Changes in temperature and silo gas composition during ensiling, storage and feeding-out grass silage

An experiment is reported in which two bunker silos were filled with about 12 t of ryegrass; one was compacted with a surface pressure of 4·1 kPa and the other was not. Both were instrumented to record temperatures and gas compositions during fermentation and feed-out. Cores of silage (100 mm diameter) were removed from the bunkers and subjected to forced aeration in insulated cylinders. Simple mathematical models were used to simulate CO₂ leakage from the silos and the associated loss of dry matter, and temperature changes during the forced aeration of silage cores. The leakage of CO₂ from the silos, as shown by the fall in CO₂ concentration with time, could be described empirically by an exponential equation and could be simulated with a simple mathematical model. Experimental evidence supported the hypothesis that permeation was the main method of gaseous exchange during fermentation and feed-out, but gas mixing by diffusion and/or convection also occurred within the silos during fermentation. Dry-matter losses by aerobic activity during storage of 120–150 d were conservatively estimated to be about 0·3% and 0·9% for the uncompacted and compacted bunkers respectively. These were considered to be lower than those that could be expected in farm silos, because the sealing was likely to be more thorough. The silages were also more stable, when subject to aeration, than others examined in this laboratory, probably because of high contents of acetic, propionic and butyric acids. Simple mathematical models were of value, but greater sophistication is needed (e.g. multi-compartmental models) to deal comprehensively with the heat and gas flows found in the complex biophysical systems of silage.     

Effects of treating lucerne with an inoculum of lactic acid bacteria or formic acid upon chemical changes during fermentation, and upon the nutritive value of the silage for lambs

A second cut of lucerne was wilted to 500 g DM kg⁻¹ and either left untreated (control) or treated with formic acid (4.5 1 fresh forage t⁻¹) or with a commercial inoculum of lactic acid bacteria (10⁵ colony forming units (cfu) g forage⁻¹). The forages were ensiled in 2-t capacity silos for 8 months, and later fed to six lambs (mean initial weight 27.7 ±1.60 kg) in a 3x3 duplicated Latin square with 27-d periods. Portions of the untreated and additive-treated forages were also ensiled in laboratory silos at 25 ° C for intervals up to 42 d. Results from the laboratory silos showed that the major increase in ammonia-N in silage occurred between 40 h and 7 d of fermentation; during this period, both formic acid and the inoculant produced a smaller increase in ammonia-N, than did the control. The pH of inoculated silage declined from 5.74 to 4.57 in 7 d, but it took 14 d for the pH of the control silage to fall below 5.0. Formic acid treatment immediately reduced the silage pH from 5.74 to 5.10 ( P < 0·01); the pH then remained unchanged until 21 d, after which it decreased slightly. When compared with control, lambs fed formic acid-treated silage consumed more ( P < 0·05) digestible organic matter; the response was associated with a trend towards decreased concentration of ammonia in plasma. Inoculation of lucerne silage did not ( P < 0·05) affect voluntary intake but increased ( P <0.05) apparent digestibility of fibre and tended to increase N retention.     

The microbiological and chemical composition of silage over the course of fermentation in round bales relative to that of silage made from unchopped and precision-chopped herbage in laboratory silos

The composition of baled silage frequently differs from that of comparable conventional silage. A factorial experiment was conducted with three wilting treatments (0, 24 or 48 h) × three ensiling systems [unchopped grass in bales, unchopped grass in laboratory silos (LS), precision-chopped grass in LS] × six stages of ensiling to (i) confirm that the fermentation of unchopped grass in LS could be used as an adequate model for baled silage fermentation, (ii) quantify the differences between baled silage and silage made from precision-chopped herbage across a range of dry-matter contents and (c) quantify the fermentation dynamics within the various treatments. The onset of fermentation as evidenced by the accumulation of fermentation products and the decline in pH were slower ( P  < 0·05) in baled silage compared with silage made from precision-chopped herbage. Furthermore the pH ( P  < 0·001) and overall concentration of fermentation acids ( P  < 0·01) were lower while ammonia-N concentration was generally higher in baled silage, making it more conducive to the activities of Clostridia , Enterobacteria and yeast. Numbers of Enterobacteria were higher ( P  < 0·001) in baled silage in the early stages of ensilage and persisted in baled silage at the end of the storage period. The implications of a slower onset of fermentation in baled herbage are greater in farm practice, as the fermentation would be further restricted by a more extensive wilting of the herbage prior to ensiling.     

