The experiment was carried out at the National Institute of Agricultural Technology (INTA) in Balcarce (37˚45'S, 58˚18'W) during September and October of 2013. Total duration of the experiment was 38 days. Procedures and animal cares were approved by the Institutional Committee for the Care and Use of Experimental Animals (CICUAE, INTA CERBAS). Thirty-six multiparous Holstein cows (548 ± 56 Kg LW) in early lactation (77 ± 43 days postpartum) were grouped based on parity and milk production measured during the first 7 days of the experiment and randomly assigned to 1 of 6 treatments (6 cows/treat- ments) in a complete randomized design. The basal (Control) diet was composed (DM) by pasture (56%) and concentrate (44%) without supplementary oils. From day 8^th of the trial, six cows remained in the Control diet while the cows in oil treatments were supplemented with SO (Glycine max), LO (Linum usitatitissimum), or their blends (%w/w) at 75 - 25 (SO75 - LO25), 50 - 50 (SO50 - LO50) and 25 - 75 (SO25 - LO75). The dose of supplemented materials was calculated to provide 4.0% of the total DM intake of cows [3] . Pure oils or blends (0.8 kg/cow.day) were manually-mixed to the concentrate during each milking time and thoroughly consumed by cows. Adaptation to oils proceeded gradually starting with 0.2 kg/cow.day over the first day, 0.3 kg during the following 2 days and 0.8 kg from day 4 until the end of the experiment including 28 days of full-dose oil supply. Cows were weighed on 2 consecutive days after the a.m. milking at the start (day 8^th) and at the end of the period of lipid supplementation. Animals grazed together on mixed pastures of fescue (Festuca arundinacea), red clover (Trifolium pratense), white clover (Trifolium repens) and bromegrass (Bromus unioloides) in a daily-strip grazing system. The area of the strip was regulated using a temporary electric fence to provide an herbage allowance of 27 kg DM/cow.day. After grazing, each strip was clipped-out of non-grazed forage to about 6 cm to allow a lean and uniform regrowth. The concentrate (16% CP) was composed by ground corn grain (35%), malt brewery waste (10%), pelletized sunflower meal (20%), soybean (10%), wheatgrass (21.48%), calcium carbonate (2%), magnesium oxide (0.4%), salt (1%), rumensin (0.02%), and a vitamin-mineral mix (0.1%). It was offered at a rate of 8 kg/cow.day in two equal feedings during milking times (06.00 and 16.00).

Milk production was daily recorded over the whole experiment. Milk samples (50 ml) were collected at a.m. and p.m. milkings twice a week on non-consecu- tive days, composited according to the corresponding volume measured at each milking time and analyzed for fat, total protein, lactose, total and not-fat solids by mid-infrared spectrophotometry (Milko Scan-Minor, Foss Electric, Hillerod, Denmark). Milk urea nitrogen (MUN) was determined using a commercial enzymatic kit (Wiener Lab., Rosario, Argentina). During the last 2 weeks of oil supplementation and from each composite sample collected to determine the chemical composition of milk, aliquots of 50 ml were frozen (−24˚C) to obtain a single pool sample per cow for the determination of milk FA composition by gas liquid chromatography (GLC) as described in [16] . Cows were weighed on two consecutive days after the morning milking at the start (days 6 and 7) and the end (days 38 and 39) of the experiment and the mean value of the 2 records was used to calculate changes in body weight (BW) gain.

The quality of the concentrate and herbage was estimated from samples taken weekly. Each sample was dried in a forced-air oven (60˚C, 48 hs), ground through a 1-mm screen (Willey mill, Philadelphia, PA) and analyzed for organic matter (OM), neutral detergent fiber (NDF) [17] , acid detergent fiber (ADF) [18] , crude protein (CP) [19] using an autoanalyzer (LECO FP-528, Leco Corp., Saint Joseph, MI, EE.UU.), water-soluble carbohydrate (WSC) [20] , ether extract (EE) [21] using an autoanalyzer (ANKOM Corp., Fairtport, NY, EE.UU.), In vitro DM digestibility (IVDMD) was estimated using the Ankom Tech. Daisy II incubator for 48 h and starch as described in [22] . Pasture DM intake was individually estimated during the last 3 days of weeks 4 and 5 of the trial by the difference method [23] . The average DM intake of the three consecutive days from each cow was computed for the statistical analysis.

