Introduction
Recent research (Enser et al., Reference Enser, Hallett, Hewett, Fursey, Wood and Harrington1998; Nürnberg et al., Reference Nürnberg, Grumbach, Zupp, Hartung, Nürnberg and Ender2001; Aurousseau et al., Reference Aurousseau, Bauchart, Calichon, Micol and Priolo2004) has demonstrated that meat from pasture-fed ruminants has a healthier fatty acid composition than meat from animals fed concentrate diets. Furthermore, consumers are demanding clear information on the food supplied to animals and increasingly focusing on the ‘green image’ of animal products (Prache and Thériez, Reference Prache and Thériez1999). It is therefore important to be able to discriminate between products obtained through different production systems, in particular, pasture feeding v. stall feeding.
Recently, efforts have been made to develop analytical tools to quantify specific compounds in the product or the animal tissues that can act as tracers of the type of food given to animals (Martin et al., Reference Martin, Cornu, Kondjoyan, Ferlay, Verdier-Metz, Pradel, Rock, Chilliard, Coulon and Berdagué2005; Prache et al., Reference Prache, Cornu, Berdagué and Priolo2005).
Carotenoid pigments have been shown to be good biomarkers of pasture feeding in sheep (Prache and Thériez, Reference Prache and Thériez1999; Prache et al., Reference Prache, Priolo, Tournadre, Jailler, Dubroeucq, Micol and Martin2002; Priolo et al., Reference Priolo, Prache, Micol and Agabriel2002). Plasma carotenoid concentration has been successfully used to discriminate between pasture-fed and stall-fed lambs (Prache et al., Reference Prache, Priolo and Grolier2003a), and Prache and Thériez (Reference Prache and Thériez1999) had proposed a mathematical analysis of the reflectance spectrum of the fat at wavelengths between 450 and 510 nm (i.e. the zone of light absorption by carotenoids) to quantify the signature of these pigments and discriminate pasture-fed from stall-fed lamb carcasses. A measurement of the reflectance spectrum of the fat is actually of obvious practical interest since it is non-invasive, takes little time and can easily be implemented in the meat industry with a portable spectrophotometer.
However, using the full data set of optical information contained in the fat reflectance spectrum, i.e. all the reflectance data at wavelengths between 400 and 700 nm, may provide further valuable information and better discrimination than only using the fat reflectance data at wavelengths between 450 and 510 nm. The purpose of this study was therefore to compare two spectral methods using a portable spectrophotometer, by assessing their reliability in terms of discriminating between carcasses from pasture-fed or concentrate-fed lambs (1) using the quantification of light absorption by carotenoid pigments at wavelengths between 450 and 510 nm and (2) using the full data set of optical information provided by the visible reflectance spectrum at wavelengths between 400 and 700 nm. The interactions with the site of measurements (perirenal v. subcutaneous caudal fat) and the time elapsed between slaughter and measurement (0 h v. 24 h post mortem) were investigated, since these factors may affect carotenoid concentration and the spectral characteristics of the fat (Kirton et al., Reference Kirton, Grane, Paterson and Clare1975; Priolo et al., Reference Priolo, Prache, Micol and Agabriel2002).
Material and methods
This study was carried out over 2 years (2004–05) at the Unité Expérimentale des Mont Dore, Site d’Orcival, an experimental farm run by the Clermont-Ferrand/Theix INRA Centre in France. The animals were handled by specialised personnel who cared for animal welfare in line with European Union directive no. 609/1986.
Animals, diets and slaughter procedures
A total of 307 Limousine-breed lambs were used; 143 were pasture-fed (group P, 81 males and 62 females) and 164 were stall-fed (group S, 86 males and 78 females). One hundred and thirty-four lambs were slaughtered in 2004 (43 S and 91 P) and 173 lambs were slaughtered in 2005 (121 S and 52 P).
The pasture-fed lambs were born in April (over 4 to 24 April 2004 and 6 to 24 April 2005, except for one lamb that was born on 17 May 2005). They were offered ad libitum a permanent pasture from 3 May in 2004 and from 12 May in 2005 until slaughter, which occurred between 4 August and 4 November in 2004 and between 17 August and 3 November in 2005.
