

    MUFA 69 None/3a derivada 0.943 0.351 1.287 7 99.36

    PUFA 71 DT/2a derivada 0.858 0.743 1.249 9 99.28

    I. Gonzalez-Martn et al. / Meat Science 69 (2005) 243248 245

  • 8/2/2019 1-s2.0-S0309174004001810-main

    4/6

    were removed using MSC, SNV, DT or SNVDT.

    MSC was first used by Geladi, Mac Dougall, and

    Martens (1985) and prevents the dispersion in the

    samples from imposing itself on the chemical signals.

    Dhanoa, Lister, and Barnes (1995) indicate that

    SNVDT were introduced not only for the reduction

    of multicollinearity but also to calculate spectral dif-

    ferences by reducing the confounding effects of base-

    line shift and curvature.

    The spectral information defines a series of charac-

    teristic absorption bands. Thus, the CH bond, which

    is a fundamental constituent of fatty acid molecules,

    absorbs clearly at wavelengths close to 1200, 1400,

    1750, 2310 and 2340 nm (Willians & Norris, 1987).

    Moreover, the 23102340 region corresponds to the

    combination bands of the CH bond, and the absorp-

    tion produced in the 17201760 region corresponds to

    the first overtone of that bond (Shenk, Workman, &

    Westerhaus, 1992). Osborne, Fearn, and Hindle

    (1993) have attributed the absorption produced at a

    wavelength of 1210 nm the second overtone of the

    CH2 bond. The absorption in the 21502190 region

    at 1680 nm indicates the presence of cis double bonds

    in the molecules; i.e., the existence of unsaturated fatty

    acids (Garrido-Varo, Carrete, & Fernandez-Cabanas,

    1998) in the samples analysed.

    Fig. 2. Correlation of the values obtained at the laboratory with respect to those predicted by NIR for measurement with a fibre-optic probe of the

    different fatty acids in Iberian pork loin samples.

    246 I. Gonzalez-Martn et al. / Meat Science 69 (2005) 243248

  • 8/2/2019 1-s2.0-S0309174004001810-main

    5/6

    3.2. Determination of the fatty acid composition

    To calibrate fatty acids in the intramuscular fat of

    Iberian pork loin, the NIRS technology with a fibre-op-

    tic probe was used, applying the probe directly onto the

    loin samples. Calculation of the statistical parameters of

    the calibration equations for each fatty acid is shown inTable 2, which also reflects the best treatments. After the

    number of principal components had been calculated,

    the detection of anomalous spectra was accomplished

    using the Mahalanobis distance (H statistic), establish-

    ing a limit value ofH= 3.0, such that the spectra whose

    H distance was greater than this figure were considered

    different from the spectral population and were dis-

    carded. The numbers of remaining samples (N) are

    shown in Table 2.

    Regarding the number of principal components, be-

    tween 7 and 9 components were used, depending on

    the fatty acid to be determined, and these accounted

    for 99.30% of the variance in the loin samples.

    We consider that the results obtained in the determi-

    nation of the fatty acids C14:0, C16:0, C16:1, C17:0,

    C17:1, C18:0, C18:1, C18:2, C18:3, RMUFA, RSFA

    and RMFA are very good. The results obtained in the

    determination of C12:0, C20:0 and C20:1 are acceptable.

    The concentration ranges and standard deviations for

    each fatty acid that it is possible to determine in the

    intramuscular fat of Iberian pork loin are shown in

    Table 3, which reveals the broad applicability of the cal-

    ibration equations obtained.

    3.3. Validation

    3.3.1. Internal validation (prediction)

    Model evaluation was performed by cross-validation.

    In this method, the set of calibration samples is divided

    into a series of subsets: in our case 6. Of these, 5 are ta-

    ken for the calibration set and one for the prediction set.

