A number of security and nutritional roles exist in the production of beef for the human diet ( Mapiye et al. , 2015 ). While in developing countries the system of cattle production is based on use of grass pasture in feeding animals, in developed countries the cattle with a high potential for lean beef production are frequently fattened on diets of concentrate, which may be unfavourable to the nutritional values in the meat as regards human health. While the contribution of saturated (SFA) and monounsaturated fatty (MUFA) acids to cardiovascular health is debatable ( Calder, 2004 ), evidence regarding the role of polyunsaturated fatty acid (PUFA) in the prevention of heart disease is convincing. Clinical intervention studies and animal experiments indicate that a lower ratio n -6/ n -3 PUFA in dietary fat is desirable in order to reduce the risk of chronic diseases ( Simopoulos, 2001 ). In this sense, Wolfram (2003) and Nuernberg et al. (2005) reported that the n -6/ n -3 PUFA ratio could be four times higher in the meat of bulls fed on concentrates indoors than those with grass-based feeding. The influence of the feed on the fatty acid composition of the beef in calves has been widely reported ( Srednicka-Tober et al. , 2016 ). Reducing concentrate in the diet of beef cattle should enhance the n -6 fatty acid concentration, because concentrates are a source of 18:2 n -6. Of the major 18:1 isomers found in beef, vaccenic acid (isomer 11 t 18:1) is the main isomer in beef from forage-based diets and has potential human health benefits ( Dilzer & Park, 2012 ), whereas 10 t 18:1, the major t -18:1 isomer in beef from grain-based diets, seems to have detrimental effects ( Wang et al. , 2012 ).

Changes in the diet of animals can lead to changes in the fatty acid composition of the meat and may also affect sensory eating quality ( Vatansever et al. , 2000 ). In fact, consistent positive correlations between major fatty acid 9 c 18:1 (oleic) in meat and the scores given by panels for odour and flavour intensity have been reported ( Sañudo et al. , 2000 ). Beef from high-concentrate fed animals is in general fatter and contains greater total proportions of unsaturated fatty acid (mainly C18:1) than that from animals finished on pasture diets, when compared at a similar age range ( Daley et al. , 2010 ). Research to find dietary strategies that increase the ratios of favourable fatty acids for human health or improve sensorial properties of meat through the animal diet can, therefore, be a useful tool in the development of beef livestock production.

A number of discrimination procedures are used as a tool to get to know the origin of the product. For example, in the meat product industry, near infrared spectroscopy had proven to be a rapid and effective tool for determining the meat quality and the origin of the products ( Fernández Cabanás et al. , 2011 ). NIR calibrations have been developed using least squared regressions models based on the use of the fatty acid profile. Besides, the effectiveness of using methodologies based on the fatty acid profile of meat in order to differentiate production systems for young bulls ( Moreno et al. , 2006 ; Dias et al. , 2008 ), lambs ( Juárez et al. , 2010 ) or goats ( Mellado-González et al. , 2009 ) has already been reported. In fact, Martínez et al. (2013) proposed that a linear discriminant analysis using only 10 fatty acids of intramuscular fat is able to distinguish beef samples according to the type of finishing diet in bulls. The present work proposes a scientific initiative to improve the control of the production systems of beef and the system to authenticate the product origin. Therefore, the main objective of this work was to develop a methodology based on the fatty acid profile of intramuscular fat in beef from young bulls, in order to discriminate between feeding systems based on the combined use of different levels of concentrate and grass or forage. This methodology could be used to certified the beef origin in the market and provide the consumer with information about the nutritional values of this product before consumption.

A total of thirty fatty acids were detected in intramuscular fat of the longissimus lumborum according to their relative retention times (Table 3). Palmitic acid (16:0), stearic acid (18:0), oleic acid (9 c 18:1) and linoleic acid (6 c 18:2) accounted for around 75% of the total fatty acids detected in the intramuscular fat of Retinta young bulls. The corresponding individual chromatograms of intramuscular fat of young bulls raised on GP, MC and HC feeding regimes is shown in Fig. 1. Under the conditions of our assay, the first group of fatty acids of interest in intramuscular fat (10:0 to 16:1) elutes at a retention time of between 10 and 27 min, although this group also includes 16:0 that elutes at 25 min. The group of fatty acids with 18 carbon atoms elutes between 31 and 39 min. and includes the commonest fatty acid (9 c 18:1), which elutes at 32.2 min. 9 c 18:1 acid expressed as mg/100g muscle was the greatest in all different feeding regimes. The peaks of long-chain fatty acids (carbon number >20) appear after 39 min. This group includes some n -6 and n -3 PUFA which are associated with human health. The two major long-chain peaks corresponding to 20:4 n -6 (arachidonic) and 22:5 n -3 (docosapentaenoic) fatty acids elute at 44.7 and 55.4 min, respectively.