The same pasture was used in both years. It was maintained at a leafy, green vegetative state. Its botanical composition (Table 1) was visually assessed in July 2005 following the method described by Daget and Poissonnet (Reference Daget and Poissonnet1971). The presence and the specific volume of each species was recorded at 300 points located on 12 transects that were located in areas representative of the paddock vegetation, with 25 points set, 1.0 m apart on each transect. For each point, an abundance score was given for each species present, so that the sum of all scores at each point equalled 6. The proportion of each species in relation to total biomass was then calculated as the ratio of the sum of the scores for each species divided by the sum of all scores. The lambs weighed 9.2 (standard deviation (s.d.) 2.21) kg, when at a mean age of 22 (s.d. 6.5) days they were turned out to pasture, except for one lamb that was born at pasture. They were weaned on 26 July in 2004 and on 25 July in 2005. Pasture-fed lambs received no supplementation at pasture.
Table 1 Botanical composition of the pasture grazed by pasture-fed lambs (% of total biomass)
Stall-fed lambs were given ad libitum indoor access to commercial concentrate and hay. The composition of the concentrate offered is given in Table 2. The stall-fed lambs slaughtered in 2004 were born between 28 November 2003 and 22 April 2004, while those slaughtered in 2005 were born between 13 November 2004 and 24 April 2005. Samples of the hay and concentrate offered to the animals were taken twice weekly to assess carotenoid concentration.
Table 2 Composition of the concentrate dispensed to stall-fed lambs (g/kg, as-fed basis)
Water and salt blocks were made constantly available in both feeding treatments. Lambs were slaughtered when they had reached a target condition score of 3 (on a scale of 0 to 5), which was manually assessed by skilled technicians according to the method described by Russel et al. (Reference Russel, Doney and Gunn1969). Animals were transported to the abattoir by truck. The abattoir was located within 25 km of the pasture and the stall. Immediately after their arrival at the slaughterhouse, the animals were electrically stunned and slaughtered by throat cut. The carcasses were placed in a chiller set at 4°C until 24 h post mortem, and they were always kept in the dark.
Measurements
Animal characteristics at slaughter
The lambs were weighed just before slaughter. Carcass weight was measured 24 h post mortem. The perirenal fat and the kidneys were then removed from the cold carcass. The fat was separated from the kidneys using a knife and then weighed. Subcutaneous fat thickness was measured at 24 h post mortem by making two incisions through the fat along lines extending 4 cm ventro-laterally from the dorsal mid-line at the last rib and, at the limit of that cut, extending 4 cm cranially. A flap of fat was raised and subcutaneous fat thickness was measured at the intersection of the incisions (Fisher and De Boer, Reference Fisher and de Boer1994).
Reflectance spectrum of perirenal and subcutaneous caudal fat
The reflectance spectrum of perirenal and subcutaneous caudal fat was measured on all lambs at wavelengths between 400 and 700 nm, using a MINOLTA CM-2002 spectrophotometer (D65 illuminant, observer angle 10°). The instrument was equipped with a protective glass visor to protect the eye of the apparatus from the fat sample. This spectrophotometer measures the proportion of light reflected at 10-nm intervals at wavelengths between 400 and 700 nm, and records the corresponding reflectance spectrum. Measurements were made in triplicate, at slaughter and at 24 h post mortem. For the measurements made on perirenal fat 24 h post mortem, a knife was used to obtain a flat surface so that the fat would adhere perfectly to the eye of the apparatus.
Methods used to discriminate pasture-fed from stall-fed lamb carcasses
In method 1, the fat reflectance spectrum data were used at wavelengths between 450 and 510 nm to calculate an index quantifying light absorption by carotenoid pigments in the fat. This index was measured as follows. The reflectance spectrum was translated to give a reflectance value at 510 nm of zero (TR). On the translated spectrum, the integral value (I 450−510) was calculated as follows:
The integral value was averaged over the three measurements, then linear discriminant analysis was performed, followed by a cross-validation procedure to classify the fat samples according to feeding treatment, using Minitab software v.13 (Minitab Inc., Paris).
