Effect of Dietary Supplementation with Lipids of Different Unsaturation Degree on Feed Efficiency and Milk Fatty Acid Profile in Dairy Sheep

Simple Summary The use of fats derived from palm is becoming very common in dairy sheep farms to increase the energy concentration of the diet and therefore the milk production. However, these fats may negatively affect the nutritional quality of milk, whereas feeding unsaturated oils may improve milk fatty acid profile. In this regard, our results in dairy sheep suggested that using palm fat had no evident disadvantage in terms of milk fatty acid composition compared with a diet without supplementation. Nevertheless, it had no positive effects on production or indicators of feed efficiency (for example, milk yield per unit of feed consumed). By contrast, supplementation with oils rich in unsaturated fatty acids (specifically olive oil and soybean oil) improved milk fatty acid profile, with stronger effects with the use of the most unsaturated fat: soybean oil. For example, the latter oil induced the greatest increases in fatty acids with potentially positive effects on human health (e.g., conjugated linoleic acid). In addition, from a practical point of view, the use of soybean oil might also be recommendable to improve the amount of milk produced per unit of feed consumed, compared with the use of palm fat. Abstract Lipids of different unsaturation degree were added to dairy ewe diet to test the hypothesis that unsaturated oils would modulate milk fatty acid (FA) profile without impairing or even improving feed efficiency. To this aim, we examined milk FA profile and efficiency metrics (feed conversion ratio (FCR), energy conversion ratio (ECR), residual feed intake (RFI), and residual energy intake (REI)) in 40 lactating ewes fed a diet with no lipid supplementation (Control) or supplemented with 3 fats rich in saturated, monounsaturated and polyunsaturated FA (i.e., purified palmitic acid (PA), olive oil (OO), and soybean oil (SBO)). Compared with PA, addition of OO decreased milk medium-chain saturated FA and improved the concentration of potentially health-promoting FA, such as cis-9 18:1, trans-11 18:1, cis-9 trans-11 CLA, and 4:0, with no impact on feed efficiency metrics. Nevertheless, FA analysis and decreases in FCR and ECR suggested that SBO supplementation would be a better nutritional strategy to further improve milk FA profile and feed efficiency in dairy ewes. The paradox of differences observed depending on the metric used to estimate feed efficiency (i.e., the lack of variation in RFI and REI vs. changes in FCR and ECR) does not allow solid conclusions to be drawn in this regard.


Introduction
In intensive dairy sheep production, feeding systems have moved away from pasturebased to high-concentrate diets, which may affect the nutritional value of milk fat, decreasing the concentration of potentially health-promoting fatty acids (FA), such as cis-9 trans-11 conjugated linoleic acid (CLA), trans-11 18:1, or 18:3n-3 [1][2][3]. In these production systems, diet supplementation with lipids is also widespread to increase the energy density

Animals and Management
Forty lactating Assaf ewes were housed in individual tie stalls and fed a total mixed ration (TMR) formulated from dehydrated alfalfa (particle size > 4 cm) and a concentrate (50:50 forage: concentrate ratio). The TMR contained molasses (4% of diet fresh matter) to hinder selection of dietary components. Clean water was always available and fresh diets were offered daily ad libitum after morning milking. Animals were milked twice daily at approximately 08:30 and 18:30 h in a single-side milking parlor with 10 stalls (DeLaval, Madrid, Spain).
After adaptation of the ewes to the TMR (for 1 month) and to the individual tie stalls (for 1 week), feed intake, body weight (BW), and dairy performance were examined over three weeks (pre-experimental period). Then, the 40 sheep were distributed into 4 groups (10 ewes/group) balanced (mean ± SE) for dry matter intake (DMI; 3.70 ± 0.08 kg/day), milk yield (2.59 ± 0.10 kg/day), milk fat and protein concentration (55.0 ± 0.8 and 49.8 ± 0.5 g/kg raw milk, respectively), BW (74.7 ± 1.4 kg), and days in milk (DIM; 61.6 ± 0.7). Groups were randomly allocated to 4 dietary treatments consisting of the basal TMR without lipid supplementation (Control) or supplemented with 2% dry matter (DM) of palm distillate FA (purified commercial product containing 98% of palmitic acid; PA treatment), 2% DM of olive oil (OO treatment) or 2% DM of soybean oil (SBO treatment). These dietary treatments were fed over 4 additional weeks (experimental period). This level of oil supplementation was selected based on their potential modulatory effects on milk FA profile [1,8] and to be practical in terms of cost.
The ingredients and chemical composition of the diets are given in Table 1. Representative samples of the 4 experimental diets were collected weekly during the pre-experimental and experimental periods (i.e., 7 samples of the basal diet and 4 samples of the supplemented diets). Samples were stored at -30 • C, freeze-dried, and again stored frozen to prevent alterations in fatty acid profile before chemical analysis.

