In recent decades consumers have become increasingly interested in foods containing compounds that provide health benefits. Thus, in these last years a significant research effort has been realised to modify the composition of milk and cheese fat in order to increase the concentration of compounds with positive effects in human health through the diet of the lactating animals (Jenkins & McGuire, Reference Jenkins and McGuire2006; Shingfield et al. Reference Shingfield, Bonnet and Scollan2013). In all these studies primary attention has been given to the amount of certain individual fatty acids (FA), particularly to conjugated linoleic acid (CLA) isomers, which may have positive biological activity in cellular and animal studies (Chilliard & Ferlay, Reference Chilliard and Ferlay2004; Parodi, Reference Parodi2005; Bauman & Lock, Reference Bauman, Lock, Fox and McSweeney2006). Most of the studies reported in the scientific literature used experimental diets based on concentrates supplemented with vegetable oils (soya, sunflower, olive; Gómez-Cortés et al. Reference Gómez-Cortés, Frutos, Mantecón, Juárez, de la Fuente and Hervás2008; Castro et al. Reference Castro, Manso, Jimeno, del Alamo and Mantecón2009), oil seeds (flax, sunflower; Luna et al. Reference Luna, Bach, Juárez and de la Fuente2008), or even fish oils (Kitessa et al. Reference Kitessa, Peake, Bencini and Williams2003) or algae (Reynolds et al. Reference Reynolds, Cannon and Loerch2006). Other researchers use feeds based on different forage:concentrate ratios, with or without oil supplements, whereas others take advantage of pastures, with or without additional concentrate and/or supplements (Sanz Sampelayo et al. Reference Sanz Sampelayo, Chilliard, Schmidely and Boza2007; Ferlay et al. Reference Ferlay, Agabriel, Sibra, Journal, Martin and Chilliard2008; Abilleira et al. Reference Abilleira, Virto, Nájera, Salmerón, Albisu, Pérez-Elortondo, Ruiz de Gordoa, de Renobales and Barron2010).
In addition, the cost of raw materials for feed production has steadily increased in the last decade. Several by-products have been used as feed components, although few studies have analysed the characteristics of the resulting milk and cheese (Chiofalo et al. Reference Chiofalo, Liotta, Zumbo and Chiofalo2004; Cabiddu et al. Reference Cabiddu, Addis, Pinna, Decandia, Sitzia, Piredda, Pirisi and Molle2006; Abbeddou et al. Reference Abbeddou, Rischkowsky, Hilali, Hess and Kreuzer2011).
In the Basque Country (Northern Spain) most of the ewe's milk produced is used for Idiazabal cheesemaking. Latxa dairy sheep production systems are based on part-time grazing during the spring milking period, with pasture feeding being supplemented with forage and concentrate to meet milk production requirements. We have explored the use of oilcakes from locally-produced sunflower and rapeseed crops in the formulation of concentrates. The goal of the present study is to assess the influence of different amounts of oilseed cake (rapeseed and sunflower) on the animal production parameters, and the milk FA concentrations, both during the winter indoor feeding of dairy ewes and in spring in combination with part-time grazing.
Materials and methods
Animals, feeds and experimental setup
Two experiments were conducted with the experimental Latxa dairy sheep flock of NEIKER-Tecnalia, in accordance with the European Council Directive 86/609/ECC for the protection of animals used for experimental and other scientific purposes. Two different levels of rapeseed or sunflower oilcakes were used to prepare the concentrates fed for the animals: 50% (Experiment I) or 30% (Experiment II). Both experiments were carried out for a total of 38 d, allowing the first 10 d as adaptation time for the sheep to get accustomed to their new feed. Thus, data were collected during 4 weeks. Experiment I was carried out from the end of February to the end of March with indoor feeding (winter), whereas Experiment II was carried out from mid April to the end of May in which time animals were taken out to graze ad libitum during 4 h after the morning milking.