Milk production and composition were evaluated by the PROC MIXED procedure of SAS/STAT® program [24] using the following model:

Y i j k = μ + T i + A ( i ) j + W j + C o v + ( T i ∗ W j ) + E ( i j k )

where: Y_ijk = the dependent variable, μ: overall mean, Cov = covariate (milk yield and composition over the first 7 days), T_i = treatment effects, A_(i)j = random effects of animal within treatments, W_j = effects of week, (T_i * W_j) = interaction effects between of treatment and sampling week, E_(ijk) = the residual error associated with the ijk observation. Data from milk FA composition, DM intake and changes in BW gain were analyzed by the PROC GLM procedure of the SAS/STAT® (2002-2010) program using the following model:

Y i j = μ + T i + A ( i ) j + E ( i j )

where: Y_ij = the dependent variable, μ: overall mean, T_i = treatment effects, A_(i)j = random effects of animal within treatments, E_(ij) = the residual error associated with the ij observation.

The daily strip-grazing system allowed to provide 29 kg DM/cow.day considered adequate to maximize pasture intake [25] . In grazing conditions, pasture intake should be maximal when herbage is offered at a rate of 45 g pasture OM/kg BW [25] . From the average BW of cows (563 kg) and the average OM content of pastures (90 g OM/100g DM, Table 1) it can be calculated that a non-limitant herbage allowance should be around 22.8 kg DM which resulted lower to that obtained in the present experiment. Maximal DM intake should be obtained when pasture allowance was 45 to 55 g DM/kg BW per day [26] . The average BW of cows (563 kg, Table 4) suggests an optimal range in pasture allowance of 25 to 31 kg DM. Thus, the herbage allowance obtained (29 kg DM/cow.day) was within the optimal range. Pasture DM content (21.85%) was over the critical range of 15% - 18% proposed to decrease voluntary intake [27] . In turn, NDF (46.23%) and CP (15.1%) contents were in the range of 40% - 50% (NDF) and 15% - 25% (CP) considered as adequate for well managed pastures [25] . In our experiment, pasture quality and quantity were sufficiently enough to maintain or increase total DM and energy intake of cows (Table 5).

SO represented a good source of oleic (21.55%) and essentially linoleic (53.55%) acids as reported in [28] [29] . In LO, linolenic acid content (41.9%, Table 2) resulted lower to the 55% value reported by others [7] [28] [30] but near to values informed in [31] [32] . Linolenic acid in pasture was nevertheless higher than reported by [33] and [30] .

In [34] , supplementing SO or LO alone or in combination at 4% of DM intake increased milk yield (16.7%) compared to Control without differences between oils. In our trial, the average increase in oil-supplemented cows over Control was somewhat moderate (9.4%) and mainly explained by both oil blends at a ratio of 75:25 (Table 3). Since milk production at SO100 and LO100 did not differ from Control, a synergic effect on milk output of both 75:25 combinations can be expected. Comparison between oil blends did not reveal a specific effect on milk production. A high frequency of favorable effects on milk production after the inclusion of unprotected vegetable oils in the diet was reported by [35] . The lack of differences between SO100 and LO100 respect to Control (Table 3) was also observed in the meta-analysis by [28] suggesting the absence of any net advantage of one or another oil over milk production. Feeding LO at 3% or 4% DM intake increased milk production in [34] a result not observed in other trials [7] [33] . Supplementary SO at 2.9% ± 1.2% of DMI (533 ± 228 g/day) did not affect milk production in the experiments reviewed by [28] or when SO was fed at 3.5% to 5% of DM intake [36] [37] [38] . In addition, LO supply (1% to 7% of DM intake) did not affect milk production in [28] [30] [33] [36] . In our experiment, the higher yield of FCM from cows in SO75 - LO25 was explained by the higher volume of milk produced since milk fat content did not change (Table 3). These results suggest that energy excreted in milk was the same across treatments as reported by [34] . Unsaturated lipid supply generally has neutral effects on yield of FCM both in non-grazing [39] as in grazing experiments [40] . In a wide dose-range of lipid supplementation (0.2 to 1.0 kg/day) it has been observed that unsaturated lipids decrease milk fat content and fat yield by 8% in grazing dairy cows [40] . The lowest milk fat content observed in the 50 - 50 treatment (Table 3) was consistent with the higher levels of trans-10 C18:1 (Table 6) because both parameters were negative correlated (Figure 2). A direct relationship between increasing milk levels of trans-10 C18:1 and the reduction of de novo mammary synthesis has been previously reported [41] a fact that contributes to explain the lower milk fat content observed.

