Alfalfa (Medicago sativa L.) and tall fescue (Festuca arundinacea Schreb.) were cultivated in the experimental field of Rikaze Grassland Station (29° 16′ latitude N, 88° 53′ longitude E, 3,836 m above sea level, Tibet, China). The HBS was the residue remaining after harvesting grain. Alfalfa was harvested at 75% bloom and tall fescue was harvested at the boot stage. The forages and HBS were chopped to the length of 2 to 3 cm with a manual forage chopper. The WHDG was obtained from a private barley wine processing company at Rikaze, and the mixed concentrates (7.5% crack corn, 20% rape cake meal, 20% cotton seed, 27.5% distillers dried grains with soluble, 20% wheat bran, 5% vitamin–mineral) were obtained from a private small-scale dairy farm in Rikaze, Tibet, China. TMRs (640 g) were ensiled using a plastic laboratory silo (1 L capacity). The chemical compositions of all used materials are shown in Table 1, and ingredients and chemical compositions of TMR are shown in Table 2. A total of 150 experimental silos (6 treatments×5 time points×5 replicates per treatment) were prepared in a completely randomized design to evaluate the following treatments: i) control (without additive), ii) L. buchneri, iii) acetic acid, iv) propionic acid, v) 1,2-propanediol, and vi) 1-propanol. The L. buchneri was supplied by Institute of Forage Ensiling and Processing of Nanjing Agricultural University and applied at 1×10⁶ colony-forming units (cfu)/g based on fresh weight (FW) [6]. The application rates of acetic and propionic acids were 0.3% FW (equals to 6.2 g/kg dry matter [DM]) [6]. The 1-propanol and 1,2-propanediol were applied at 0.5% FW. All chemicals used were of analytical grade. The prepared silos were stored at ambient temperature, opened after 90 days and then subjected to an aerobic stability test for 14 days.

Fresh forages, pre-ensiled TMR and fermented TMR were analyzed for chemical and microbiological compositions. Approximately 200 g of sample was oven-dried at 60°C for 48 h to determine DM content and then ground to pass 1-mm screen with laboratory knife mills (FW100, Taisite Instrument Co., Ltd., Tianjin, China) for other chemical composition analysis. Total nitrogen (TN, 978.04), ether extract (EE, 920.39), and crude ash (Ash, 942.05) were measured according to the methods of Association of Official Analytical Chemists [7]. Crude protein (CP) was calculated as TN×6.25. Water soluble carbohydrates (WSC) was determined by colorimetric after reaction with anthrone reagent [2]. The contents of neutral detergent fibre (aNDFom) and acid detergent fibre (ADFom) were measured by the procedures of Van Soest et al [8], the heat stable amylase and sodium sulphite were used for NDF procedure.

For microbiological composition analysis, 10 grams of sample was blended with 90 mL of sterilized water, and serially diluted in sterilized water. The lactic acid bacteria (LAB) was counted on de Man, Rogosa, and Sharpe agar medium, incubated in an anaerobic incubator at 30°C for 2 days. Yeasts and aerobic bacteria were enumerated on potato dextrose agar (Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) and nutrient agar (Shanghai Sincere Biochemical Technology Co., Ltd., Shanghai, China) under aerobic conditions.

About 35 grams of sample was blended with 60 mL distilled water and macerated for 24 h at 4°C. The extract was filtered through 2 layers of cheesecloth and a filter paper (Xinhua Co, Ltd., Hangzhou, China). The filtrate was used for pH, organic acids and ammonia nitrogen (NH₃-N) determinations. The pH was measured with a HANNA HI 2221 pH meter (Hanna Instruments Italia Srl, Villafranca Padovana, Italy). The NH₃-N was determined using the phenol-hypochlorite reaction method [9]. Buffering capacity was determined according to the method of Chen et al [10]. The organic acids (including lactic, acetic, and propionic acids) and alcohols (including ethanol, 1,2-propanediol and 1-propanol) were quantified using an Agilent 1260 HPLC system equipped with a refractive index detector (Carbomix H-NP5 column, 2.5 mM H₂SO₄, 0.5 mL/min).

In vitro fermentation was conducted in serum bottles following the method of Contreras-Govea et al [11] with some modifications. Briefly, approximately 1 g of ground sample was placed in 130-mL serum bottles. The rumen fluid was obtained through a rumen fistula before morning feeding from four dry Boer goats fed with diet consisting of 6% alfalfa, 59% guinea grass, and 35% concentrate at 1.3 times of the maintenance level. Rumen fluid was filtered through 4 layers of gauze and mixed in the ratio of 1:2 (v/v) with buffer, and 60 mL of the mixture was transferred into each serum bottle. Each serum bottle was flushed with CO₂ and kept in a water bath at 39°C, after being capped with a butyl rubber stopper and sealed with an aluminum crimp. Gas production was measured at 4, 8, 12, 24, 48, and 72 h using a pressure transducer technique and corrected with blank bottles. After 72 h of incubation, undigested solids were precipitated by centrifugation at 1,000 g for 10 min at room temperature, dried in an aerated oven at 65°C for 48 h and then assayed for DM and aNDF. The in vitro digestibility of DM (IVDMD) and NDF (IVNDFD) were calculated based on the differences in their respective weight before and after incubation.