In method 2, the reflectance spectrum of the fat was explored further, by using the full reflectance data set at wavelengths between 400 and 700 nm to discriminate P lambs from S lambs. The reflectance data (R) at wavelengths between 400 and 700 nm were averaged over the three replicates, then transformed (log (1/R)) and exported into Win ISI II version 1.5 software (Infrasoft International, Port Matilda, PA, USA) for multivariate analysis. The raw reflectance spectra of each tissue representing the two feeding treatments were submitted to discriminant analysis using a partial least squares discriminant analysis (PLS-DA) approach. PLS-DA consists of a PLS regression where the dependent variable is a set of categorical variables describing the different classes of observations. Each sample was assigned a dummy dependent variable according to feeding treatment (1 for P lambs, 2 for S lambs), using a cut-off value of 1.5 to classify samples according to the feeding treatment. A principal component analysis (PCA) performed beforehand was used to rank the reflectance spectra from each feeding treatment according to the Mahalanobis distance (H) to the average reflectance spectrum, in order to detect sample outliers (H > 3). No outliers were found. The models were tested via a cross-validation procedure, in which 66 randomly chosen samples were temporally removed from the initial data set to be used for validation (i.e. a quarter of all data samples). The PLS-DA model was developed based on the other samples (calibration samples) and used to classify the validation samples. This procedure was repeated four times, i.e. until all data set samples had been used through the validation procedure. The cross-validation error of the models was calculated.
Carotenoid concentration in feed
Carotenoids were extracted from concentrate and hay using the procedure described by Cardinault et al. (Reference Cardinault, Doreau, Poncet and Nozière2006). Lipophilic components of 50 mg of lyophilised and ground food were first extracted with acetone and then purified with diethyl ether containing echinenone as internal standard. After being saponified and rinsed with water, the carotenoids were then analysed by high-performance liquid chromatography (HPLC) as described by Lyan et al. (Reference Lyan, Azais-Braesco, Cardinault, Tyssandier, Borel, Alexandre-Gouabau and Grolier2001). The HPLC apparatus consisted of a Waters Alliance 2996 HPLC system with photodiode array detector monitoring between 280 and 600 nm. Carotenoids were separated on a 150 × 4.6 mm, RP C18, 3 μm Nucleosil column coupled with a 250 × 4.6 mm, RP C18, 5 μm Vydac TP 54 column (Interchim, Montluçon, France). Millenium 32 software published by Waters SA (Saint-Quentin-en-Yvelines, France) was used for instrument control, data acquisition and data processing. Detection wavelength for carotenoids was 450 nm, and the compounds were identified by comparing retention times and spectral analyses with those of pure standards (>95%). Concentrations of each compound were calculated using external standard curves and were then adjusted by per cent recovery of the added internal standard.
Crude estimation of total plasma carotenoids
Blood was obtained from all lambs on the day before slaughter in order to measure plasma carotenoid concentration. Blood samples were taken from the jugular vein of each lamb in the morning, and the plasma was stored at −20°C until assay. Carotenoids were extracted from the plasma within 3 months after sampling.
A crude estimation of total carotenoids in the plasma was obtained by a spectrophotometric procedure using the following method. Protein from 3 ml of plasma diluted with 2 ml of distilled water was precipitated with 4 ml of ethanol. Carotenoids were then extracted with 4 ml of hexane. Absorption of the upper layer obtained after centrifugation at 5000 × g for 5 min was measured between 600 and 400 nm using a Kontron Uvikon 860 spectrophotometer (Kontron Instruments S. A., Montigny-le-Bretonneux, France). The concentration of total carotenoids was calculated from absorption maxima (Karijord, Reference Karijord1978), assuming a value of 2500 for the E1% extinction coefficient (Patterson, Reference Patterson1965; Karijord, Reference Karijord1978) and allowing for the dilution of the original sample. Care was taken throughout the experimental and analytical procedure to protect samples from natural light (tubes wrapped in aluminium foil to keep light out, and extraction under dim artificial light). Linear discriminant analysis was performed followed by a cross-validation procedure to classify the plasma samples according to feeding treatment, using Minitab software v.13 (Minitab Inc., Paris).
Statistical analysis
Animal performances from birth to slaughter, carcass characteristics, I 450−510 and plasma carotenoid concentration were subjected to analysis of variance using the GLM procedure of the Statistical Analysis Systems Institute (SAS, 1999) software suite to compare feeding treatments. Variances in I 450−510 and plasma carotenoid concentration were stabilised beforehand using logarithmic transformation.