Microscopic and chemical changes during the first 22 days in Italian ryegrass and cocksfoot silages made in laboratory silos

Although many aspects of grass silage have received intensive study, the changes that take place within the grass blades during ensilage have received little attention.
In two factorial experiments Italian ryegrass (cv. Lemtal) and cocksfoot (indigenous) were ensiled in laboratory silos (capacity 0·2 kg) with the grass under pressure (700 Pa) and subjected to two treatments (with or without 3·31 t^-1 85% formic acid) and two silos from each treatment opened on eight sampling dates(1,2,3,4,7,10,14 and 21 d).
For each sampling date transverse sections of grass laminae were examined and changes in the cross sections of the protoplasts and of cells, as defined by the cell walls, were recorded. These data were used to calculate the ratios of protoplast to cell volumes at each sampling date.
Each silage was analysed for dry matter, pH, lactic acid, VFA, ammonia and glucose. The distance settled by the silage (a measure of decrease in silage volume) and the silage conductivity were also recorded.
Both of the untreated grasses yielded silages with typical lactic acid levels and pH values. In all of the silages there was a marked shrinkage of the plast within the space defined by the cell wall. The ratio of protoplast to cell volume eventually stabilized at 0·4. The formic acid-treated grasses reached this level by day 1 but a longer period was required by untreated grasses.
It is suggested that the decrease in silage volume is related to the lactic acid content of the liquid phase due to their close relationship in silages produced from untreated grasses ( r = 0·97***).     

Evaluation studies in the development of a commercial bacterial inoculant as an additive for grass silage

Two 2×2 factorial experiments are described in which a bacterial inoculant being developed as a silage additive and containing a strain of Lacto-bacillus plantarum (Ecosyl, ICI plc) was evaluated at two harvests (18 July and 30 September 1985) of two swards (perennial ryegrass and permanent pasture) in difficult ensiling conditions. On each occasion erbage was ensiled with and without inoculant using two 0·5–t capacity steel tower silos per treatment. The contents of the two replicate silos per treatment were combined for feeding to cross-bred wethers in digestibility and metabolizable energy (ME) partition studies.
Overall, inoculated herbage declined in pH post-harvest at a faster rate than control herbage (p<0·001) and three out of the four inoculated silages had lower pH, ammonia-N, acetate and alcohol and higher residual soluble carbohydrate content (p<0·001) than control. Significantly higher digestibility of nutrients (P<0·05) was found in three of the inoculant-treated silages and these also had significantly higher ME values than control (P<0·001), (10·58 and 8·77 MJ kg tol DM⁻¹ for the treated and untreated silages respectively). The use of inoculant on herbage of only moderate ensiling potential therefore, produced significant improvements in fermentation quality and feeding value over control.     

The effect upon fermentation and nutritive value of silages produced after treatment by three different inoculants of lactic acid bacteria applied alone or in combination

Silages were made from the first cut of a predominantly perennial ryegrass sward. The silages were either untreated (W) or treated with formic acid (31 t⁻¹, F) or with 10⁶ lactic acid bacteria (LAB) g⁻¹ grass of each of three strains alone (A. Lactobacillus plantarum MTD1; B, Pediococcus species 6A2; C, L. plantarum 6A6) or in combination (AB. AC) to give seven treatments. The silage fermentation in 10-kg silos was followed chemically and microbiologically and the nutritive value of selected treatments evaluated using 2-t silos.
The control silage (W) fermented well. Addition of formic acid restricted fermentation and produced a silage with a high ethanol concentration. After day 4, all inoculated silages had lower pH values and higher lactic acid concentrations and a higher ratio of lactic acid to acetic acid than the control silage. Chemically there was little difference between the inoculated silages in terms of final composition. Microbiologically the LAB applied in treatments B and C dominated the LAB populations in those silages when applied alone; however, they were suppressed when applied in combination with inoculant A.
When fed to sheep, the intake of the formic acid-treated silage was significantly ( P < 0·01) lower than that of the other silages and the intake of silage treated with inoculant A significantly ( P < 0·001) higher than that of silages treated with inoculants B and C. The apparent organic matter ( P < 0·001) and nitrogen ( P < 0·01) digestibilities of the formic acid-treated silage were also significantly lower than those of the other silages.     