Cumulative gas production (GP) data were fitted to the exponential equation: y = b(1−e^−ct), where y is the volume of gas produced at time t, b is the GP from the insoluble fraction (mL), c is the GP rate constant, t is the incubation time (h).

After 90 days of ensiling, fermented TMR from each silo was taken out, fully mixed and loosely placed into a bigger 15 L open-top and sterile polyethylene bottle. Each bottle was covered with a double layer of gauze and stored at ambient temperature (24°C to 27°C). During the test, TMR were sampled for pH, organic acids, NH₃-N, WSC, and microbes count analyses at 0, 3, 6, 9, and 14 days. Aerobic stability is defined as a rise in pH value of TMR by 0.5 unit above the initial pH value at silos opening [12].

Analysis of variance (ANOVA) was performed using the general linear model procedure of SAS rev. 9.2. The data related to fermentation variables were subjected to one-way ANOVA, with fixed effect of treatments. While data related to chemical and microbial composition during aerobic exposure were analyzed using the following model: Y_ij = μ+S_i+ A_j+S_i×A_j+ɛ, where: Y_ij = the response variable; S_i = treatment; A_j = aerobic exposure; S_i×A_j = treatment×aerobic exposure; ɛ = random errors. Duncan’s multiple range test was used to separate means when significant effects (p<0.05) were detected.

As shown in Table 1, two roughages contained similar ADF, ash, and EE. Compared with tall fescue, alfalfa was lower in DM, WSC, and NDF contents, while higher in CP and buffering capacity. The chemical and microbial compositions of the TMR before ensiling are presented in Table 2. The DM, NDF, CP, and WSC contents of TMR were 485, 418, 221, and 90.5 g/kg DM, respectively. The LAB, aerobic bacteria and yeast counts were 5.37, 6.44, and 5.21 log₁₀ cfu/g FW, respectively.

The fermentation quality and microbial composition of TMR after 90 days of ensiling are given in Table 3. Additives affected all fermentation parameters (p<0.05), except for butyric acid contents and LAB number. Treating L. buchneri, acetic acid decreased propionic acid contents, whereas increased (p< 0.05) pH, acetic acid and ethanol contents. Adding 1,2-propanediol and 1-propanol decreased (p<0.05) lactic acid and propionic acid contents, whereas increased pH, acetic acid, ethanol and NH₃-N contents. The 1-propanol and 1,2-propanediol were only accumulated greatly in their respectively treated TMR. Addition of acetic acid reduced (p<0.05) the aerobic bacteria counts. Compared with other TMR, the numbers of yeasts were much lower (p<0.05) in TMR ensiled with L. buchneri, acetic acid, 1,2-propanediol, and 1-propanol.

The chemical compositions of ensiled TMR after 90 days of ensiling is shown in Table 4. With respect to chemical compositions, only DM was influenced by the additives (p<0.05). Compared with control, TMR ensiled with 1,2-propanediol and 1-propanol showed lower (p<0.05) DM contents and higher (p<0.05) DM loss. The measured or estimated in vitro parameters are presented in Table 5 and Figure 1. The potential GP ranged from 158 to 200 mL/g DM. Additives did not affect in vitro parameters including GP₂₄, GP rate constant, potential GP, IVDMD, and IVNDFD.

The changes in fermentative characteristics and microbial compositions of ensiled TMR during aerobic exposure are given in Table 6 and 7, respectively. The control began to spoil after 6 days of aerobic exposure, with rises in pH and declines in lactic acid contents. All additives improved the aerobic stability of ensiled TMR to different extents. Of the additives, L. buchneri, acetic acid, 1,2-propanediol, and 1-propanol had superior abilities to propionic acid at improving aerobic stability, indicated by stable pH and lactic acid content. Throughout the aerobic stability test, acetic acid, and ethanol contents were always greater (p<0.05) in L. buchneri, acetic acid, 1,2-propanediol, and 1-propanol-treated TMR relative to other TMR. Table 7 shows the changes in chemical compositions of ensiled TMR during aerobic exposure. The DM, WSC, and NH₃-N contents fluctuated during the test period and did not differ among the TMRs at most intervals of the aerobic stability test. The changes in microbial composition of ensiled TMR during aerobic exposure are displayed in Table 8. TMR ensiled with L. buchneri, acetic acid, 1,2-propanediol, and 1-propanol showed numerically, or significantly lower yeast and aerobic bacteria counts than other TMR during the aerobic exposure.