The proportion of correctly classified carcasses was analysed using the CATMOD procedure of the SAS (1999) software suite using a four-factor model (feeding treatment – pasture v. stall feeding –; discrimination method used in the fat – method 1 v. method 2 –; site of measurement – perirenal v. subcutaneous caudal –; and time of measurement – slaughter v. 24 h post mortem), with repeated measures on the three last factors. The individual animal was considered as the statistical unit.
Results
Animal performances from birth to slaughter and carcass characteristics for P and S lambs are given in Table 3.
Table 3 Animal performances and carcass characteristics: mean and s.d.
†NS = non-significant.
For P lambs, the duration of the grazing period averaged 143 days (s.d. 28.3), ranging from 94 to 187 days. Carotenoid concentrations in the concentrate and hay fed to S lambs were 4.3 and 43.6 μg/g dry matter, respectively (Table 4).
Table 4 Carotenoid concentrations (μg/g dry matter) of the feed offered to stall-fed lambs
Plasma carotenoid concentration
Plasma carotenoid concentration at slaughter (PCCS) was higher for P than for S lambs (P < 0.001). Plasma carotenoid concentration averaged 75 (standard deviation (s.d.) 33.2)μg/l for P lambs, ranging from 27 to 194 μg/l, whereas it averaged only 10 (s.d. 8.5) μg/l for S lambs, ranging from 0 to 43 μg/l (Figure 1). The proportion of P lambs with PCCS greater than or equal to 43 μg/l was 83.8%, whereas the PCCS in 99.1% of the S lambs did not even reach this threshold. This threshold therefore allowed for the correct discrimination of 90.7% of the plasma samples.
Figure 1 Frequency distribution of plasma carotenoid concentration at slaughter in pasture-fed and stall-fed lambs.
Reflectance spectrum of the fat
Reflectance spectra of the fat from P and S lambs are reported in Figures 2 and 3 for measurements made on perirenal and subcutaneous caudal fat, respectively. All spectra showed peaks at 410 to 420 nm wavelengths. The spectra generally showed absorbance at wavelengths 540 nm and 570 to 580 nm, except for the spectra for perirenal fat at slaughter where absorbance occurred at 555 nm for both feeding treatments. The P samples generally showed lower absorbance than S samples throughout the full visible spectrum. The shape of the reflectance spectra also differed between P and S in the area of light absorption by carotenoid pigments. The first derivative of the reflectance spectrum (illustrated in Figure 4 for perirenal fat at 24 h post mortem) was actually higher for P lambs than for S lambs at wavelengths between 510 and 480.
Figure 2 Mean reflectance spectrum of perirenal fat. Pasture-fed lambs (—) v. stall-fed lambs (- - -). Measurements made (a) at slaughter and (b) 24 h post mortem.
Figure 3 Mean reflectance spectrum of subcutaneous caudal fat. Pasture-fed lambs (—) v. stall-fed lambs (- - -). Measurements made (a) at slaughter and (b) 24 h post mortem.
Figure 4 First derivative of the reflectance spectrum of perirenal fat. Pasture-fed lambs (—) v. stall-fed lambs (- - -). Measurements made 24 h post mortem. First derivative at wavelength 410 nm is calculated as follows: (reflectance value at wavelength 410 nm—reflectance value at wavelength 400 nm)/10.
Reflectance spectrum of the fat at wavelengths between 450 and 510 nm (method 1)
The first method tested quantifies light absorption by the carotenoid pigments stored in the fat, using the shape of the reflectance spectrum at wavelengths between 450 and 510 nm.
Mean I 450−510 of perirenal fat at slaughter was significantly different between P and S lambs (P < 0.001), averaging −112.31 and −28.81 units for P and S lambs, respectively. It ranged from −300.05 to 30.77 units for P lambs and from −156.85 to 54.27 units for S lambs. The frequency distribution of P and S lambs in the different classes of I 450−510 values measured on perirenal fat at slaughter is reported in Figure 5a. The proportion of P lambs with a mean I 450−510 greater than or equal to −70 units was 28.7%, whereas the proportion of S lambs with a mean I 450−510 lower than −70 units was 13.4%. This threshold therefore allowed for the correct classification of 79.5% of the lambs when the measurement was made on perirenal fat at slaughter.