The effects of pressure and stage of ensilage on the mechanical properties and effluent production potential of grass silage

Direct-cut grass silage was ensiled without compression in laboratory silos for 0–75 d. On occasions during this period, the silage was subjected to creep compression tests at three pressure levels for a period of 5 h and effluent production was measured. Precision-chopped Italian ryegrass ( Lolium multiflorum ) was ensiled in the first experiment, whereas flail-harvested perennial ryegrass ( Lolium perenne ) was ensiled in the second. Pressure levels were 11·3, 16·9 and 22·5 kPa for Experiment 1 and 5·7, 11·3 and 16·9 kPa for Experiment 2. Moisture contents of the ensiled herbage were 815 and 856 g kg⁻¹ for Experiments 1 and 2 respectively. The consolidation of the grass silage was described by a Burgers body model. Effluent production was more closely related to strain than to compressibility. Linear regression equations for the relationship between strain and effluent production are presented. There was a significant positive linear relationship between pressure and effluent production at each silo opening time in both experiments. The time course of effluent production was fitted to a negative exponential curve. The time that elapsed before effluent release in each experiment was a function of both pressure and time after ensilage. The results of the experiments were compared with the predictions of two models of effluent production. Reasonable agreement between predicted and actual effluent production could be obtained provided the measured material parameters were used and immediate saturation of the forage was assumed. Using the models highlighted the need for a better understanding of saturation development in the silage.     

Enzymes as silage additives. 1. Silage quality, digestion, digestibility and performance in growing cattle

Four grass silages were made from perennial ryegrass ensiled after a 1h wilt in 2-t silos without additive application, with application of formic acid or with one of two enzyme mixtures of hemicellulases and cellulases (enzyme 1 and enzyme 2). Effluent losses were monitored over the ensiling period (130 d).
Analyses of the silage showed that formic acid-treated silage had lower concentrations of lactic acid than the other silages. Both enzyme-treated silages had lower levels of cellulose, acid detergent fibre (ADF) and neutral-detergent fibre (NDF) than the untreated and formic acid treated silages. Effluent production was highest with enzyme-treated silages.
The silages were subsequently fed to growing steers equipped with rumen cannulae and T-piece duodenal cannulae. Apparent whole-tract digestibilities of organic matter constituents were significantly lower ( P < 0·05) with both enzyme-treated silages (untreated; 0·736, formic acid; 0·722, enzyme 1; 0·694, enzyme 2; 0·703). Both untreated and enzyme 2-treated silages sustained higher nitrogen digestibilities (g g⁻¹ intake) (untreated; 0·675, formic acid; 0·636, enzyme 1; 0·630, enzyme 2; 0·662) and N retentions (g d⁻¹) untreated; 16·0, formic acid; 14·0, enzyme 1; 11·6, enzyme 2; 16·6), but none of these differences was significant. When formic acid-treated silage was offered, there was a greater amount of organic matter apparently digested in the rumen (ADOMR). Non-ammonia nitrogen and microbial nitrogen flows at the duodenum were similar on all diets. The efficiency of microbial protein synthesis was highest with enzyme 2-treated silage and lowest with formic acid-treated silage (untreated, 35·4; formic acid, 25·2; enzyme 1, 30·4; enzyme 2, 39·4), but none of these differences were significant.     

Influence of bacterial inoculant and substrate addition to lucerne ensiled at different dry matter contents

Field-wilted lucerne was chopped with a forage harvester at 33 ± 1·5, 43 ± 2·0 and 54 ± 1·8% dry matter, treated and ensiled in laboratory silos during four cuttings in each of two years. Treatments were control (C), sugar addition at 2% of fresh weight (S), inoculum applied at 3 × 10⁵ bacteria g⁻¹ herbage (I), and sugar and inoculum combined (IS). Duplicate silos were opened and analysed after 1, 2, 3, (4 or 5), 7, 14 and 60 d of fermentation. The initial rate of proteolysis of lucerne protein decreased with increasing dry matter (DM) content of the lucerne, and was not influenced by the year, cutting or silage treatment. Inoculation increased ( P <0·05) the rate of pH decline for all silage dry matters, and shortened the lag time prior to pH decline with 33 and 43% dry matter silages. Sugar addition had no effect on rate of pH decline or lag time. Inoculation and sugar addition both lowered final pH, acetic acid, ammonia (NH₃), free amino acids (FAA) and soluble non-protein N (NPN) in silages ( P <0·01) and increased lactic acid content with 33 and 43% dry matter silages. Only inoculation was beneficial at 54% DM with no difference between I and IS. The influence of forage characteristics (epiphytic lactic acid bacteria, buffer capacity and sugar:buffer capacity ratio) on treatment effectiveness varied with dry matter content.     

