Cows, experimental design and treatments

The experiment complied with the guidelines of the Danish Ministry of Justice Law No. 726 (9th September 1993) concerning experiments with animals and care of experimental animals.

Fourty-four Jersey cows (mean ± SD; 473 ± 52 kg body weight; 134 ± 108 d in milk; 29·8 ± 5·9 kg energy corrected milk; 20 primiparous and 24 multiparous) were used in the experiment. The cows were blocked according to parity (first parity and older), calving date, and milk yield and assigned to one of four different treatments in a balanced 4 × 4 Latin square design with a 2 × 2 factorial arrangement of treatments. Halfway through the experiment eight cows (one block of first parity and one older) in late lactation were exchanged with similar parity cows in early lactation, therefore only 36 cows were in experiment at a certain time. Cows were fed the experimental diets for periods of 3 weeks; 2 weeks adaptation and 1 week data collection. Cows were kept in a loose-housing system with slatted floors and cubicles with mattresses. The cows were milked in automatic milking unit (AMU) (DeLaval AB, Tumba, Sweden).

The four treatments were grass/clover silage made of spring grass/clover with or without rapeseed supplementation and grass/clover silage made of autumn grass/clover with or without rapeseed supplementation. The silages were mixtures of red (Trifolium pratense) and white clover (Trifolium repens), perennial ryegrass (Lolium perenne), and hybrid ryegrass (Lolium × boucheanum Kunth) harvested at the same field at AU-Foulum (56°N, 9°E). The spring silage was a first growth harvested May 23rd 2012 and contained 81, 9 and 10% of grass, white clover, and red clover on dry matter (DM) basis, respectively. The two autumn silages were a third and fourth regrowth and they were harvested September 8th and October 11th 2012, respectively. Grass, white clover and red clover concentrations were 77, 15 and 8% on DM basis for the third regrowth and 87, 8 and 6% on DM basis for the fourth regrowth. A mixture of equal amounts of 3rd and 4th regrowth was used in the feeding experiment to ensure sufficient amounts of silage during the experimental period. After prewilting, grass/clover was precision chopped with a forage harvester (Claas Jaguar 870, Claas KGaA mbH, Harsewinkel, Germany) and ensiled in bunker silos without any ensiling additives. The rapeseed was a double-00 variety and it was ground before inclusion in the rations.

The cows had ad libitum access to the four partial mixed rations (PMR; Table 1) which were mixed once daily in a vertical mixer wagon (JF-Stoll, Sønderborg, Denmark). The daily feed intake was automatically recorded by the Insentec RIC system (Insentec, Marknesse, The Netherlands). The cows were offered up to 3·0 kg of concentrate in the AMU daily. The actual intake (offer and left over) of AMU concentrate was registered. For more details see Bossen et al. (Reference Bossen, Weisbjerg, Munksgaard and Højsgaard2009). Cows had free access to drinking water at all time.

Table 1. Composition of partial mixed rations in g/kg dry matter for cows fed either silage made of spring or autumn grass/clover without or with rapeseed supplementation

Sampling

Representative samples of silages and concentrates were taken weekly and stored at −20 °C. Samples of silages and AMU concentrate were pooled for 6-week periods and all other ingredients were pooled for the whole experiment for chemical analysis. Milk yield was measured at every visit in the AMU using the DeLaval Free Flow meter MM25 (DeLaval AB, Tumba, Sweden) based on optical milk flow measurement. For determination of ECM yield representative milk samples were taken at each milking during a 48-h period in the last week of each period. These samples were sent directly to an external laboratory for analysis of protein, fat, and lactose concentrations. This was followed by another 24-h sampling period where samples for fatty acids, β-carotene, α-tocopherol, and riboflavin were taken. These samples were stored at at −20 °C until analysis.

Chemical analyses

Dry matter of ingredients and feed samples was determined at 60 °C for 48 h. Ash was determined by combustion at 525 °C for 6 h (AOAC, 1990). Nitrogen was determened by the Dumas principle (Hansen, Reference Hansen1989) using a Vario Max CN (Elementar Analysesysteme GmbH, Hanau, Germany). Crude protein concentration was calculated as N × 6·25. Crude fat was extraced with petroleum ether (Soxtec 2050, Foss analytical, Hillerød, Denmark) after hydrolysing with HCl (Stoldt, Reference Stoldt1952). Total sugars were analysed by the Luff-Schoorl method (European Community, 2012, 71/250/EEC). Starch was analysed by an enzymatic calorimetric technique (Knudsen et al. Reference Knudsen, Åman and Eggum1987). Neutral detergent fibre (NDF) was analysed by neutral detergent extraction using heat stable amylase and reported as ashfree according to Mertens (Reference Mertens2002) using a FibertecTM M6 system (Foss Analytical, Hillerød, Denmark). In vitro organic matter digestbility of silage followed Tilley & Terry (Reference Tilley and Terry1963). The digestibility of concentrate was determined in vitro by enzymatic digestion. For further details and prediction of in vivo organic digestibility of silages and concentrate, see Åkerlind et al. (Reference Åkerlind, Weisbjerg, Eriksson, Tøgersen, Udén, Ólafsson, Harstad, Volden and Volden2011). Net energy for lactation was calculated from Weisbjerg & Hvelplund (Reference Weisbjerg and Hvelplund1993). The concentration of acetate and lactate was analysed according to the method described by Canibe et al. (Reference Canibe, Højberg, Badsberg and Jensen2007). Ammonium nitrogen was determined by alkalisation of the sample with KOH and the NH₃ was determined by titration after distillation using a Kjeltec 2400 (Foss Analytical, Hillerød, Denmark). The concentrations of fat, protein, and lactose in milk was determined on a Milkoscan 4000 infrared analyser at Eurofins Steins (Holstebro, Denmark).