Animal Performance and Feed Efficiency Indicators
To estimate the individual feed efficiency at the pre-experimental and experimental periods, animal performance was monitored over the whole experiment. The BW of each sheep was recorded once weekly.
The DMI was calculated by weighing the amounts of feed offered and refused by each animal. Then, the net energy content of experimental diets (NE D ) was estimated using the INRA [25] tables of nutritive values of feeds and employed to calculate the net energy intake (NEI = DMI × NE D ), which is expressed as MJ of net energy/day. Total milk produced by each ewe at morning and evening milkings was collected and weighed to calculate milk yield. Composite samples of the daily milk produced by each sheep were prepared according to individual yields in morning and evening milkings twice per week (and three times on the last week of each period). One aliquot of that composite milk was preserved with bronopol (D&F Control Systems Inc., San Ramon, CA, USA) and stored at 4 • C until analysis for fat, protein, lactose, and total solid concentrations (within 24-72 h after collection).
On each period, milk yield and milk composition data were used to estimate energycorrected milk [ECM = kg/d of milk yield × [(0.0071 × g/kg of milk fat) + (0.0043 × g/kg of milk protein) + 0.2224], and net energy requirements for lactation (NE L = 0.686 × ECM, and expressed as MJ of net energy/day), according to INRA [25] equations for sheep. Requirements of protein digestible in the small intestine (PDI) were also estimated according to INRA [25].
The feed conversion ratio (FCR) was calculated as the relationship between mean DMI and ECM on each period, whereas the energy conversion ratio (ECR) was obtained as the relationship between the mean NEI and NE L [26].
Residual feed intake (RFI) on each period was estimated as the residuals of the following regression model [27] using the GLM procedure of the SAS software package (version 9.4; SAS Institute Inc., Cary, NC, USA): where DMI represents the mean dry matter intake over the period (kg/day); µ is the intercept; ECM is the energy-corrected milk (kg/day); MBW is the mean metabolic body weight (BW 0.75 ; kg); BWC is body weight change over the period (kg); DIM are days in milk; RFI is the residuals; and a, b, c, and d are the regression coefficients.
The same procedure was used to estimate the residual energy intake (REI) as the residuals of the following regression model: where NEI represents the mean net energy intake over the period (MJ/day); µ is the intercept; ECM is the energy-corrected milk (kg/day); MBW is the mean metabolic body weight (BW 0.75 ; kg); BWC is body weight change over the period (kg); DIM are days in milk; REI is the residuals; and a, b, c, and d are the regression coefficients.

Milk FA Composition
On the last week of each period, aliquots of composite milk from each ewe were collected on 3 consecutive days and stored without preservative at −30 • C until fat extraction for FA composition analysis.
The fatty acid methyl esters (FAME) of lipid in freeze-dried TMR samples were prepared in a 1-step extraction-transesterification procedure [28], adding 1 mg of cis-12 13:1 (10-1301-9, Larodan Fine Chemicals AB, Solna, Sweden) as an internal standard. The methyl esters were separated and quantified using a gas chromatograph (Agilent 7890A GC System, Santa Clara, CA, USA) equipped with a flame ionization detector and a 100 m fused silica capillary column (0.25 mm i.d., 0.2 µm film thickness; CP-SIL 88, Varian Ibérica S.A., Madrid, Spain), and hydrogen as fuel and carrier gas (207 kPa, 2.1 mL/min). Total FAME profile in a 2 µL sample volume at a split ratio of 1:50 was determined using a temperature gradient program [28]: following sample injection, column temperature was maintained at 70 • C for 4 min, increased at a rate of 8 • C/min to 110 • C, raised to 170 • C at a rate of 5 • C/min, held at 170 • C for 10 min, increased at 4 • C/min to a final temperature of 240 • C that was maintained for 14.5 min. Peaks were identified based on retention time comparisons with commercially available standards (GLC463, Nu-Chek Prep, Elysian, MN, USA; 18919-1AMP Supelco, Sigma-Aldrich, Madrid, Spain).
Lipids in 1 mL of milk were extracted and converted to FAME by base-catalyzed transesterification [28]. Total FAME profile was determined using the same chromatograph and temperature gradient program applied for the analysis of feed, but isomers of 18:1 were further resolved in a separate analysis under isothermal conditions at 170 • C [26].  [29,30], and with chromatograms reported in the literature [28].