Determination of microbial protein in perennial ryegrass silage

The microbial matter fraction was determined in perennial ryegrass silages of different dry-matter (DM) contents, ensiled with or without Lactobacillus plantarum . ¹⁵N-Leucine and the bacterial cell wall constituent diaminopimelic acid (DAPA) were used as markers for microbial-N. Perennial ryegrass crops with DM contents of 202, 280 or 366 g kg⁻¹ fresh weight were ensiled in laboratory-scale silos and stored for 3 to 4 months. At different times after ensiling, silages were analysed and microbial fractions were isolated. Microbial-N concentration determined with ¹⁵N-leucine reached a maximum during the first week of ensilage. It remained unchanged thereafter, except in silage with a DM content of 280 g kg⁻¹ in which it decreased ( P  < 0·01) by 32% during storage. After 3 to 4 months ensilage, microbial-N concentration varied from ≈0·3 to ≈1·7 g kg⁻¹ DM. A negative relationship was observed between microbial-N concentration and silage DM content. Inoculation resulted in an approximately twofold increase ( P  < 0·001) in microbial-N concentration. Microbial-N concentrations determined with DAPA were 1·14–2·07 times higher than those determined with ¹⁵N-leucine. However, 19–35% of the DAPA in silage occurred in a soluble form, indicating that this fraction of DAPA was not associated with intact bacteria.     

A comparison of unwilted and wilted grass silages offered to beef cattle without and with monensin sodium

A randomized block experiment was conducted to compare unwilted and wilted grass silages and the effects of the feed additive monensin sodium on the silage intake and performance of finishing beef cattle. Two regrowths from a predominantly perennial ryegrass (cv. S24) sward were ensiled either without wilting or after field wilting for 3 d (dry matter (DM) concentrations 161 and 266 g kg⁻¹ respectively). Both silages were treated with formic acid (2·6 and 30 litre t⁻¹ respectively) and were well preserved. The silages were offered ad libitum to forty-eight Charolais-cross cattle (thirty-two steers and sixteen heifers, mean initial live weight 351 kg) for 145 d. All animals received 2·2 kg concentrates per head daily and half of those on each silage treatment received in addition 200 mg monensin sodium per head daily. Silage DM intake was 5.04, 504. 5·48, 5·63 ± 0.134 kg d^−l; fasted liveweight gain was 0·69, 0·77. 0·64 and 0·73 ± 0.033 kg d^−l and carcass gain was 0·47, 0·50, 0·40 and 0·45 ± 0·020 kg d⁻¹ for the unwilted silage without and with monensin and the wilted silage without and with monensin respectively. It is concluded that wilting grass of low DM concentration for 3 d prior to ensiling reduced the performance of finishing beef cattle below that obtained from well-preserved unwilted silage in spite of a higher DM intake being achieved with the wilted silage. The inclusion of monensin sodium in a silage-based diet increased performance without significantly affecting feed intake.     

The effect of addition of sugar and inoculation with two commercial inoculants on the fermentation of lucerne silage in laboratory silos

Lucerne (DM 236 g kg^-1, WSC 49 g (kg DM)^-1) was ensiled in test-tube silos with or without either glucose or fructose and with or without one of two commercial inoculants. The WSC content of the forage as ensiled was too low to obtain a well preserved untreated silage. By day 4 the pH values of the silages with added sugar or inoculant were significantly lower (P< 0·001) than the control silage. A satisfactory fermentation was attained only in the silages to which sugar and an inoculant had been added. These silages had a lower pH, more protein-N (P< 0·001), less ammonia-N (P<0·001), a faster increase in counts of lactic acid bacteria, and decrease in counts of coliforms than the other silages. Homo-fermentative lactic acid bacteria dominated the fermentation in the inoculated silages while leuconostocs dominated the early stages of fermentation in the control silages. The results indicate that if there is insufficient sugar in the original crop, then the bacteria in an inoculant will not be able to produce enough lactic acid to lower the pH to an acceptable level. This has important implications for the ensilage of lucerne and other highly buffered low sugar crops.     