The lack of negative effects of oil supply on milk protein content (Table 3) is a relevant result as this parameter not only affects milk price but also determines the speed and quality of coagulation in cheese production. In confined production systems, supplementation with unprotected lipids often decrease milk protein content [35] [39] [42] while in pasture based diets this parameter is often not affected [40] [43] . Feeding LO does not appear to affect either milk fat [28] [34] nor milk protein contents or production [7] [33] [34] .

The lack of differences in BW gain (Table 4) was consistent with [30] [44] [45] [46] and the similar plasma NEFA concentrations (data not shown). In fact, supplementation with unsaturated lipids does not appear to reduce BW loss in lactating cows or favor the reconstitution of body reserves in lactating cows [47] .

In our trial, the effect of supplemental fat on DM intake showed different responses (Table 5) depending on the specific oil blend consumed. Pasture intake slightly increased in SO100 and SO75 while it was decreased in others treatments (SO50 - LO50; SO25 - LO75 and AL100) while total DM intake resulted higher in the SO100 and SO75 (Table 5). The inclusion of LO (3.2% ± 1.7% DM intake) or SO (2.9% ± 1.2% DM intake) in the ration did not affect DM intake of cows in the meta-analysis by [28] nor in [33] [34] . Feeding unsaturated FA is more likely to reduce feed intake than saturated FA owing to their potential negative effects on ruminal digestion. However the results are variable including negative [48] , neutral [49] or even positive [50] [51] effects on rumen function. The forage concentrate ratio (F/C) is relevant because when LO was included at 3% of DM intake in a 65/35 F/C diet, positive effects were reported on FDN digestion with an opposite result when the F/C was 35:65 [51] . In the present trial, the F/C averaged 54:46 (Table 5).

The increase in mammary uptake of circulating FA after oil supply [52] may explain the changes in milk FA composition compared to Control treatment (Table 6) confirming ruminant milk fat plasticity [3] [28] . The consistency in content of butyric acid after oil intake (Table 6) is a frequently reported result [3] [28] which is of interest for its potential beneficial role in human health [3] . Butyric acid can be synthesized by an independent malonyl-CoA pathway and therefore not dependent on the activity of the acetyl CoA carboxylase that is inhibited by the uptake of the exogenous FA supplied by oils [3] [6] .

The decrease in the total content of de novo synthesized FA (C4:0 to C15:1) was similar between the pure oils and their mixtures (Table 6) as reported in [34] . The effect is explained by the inhibition in the activity of lipogenic mammary enzymes such as Acetyl-CoA carboxylase [53] [54] . Antonacci et al. [55] also reported a reduction (−17.8%) in the total de novo synthesized FA content (from 22.49 to 18.48 g/100g FA) after feeding 0.7 kg of an SO70-LO30 blend to grazing dairy cows. In our study, the decrease in milk fat content (Table 3) was negatively correlated to trans-10 C18:1 content (Figure 2) in agreement with [41] . A high content of trans-10 C18: 1 or related metabolites like trans-10, cis-12 C18:2 in milk has been associated with dysfunctions in lipoprotein lipase (LPL) and stearoyl CoA desaturase (SCD) enzymes involved in milk fat uptake (LPL) and synthesis explaining the decrease in the fatty content of milk [56] . In our study, the reduction in milk fat content (Table 3) occurred in part at the expense of the amount of hypercholesterolemic FA (Table 6) which improves the healthy value of milk and contributes to decrease the atherogenic potential of milk fat. In grazing dairy cows, supplementation with 0.7 kg/cow/day of an SO70 - LO30 mixture reduced the atherogenicity index of milk from 1.6 to 1.25 [55] .