Statical Analysis
Statistical analysis was performed using the MIXED procedure of SAS software package (version 9.4).
Data were analyzed by one-way analysis of covariance with a model that included the fixed effect of the 4 experimental treatments (Control, PA, OO and SBO) and measurements on the pre-experimental period as a covariate, as follows: where y ijk is the dependent variable measured at time k (experimental period) on the jth animal assigned to the ith diet, µ the overall mean effect, α i the ith fixed diet effect, d j(i) the random effect of the jth animal within the ith diet, b the common regression coefficient of initial value of x ij , ϕ j the slope deviation of the ith diet from common slope b, x ij the initial record measure (pre-experimental period) of the jth animal on the ith diet, and e ijk the random error associated with the jth animal assigned to the ith diet at time k.
Means were separated through the pairwise differences (pdiff) option of the least squares means (lsmeans) statement of the MIXED procedure and adjusted for multiple comparisons using Bonferroni's method. Differences were declared significant at p < 0.05 and considered a trend toward significance at 0.05 ≤ p < 0.10. Least squares means are reported.

Animal Performance and Feed Efficiency Indicators
As shown in Table 2, diet supplementation with lipids affected FCR, with a 12% decrease in SBO treatment compared with the Control and PA (p = 0.012). Similarly, ECR tended to be 11% lower in SBO than in PA treatment (p = 0.052). On the contrary, residual traits (RFI and REI) were not significantly modified by the inclusion of lipids in the TMR (p > 0.10). a-c Within a row, different superscripts indicate differences (p < 0.05) due to the effect of diet. 1 SED = standard error of the difference. 2 In the pairwise analysis, no significant differences were found after adjustment for multiple comparisons using Bonferroni's method.
Feed intake tended to be affected by diet (p = 0.066), but no differences or trends toward difference were observed in pairwise comparisons after adjustment using Bonferroni's method. Body weight and yields of milk, ECM, protein, lactose and total solids remained unaffected by treatment (p > 0.10). However, compared with the Control, milk fat yield was 12% greater in OO (p = 0.039), with an increase in the molar production of <C16 and >C16 FA (p < 0.01). Supplementation with SBO also improved milk > C16 FA yield compared with the Control (p < 0.001), whereas the production of C16 FA was greater in PA than in other treatments (p < 0.001). In addition, milk fat content was 10 and 13% higher in OO and SBO, respectively, compared with the Control (p = 0.001), but no significant effects were observed in the concentration of milk protein, lactose, and total solids.
Protein balance was positive in the four experimental treatments: ewes consumed on average 127 ± 3% of their estimated PDI requirements. Table 3 reports the content of milk short-and medium-chain FA, which were differently affected by lipid supplementation. Specifically, ewes on PA treatment showed the greatest proportions of 16:0 and cis-9 16:1 in milk (p < 0.001), but 12:0, 14:0, and cis-7 14:1 concentrations were lower than in the Control (p < 0.001). Reductions in these mediumchain FAs were greater in OO and SBO treatments, which showed the lowest content of most FAs with 10 to 16 carbon atoms, such as 10:0, cis-9 12:1, and 16:0 (p < 0.05), except for the increase in trans-9 16:1 in SBO relative to other diets (p < 0.001) and the lack of variation in cis-9 10:1 and trans-5 to -8 16:1 (p > 0.10). Compared with the Control, the milk concentration of 4:0 was increased in OO and SBO (p < 0.001), and that of 6:0 in SBO (p = 0.001). On average, lipid supplements caused an 11% decrease in the sum of saturated C4-C14 FA (i.e., those mostly derived from mammary de novo synthesis) relative to the Control (p < 0.001). Table 3. Milk short-and medium-chain fatty acids (g/100 g of total fatty acids) in dairy ewes fed a total mixed ration without lipid supplementation (Control) or supplemented with 2% dry matter (DM) of palm distillate fatty acids (PA), olive oil (OO), and soybean oil (SBO). a-c Within a row, different superscripts indicate differences (p < 0.05) due to the effect of diet. 1 SED = standard error of the difference. 2 In the pairwise analysis, no significant differences were found after adjustment for multiple comparisons using Bonferroni's method.