Figure 5 Frequency distribution of the pasture-fed and stall-fed lambs in the different classes of the integral value from the translated spectrum of perirenal fat between 450 and 510 nm (I 450−510). Measurements made (a) at slaughter and (b) 24 h post mortem.
Mean I 450−510 of perirenal fat measured 24 h post mortem was significantly different between P and S lambs (P < 0.001), averaging −265.32 and −128.65 units for P and S lambs, respectively. It ranged from −487.97 to −78.85 units for P lambs and from −283.27 to −8.37 units for S lambs. The frequency distribution of P and S lambs in the different classes of I 450−510 values measured on perirenal fat 24 h post mortem is reported in Figure 5b. The proportion of P lambs with a mean I 450−510 greater than or equal to −197 units was 17.5%, whereas the proportion of S lambs with a mean I 450−510 lower than −197 units was 6.1%. This threshold therefore allowed for the correct classification of 88.6% of the lambs when the measurement was made on perirenal fat 24 h post mortem.
Mean I 450−510 of subcutaneous caudal fat at slaughter was significantly different between P and S lambs (P < 0.001), averaging −111.69 and −69.55 units for P and S lambs, respectively. It ranged from −248.15 to −0.43 units for P lambs and from −228.62 to 66.77 units for S lambs. The frequency distribution of P and S lambs in the different classes of I 450−510 values measured on subcutaneous caudal fat at slaughter is reported in Figure 6a. The proportion of P lambs with a mean I 450−510 greater than or equal to −90 units was 41.3%, whereas the proportion of S lambs with a mean I 450−510 lower than −90 units was 29.9%. This threshold therefore allowed for the correct classification of 64.8% of the lambs when the measurement was made on subcutaneous caudal fat at slaughter.
Figure 6 Frequency distribution of the pasture-fed and stall-fed lambs in the different classes of the integral value from the translated spectrum of subcutaneous caudal fat between 450 and 510 nm (I 450−510). Measurements made (a) at slaughter and (b) 24 h post mortem.
Mean I 450−510 of subcutaneous caudal fat 24 h post mortem was significantly different between P and S lambs (P < 0.001), averaging −236.88 and −155.39 units for P and S lambs, respectively. It ranged from −381.60 to −51.12 units for P lambs and from −289.23 to −36.47 units for S lambs. The frequency distribution of P and S lambs in the different classes of I 450−510 values measured on subcutaneous caudal fat 24 h post mortem is reported in Figure 6b. The proportion of P lambs with a mean I 450−510 greater than or equal to −196 units was 25.2%, whereas the proportion of S lambs with a mean I 450−510 lower than −196 units was 17.1%. This threshold therefore allowed for the correct classification of 79.2% of the lambs when the measurement was made on subcutaneous caudal fat 24 h post mortem.
Reflectance spectrum at wavelengths between 400 and 700 nm (method 2)
Method 2 utilised all the reflectance spectrum optical data for wavelengths between 400 and 700 nm.
More than 90% of the total variance of the reflectance data for all 307 lambs was explained by the three first principal components (PC) axes regardless of site or time of measurement. The score plots of the first two PC axes and of the first two factors of the PLS-DA model for perirenal fat measured at 24 h post mortem are shown in Figures 7 and 8, respectively. PC1 and PC2 axes explained 52% and 27% of variability, respectively. The PCA loadings of the two PC axes for perirenal fat measured at 24 h post mortem are shown in Figure 9; the highest loading was situated around 460 to 480 nm for PC1. The PLS-DA loadings (data not shown) were similar to those observed in the PC analysis.
Figure 7 Score plot of the PC1 and PC2 axes for perirenal fat samples measured 24 h post mortem according to feeding treatment (black and white symbols refer to pasture-fed and stall-fed lambs, respectively).
Figure 8 Score plot of the two first partial least square (PLS) factors for perirenal fat samples measured 24 h post mortem according to feeding treatment (black and white symbols refer to pasture-fed and stall-fed lambs, respectively).
Figure 9 Loadings for the first two principal components axes for perirenal fat samples measured 24 h post mortem.