The fertility of heifers consuming red clover silage

The fertility of forty-two heifers offered either red clover silage or grass silage prior to and during the period of insemination was compared. Pregnancy rate to first service was significantly higher ( P < 0·25) on red clover silage at 76% compared with 43% on grass silage. Pregnancy rates from a mating period covering three oestrus cycles were similar on the two silage diets. The ratio of services to pregnancies was lower on red clover silage (1·2) than on grass silage (2·2). In a subsidiary trial with twenty-three heifers, pregnancy rate to first service on red clover silage was 78%.
There was no evidence to indicate that herd fertility is depressed by red clover silage.     

A meta‐analysis comparing standard polyethylene and oxygen barrier film in terms of losses during storage and aerobic stability of silage

A meta‐analysis was undertaken of 51 comparisons of standard polyethylene film with oxygen barrier (OB) film in covering systems for bunker silos, unwalled clamp silos and bales. Mean losses of DM or OM during storage from the top 10 to 60 cm of bunker and clamp silos were 195 g kg^?1 for standard film and 114 g kg^?1 for OB film systems (41 sets of data, P < 0·001), while mean total losses of DM from baled silage were 76·8 g kg^?1 for standard film and 45·6 g kg^?1 for OB film systems (10 sets of data, P < 0·001). Top surface silage judged subjectively to be inedible was 107 and 29·6 g kg^?1 for standard film and OB film systems respectively (5 sets of data, P = 0·02). Aerobic stability was 75 h for silage stored under standard film system and 135 h for silage stored under OB film system (11 sets of data, P = 0·001). It is concluded that the OB film system reduces losses from the outer layers of silos and from bales and increases the aerobic stability of silage in the outer layers of silos.     

Protein supplementation of formic acid- and enzyme-treated silages. 2. Nitrogen and amino acid digestion

Two silages were made from primary growth perennial ryegrass and ensiled after the application of either formic acid or an enzyme mixture of cellulase and hemicellulase. Silage analysis showed both silages to be well preserved with low pH of 3·70 and 3·62 for the formic and enzyme treatments respectively. Formic acid-treated silage had a higher total amino acid concentration than enzyme-treated silage. The silages were offered to growing steers either as the sole diet or supplemented with rapeseed meal at 60 g or 120 g fresh weight kg⁻¹ silage DM offered, in a 6 × 6 Latin square arrangement.
Non-ammonia nitrogen and microbial nitrogen flows at the duodenum (g d⁻¹) were significantly ( P < 0·05) increased by supplementation of enzyme-treated silage compared with formic acid-treated silage (enzyme, 83·6, 58·7; enzyme + 60 g, 101·7, 75·3; enzyme + 120 g, 112·5, 80·7; formic, 91·9, 63·7; formic + 60g, 88·3, 67·9; formic + 120 g, 95·5, 67·1) respectively. Efficiencies of microbial protein synthesis were increased for supplemented enzyme-treated silage diets and values were reduced for supplemented formic acid-treated silage diets compared with the silage only diets (enzyme, 27·9; enzyme + 60 37·7; enzyme + 120 g, 38·6; formic, 33·7; formic + 60g, 31·2; formic + 120 g, 28·8). Total amino acid flow at the duodenum increased with supplementation of both silages; however, microbial amino acid flow increased significantly ( P < 0·05) with supplementation of enzyme-treated silage compared with formic acid-treated silage diets. Significantly greater amounts of cystine, methionine, alanine, valine and aspartic acid entered the small intestines of animals receiving supplemented enzyme silages compared with supplemented formic acid silages.     

LOSSES IN THE CONSERVATION OF HEAVILY-WILTED HERBAGE SEALED IN POLYTHENE FILM IN LINED TRENCH SILOS

A conservation experiment is described in which the losses in ensiling heavily-wilted herbage, of approximately 50% dry matter, in lined trench silos, with and without a polyethylene film seal, are compared. The composition and digestibility of the silages produced in the sealed silos were comparable with those of the original herbage. The content of organic acids in the sealed silos was low, suggesting that fermentation was suppressed by the partially gas-tight seal. In the unsealed silos there was considerable spoilage, 70% of the material being inedible. In the 2 sealed silos losses of dry matter were 8·2 and 5·2% and losses of SE 11·5 and 7·8%, respectively. The losses recorded are compared with those usually encountered in lined trench silos.     