Diet
Finally, the ratio between saturated C4-C14 FA and cis-9 18:1 was 34% lower in OO and SBO than in Control and PA treatments (p < 0.001).

Discussion
In this study, lipids of different unsaturation degree were added to dairy ewe diet to test the hypothesis that unsaturated oils would modulate milk FA profile without impairing or even improving feed efficiency. To this aim, we examined the responses to 3 vegetable fats rich in saturated, monounsaturated, and polyunsaturated FA (i.e., 16:0, cis-9 18:1, and 18:2n-6, respectively). Although their main effects on milk FA profile have been previously described [8,9,15], we report a comprehensive FA composition because available profiles in the literature are often poorly detailed, especially in terms of minor C18, odd-, branched-, and very long-chain FA. Although their biological effects are largely unknown [31][32][33], a lack of detail in presentation of results may limit the future advancement of knowledge or the potential application of FA as noninvasive biomarkers [22,34,35].
The use of 16:0-rich supplements, widely spread in cattle production, is increasingly common in dairy sheep farms under intensive conditions [5,15]. These fats are very effective at improving the energy density of the ration without negatively affecting nutrient digestibility [11,36], but their effects on milk FA profile might offer some drawbacks [7,16,36]. In our study, we observed an increment in the milk concentration of 16:0 with PA, consistent with expectations [7,15]. Although increasing 16:0 consumption might pose a greater risk of cardiovascular disease for human consumers [37,38], such effect might be counteracted by the inversely proportional impact of PA on milk 14:0 and 12:0, which have also been reported to be atherogenic [39]. In addition, PA caused virtually no alteration in the concentration of other bioactive FA in milk, either potentially negative (e.g., trans-9 and trans-10 18:1) or positive (e.g., cis-9 trans-11 CLA and trans-11 18:1), in agreement with its potential inertness in the rumen and lower toxicity for microbiota than unsaturated FA [36,40,41]. Thus, our results would support that using palmitic-rich products in dairy sheep feeding has no evident disadvantage in terms of milk fat quality. Nevertheless, it does not appear to offer any advantage in terms of efficiency of feed utilization, according to the lack of variation in the studied metrics compared with the control, both in ratio traits (i.e., FCR and ECR) and in residual traits (i.e., RFI and REI).
Similarly, OO treatment had neither positive nor negative consequences on feed efficiency indicators, despite improvements in milk fat concentration and yield. We used olive oil as a model of fat rich in monounsaturated FA (specifically, cis-9 18:1), due to its easy and ready availability in most intensive dairy sheep production areas (in particular, in the Mediterranean basin) and its close FA profile to that of other lipid supplements widely studied in ruminant nutrition (e.g., rapeseed oil) [42][43][44]. Regarding the impact of OO on milk fat composition, it is worth highlighting some desirable effects, such as the decrease in medium-chain saturated FA and the increase in some potentially healthpromoting compounds, specifically 4:0, trans-11 18:1, cis-9 trans-11 CLA, and cis-9 18:1 [1,39]. The large variation in the latter would derive not only from dietary cis-9 18:1 supply, but also from its extensive saturation in the rumen [45,46], enhancing the availability of 18:0 for mammary ∆ 9 -desaturation [47]. Ruminal cis-9 18:1 metabolism also involves isomerization and hydration/oxidation processes [45,46], which would partly explain the increments in milk trans 18:1 and 10-oxo-18:0, respectively. In addition, changes in 18:1 isomers may also derive from a greater biohydrogenation extent of 18:2n-6 and 18:3n-3, as suggested by the drop in their milk concentration. This effect on biohydrogenation extent has been consistently described in studies on ruminal metabolism when unsaturated FA supplements are provided [46,48]. On the contrary, certain effects of OO on milk FA profile were less desirable, in particular the increase in trans-9 and trans-10 18:1 or the decrease in branchedchain FA, which would be explained by direct isomerization of cis-9 18:1 in the rumen or inhibition of microbial de novo FA synthesis, respectively [35,40,45].