Fatty acids

Fatty acid analysis in milk was performed based on Larsen et al. (Reference Larsen, Kidmose, Kristensen, Beaumont and Mortensen2013), where fat was separated from milk by centrifugation, and fatty acids were methylated using sodium methylate. Fatty acid methyl esters (FAME) were quantified by the use of external standards (Supelco FAME mix C4–C24, Bellefonte, USA; PA and GLC 469 methyl ester standard from Nu-Chek Prep Inc. Elysian, MN, USA) and the concentrations were calculated in g/kg of identified milk fatty acids. Fatty acids from dried feed samples were analysed based on Jenkins (Reference Jenkins2010): 1 ml of heptane including internal standard (C12 : 1cis11 triglyceride, 0·4 mg/ml) was added to 0·2–0·3 g of silage or 30 mg of rapeseed or 0·1 g of the other feeds and mixed for 10 s. Then 0·2 ml of sodium-methylate (25%) was added to the solution and mixed for 10 sec and incubated at 50 °C for 10 min. After cooling on ice, 1·5 ml of methanolic hydrochloric acid (10%) was added and samples were incubated at 90 °C for 30 min. After cooling on ice, 1 ml of heptane and 3 ml of potassium carbonate (10%) were added. Samples were centrifuged and the heptane phase was used for GC analysis. Fatty acids were identified as above and based on the internal standard fatty acids in feed and were quantified and expressed as mg FA/g DM.

Analysis of β-carotene and α-tocopherol

Milk or dried feed samples were saponified prior to extraction of β-carotene or α-tocopherol for HPLC analysis (Slots et al. Reference Slots, Butler, Leifert, Kristensen, Skibsted and Nielsen2009; Larsen et al. Reference Larsen, Kidmose, Kristensen, Beaumont and Mortensen2013). External standards of β-carotene and α-tocopherol were used for quantification.

Riboflavin analysis

Milk proteins were precipitated and serum was used for HPLC analysis of riboflavin (Poulsen et al. Reference Poulsen, Rybicka, Poulsen, Larsen, Andersen and Larsen2015).

Calculations and statistical analysis

Energy corrected milk (ECM, 3·14 MJ/kg) was calculated according to Sjaunja et al. (Reference Sjaunja, Baevre, Junkkarinen, Pedersen and Setala1991).

Apparent recovery of C18 : 2n6 and C18 : 3n3 was calculated as the ratio between total amount excreted in milk and total amount ingested from feed. The degree of biohydrogenation of C18 : 2n6 and C18 : 3n3 was estimated using the following formula (where C18 : a is C18 : 2n6 or C18 : 3n3) (Larsen et al. Reference Larsen, Hymøller, Brask-Pedersen and Weisbjerg2012):

$${\rm BH(C}18:a) = 1 - \displaystyle{{{\rm (C}18:a/{\rm Total}\;{\rm C}18)\,{\rm in}\;{\rm milk}} \over {{\rm (C}18:a/{\rm Total}\;{\rm C}18)\;{\rm in}\;{\rm feed}}}$$

Data were analysed using PROC MIXED in SAS (SAS^® version 9.2, Cary, NC, USA) with the cow as experimental unit using the following model:

$${X_{ijklm}} = \mu + {\alpha _i} + {\beta _j} + {\delta _\kappa } + {\gamma _l} + {\alpha _\iota }{\beta _j} + {\beta _j}{\delta _\kappa } + {\alpha _i}{\delta _k} + {E_{ijklm}} + {e_{ijklm}}$$

Where X _ijklm was the dependent variable, μ was the overall mean, α _i was the fixed effect of grass/clover silage type i (spring, autumn), β _j was the fixed effect of the rapeseed j (−, + rapeseed), δ_k was the fixed effect of parity k (first parity or older), γ _l was the fixed effect of period l (1–4), E _ijklm was the random effect of the cow, e _ijkm was the random residual error. Degree of freedom was estimated by the Satterthwaite procedure. Results were reported as least square means and standard error of mean (SEM). Results were considered to differ significantly if the P-value of the model was less than 0·05. Pairwise comparisons of LS means for significant effects were performed using the PDIFF option adjusted with Turkey-Kramer. Four cows were considered as outliers and omitted from the analyses because either the cows were sick or cows had dubious feed intake registrations.