The statistical results generated by PLS-DA models are shown in Table 5. The PLS-DA analysis allowed for the correct classification of 79.7% and 95.1% of the P and S lambs, respectively (i.e. 87.4% on average), when the measurement was made on perirenal fat at slaughter, and 90.2% and 97.6% of the P and S lambs, respectively (i.e. 93.9% on average), when the measurement was made on perirenal fat 24 h post mortem. This method correctly classified 94.4% and 91.5% of the P and S lambs, respectively (i.e. 92.9% on average), when the measurement was made on subcutaneous fat at slaughter, and 88.1% and 93.9% of the P and S lambs, respectively (i.e. 91.0% on average), when the measurement was made on subcutaneous fat 24 h post mortem.
Table 5 Statistical results from the PLS-DA models for classifying perirenal and caudal fat samples measured at slaughter or 24 h post mortem according to feeding treatment†
†Abbreviations are: PLS-DA = partial least squares discriminant analysis; SECV = standard error of cross-validation; R 2 = coefficient of determination in cross-validation; n = number of factors used in the PLS-DA calibration model; % CCS = proportion of correctly classified samples.
Comparison between both reflectance methods applied on fat
Method 2, which used all the reflectance spectrum data at wavelengths between 400 to 700, showed a higher performance (P < 0.001) than method 1, which only used the reflectance spectrum data at wavelengths ranging from 450 to 510 nm to calculate I 450−510. Taking both sites, both measurement times and both feeding treatments together, the overall proportion of correctly classified lambs was actually 77.6% for method 1 and 91.3% for method 2.
The proportion of correctly classified lambs also differed between feeding treatments (80.0% and 88.9% for P and S lambs respectively, P < 0.001), measurement sites (87.1% and 81.8% for perirenal and subcutaneous caudal fat, respectively, P < 0.001) and measurement times (80.9% and 88.0% for measurements made at slaughter and 24 h post mortem, respectively, P < 0.001). There were significant interactions between experimental factors. Figures 10 and 11 show the results obtained for P and S lambs using both methods applied on perirenal and subcutaneous fat at slaughter and 24 h post mortem. For P lambs, using method 2 yielded a higher proportion of correctly classified lambs compared with method 1 (from 7.7 to 35.7 points, P < 0.05 to 0.001). For S lambs, method 2 yielded a higher proportion of correctly classified lambs compared with method 1 (from 8.5 to 21.3 points, P < 0.001), except for the measurements made at 24 h post mortem on perirenal fat, where the results of the two methods were not significantly different (93.9% and 97.6% for methods 1 and 2 respectively, i.e. P>0.05). The greatest difference between methods occurred for the measurement made at slaughter on the subcutaneous caudal fat of P lambs, where the proportion of correctly classified lambs was only 58.7% with method 1, compared with 94.4% with method 2.
Figure 10 Proportion of correctly classified lambs using method 1 (integral value I 450−510, black symbols) and method 2 (full reflectance spectrum data sets at wavelengths between 400 and 700 nm, white symbols) on perirenal and subcutaneous caudal fat measured at slaughter. Perirenal Pasture, caudal Pasture, perirenal Stall and caudal Stall refer to measurements made on perirenal fat from pasture-fed lambs, on subcutaneous caudal fat from pasture-fed lambs, on perirenal fat from stall-fed lambs and on subcutaneous caudal fat from stall-fed lambs, respectively. *, **, ***: the proportion was significantly different between methods at P < 0.05, P < 0.01 and P < 0.001, respectively.
Figure 11 Proportion of correctly classified lambs using method 1 (integral value I 450−510, black symbols) and method 2 (full reflectance spectrum data sets at wavelengths between 400 and 700 nm, white symbols) on perirenal and subcutaneous caudal fat measured 24 h post mortem. Perirenal Pasture, caudal Pasture, perirenal Stall and caudal Stall refer to measurement made on perirenal fat from pasture-fed lambs, on subcutaneous caudal fat from pasture-fed lambs, on perirenal fat from stall-fed lambs and on subcutaneous caudal fat from stall-fed lambs, respectively. *, **, ***: the proportion was significantly different between methods at P < 0.05, P < 0.01 and P < 0.001, respectively.