The fermentation quality of silage depends on the chemical and microbial properties of the material to be ensiled. To ensure satisfactory silage quality (i.e., extensive lactic fermentation and high recovery of DM and energy), a feedstuff must have adequate WSC content (>60 g/kg DM) and population of LAB (>5.00 log₁₀ cfu/g FW) as well as a proper DM (300 to 400 g/kg DM) [13]. These criteria were almost met in the TMR at the time of ensiling, despite the DM being slightly higher than the normal. After 90 days of ensiling, the butyric acid was undetected and lactic acid was dominant among the fermentation products in all ensiled TMRs (Table 3). Treating with L. buchneri and acetic acid increased pH, acetic acid, and ethanol production in the TMRs. L. buchneri is well known as a heterofermentative LAB species. Altered fermentation in L. buchneri-inoculated silage might be attributed to the increased dominance of L. buchneri in epiphytic microflora that shifted the metabolism to a more heterofermentative process, while that in acetic acid-treated silage might be related to the depression of homofermentative LAB species considering that heterofermentative LAB are more tolerant to acetic acid than homofermentative LAB [14]. Similarly, Ren et al [15] also found that addition of acetic acid weakened the intensity of lactic fermentation and lowered the ratio of lactic to total organic acids in silages prepared with dried corn stover and cabbage waste. Interestingly, it was observed that adding 1,2-propanediol and 1-propanol also increased the pH, acetic acid and ethanol concentrations in the ensiled TMR. The effects of 1,2-propanediol and 1-propanol addition on fermentation have been rarely reported in the literature. However, Mukdsi et al [16] reported that some lactobacilli are capable of synthesizing esters with lactic acid and alcohol as the precursors. A possible explanation might be esterification of lactic acid and the alcohol which slowed the pH decline and increased the activities of acetic acid and ethanol-producing microorganisms, such as enterobacteria and yeasts, at early stages of ensiling. NH₃-N is a sensitive indicator for silage proteolysis and its production is closely related to the pH decline during ensiling [17]. Higher NH₃-N concentrations demonstrated the slower pH decline in 1,2-propanediol and 1-propanol-treated TMR in comparison to control.

Krooneman et al [18] previously isolated two strains of L. diolivorans from corn silage and demonstrated their abilities of fermenting 1,2-propanediol to propionic acid and 1-propanol. As propionic acid exhibits antimicrobial activity, this fermentative process also contributes to the improvement in aerobic stability of L. buchneri-inoculated silage [4]. In the experiment, adding 1,2-propanediol did not result in increases in propionic acid and 1-propanol concentrations suggesting the absence of L. diolivorans in the TMR. In addition, propionic acid addition did not elicit a significant effect on the suppression of aerobic bacteria and yeast when compared with control in this study (Table 3). This might be linked to the presence of microorganism species resistant to low concentrates of propionic acid. Crawshaw et al [19] found some yeasts still flourished in grass silage when treated with propionic acid at a level of 6 litres/t FW (equals to 8 g/kg DM).

During ensiling process DM and nutrient losses are unavoidable and mainly results from plant respiration and activities of microorganisms [2]. Addition of 1,2-propanediol and 1-propanediol did not efficiently decline pH and resulted in lack of preventing microbial activity. This probably caused higher DM loss relative to other treatments (Table 4).

In vitro GP is often used as an important indicator for rumen digestibility potential [11]. In the present study, despite greater DM loss of 1,2-propanediol or 1-propanol treated TMR than others, in vitro digestibility parameters among the silages did not differ, suggesting none of these additives substantially affected ruminal digestion of the TMR.

Aerobic deterioration is initiated, in most cases, by acid-tolerant yeasts [20]. Yeasts can oxidize the fermentation products, leading to pH rises and proliferation of aerobic microorganisms in air-exposed silage. Aerobic deterioration of silage does not only cause losses of nutritional value, it also negatively affects the safety of silage with an increased risk of pathogenic microorganisms [21]. Well-fermented silages are easily subject to aerobic spoilage when exposed to aerobic condition because of degradation of lactic acid by yeasts and aerobic microbes [22,23]. It is generally believed that silages are prone to deteriorate when the yeast population exceeds 5.00 log₁₀ cfu/g FW [24]. In the experiment, control began to spoil after 6 days and propionic acid-treated TMR became unstable after 14 days of aerobic exposure (Table 6). This may be linked with high yeast numbers (>5.00 log₁₀ cfu/g FW) and accumulations of lactic acid (>72 g/kg DM) at the silos opening (Table 3). The TMRs treated with L. buchneri, acetic acid, 1,2-propanediol, and 1-propanol kept stable throughout the aerobic stability test (Table 6). 1,2-Propanediol itself has no inhibitory effects on yeast growth in silage, and 1-propanol only causes weak inhibition at concentrations of more than 20.0 g/kg DM [25]. While acetic acid is well known as an active substance for suppressing the proliferation of yeast, mold and fungi during aerobic exposure. This acid is typically produced in greater quantities than propionic acid during ensiling due to the hetero-fermentative metabolism of LAB [26]. High accumulations of acetic acid in these TMRs may be directly responsible for the increased aerobic stability. Furthermore, ethanol also exhibits antimicrobial activity and reportedly has the ability to potentiate the effect of acetic acid with respect to inhibition of fermenting yeasts [27]. High ethanol concentrations were also supposed to play an important role in preventing aerobic deterioration.