Two weeks before the initiation of each Experiment, sheep were separated into three homogeneous groups of 12 multiparous sheep each according to the month of lactation, daily milk yield, live weight (LW), and body condition score (BCS). Groups of animals were randomly assigned to one of the concentrates (control, rapeseed or sunflower) described in Table 1. Animals in each group were not the same in both Experiments, in an attempt to partially simulate conditions in commercial flocks in which the group of animals in milk production varies throughout the season. Soya and cereals were the main protein and energy sources in the commercial concentrate used in control groups (with or without grazing), whereas oilcakes were the main protein and energy sources in the experimental concentrates (rapeseed or sunflower). Soya was replaced with different amounts of either rapeseed or sunflower oilcake (30 or 50% by weight of the total concentrate) in the experimental concentrates. Oilcakes were the by-product of the cold extraction of oil from locally produced rapeseed and sunflower seeds.
Table 1. Gross nutritional characteristics of the oilcakes, concentrates and forages

† Unités Fourrage Lait
‡ Per cent oilcake in the concentrates: Experiment I, 50%; Experiment II, 30%
Animals in Experiment I (50% oilcakes, indoor feeding) received 840 g dry matter (DM)/d concentrate per ewe, whereas those in Experiment II (30% oilcakes, outdoor grazing) received 650 g DM/d concentrate per ewe. Diets were adjusted to supply the necessary nutritional requirements for the animals at the different lactation stages and indoor feeding or grazing, when the experiments were carried out. In all cases, concentrate was divided into two halves and provided during the morning and evening milkings. In addition, Festuca hay was offered ad libitum twice a day in Experiment I (after each milking) and only once a day in Experiment II (after the afternoon milking).
Daily intake of indoor given feed (concentrates and hay) was estimated for each group as the difference between amounts offered and amounts refused. Groups receiving 30% oilcakes were free to graze during 4 h after the morning milking on an existing polyphite pasture (composed mainly of Lolium perenne, Dactylis glomerata, and Trifolium repens). Stocking rate during Experiment II was 12 animals per 0·2 Ha plot for a total of 4 h. Actual grazing time was visually monitored and pasture intake (PI) was estimated by comparing the faecal recovery of external C32 n-alkane with that of grass-contained C33 n-alkane, essentially as described by Mayes et al. (Reference Mayes, Lamb and Colgrove1986). Pellets with C32 n-alkane were administered orally to six ewes/group during the morning milking for 10 d (during weeks three and four of Experiment II). Individual faecal samples were collected for the last 3 d (during morning and evening milkings), all faeces were pooled and faecal n-alkanes were determined by gas chromatography.
During each experiment, animals were milked twice daily. The average daily milk production per sheep (ADMP) and per week was determined by dividing the total daily milk amount of each group by the number of animals. Sheep were weighed once a week and BCS was monitored simultaneously as described by Russel et al. (Reference Russel, Doney and Gunn1969).
Milk sampling and analytical procedures
Duplicate milk samples (1·5 L) for analysis (crude protein, crude fat and FA) were taken from each entire group (bulk tank milk) once a week, but not during the ten adaptation days. Crude protein (CP), crude fat (CF) and DM were determined using the method PE/ALVO/02 (2005) (de Renobales et al. Reference de Renobales, Amores, Arranz, Virto, Barrón, Bustamante, Ruiz de Gordoa, Nájera, Valdivielso, Abilleira, Beltrán de Heredia, Pérez-Elortondo, Ruiz, Albisu and Mandaluniz2012). The cream was removed by centrifugation at 2000 g and 4 °C for 30 min in a Sorvall refrigerated high-speed centrifuge, and both the cream and the skim milk were kept at −80 °C until analysed.
For FA analysis, milk fat of bulk milk was extracted from 1 g cream by a modified Folch method (Folch et al. Reference Folch, Lees and Stanley1957), as described by Virto et al. (Reference Virto, Bustamante, Ruiz de Gordoa, Amores, Fernández-Caballero, Mandaluniz, Arranz, Nájera, Albisu, Pérez-Elortondo, Barron and de Renobales2012). Extracted fat was dissolved in toluene and glycerides were trans-esterified to the corresponding fatty acids methyl esters (FAME) with a solution of sodium methoxide in methanol as described by Christie (Reference Christie1982). FAME were separated by gas chromatography (Agilent 7890A, Palo Alto, CA) and identified as described by de Renobales et al. (Reference de Renobales, Amores, Arranz, Virto, Barrón, Bustamante, Ruiz de Gordoa, Nájera, Valdivielso, Abilleira, Beltrán de Heredia, Pérez-Elortondo, Ruiz, Albisu and Mandaluniz2012). Quantification was done using n-pentanoic, n-nonanoic, n-heptadecenoic (cis-10) and n-nonadecanoic acids (Sigma-Aldrich, Madrid, Spain) as internal standards, added to the sample at the time of extraction. Results were expressed as μmol FA/g fat.