    The process is repeated as many times as there are sets,

    such that all pass through the calibration set and the

    prediction set. Using this process, we validated the model

    used and checked its prediction capacity. This process

    was carried out for the validation of fatty acids in intact

    loin samples.The predicted values gave validation errors that were

    combined into SEP(C). Fig. 2 shows the correlation of

    the values obtained at the laboratory with respect to

    those predicted by NIR with a fibre-optic probe for

    the fatty acids C14:0, C16:0, C16:1, C17:0, C17:1,

    C18:0, C18:1, C18:2, C18:3, RSFA, RMUFA, RPUFA

    in the intramuscular fat of Iberian pork loin, together

    with the statistical descriptors of prediction. From this

    information it may be deduced that the NIRS technique,

    using a fibre-optic probe, is a good alternative for the

    determination of the content in fatty acids in samples

    of intramuscular fat from Iberian breed swine. In later

    studies the method will be use to analyse the composi-

    tion in fatty acids in the intramuscular fat of the loin

    by NIRS and a fibre-optic probe in order to classify

    the animals on the basis of their feeding regimen in

    the fattening stage (acorn, recebo and feed).

    3.3.2. External validation

    We checked the robustness of the method by applying

    the fibre-optic probe to 15 new samples in a slaughter-

    house. The procedure was as follows: spectra were re-

    corded in triplicate and the spectral mean was taken;

    the calibration equations obtained during the work were

    applied and the predicted values were compared withthose obtained later using gas chromatography. Table 4

    shows the differences found between the reference

    Table 3

    Applicability of the calibration equations relating to fatty acids in

    intramuscular fat of Iberian pork loin

    Fatty acid Est. min. Est. max Mean SD

    12:0 0.04 0.10 0.07 0.01

    14:0 0.77 1.81 1.29 0.17

    16:0 19.17 28.01 23.59 1.47

    16:1 1.95 5.49 3.72 0.59

    17:0 0.04 0.28 0.16 0.04

    171 0.09 0.33 0.21 0.04

    18:0 7.73 14.39 11.06 1.11

    18:1 48.41 57.29 52.85 1.48

    18:2 0.06 10.94 5.50 1.81

    18:3 0.00 0.82 0.36 0.15

    20:0 0.04 0.32 0.18 0.05

    20:1 0.63 1.33 0.98 0.12

    SFA 28.63 42.88 36.25 2.54

    MUFA 53.29 62.10 57.69 1.47

    PUFA 0.00 11.78 5.87 1.97

    Table 4

    External validation for the determination of fatty acids in the

    intramuscular fat of Iberian breed swine

    Fatty acid Range Differences (%)

    C12:0 0.060.09 9.93

    C14:0 1.111.61 12.15

    C16:0 21.6925.80 4.38

    C16:1 2.964.97 11.62

    C17:0 0.110.21 14.83

    C17:1 0.150.24 8.24

    C18:0 10.0613.90 5.91

    C18:1 49.7854.24 2.09

    C18:2 2.719.53 13.69

    C18:3 0.160.67 26.39

    C20:0 0.130.26 17.90

    C20:1 0.831.14 7.35

    SFA 34.1841.57 3.85

    MUFA 54.8059.19 2.37

    PUFA 2.8710.20 13.67

    Differences between the reference method (gas chromatography) and

    the NIRS technique.

    I. Gonzalez-Martn et al. / Meat Science 69 (2005) 243248 247

  • 8/2/2019 1-s2.0-S0309174004001810-main

    6/6

    method (gas chromatography) and the NIRS technique

    in the external validation.

    In view of the present findings, it may be concluded

    that the NIRS technique with a fibre-optic probe is good

    for the determination of fatty acids in the intramuscular

    fat of Iberian pork loin. Furthermore, use of the fibre-

    optic probe enables determinations to be made simplyby placing the probe on the sample itself, with no previ-

    ous treatment. This has evident advantages as regards

    the possible speed of the analyses.

    Acknowledgements

    The authors thank the Company Ganaderos Sal-

    mantinos de Porcino Iberica S.L, Salamanca (Spain),

    for providing the samples and the participation of INIA,

    Ref CAL02-053 thanks to which this work was possible.