Discussion
Plasma carotenoid concentration
This study confirmed that carotenoid pigments can be used to discriminate pasture-fed from stall-fed lambs, which is in line with previous studies (Prache et al., Reference Prache, Priolo and Grolier2003a). PCCS allowed for the correct discrimination of feeding treatments for 90.7% of the plasma samples. The quality of the discrimination was lower than that reported in Prache et al. (Reference Prache, Priolo and Grolier2003a), where there was no overlapping in the frequency distribution of the PCCS of pasture-fed lambs and stall-fed lambs, although the mean values were similar between both studies (75 and 10 μg/l for P and S lambs, respectively, in this study, compared with 63 and 12 μg/l for P and S lambs, respectively, in Prache et al., Reference Prache, Priolo and Grolier2003a). The discrimination quality differences between studies are probably due to differences in between-sample variability, with PCCS ranging from 27 to 194 μg/l for P lambs and from 0 to 43 μg/l for S lambs in this study, compared with 37 to 105 μg/l for P lambs and 6 to 21 μg/l for S lambs in Prache et al. (Reference Prache, Priolo and Grolier2003a). Possible explanations for this variability are variability in carotenoid intake level and between-animal variability in carotenoid absorption and metabolism. Carotenoid intake level probably varies with carotenoid concentration in the herbage, which may have varied widely in this study, since blood samples were collected from the beginning of August to the beginning of November, compared with from the end of July to the end of August in Prache et al. (Reference Prache, Priolo and Grolier2003a). However, we still observed a broad variability among animals at any given date, despite the fact that the lambs were grazing the same paddock. This may have been due to inter-individual variations in forage intake levels and in carotenoid absorption and metabolism (Rock, Reference Rock1997). On heterogeneous swards, variations can also be due to inter-individual variability in dietary choices (Prache and Damasceno, Reference Prache and Damasceno2006).
Reflectance spectrum at wavelengths between 450 and 510 nm
Perirenal fat
Discrimination between feeding treatments using perirenal fat I 450−510 was more reliable when measurements were made 24 h post mortem (88.6%) than at slaughter (79.5%). This is in disagreement with the results of Priolo et al. (Reference Priolo, Prache, Micol and Agabriel2002), who reported lower discrimination reliability 24 h post mortem than at slaughter when using the same method. This discrepancy may be due to the fact that we cut a flat surface on the fat before the measurements to ensure a perfect adherence of the eye of the apparatus, whereas this was not done in Priolo et al. (Reference Priolo, Prache, Micol and Agabriel2002).
Subcutaneous caudal fat
Discrimination between feeding treatments using subcutaneous caudal fat I 450−510 was more reliable when measurements were made 24 h post mortem (79.2%) than at slaughter (64.8%). These results are in agreement with Priolo et al. (Reference Priolo, Prache, Micol and Agabriel2002), who reported higher discrimination reliability 24 h post mortem than at slaughter.
The results from the present study therefore confirm that perirenal fat is better than subcutaneous caudal fat for discriminating pasture-fed from stall-fed lambs based on the quantification of light absorption by carotenoid pigments at wavelengths ranging from 450 to 510 nm of the reflectance spectrum. The greater reliability after 24 h of cooling is likely to be due to an increase in fat-based carotenoid concentration due to water evaporation. The quality of the discrimination was, however, lower in this study than in the previous studies by Priolo et al. (Reference Priolo, Prache, Micol and Agabriel2002) and Prache et al. (Reference Prache, Priolo and Grolier2003b). This may be due to variations across breeds (Limousine breed used in this study whereas Priolo et al., Reference Priolo, Prache, Micol and Agabriel2002 and Prache et al., Reference Prache, Priolo and Grolier2003b used the Ile de France breed), as animal tissues may store different amounts of carotenoid pigments depending on the breed (Nozière et al., Reference Nozière, Graulet, Lucas, Martin, Grolier and Doreau2006). It may also be due to a higher variability in the carotenoid concentration of the herbage and (or) to a higher between-animal variability, given the large number of animals used in the present study (143 P and 164 S Limousine lambs) compared with the previous studies with Ile de France lambs (24 P and 24 S Ile de France lambs, after pooling the data from Priolo et al., Reference Priolo, Prache, Micol and Agabriel2002 and Prache et al., Reference Prache, Priolo and Grolier2003b together). This large between-animal variability highlights the essential role of the validation procedures on large number of animals. Given that the tendency to accumulate carotenoids has a genetic component (Sheath et al., Reference Sheath, Coulon and Young2001), the range of variation of the plasma carotenoid concentration and of I 450−510 in P and S lambs probably varies across breeds. The reliability of the corresponding methods for discriminating P and S lambs in different breeds therefore requires further experimental evaluation. Consequently, we have launched a validation procedure with a large cohort of animals of different breeds soundly to assess variability within and across breeds.
