Pure sudangrass (Sorghum sudanense (Piper) Stapf) was grown in experimental plots at Sanping practice base in Xinjiang Agricultural University (latitude 43°54′N, longitude 87°19′E), which is situated in Xinjiang province, China, from April to July 2019, using Xinsu sudangrass variety No. 3 (Ministry of Education Key Laboratory for Western Arid Region Grassland Resources and Ecology, Xinjiang, China). Triplicate randomized experimental plots were used, each plot with an area of 5 m×10 m. Approximately 45 kg/hm² of cow manure fertiliser being applied during the sudangrass growing. The soil type in this area is typically loamy soil and the average soil organic matter in the experimental plots was 15.6 g/kg; effective N, P, and K in the soil were 91.3, 23.7, and 103.1 mg/kg, respectively. The sudangrass was sown in 25 April 2019 at a seeding rate of 105 kg/hm² in each block and the first-cut sudangrass was manually harvested in 6 July 2019 by sickle at the early-heading stage from three replicated plots. The plant materials were chopped to a 20-mm nominal length and then sampled in triplicate and stored at −20°C prior to chemical analyses.

The ensiling experiment was carried out according to a 2×4 (wilted stages×additives) factorial treatment structure. The sudangrass was direct wilted at the field in the sun for 8 h (direct-cut sudangrass) and turning of the forage every 2 h. The factors associated with the wilted stage were fresh or wilted (8 h), and the 4 additive treatments were i) no additive (control, CK); ii) molasses (M) (Tianshun Bioengineering Co., Ltd., Urumqi, China) applied at 5% fresh weight (FW); iii) L. plantarum (LP) (strain 10474, China General Microbiological Culture Collection Center, Beijing, China) applied at 1×10⁵ colony forming units (CFU)/g FW; iv) molasses (M, Tianshun Bioengineering Co., Ltd, Urumqi, China) applied at 5% FW + L. plantarum (LP, strain 10474, China General Microbiological Culture Collection Center, Beijing, China) applied at 1×10⁵ CFU/g FW. Based on the manufacturer’s information, the molasses contained 45.2% moisture, 59.2% total sugar DM, 38.4% sucrose and 12.4% reducing sugars. According to the set additive amount, the molasses, L. plantarum, and molasses + L. plantarum were diluted to 0.5 L with water, respectively. The same amount of water was added to the control. The additives were diluted in deionized water and each treatment was sprayed in a fine mist on the sudangrass from each of the appropriate replicate plots separately using a hand sprayer (Sail Multipurpose sprayer, Sail Corp., Beijing, China). Hand forks were used to mix the forage from each treatment. The materials (3,200 g) were individually packed into 5.0-L plastic jars (Hewanglan Paper and Plastic Products Factory, Beijing, China) and sealed with plastic tape and caps. Each treatment stored at an ambient temperature of approximately 26.0°C±3.7°C. After 60 days of ensiling, the jars were opened and sampled; the samples were stored at −20°C for subsequent chemical analysis of the DM, crude protein (CP), neutral detergent fiber (NDF), acid detergent fiber (ADF), ether extract (EE), ash, pH, LA, acetic acid (AA), propionic acid (PA), BA, WSC, NH₃-N (ammonia nitrogen), and in vitro batch cultures.

Following the method described by Owens et al [8], 20 g of each fresh silage sample was homogenized in a blender with 180 mL of distilled water for 1 min and then filtered through four layers of cheesecloth. Then, the pH of the filtrate was measured immediately using a glass electrode pH meter (PHS-3C, INESA Scientific Instrument Co., Ltd., Shanghai, China). The DM content was determined by oven drying at 60°C for 48 h. The DM recovery was calculated as (1 – [(ensiled forage DM – forage DM at silo opening)/ensiled forage DM])×100. NDF and ADF levels were measured as described by Mertens [9]. The ash content was determined by ignition at 550°C for 3 h, and the EE was examined according to the methods described by the Association of Official Analytical Chemists (AOAC) [10]. The buffering capacity (BC) of the fresh matter (FM) was determined by titration [11]. The WSC levels were determined using the anthrone method [12]. The nitrogen (N) content was analyzed according to the methods of the AOAC [10], and the CP content was estimated by multiplying the N content by 6.25.

The pH values of the in vitro ruminal fluid samples were measured using a pH meter as described above. The NH₃-N concentrations in the silage were determined by the method described by Broderick and Kang [13] and expressed as g/kg of total nitrogen (TN). The levels of organic acids, including lactic acid, AA, PA, and BA, in the silage and in vitro ruminal fluid samples were determined by high-performance liquid chromatography (LC-20A; Shimadzu, Tokyo, Japan). The analytical conditions were as follows: column, Shodex RSpak KC-811S-DVB gel C (8.0 mm×30 cm, Shimadzu, Japan); oven temperature, 50°C; mobile phase, 3 mmol/L HClO4; flow rate, 1.0 mL/min; injection volume 5 μL; detector, SPD-M20AVP (Shimadzu, Japan). To assess the ensiling quality, the V-score from the NH₃-N and VFA concentrations were calculated [14].