The different feeding regimes caused large differences in fatty acid profiles of intramuscular muscle fatty acids as shown in Table 3. The univariate test of equality between groups showed that 28 of the 30 fatty acids detected were affected ( p <0.05) by the feeding regime. In fact, 20:3 n -3 and 24:1 content were the only fatty acids not affected by dietary treatment ( p >0.05). Intramuscular fat of regimes based on grain feed (MC and HC) showed a higher content of 9 c 18:1, 18:0, and 6 c 18:2 than feeding regimes based on grass (GP) ( p <0.05), while values of saturated hypercholesterolemic fatty acids (lauric, 12:0; myristic, 14:0 and palmitic, 16:0) were lower in young bulls reared on grass pasture ( p <0.05) than young grain-fed bulls. However, the intramuscular fat of meat from GP showed higher polyunsaturated fatty acid content as 20:4 n -6; 20:5 n -3, 22:5 n -3 and 22:6 n -3 than the MC and HC feeding regimes.

In reference to the fat content in meat, the young bulls fed diets of concentrate (MC and HC) showed a higher fat content (1.86 and 1.58% of fresh meat for MC and HC, respectively) than meat from grazing young bulls (1.04% of fresh meat; p <0.05).

The stepwise procedure provided a final model with only two fatty acids, 9 c 18:1 and 22:5 n -3 (Table 4). The classification model provided two linear discriminant functions (LDF1 and LDF2) as follows: LDF1 accounted for 81.9% of the inter-group variance, while LDF2 accounted for 18.1%. The great ability of LDF1 and LDF2 to separate the three groups is shown in Fig. 2, where the discriminant scores are graphically depicted and each class is identified separately. The class means for each group in the canonical observation scores are presented in Table 4. The highest scores for the first discriminant function were associated with the GP system, while the lowest scores for discriminant function were associated with the MC and HC systems.

The three linear classification functions obtained to discriminate feeding regimes are shown in Table 5. At the modelling stage, the recognition ability of the discriminant model was evaluated by 100% correct classifications following the “leave-one-out” procedure, allowing the three production systems to be differentiated. The 9 c 18:1 and 22:5 n -3 fatty acid contents were selected to discriminate between groups with different feeding regimes (Fig. 3).

The fatty acid profile of intramuscular fat from Retinta young bulls in this study were in line with those reported by other authors who studied intramuscular fat from other lean young bulls of European continental breeds, such as Limousin ( Aldai et al. , 2012 ) Pirenaica ( Indurain et al. , 2006 ) or Serrana de Teruel ( Ripoll et al. , 2016 ) slaughtered at the same live weight. Saturated fatty acids 16:0 and 18:0, monounsaturated fatty acid 9 c 18:1 and polyunsaturated fatty acid 6 c 18:2 were the most abundant fatty acids in the intramuscular fat and this concurs with results reported for meat from young bulls of different origins ( i.e. Belgian Blue, Limousin, Irish, Argentine and Pirenaica) raised in traditional production systems ( Raes et al. , 2003 ; Albertí et al. , 2014 ). In fact, 9 c 18:1 and 16:0 fatty acids were the commonest fatty acids in the intramuscular fat of the young bulls (around 30 and 25% of the total of fatty acid detected, respectively), as was described by other authors in beef ( Alfaia et al. , 2009 ; Martínez et al. , 2013 ). The group of long-chain fatty acids detected in intramuscular fat of Retinta young bulls includes some n -6 (18:3 n -6; 20:2 n -6; 20:4 n -6) PUFA and n -3 (18:3 n -3; 20:3 n -3; 20:5 n -3; 22:5 n -3; 22:6 n -3) PUFA, which are regarded as being beneficial for human health ( Wood et al. , 2004 ). In fact, nutritionists recommend a higher intake of long-chain PUFA, especially n -3 PUFA rather than n -6 PUFA ( Kamihiro et al. , 2015 ) to prevent coronary diseases.