Reflectance spectrum at wavelengths between 400 and 700 nm
The main result of this study is that taking all the reflectance spectral data within the wavelengths range 400 to 700 nm into account improves the reliability of the discrimination based on the fat reflectance spectrum. Performing discriminant analysis on all the optical data generated at wavelengths between 400 and 700 nm made it possible to correctly classify 87.4% of the perirenal fat samples and 92.9% of the subcutaneous fat samples when measurements were made at slaughter, and 93.9% of the perirenal fat samples and 91.0% of the subcutaneous fat samples when measurements were made 24 h post mortem. The improvement in discrimination reliability was therefore greater for subcutaneous than for perirenal fat, and for measurements made at slaughter compared with those made 24 h post mortem. The underlying mechanisms involved remain an open question. The differences in the shape of the reflectance spectrum between P and S lambs at wavelengths between 450 and 510 nm, and the loading values of the PC1 axis in this area clearly underline the importance of carotenoid pigments in discriminating between P and S lambs. Beyond carotenoid pigments, the visible region is also characteristic for the respiratory pigments, whose absorption bands in the fat reflectance spectrum (Soret band at wavelengths 415 to 435 nm, absorption bands at wavelengths 540 to 580 nm) are probably linked to residual haemoglobin after slaughter (Swatland, Reference Swatland1989; Prache et al., Reference Prache, Aurousseau, Thériez and Renerre1990; Irie, Reference Irie2001). In haemoglobin derivates, the absorbance bands are located at wavelengths 430 and 555 nm for deoxyhaemoglobin and 418, 540 to 542 and 576 to 578 nm for oxyhaemoglobin, the loss of oxygen being associated with a shift in the Soret band to a slightly higher wavelength and the loss of the bicuspid shape of the secondary peak around 555 nm (Irie, Reference Irie2001). The PC1 and PC2 axis loading values at wavelengths around 410 to 420, 440 and 560 nm may indicate that haem pigments could also be involved in the discrimination between P and S lambs.
Conclusions
Research on diet authentication in herbivore products is conducted within a general context of increasing consumer concern regarding the mode of animal production. This study reports methodological developments in the use of visible reflectance spectroscopy for discriminating between carcasses from pasture-fed or concentrate-fed lambs. The main result is that the reliability of the discrimination between pasture-fed and stall-fed lambs based on the fat reflectance spectrum was generally higher when using the full set of optical data at wavelengths between 400 and 700 nm than when only quantifying light absorption at wavelengths ranging from 450 to 510 nm, as previously proposed in the literature. The proportion of lambs correctly classified via multivariate analysis factoring in the full set of optical data at wavelengths between 400 and 700 nm was 87.4% and 92.9% for measurements made on perirenal and caudal fat at slaughter, and 93.9% and 91.0% for measurements made on perirenal and caudal fat at 24 h post mortem. This methodological development is of practical interest because the measurement can still be made quickly to keep in pace with processing lines in commercial slaughterhouses. This study also confirmed that plasma carotenoid concentration can be used to discriminate pasture-fed from stall-fed lambs, as it enabled the correct classification of 90.7% of lambs. Future research will be targeted at evaluating these methods in different feeding conditions in interaction with animal characteristics.
Acknowledgements
We thank J. Ballet, R. Jailler, J. Pourrat, and all the staff of the INRA, UE1153 Unité Expérimentale des Mont Dore and the Abattoir staff for their cooperation on the study. We also thank P. Gasqui for valuable advice on data analysis. P.H.M. Dian thanks the Brazilian Ministry of the Education ‘Coordenação de Aperfeiçoamento de Pessoal de Nível Superior’ (CAPES-PDEE) for financial assistance given under a doctoral grant.