Among the treatments studied, SBO showed the best potential to modulate milk FA profile. Compared with OO, it induced even greater improvements in cis-9 trans-11 CLA and trans-11 18:1 concentrations, with similar variations in medium-chain saturates and other potentially bioactive FA (e.g., 4:0 and trans-10 18:1). Moreover, SBO improved 18:2n-6 and had no negative effect on 18:3n-3. Although this may increase the n-6/n-3 FA ratio in milk, the implications of this index for human health are under debate, and focusing attention on improving the consumption of both types of polyunsaturated FA is increasingly encouraged [49,50]. On the other hand, SBO was the only treatment that raised milk trans-10 cis-12 CLA content, but its final proportion was actually marginal (0.006% of total FA) and, therefore, no MFD was induced. A recent meta-analysis has indeed shown that much higher trans-10 cis-12 CLA concentrations may be reached (~0.031% of total FA) without risk of MFD in sheep fed high-concentrate diets and plant oils, given their ability to compensate the inhibition of de novo FA synthesis by enhanced preformed FA yield [51].
In addition, the reduction in FCR with SBO suggests an improvement in the efficiency of feed utilization compared with the Control and PA treatments. In this regard, the comparison between SBO and PA is particularly interesting, as they are isoenergetic diets. This would explain the consistency in the SBO vs. PA comparison when the ECR was employed, an indicator that is estimated using the net energy intake, whereas the FCR is based on DM intake [18,26]. Thus, ECR seems more convenient in our study because it avoids the bias associated to the different energy density of our experimental diets [26]. However, when residual traits (RFI and REI) were examined, no variation was detected and responses to supplemented treatments did not follow a similar pattern to that observed with ratio traits.
Residual traits are currently more recommended and used as indicators of feed efficiency in genetic selection [19,22,27]; their interest deriving from their potential relationship with basic metabolic processes [52,53]. In Australia, steers from low-RFI selection lines have been shown to consume less feed for the same level of growth performance and, thus, improve the profitability of farms [24]. Nevertheless, from a productive point of view and with the perspective of a direct application in the dairy sector, decreased FCR and ECR would also entail economic advantages for farmers, thus the potentially positive implications of our findings. Furthermore, animal performance data suggest that the lower FCR and ECR in SBO would partly be explained by increased FA yield, which supports a key role of lipid metabolism in underlying feed efficiency mechanisms [22,54]. Further, note that our results did not seem to be explained by mobilization of body reserves, since all treatments showed improved body weight during the trial.
Finally, regarding a validation of previously suggested biomarkers of feed efficiency in dairy ewes (e.g., saturated C4-C14 FA, saturated C4-C14 fatty acids/cis-9 18:1 ratio or C20-22 n-6 polyunsaturated FA in milk) [22], no solid conclusions can be drawn. The reason is none other than the divergent effects of experimental diets on the milk concentration of these biomarkers. Thus, for example, OO and SBO treatments caused both increases and decreases in individual even-chain saturated C4-C14 FA, which may bias their total amount in milk. In addition, the improvement in cis-9 18:1 concentration in the same treatments would be explained by the additional dietary supply of this monounsaturated FA and the greater mammary availability of its precursor, 18:0, rather than a greater mobilization of adipose tissue (rich in cis-9 18:1) in animals under negative energy balance [55,56]. Therefore, treatment differences in saturated C4-C14 fatty acids/cis-9 18:1 ratio cannot actually be related to potential variations in feed efficiency when animals fed different lipid supplements are compared [22]. In any event, the results of this experimental trial would not undermine the application of suggested biomarkers in dairy sheep farms, where all lactating ewes would be offered the same diet. Otherwise, discriminating animals by feed efficiency level should be conducted independently within each dietary condition.

Conclusions
Overall, our results support the initial hypothesis that unsaturated lipid supplements modulate milk FA profile in dairy sheep without impairing or even improving feed efficiency. Compared with a saturated fat rich in 16:0 (palm distillate FA), addition of a source of monounsaturated FA (olive oil) decreases medium-chain saturated FA in milk and improves the concentration of potentially health promoting FA, such as cis-9 18:1, trans-11 18:1, cis-9 trans-11 CLA, and 4:0, with no impact on feed efficiency indicators. Nevertheless, results of FA analysis and decreases in FCR and ECR suggest that using soybean oil supplementation would be a more convenient nutritional strategy to achieve further improvements in milk FA profile and also in feed efficiency in dairy ewes. However, the paradox of differences observed depending on the metric used to estimate feed efficiency (i.e., the lack of variation in residual traits-RFI and REI-vs. changes in ratio traits-FCR and ECR) does not allow solid conclusions to be drawn in this regard.

Conflicts of Interest:
The authors declare no conflict of interest.