The LAB counts were determined using MRS agar and violet red bile agar after incubation at 30°C for 2 days. Aerobic bacterial counts were determined after growth on nutrient agar at 37°C for 3 days under aerobic conditions. All yeasts and moulds were enumerated on spread plates containing yeast extract/peptone/dextrose agar and salt Czapek Dox agar, respectively, after incubation at 28°C for 3 to 5 days. The four media were obtained from Beijing Aoboxing Biotech (Beijing, China). All microbial data were transformed to log10 values and presented on a wet weight basis.

Silage samples were oven-dried at 65°C, ground, and sieved through a 1-mm screen. The dried silages (500 mg) were weighed into 120-mL glass bottles with butyl rubber stoppers and Hungate’s screw caps. Five bottles per silage sample and 30 bottles per treatment were arranged. Fifty milliliters of a freshly prepared buffer solution (pH 6.85) and 25 mL of filtered rumen fluid collected from three rumen-fistulated lactating Holstein dairy cows (423±18 kg body weight, the animals were restricted to a supply of the same amount of a standardized diet containing 5.0 kg of alfalfa hay, 2.0 kg of wheat straw and 6.0 kg of commercial concentrate supplement per day) were added to the bottles before the morning feeding. The bottles were purged anaerobically with CO₂ for 5 s, sealed with the butyl rubber stopper and Hungate’s screw caps, and individually connected with medical plastic infusion pipes to gas inlets of an automated trace gas recording system to continuously record cumulative GP (0, 2, 4, 8, 12, 24, 36, 48 h). All bottles were incubated at 39°C for 48 h, and the entire experiment was repeated for three runs. At the end of incubation, the culture fluids in each bottle were individually filtered with pre-weighed nylon bags (8×12 cm, 42-μm pore size). The pH value, NH₃-N content and organic acid content of the filtrate were determined according to the method described above. The bags were thoroughly rinsed and dried at 65°C for 48 h to a constant weight. The differences between DM, CP, NDF, and ADF at initial incubation and the residual DM, CP, NDF, and ADF, respectively, corrected with the blanks after incubation were calculated to determine the in vitro DM, CP, NDF, and ADF disappearance (IVDMD, IVCPD, IVNDFD, and IVADFD, respectively) values.

The cumulative GP (t, mL/g DM) at time (t) for each fermentation bottle was fitted to an exponential model [15]:

where ‘e’ is the base of a natural logarithm; A represents the asymptotic GP generated at a constant fractional rate (c) per unit time; t is the gas recording time, and Lag represents a lag time phase before commencement of GP. The average GP rate (AGPR, mL/h) was defined as the rate between the start of incubation and the time at which the GP was half of its asymptotic value according to the following equation [16]:

The ratio of non-glucogenic to glucogenic acids (NGR) was calculated as described by Orskov [17] as follows:

where acetate, propionate, butyrate and valerate were expressed in molar proportions.

The fermentative characteristics, microbial counts, and in vitro and chemical compositional parameters were analyzed for significant differences via analysis of variance (ANOVA), with significance reported at a probability level of 0.05, using the general linear model in SPSS (version 21.0, SPSS Inc., Chicago, IL, USA). All data were subjected to ANOVA using the general linear model procedure in SPSS, based on the following model:

where Y_ij is the response variable (e.g., GP, fermentation kinetics parameters); μ is the overall mean; W_i is the wilted stage of the sudangrass (i = fresh or wilted); A_j is the additive (j = control, molasses, L. plantarum or molasses + L. plantarum); (W×A)_ij is the interaction term between wilted and additives; and e_ij is the residual error. Differences between treatment means were compared using the least square means method and Tukey’s multiple comparison test. The results are reported as least square means and the associated standard errors.

The included as a chemical composition of the sudangrass before ensiling are presented in Table 1. The DM, ADF, and ash content of the sudangrass after wilting was higher than that fresh. In contrast, the WSC and BC content of the sudangrass after wilting was lower than that fresh. The fermentation coefficient was lower than 25 for both treatments. The number of epiphytic LAB on the sudangrass was less than 1.0×10⁵ CFU/g FM, and the number of yeast cells was approximately 1.0×10³ CFU/g FM.