The feeding regime had a major impact on the individual fatty acids of the intramuscular fat. These results are in accordance with Alfaia et al. (2009) , who reported the influence of the feeding system in 27 of the 36 fatty acids analysed in intramuscular fat of the longissimus lumborum muscle of young bulls of the Alentejana breed. Walshe et al. (2006) and Srednicka-Tober et al. (2016) reported that greater differences in the fatty acid profiles of intramuscular fat of beef obtained from pasture- and grain-fed were mainly due to the PUFA content, while the concentration of SFA or MUFA was similar or slightly lower, respectively, in pasture compared with meat from a diet of concentrate. However, in our study, the 9 c 18:1 content in the intramuscular fat of Retinta breed young bulls fed on grain-based regimes was approximatively twice that of pasture-raised young bulls. An increase in 9 c 18:1 in meat from concentrate-feeding is expected because the content of 9 c 18:1 in cereal grains was high (Table 2). On the other hand, according to Urrutia et al. (2015) accumulation of oleic fatty acid in meat from concentrate-feeding could be linked to increased stearoyl-CoA desaturase activity in grain feeding. Stearoyl-CoA desaturase (SCD1 gene) catalyzes the main reaction to synthesize oleic acid (C18:1 c 9) from stearic acid (C18:0). SCD1 inserts the cis-9 double bond in stearic acid to make oleic acid. In fact, Hiller et al. (2011) reported an increase in SCD1 gene expression in bovine muscle after feeding bulls a grain-based diet, while Dervishi et al. (2011) and Corazzin et al. (2013) reported that SCD gene expression could be reduced by grazing.

The percentages of n -6 and n -3 fatty acids were different between the feeding regimes ( p <0.001). Some studies reported that increases in n -3 PUFA are matched by reductions in n -6 PUFA, because competition occurs between these fatty acids for the same set of elongation and desaturation enzymes ( Lorenz, 2004 ). However, Nuernberg et al. (2005) in beef muscle from German Simmental and German Holstein showed no evidence of this inverse relationship. In our experiment, evidence of this inverse relationship was not found and the highest percentages of n -6 PUFA and n -3 PUFA were observed in the GP system ( p <0.05). The percentages of long-chain fatty acids beneficial to human health n -6 (20:4 n -6) and n -3 (20:5 n -3; 22:5 n -3 and 22:6 n -3) were higher in young bulls fed with grass pasture compared to concentrate-fed animals ( p <0.05). As regards the n -6/ n -3 ratio calculated from the content of individual fatty acid n -6 and n -3, in this experiment we have calculated the ratio n -6/ n -3 from the percentage of n -6 and n -3 to total fatty acid detected and we could observe a less favourable nutritional value in meat from the highest concentrate diet (HC; 12.70) than meat from the lowest concentrate diet (6.38 and 5.71 for MC and GP, respectively). Several studies using young bulls ( French et al. , 2000 ; Razminowicz et al. , 2006 ) have demonstrated that increasing the proportion of grass in the diet and decreasing the concentrate intake caused a decrease in the n -6/ n -3 ratio. As reported by Nuernberg et al. (2005) , the availability of 18:3 fatty acid in the grass-based or reduced concentrate diet resulted in an increase in the synthesis of the C20:5 n -3, C22:5 n -3 and C22:6 n -3 fatty acid concentration in the animals' muscle.

In order to discriminate the fatty acids involved in the identification of feeding regimes, a stepwise linear discriminant analysis with individual fatty acids of the intramuscular fat was proposed, which offers both a classification rule and a way to explain the differences between groups. The stepwise linear discriminant analysis previously described in the Statistical Analysis section was applied on the set of the thirty fatty acids of intramuscular fat (Table 3). This procedure identified two fatty acids (9 c 18:1 and 22:5 n -3), and therefore a linear discriminant analysis based on these two variables was built to discriminate among the feeding regimes from Retinta young bulls (Fig. 1). The ability of 9 c 18:1 to discriminate between beef meat from animals raised under different feeding regimes has been suggested previously. In fact, the discriminant capacity of this fatty acid has been reported by Moreno et al. (2006) to separate young bulls weaned and not weaned before slaughter or to classify bulls according to the feeding system based on the use of concentrate for two or four months during the finishing period ( Alfaia et al. , 2009 ). However, there are no reported references to 22:5 n -3 regarding its discriminant capacity to differentiate between production systems based mainly on different dietary treatments. Other PUFA ( i.e. 18:3 n -3; 20:3 n -3; 22:6 n -3) have been proposed as predictors of feeding systems based on grass feeding or concentrate feed in young bulls ( Moreno et al. , 2006 ), but this is not relevant to the present study.