Statistical analysis
IBM SPSS statistical package version 20.0 (IBM SPSS Inc., Chicago, Il) was used for data treatment. ANOVA was done separately for Experiments I and II to determine the presence of significant differences (P⩽0·05) in production parameters and milk FA composition of the groups of animals. General lineal model was used including ‘feeding’ and ‘week’ as fixed effects. Means for the different feeding groups were compared using Tukey's test. Mean FA composition for rapeseed and sunflower cakes were compared using Student's t-test. Stepwise discriminant analysis using milk FA composition was applied to all groups of animals from both Experiments I and II (six ‘feeding’ groups). Wilk's lambda criterion was used to select discriminant variables, and milk sample distribution for the ‘feeding’ groups was plotted in two-dimensional graphs defined by canonical discriminant functions.
Results and discussion
The FA composition of the oilcakes is described in Table 2. Rapeseed oilcake had significantly (P⩽0·05) higher amounts of oleic (C18:1cis-9; 7·2-fold higher), linolenic (C18:3cis-9, cis-12, cis-15; 70-fold higher), and C18:1trans-9+C18:1cis-11 (12·5-fold higher) acids than sunflower oilcake, which had significantly higher concentration of linoleic acid (C18:2cis-9, cis-12; 1·9-fold higher). Although their individual FA concentrations were statistically different (P⩽0·05), both oilseed cakes had the same per cent of total unsaturated FA (Table 2).
Table 2. Mean and sd of the fatty acid (FA) content (μmol/g) of rapeseed and sunflower oilseed cakes (n=4)

a,bDifferent superscripts in the same row indicate statistically significant (P⩽0·05) differences
None of the groups refused any amount of concentrate or hay during the experiments; average hay intake per ewe was 1500–1600 g DM/d in Experiment I and 1200–1300 g DM/D in Experiment II. Pasture intake per ewe in Experiment II ranged between 560 and 650 g DM/d, which represented approximately 25% of the total daily feed intake (Table 3). According to these results, forage-to-concentrate ratios were 66/34 in Experiment I and 76/24 in Experiment II.
Table 3. Effect of the type and concentration of oilseed cake on average daily milk production per sheep (ADMP), crude fat (CF) and crude protein (CP) content and CF/CP ratio of the milk, and live weight (LW), body condition score (BCS), grazing time (GT) and pasture intake (PI) of the animals. Values presented are the mean and sd of data collected during 4 weeks

a,bDifferent superscripts for each Experiment in columns indicate statistically significant (P⩽0·05) differences
To ascertain that 10 d for adaptation to oilcake diets were sufficient, the FA composition of the milk samples taken during the first week were compared with those of the last week, showing no statistically significant differences (data not shown).