The six sudangrass silages were well preserved with additives, as indicated by the low pH value and NH₃-N proportion and high LA content and V-score compared to the control (Table 2). The pH, LA, AA, PA, BA, WSC, NH₃-N levels and V-score were significantly influenced by wilted, the additives, and the interaction of wilted with the additives (p<0.05). No significant interaction between wilted and additives was observed in terms of the counts of yeasts (p = 0.150) and moulds (p = 0.158). The BA content was less than 5.0 g/kg DM in all the treatments except the control of the fresh. Except for the control of the fresh, all the silages were fermented, with the V-score in all treatments ranging from 51 to 89 after 60 days of ensiling. The M+LP treatment at wilted exhibited significantly increased LA content and V-score and decreased pH and BA values compared to the other treatments.

The chemical composition of the silages is listed in Table 3. After silage, the DM, DM recovery, ADF, and ash levels were higher at wilted than fresh. The NDF of the wilted-treated was not significantly different from that of the fresh-treated. The additive-treated groups had lower ADF levels than the control in the wilted-treated (p<0.05), however, there was no significant difference of effects of wilted, additives and interaction between wilted and the additives on CP. The interaction between wilted and the additives was significant for EE (p<0.001), ash (p<0.001) and non-fibrous carbohydrate (NFC) (p = 0.013). At fresh, after silage-treated had lower DM and CP contents and higher NFC than the before silage, there was no significant difference for NDF before and after the fermentation of silage. At wilted, after ensiling had lower DM contents and higher NFC than before ensiling, in contrast, there was no significance difference of CP before and after ensiling.

The GP profiles of the sudangrass silages are presented in Figure 1. Significant differences were observed among treatments in terms of gas volume. As shown in Table 4, the IVADFD values were higher in additive-treated silages than in the control. Significant interactions were observed between wilted and additives in terms of in vitro GP (p<0.001) at 48 h of incubation. In the Kinetic PG model, significant interactions were observed between wilted and additives of the asymptotic GP (A) (p<0.001), GP rate (p = 0.005), half time (p<0.001) and AGPR (p<0.001), however, there was no effect on lag time (p = 0.341). The total 48-h GP was increased by supplementation with additives and was higher than in the control. The estimated values for A were in highest for the M+LP silage at wilted.

The NH₃-N, NGR, and total VFA levels were affected by wilted, the additives and the interaction of wilted with the additives (p<0.05; Table 5). The total VFA content in the additive treatments was higher than that in the control. All the additive treatments decreased the pH levels compared to the control. The highest total VFA content (115 mmol/L) was observed in the LP treatment at wilted. The interaction between wilted and the additives affected the VFA pattern, including the concentrations of acetate (p = 0.023), propionate (p<0.001), butyrate (p<0.001), isobutyrate (p = 0.022), valerate (p<0.001), and isovalerate (p<0.001).

The most significant change of forage chemical composition caused by wilting is to reduce the water content. Low water content can effectively inhibit the fermentation of clostridium, to improve the smell of silage and effectively inhibit the occurrence of mildew. The DM content of grass strongly influences the rate and extent of ensiling. Grasses with a low DM and sugar levels at pre-ensiling may exhibit clostridial fermentation and subsequent poor acceptance of the silage by animals [18]. McDonald wrote that wilting can increase the content of the DM and WSC [7], however, similar results were not observed in this experiment. This may be due to the high lignification of the sudangrass stalk, which is difficult to break during processing, and with the loss of the water the content of WSC will also decreased. According to McDonald [7], a material can be considered to be successfully ensiled if it exhibits favorable DM levels (280 to 340 g/kg DM), fermentable sugar concentrations ranging from 30 to 50 g/kg DM, a low BC and LAB abundances greater than 1×10⁵ CFU/g FM. In this study, the DM of the sudangrass in fresh before ensiling was lower than this standard, suggesting that it might be difficult to obtain high-quality silage by direct ensiling of sudangrass.

Silage pH is one of the main factors that influences the extent of fermentation and ensiling quality [19]. The low final pH (4.70 or less) in the wilted and additive-treated sudangrass silages suggested the recommended standard for high-quality silage (4.30 to 4.70) [20]. As expected, wilted sudangrass silage supplemented with L. plantarum improved the LA levels under the increased the microorganism concentration. After wilting, the sudangrass silages supplemented with additives were all well preserved and exhibited moderate fermentation quality, despite the low WSC content and numbers of epiphytic LAB in the initial herbage. The fermentation of the wilted silage exhibited a higher target ensiling DM due to elevated total VFA and LA concentrations. The ensiling quality at wilted is usually better than that fresh [21], which is consistent with the results of this study. The presence of BA in silage is undesirable given that its generation is an energy-waste metabolism and BA>5 g/kg DM is an indicator of substantial clostridial activity reducing feed intake and causing health issues [22]. In the present study, the high BA content (12.3 g/kg DM) indicated extensive clostridial fermentation in control-treated fresh silage. Wilting and additives could help to inhibit the activity of undesirable organisms in silage. It is supported by the decrease of BA content, where BA were not higher than 5 g/kg DM in those treatments. In addition, wilting and L. plantarum can reduce the content of AA and BA, which can improve palatability of the silage. Similar results have been reported by Zhang et al [23], who found wilting and L. plantarum decreased AA and BA contents of mulberry silage.