When pooled in a canonical structure (Table 4), they showed that both 9 c 18:1 and 22:5 n -3 were closely correlated with LDF1, the former negatively and the latter positively, whereas moderate and positive correlations where found for LDF2. The greater discriminating ability of 9 c 18:1 and 22:5 n -3 ties in with the findings of Walshe et al. (2006) in relation to young bulls feeding, given that the MUFA content is negatively related to the grazing system, while the PUFA content is positively related to that system. By contrast, the MC system based on forage intake (60% of the diet) provides higher 18:3 n -3 and its long chain derivatives (including 22:5 n -3) than the HC system. Similar results were reported by Kamihiro et al. (2015) , who showed an increase of 12% of 22:5 n -3 in meat from grass pasture-fed compared to grain-fed lean beef.

The great ability of LDF1 and LDF2 to separate the three groups suggests that the feeding regime associated with grazing or different levels of concentrated feeding can be considered as a main factor for discriminating the beef fatty acid profile. One fatty acid selected as a predictor was of primarily dietary origin (9 c 18:1) and was deposited with minimal changes in intramuscular fat, while a significant amount of 22:5 n -3 could be of microbial origin from 18:3 n -3 fatty acid ( French et al. , 2000 ). The increased concentration of 22:5 n -3 in meat from young bulls fed on grass pasture suggests that the high availability of 18:3 n -3 in the pasture diet has resulted in an enhanced synthesis of these n -3 long-chain PUFAs in intramuscular fat ( Nuernberg et al. , 2002 ).

A number of studies on the influence of diet on the fatty acid profiles of meat have concluded that beef from animals fed on pasture has greater amounts of n -3 PUFA compared to beef from animals fed a concentrate-based diet ( French et al. , 2003 ; Ponnampalam et al. , 2006 ). In fact, Turner et al. (2015) reported a greater proportion of 22:5 n -3 and other long chain PUFAs (20:5 n -3 and 22:6 n -3) in meat from pasture regime than beef from high concentrate feed. The results show that it is possible to use 22:5 n -3 to discriminate between the three feeding regimes considered in this study and, indeed, the two selected variables in the stepwise linear discriminant analysis (including 22:5 n -3 and 9 c 18:1) show a clear separation between the three production systems (Fig. 3). The GP group is located at a distance from the MC and HC groups, while the MC and HC groups are close to each other. The GP group is located at the highest point, which is indicative of the highest level of 22:5 n -3, whereas the MC and HC groups are located on the right-hand side (Fig. 3), indicating the higher level of 9 c 18:1 in concentrate regimes. The proximity between MC and HC can be observed because both production systems include a significant amount of 9 c 18:1 in the animals' diet, whereas the GP system is associated with a high percentage of 22:5 n -3 from the diet of grass. The MC group was also found to have a higher percentage of 22:5 n -3 than the HC group. Thus, evidence can be found to support the classification of origin of meat according to the 9 c 18:1 and 22:5 n -3 fatty acid content.

According to the "leave-one-out” procedure, the final model offered a success rate of 100% for all the three groups studied. Our results are in agreement with other authors who obtained 77.8-100% ( García et al. , 2008 ) or 100% ( Alfaia et al. , 2009 ) of cases correctly classified using the fatty acid profiles of intramuscular fat when different diets (pasture only, pasture plus concentrate and concentrate only) were tested. The results indicate that only two fatty acids found in meat (9 c 18:1 and 22:5 n -3) are needed to assign meat very accurately to either of these feeding regimes for young bulls.

In summary, the results obtained from this study show that the fatty acid profile of intramuscular fat could be proposed as a way of differentiating meat from young bulls according to distinctive feeding regimes based on use of grain or grass. In addition to the usual practices carried out by companies to guarantee their labelling, the levels of two fatty acids (9 c 18:1 and 22:5 n -3) in the meat could be proposed as an effective tool for authenticating the origin of the beef and contribute to the traceability of the meat production system. These results provide consumers, the medical profession and meat producers with more information on the characteristics of beef and a scientific method for authenticating this product in the market. This approach would help not only to assure consumers of the origin of the meat they purchase but also to reinforce their confidence in this type of product.

The authors wish to thank the farm staff of the Divino Salvador Cooperative (Cádiz), the Diputación de Cádiz (Cádiz) and La Orden Valdesequera (Badajoz) for their technical support.