Materials and methodsThe study was

Materials and methodsThe study was conducted in December 2012, in the facilities of the UMR1213 Research Unit and the UE1354 ExperimentalUnit of INRA Auvergne-Rhône-Alpes in Central France (45◦42N, 03◦30E). All the procedures were conducted in accordancewith the European Union Directive No. 609/1986 and French Guidelines for the use of experimental animals and compliantwith animal welfare and good practice (Veissier, 1999).2.1. Experimental designThis article provides the in vitro rumen fermentation patterns of pure and mixed silages for which the preservationparameters in mini-silos and chemical composition have been previously described (Copani et al., 2014). Details on plantmaterial, silage preparation and chemical analysis of silage samples were provided in this previous study. Briefly, timothy(T, Phleum pratense, cv. Liglory), sainfoin (SF, cv. Perly) and red clover (RC, cv. Mervius) were sown in plots divided into threesubplots, which served as replicates. Six treatments were tested in triplicate (n = 18) as follows: pure T was the grass speciesused as control without bioactive compound; pure SF containing CT (35.6 g/kg DM of fresh plant) and pure RC with PPOactivity (OD420/min = 4.0 for fresh material using 4-methylcatechol as phenolic substrate) were used as bioactive legumesspecies. Mini-silos containing binary grass-bioactive legume mixtures T–SF and T–RC were prepared by mixing T and SF orRC in equal proportions on a DM basis. The ternary mixture T–SF–RC contained 500 g/kg of T, 250 g/kg of SF and 250 g/kgof RC. After stabilisation and storage of silages during 145 days, the mini-silos were opened and samples were taken. Thesesamples were then stored at −20◦C and freeze-dried for the in vitro trial. All the samples (n = 18) were incubated in batchcultures with buffered rumen fluid three times by conducting one 24 h-run of fermentation per week over three weeks.2.2. In vitro batch fermentationRumen fluid was collected in equal proportion from three fistulated sheep (Texel, adult castrated males, 61.2 ± 9 kg onaverage). One month before the beginning of the trial, the animals were fed daily 1200 g of a diet composed of, per kg (asfed), cocksfoot hay (800 g), barley (60 g), beet pulp (50 g), corn (34 g), soya bean meal (28 g), wheat bran (12 g), molasses andminerals (16 g). The daily diet was divided into two equal meals given at 08:00 h and 16:00 h. Animals had free access towater and salt block (Sel’pur, Salins, Paris, France). For each day of incubation, a sample of rumen content was withdrawnfrom the rumen canula prior to the first meal in the morning. The rumen contents from the three sheep were pooled inequal proportions into a closed container, transferred to the laboratory and squeezed through a polyester monofilamentfabric (mesh opening 800 m) in order to obtain the rumen fluid used as inoculum for the in vitro incubations. The timebetween rumen content withdrawal and inoculation of the fermenters did not exceed 25 min. Fermentation was conductedas described by Niderkorn et al. (2012). In brief, 600 ± 0.5 mg of freeze dried silage material was placed in 120 ml serumbottles, pre-warmed at 39◦C and flushed with N2; 40 ml of buffered rumen fluid (strained rumen fluid diluted 1:2 (v/v) in ananaerobic phosphate:carbonate buffer solution, initial pH 7.04 ± 0.01) was added, then the bottles were sealed hermeticallywith butyl rubber stopper and aluminium crimp seals. Blanks without any plant substrate (only buffered rumen fluid) wereincubated during the different runs. Samples of buffered rumen fluid were also taken at time 0 to determine net productionof VFA and NH3-N. All the bottles were incubated in a shaking water bath at 39◦C. The gas production was recorded at24 h using the pressure transducer technique, as described by Theodorou et al. (1994). Gas samples were taken from theheadspace of the serum bottles for determination of gas composition (CH4, CO2and H2). After 24 h, the fermentation wasstopped and the content of each bottle was treated and centrifuged as described by Niderkorn et al. (2012) to determinepH, VFA and NH3-N in the medium. The apparent degradation of DM was determined by difference between DM of plantmaterial before the fermentation and DM of residue after 24 h of fermentation.2.3. Laboratory analysisThe composition of fermentation gases were determined by gas chromatography using a Micro-GC (CP2003, Chrompack,The Netherlands) as described by Macheboeuf et al. (2008). The NH3-N concentration in the supernatant was determined asdescribed by Weatherburn (1967). The total concentration and profiles of VFA in the supernatant were determined by gas