Silage and milk production: comparisons between hays of different digestibilities as silage supplements

In three separate feeding experiments using a total of thirty individually-housed Ayrshire cows three silages made from perennial ryegrass were given ad libitum together with supplements of four different hays in the long form. The in vitro D-values of the silages ranged from 0·298 to 0·283, and the hays from 0·280 to 0·200. The daily intake of hay DM varied from 0·2 to 4·2 kg per cow and was given either without or with a daily maximum of 2·2 kg concentrate DM containing 379–527 g CP per kg DM. On average, 1 kg hay DM decreased silage intake by 0·24 kg DM with a range of 0·21–1·20 kg. The hay supplements had only small and non-significant effects on total DM intake, milk yield and milk composition, but increased the daily intake of drinking water. In three behavioural studies, the eating and ruminating times expressed as min per kg DM did not differ significantly between the various supplement treatments. It is concluded that hay has only a marginal value as a supplement for grass silage, although the hay could serve as a useful 'buffer' feed if the amount of silage was limited.     

The antimicrobial spectra of some salts of organic acids and glutaraldehyde in respect to their potential as silage additives

Salts of formic, acetic and propionic acids and glutaraldehyde were subjected to an assay with representatives of the major groups of silage micro-organisms at pH 4·0, 4·5, 5·0 and 6·0. The antimicrobial spectra of the salts suggested that they are most effective against the least desirable components of the bacterial flora of silage, coliforms, Clostridia and bacilli, and would create conditions for a desirable fermentation in practice. In many instances the activity of the salts was better than the corresponding free acid and it was postulated that, if supplemented with acid, aerobic stability of silage might also be improved. Glutaraldehyde, whilst being less selective as an antimicrobial agent, possesses properties which would encourage a lactic-acid fermentation in silage.     

An evaluation of effluent from grass silage as a feed for beef cattle offered silage-based diets

Four experiments were carried out to evaluate effluent from grass silage as a feed for beef cattle. Castrated male cattle (steers) with mean initial live weights of 380,460 and 400 kg in Experiments 1–3 were offered effluent, in addition to a diet of grass silage given ad libitum and supplemented with cereal-based concentrates. The cattle readily consumed effluent in preference to fresh water, the mean intake over the three experiments being 12·71 per head daily containing 0·77 kg dry matter (DM). Supplementation with effluent did not affect silage DM intake and increased total DM intake by 10%. Digestible energy concentrations of 13·1 and 16·6 MJ kg-^-1 DM and organic matter, respectively, were determined for effluent in Experiment 4. There were no problems of ill health in any of the 69 cattle which were offered effluent. It is concluded that fresh or well-preserved stored effluent from grass silage is readily consumed by beef cattle, is of high nutritive value and could enable savings to be made in total feed costs.     

Silage and milk production: a comparison between three grass silages of different digestibilities

Three first-harvest grass silages made from S23 perennial ryegrass cut on 25 May, 13 June and 25 June, and termed early, medium and late silages respectively, were compared in a 16-week feeding experiment with fourteen Ayrshire cows. The early, medium and late silages had D-values of 71·2, 65·0 and 62·5 respectively. The early silage was offered alone, whereas the medium and late silages were supplemented with a concentrate containing 208 g crude protein per kg DM at rates of 2, 3 and 4 kg per 10 kg milk. Silage DM intake was 12·8 kg per cow per d on the early silage treatment, and decreased progressively as concentrate intake increased on the other silage treatments. The mean daily milk yields were 16·0 kg per cow in the early silage treatment, 17·0, 18·4 and 20·4 kg per cow in the medium silage treatments, and 16·8, 18·1 and 20·2 kg per cow in the late silage treatments on the 2-, 3- and 4-kg concentrate treatments respectively. Fat concentration in the milk was not affected significantly by treatment, whereas the CP and SNF concentrations increased progressively and significantly as supplementary feeding increased. From the relationship between milk yield and concentrate intake it was calculated that the medium and late silages required a daily concentrate supplement of 2·1 kg DM per cow to give the same daily milk yields as the early silage.