The prediction of silage feeding value from the analysis of herbage at ensiling and the effects of rate of nitrogen (N) fertilizer, date of harvest and treatment additive at ensiling on fermentation, dry matter (DM) recovery and effluent production were evaluated in a factorial experiment. Herbage was harvested from 75 plots, laid out in three replicate blocks of 25, from the primary growth of a predominantly perennial ryegrass sward. The plots received 72, 96, 120, 144 or 168 kg N/ha and were harvested on 10, 17, 24 or 31 May or 7 June. Herbage was ensiled for 176 days in laboratory silos (6 kg capacity), either untreated or treated with formic acid or a bacterial inoculant, each applied at 3 ml/kg herbage. Increasing the rate of N fertilizer and delaying harvest date increased herbage DM yield but decreased silage digestible organic matter digestibility (D-value) and intake potential (P < 0·001).

Increasing the rate of N fertilizer increased crude protein (CP), acid insoluble N (AIN) (P < 0·001), acid detergent fibre (ADF) (P < 0·01) and acid detergent lignin (ADL) concentrations but had no statistically significant effect (P > 0·05) on pH or the concentrations of ammonia N, lactic or volatile fatty acids. Delaying harvesting date decreased CP, AIN and ADF concentrations (P < 0·001) and effluent DM losses (P < 0·001). Treatment with either formic acid or inoculant decreased silage pH and the concentrations of ammonia N, acetic acid and propanol and increased AIN, ethanol, neutral detergent fibre (NDF), ADF and cellulose concentrations, effluent volume, DM losses and intake potential. Formic acid treatment decreased D-value (P < 0·001), whereas inoculant treatment was not statistically significant (P > 0·05). For potential silage and metabolizable energy (ME) intakes and D- value, strong positive correlations (P < 0·001) were identified with herbage pH, buffering capacity and concentrations of CP, AIN, true protein nitrogen (TP), while negative correlations (P < 0·001) were identified with herbage yield and concentrations of NDF and hemicellulose. With untreated silages, strong positive correlations (P < 0·001) were identified between silage pH and herbage pH and concentrations of AIN, buffering capacity, ash, TP and negative correlations (P < 0·001) were identified with herbage DM yield and concentrations of DM and WSC. Herbage yield and concentrations of NDF, TP, DM and nitrate N provided the best linear relationship for predicting potential ME intake of the resultant untreated silages (R2 = 0·94). It is concluded that delaying harvest date had the most detrimental effect on silage feeding value, although this effect was partially overcome by reducing the rate of fertilizer N and use of either an inoculant or formic acid additive. Silage feed value was also highly correlated with the protein and fibre fractions, ash, buffering capacity and pH of the herbage at ensiling.