Taking into account preliminary results of our research group on oilseed cakes (Barron et al. Reference Barron, Amores, Abilleira, Virto, Arranz, Nájera, Beltrán de Heredia, Ruiz de Gordoa, Ruiz, Albisu, Pérez-Elortondo, Mandaluniz and de Renobales2011), it was decided to use 50% oilcakes for the indoor period when no pasture was available (Experiment I), and to reduce the amount of oilcakes to 30% because animals were allowed to graze outdoors (Experiment II). Table 3 summarises the values of production parameters for both Experiments. The observed differences between Experiment I and Experiment II in some production parameters such as ADMP, CF and LW are most likely due to the different stage of lactation, with Experiment I occurring at the peak of milk production for this flock. In Experiment I, addition of 50% of either oilcake did not modify ADMP, CP, LW, or BCS. Whereas rapeseed cake did not alter milk CP or the CF/CP ratio, sunflower cake did cause a significant (P⩽0·05) reduction in milk CF and the CF/CP ratio with respect to the control value. This decrease in milk fat was probably induced by the higher amount of linoleic acid in the sunflower oilcake cake (Table 2). The milk fat depression syndrome has been associated with diets rich in vegetable or fish oils containing high concentrations of linoleic acid (Griinari & Bauman, Reference Griinari, Bauman, Sjersen, Hvelplund and Nielsen2006). The relative amounts of CF and CP are among the milk parameters which most affect cheese yield (Abd El-Gawad & Ahmed, Reference Abd El-Gawad and Ahmed2011). Yet, the observed low value of 1·14 for the CF/CP ratio obtained with 50% sunflower oilcake is within the range of values of the CF/CP ratio for several commercial Latxa sheep flocks. Depending on the lactation stage, CF/CP ratios for Latxa sheep flocks have been observed to vary from 1·06 (Virto et al. Reference Virto, Bustamante, Ruiz de Gordoa, Amores, Fernández-Caballero, Mandaluniz, Arranz, Nájera, Albisu, Pérez-Elortondo, Barron and de Renobales2012) to 1·45 (Abilleira et al. Reference Abilleira, Virto, Nájera, Salmerón, Albisu, Pérez-Elortondo, Ruiz de Gordoa, de Renobales and Barron2010). Currently in the Basque Country, cooperatives making cheese under the Idiazabal DO buy milk with CF/CP ratios higher than 1·05 (Latxa Esnea Kooperatiba and Esnetik Kooperatiba, personal communication, 2014). Thus, the milk obtained with 50% sunflower cake would be marketable for cheesemaking in spite of its lower CF/CP ratio. Although in the work described herein coagulation parameters were not determined, good rennet coagulation times for milks with CF/CP values up to 1·19 were reported by Abilleira et al. (Reference Abilleira, Virto, Nájera, Salmerón, Albisu, Pérez-Elortondo, Ruiz de Gordoa, de Renobales and Barron2010). In Awassi ewes fed Mediterranean food industry by-products, milk with CF/CP ratios between 1·08 and 1·18 gave cheese yields between 0·30 and 0·34 kg cheese/kg milk (Abbeddou et al. Reference Abbeddou, Rischkowsky, Hilali, Hess and Kreuzer2011). Taken together, these values indicate that although 50% sunflower oilcake did reduce the total fat content of the milk, the technological value of this milk for cheesemaking may not have been compromised. Research is currently being carried out to ascertain this point.
The FA concentrations in the milk of animals fed 50% oilcake (Experiment I) are shown in Table 4. Both rapeseed and sunflower oilcakes caused an average reduction of 35% (10–71%) in the concentrations of all FA with chain lengths below 17 carbon atoms (mostly saturated) with respect to the corresponding values of the control group. In contrast, the concentrations of all C18 FA (mostly mono- and polyunsaturated) in the oilcake groups increased between 6 and 203% with respect to those of the control group. The concentrations of the CLA isomer C18:2cis-9, trans-11 and vaccenic acid (C18:1trans-11) were 150 and 163% higher, respectively, in the milk of animals fed sunflower oilcake, whereas in the rapeseed oilcake group the concentrations of these FA increased only by 30 and 68%, respectively, when compared with the control group. The increases observed in the sunflower oilcake group for these FA are most likely due to the significantly (P⩽0·05) higher concentration of linoleic acid in the sunflower oilcake (Table 2) which is one of the main substrates for biohydrogenation reactions in the rumen (Luna et al. Reference Luna, Bach, Juárez and de la Fuente2008; Castro et al. Reference Castro, Manso, Jimeno, del Alamo and Mantecón2009). With both oilcakes, the atherogenicity index (Ulbricht & Southgate, Reference Ulbricht and Southgate1991) of the milk decreased by an average value of 50% with respect to that of the control milk. These results indicate that the use of 50% rapeseed or sunflower oilcakes during the indoor feeding period yields milk with a healthier FA profile than that obtained with the habitual concentrate formulation.