NH₃-N production reflects the extent of proteolysis in silage. Well-preserved silages should have less than 100 g/kg TN [22]. In the present study, the value of NH₃-N was less than 94.2 g/kg TN at wilted treatment, indicating that extensive proteolysis may not have occurred. Microorganisms exhibit proteolytic activity mainly in low-DM silages at high pH values (i.e., >4.2) when the growth and activity of the microorganisms is not totally inhibited under such conditions [7], which may partly explain why the NH₃-N levels were lower at wilted than fresh (higher DM and lower pH values were observed at wilted than fresh).

Molasses and the LAB inoculants used in this study are commonly used to improve the fermentation quality of sudangrass silage. Furthermore, the increased LAB counts and decreased mould counts after supplementation with additives suggest that high-quality silage was obtained, which might considerably improve the dry matter intake (DMI) of ruminants. It is well known that adequate WSC levels and LAB counts are important for rapid establishment and growth of LAB. Molasses were added to provide the WSC levels necessary for ensiling fermentation. Microbial inoculation, especially of homofermentative LAB, is a common practice for acceleration of fermentation, resulting in high-quality silage and reduced DM loss. Adesogan et al [24] found that addition of exogenous L. plantarum accelerated homofermentation, allowing increased production of LA by L. plantarum. Thus, the increased LA content in the M+LP treatment at wilted might be attributed mainly to using molasses together with L. plantarum, which improved the WSC and LAB content before ensiling. Muck et al [25] concluded that microbial inoculation lowered the pH and improved the LA content, and Jahanzad et al [26] confirmed that molasses can enhance the activity of homofermentative bacteria, which convert WSC to LA in millet and soybean silage. L. plantarum, used for inoculation in this study, is a homofermentative lactobacillus that can ferment a wide variety of substrates and rapidly produce large amounts of lactic acid. L. plantarum had positive effects on grass silage fermentation characteristics by lowering the pH and shifting fermentation towards LA, which is consistent with the results of this study.

From a practical standpoint, wilting can reduce risk and increase DM levels during ensiling, which can be used to improve silage quality. The results in this study demonstrated that wilting is an effective measure.

In vitro GP is associated with the availability of NFC and fermentable carbohydrate content from the substrate. Evaluation of in vitro GP is also an indicator of rumen digestion, as in vitro GP affects the rumen passage rate and DMI. In this study, the IVDMD was higher in M+LP treatment at wilted than in other groups, which might be attributed to the decreased DM loss in silage after supplementation with molasses and L. plantarum. Thus, the IVDMD of the rumen was elevated. Additionally, the M+LP treatment at wilted provided and preserved the WSC and LA content in silage, which further benefited in vitro rumen degradation. In the present study, the potential GP, A and GP rates of additive-treated silages were higher than those of the control, probably due to the additive-treated reduced loss of nutrients after additive supplementation.

GP has been reported to be positively correlated with VFA production. The total VFA concentrations were significant difference between the control and the M+LP-treated silage at wilted at 48 h of fermentation. GP is a direct result of microbial degradation of feed and an indirect result of the buffering of acids generated by fermentation [27]. Moreover, additional gas is produced simply by contact between acids (VFAs and LA) produced in the silo and the buffer inoculum. Therefore, indirect GP from acids produced in the jars and during in vitro fermentation might explain the positive relationship between GP and VFA concentration in the present study.

The total VFA is one of the main energy sources for ruminant growth and development, which provides about 70% to 80% of the energy required by ruminants [28]. Although the composition and concentration of VFA in rumen were affected by feed concentrate/roughage ratio, processing method and additives, the highest content of VFA in rumen was the acetate [29]. The present study had similar results, which supports the previous studies. The VFA pattern of in vitro rumen fermentation was affected by wilted, the additives and the interactions between wilted and the additives in the present study. In vitro ruminal fermentation, the treatment after wilting and additives, affected the contents of AA, PA, and BA so that they were increased in comparison with the control. It could be explained that wilting and additives can improve the fermentation quality of sudangrass silage, especially the content of lactic acid, which can be degraded into acetate, propionate, and butyrate [30]. Thus, the VFA content of the rumen was elevated.