The reduction (33%) in myristic acid content (Table 6) is an important result because the pro-atherogenic role of C14:0 is considered to be very potent [1] . The reductions of 35% for C12:0 and 24% for C16:0 (Table 6) were comparable to those obtained in [55] and contribute to avoid an excessive consumption of unhealthy saturated fat. In the experiment of [34] , the reductions of these three FA after supplementation at 4% DM intake with a SO50 - LO50 mixture or pure oils did not differ between treatments. In the present work, the reductions were within the range estimated from the meta-analysis performed by [28] for supplements with SO and LO with values of 42% - 37% (lauric), 23% - 24% (myristic) and 30% - 17% (palmitic).

Milk content of stearic acid increased only when LO was present at 75% and 100% of the blend without differences in treatments with a higher proportion of SO (Table 6). The results were consistent with that reported by [34] and could be linked to some possible inhibition of biohydrogenation from VA to C18:0 when high contents of linoleic acid are available [9] [10] . The effect of the oil blends on the content of C18:0 in milk was inconsistent, which agrees with other experiments that reported the lack of differences in milk stearic content in dairy cows supplemented with oils rich in C18:2n − 6 or C18:3n − 3 [55] [57] .

In the meta-analysis by [28] all polyunsaturated FA supplements generate similar increases in the content of stearic and oleic acids in milk. In our trial, content of oleic acid numerically increased with oil supply but differed from Control only in LO100 (+7%). The increase in oleic acid after the addition of sunflower or soybean oils to the diet is a well-documented result [7] [28] [58] also observed when supplementing with LO [28] [34] [59] [60] . Oleic acid content in milk did not increase after the intake of an SO70 - LO30 mixture at 0.7 kg/cow/day in grazing dairy cows [55] .

Linoleic acid content in milk from Control cows (2.96 g/100g FA, Table 6) was within the range (2% - 3%) suggested by [3] . In oil-enriched diets, linoleic acid content increased up to 2.74 - 3.50 g/100 g FA (Table 6) remaining below the 4% as reported by [3] and observed in [55] (3.25 to 3.92 g/100g FA) after supplying 0.7 kg/cow/day of an oil blend (SO70 - LO30).

The levels of RA achieved in treatments with pure oils (2.96 to 3.03 g/100g FA) were higher than those of 1.60 - 2.39 g/100g FA reported in [34] when rations with a high forage content (59%) were supplemented with oils at 4% of DM intake. These authors [34] obtained greater increases for both VA and RA using SO compared to LO with additive responses of the 50:50 blend but always lower to oils utilized in their pure form suggesting no synergistic effects. A higher and more complete ruminal biohydrogenation of PUFAs in animals that consumed LO would explain the response obtained [34] . In our trial, milk RA content in oil supplemented cows (2.33 to 3.24 g/100g) were higher than values reported in the meta-analysis by [28] when cows were suplemented with SO (1.02 ± 0.36 g/100g FA) or LO (1.75 ± 0.84 g/100g FA) and also to those obtained by [61] supplementing with 500 g/day of sunflower oil or SO (2.02 gRA/100g FA) to grazing dairy cows. They were also higher than observed in [62] using 0.9 kg/day of FA calcium salts (0.9 kg/cow/day ) containing 30% linoleic acid but close to those reported in [55] (3.13 g/100 g for AR) using the mixture SO70:LO30 (0.7 kg/cow.day) in grazing dairy cows.