    References

    Cabeza de Vaca, F., Esparrago, F., Fallola, A., & Vazquez, F. Ma.

    (1992). Coste de la calidad en las producciones de cerdo Iberico:

    Montanera, recebo y pienso. Jornadas tecnicas sobre obtencion de

    productos ganaderos naturales en el ecosistema de la dehesa..

    De Pedro, E. J., Garrido, A., Bares, I., Casillas, M., & y Murrai, I.

    (1992). In K. I. Hildrum, et al. (Eds.), Near infrared spectroscopy

    bridging the gap between data analysis and NIR applications

    (pp. 341347). NY: Ellis Horwood.

    Dhanoa, M. S., Lister, S. J., & Barnes, R. J. (1995). On the scales

    associated with near-infrared reflectance difference spectra.Applied

    Spectroscopy, 49(6), 765.

    Folch, J., Lees, M., & Stanley, G. H. S. (1957). A simple method forthe isolation and purification of total lipids of animal tissue.

    Journal of Biological Chemistry, 226, 497.

    Garca Olmo, J. (1999). Clasificacion y autentificacion de canales de

    cerdo Iberico mediante espectroscop a en el infrarrojo cercano

    (NIRS). Anaporc, 192, 113141.

    Garrido-Varo, A., Carrete, R., & Fernandez-Cabanas, J. (1998). Use

    of difference near infrared reflectance spectra to extract relevant

    information from the spectra of agro-food products. Journal of

    Near Infrared Spectroscopy, 6, 8995.

    Geladi, P., Mac Dougall, D., & Martens, H. (1985). Linearizacion and

    scatter-correction for near-infrared reflectance spectra of meat.

    Applied Spectroscopy, 39(3), 491500.

    Gonzalez Martn, I., Gonzalez Perez, C., Hernandez Mendez, J., &

    Alvarez Garca, N. (2002a). Mineral analysis (Fe, Zn, Ca, Na, K)

    of fresh Iberian pork loin by near infrared spectrometry. Deter-

    mination of Fe Na and K with a remote fibre-optics reflectance

    probe. Analytica Chimica Acta, 468, 293301.

    Gonzalez Martn, I., Gonzalez Perez, C., Hernandez Mendez, J., &

    Alvarez Garca, N. (2002b). Determinacion de acidos grasos en

    grasa subcutanea de cerdo iberico mediante espectroscopia en el

    infrarrojo cercano (NIRS) con sonda de fibra optica. Eurocarne,

    109, 991007.

    Gonzalez Martn, I., Gonzalez Perez, C., Hernandez Mendez, J., &

    Alvarez Garca, N. (2003). Determination of fatty acids in the

    subcutaneous fat of Iberian breed swine by near infrared spectr-

    oscopy (NIRS) with a fibre-optic probe. Meat Science, 65, 713719.

    Gonzalez Martn, I., Gonzalez Perez, C., Hernandez Mendez, J.,

    Alvarez Garca, N., & Hernandez Andaluz, J. L. (2001). Not

    destructive determination of proteins and infiltred fat from Iberian

    breed swines loin with the help of the fibre optics of remote

    reflectance. Analytica Chimica Acta, 453, 281288.

    Ministerio de Agricultura, Pesca y Alimentacion. (1984). Servicio de

    publicaciones agrarias. Madrid: MAPA.

    Osborne, B. G., Fearn, T., & Hindle, P. H. (1993). Practical NIR

    spectroscopy with applications in food and beverage analysis. Essex,

    UK: Longman Scientific & Technical.

    Shenk, J. S., Workman, J. J., & Westerhaus, M. O. (1992). In D. A.

    Burns & E. W. Ciurczak (Eds.), Handbook of near-infrared analysis

    (pp. 181). New York: Marcel Dekker.

    Willians, P. C., & Norris, K. H. (Eds.). (1987). Near-infraredtechnology in the agricultural and food industries (p. 330). St. Paul,

    MN: American Association of Cereal Chemists.

    248 I. Gonzalez-Martn et al. / Meat Science 69 (2005) 243248