Table 4. Mean and sd of the concentration of fatty acids (FA) (μmol/g fat) in milk from Experiment I (n=4). Ewes were fed control feed or feed containing 50% sunflower or rapeseed oilcake indoors

a,b,cDifferent superscripts on the same row indicate statistically significant differences (P⩽0·05)
† Trace amounts, below the level of quantification
In Experiment II the concentrations of the individual FA in the milks from animals which received either oilcake were very similar for most FA (Table 5). When compared with those of the control group, the concentrations of FA C6 to C16:1 decreased by an average amount of 14%, whereas those of the remaining FA increased modestly by an average of 7%. Exceptions were linoleic, vaccenic, and the CLA isomer C18:2cis-9, trans-11, all of which were significantly (P⩽0·05) higher in the sunflower oilcake group, most likely due to the FA composition of the sunflower oilcake (Tables 2 and 5). In both groups, the amounts of all saturated or unsaturated FA were essentially identical. The atherogenicity indexes (Ulbricht & Southgate, Reference Ulbricht and Southgate1991) of both milks were also indistinguishable, but significantly lower than that of the milk of the control group. Grazing increases the milk concentrations of unsaturated FA and decreases its atherogenicity index (de Renobales et al. Reference de Renobales, Amores, Arranz, Virto, Barrón, Bustamante, Ruiz de Gordoa, Nájera, Valdivielso, Abilleira, Beltrán de Heredia, Pérez-Elortondo, Ruiz, Albisu and Mandaluniz2012; Virto et al. Reference Virto, Bustamante, Ruiz de Gordoa, Amores, Fernández-Caballero, Mandaluniz, Arranz, Nájera, Albisu, Pérez-Elortondo, Barron and de Renobales2012), but adding rapeseed or sunflower oilcake to the feed yields milk with a still healthier FA profile.
Table 5. Mean and sd of the concentration of fatty acids (FA) (μmol/g fat) in milk from Experiment II (n=4). Ewes were fed control feed or feed containing 30% sunflower or rapeseed oilcake. All animals were allowed to graze during 4 h

a,b,cDifferent superscripts on the same row indicate statistically significant differences (P⩽0·05)
† Trace amounts, below the level of quantification
A stepwise discriminant analysis on milk FA composition was applied to the six feeding groups to determine how similar, or how different, their milk FA compositions were. The 5 discriminant functions correctly classified the milk samples into the 6 feeding groups (100% of the cases) (Fig. 1 shows the bi-plot graph of discriminant functions 2 and 3). Discriminant Function 3 which showed the highest correlations for vaccenic acid, saturated and unsaturated FA, and atherogenicity index indicated that the FA composition of milk samples from animals fed 50% oilcake-containing concentrates during the indoor feeding period (Experiment I) appeared to be closer to those from grazing animals (Experiment II). Function 3 discriminated the indoor control group from the other animal groups. Function 2, which mainly correlated the CLA isomer C18:2cis-9, trans-11, showed slight differences among grazing animal groups (Experiment II) and those with animals fed 50% oilcake-containing concentrates (Experiment I). So, using a high percent of oilcake for indoors-kept animals yields milks with FA concentrations comparable with those obtained during grazing with or without oilcake supplementation, resulting in milks with a healthier FA composition.

Fig. 1. Bi-plot graph for discriminant functions corresponding to the stepwise discriminant analysis on milk FA composition applied to all groups of animals from both Experiments I and II. ●, indoor control flock; ○, outdoor control flock; ■, indoor 50% rapeseed cake; □, outdoor 30% rapeseed cake; ▲, indoor 50% sunflower cake; △, outdoor 30% sunflower cake.
Conclusions
The results presented in this paper indicate that the inclusion of oilcakes in concentrates for dairy sheep feeding does not compromise milk production parameters, or an important cheesemaking technical parameter such as the CF/CP ratio. The use of oilcakes during the indoor feeding period, when pasture is not available, increased the concentrations of nutritionally interesting FA yielding milk with a healthier FA profile, similar to that obtained when fed fresh pasture. Thus, these oilcakes are a viable way to use byproducts of locally produced crops without compromising dairy sheep systems production.
The authors thank Miriam Uranga for expert technical assistance. Financial support was provided by the Spanish National Institute for Agricultural and Food Research and Technology (INIA, RTA2006-00100-C01 and RTA2006-00100-C02), the Fund for Cooperation Euskadi-Aquitaine (28/2010), the Department of Industry (S-PE10UN94) and the Department of Education, Language Policy and Culture (IT766-13) of the Basque Government.