Materials and silage preparation
Experiment was conducted at the Hongyuan experimental base of Sichuan Academy of Grassland Sciences (44°53'N, 7°41'E, altitude 3500 m), which locates on the Qinghai Tibetan Plateau in P. R. China. Oat at milk stage was harvested as ensiling material at September 10, 2020. The harvested oat was chopped using a chopper, and randomly divided into thirty-six 10-kg piles to obtain nine replications per treatment. The treatments were as follows: control without additives (CK); LB, L. buchneri isolated from natural-fermented silage; nLP, low temperature tolerant L. plantarum isolated from natural fermented-reed canary grass silage [3]; pLP, phenyllactic acid-producing L. plantarum isolated from natural fermented-oat silage and preserved at China General Microbiological Cultural Collection Center (No.14117). To reduce negative effects from addition amounts, each LAB was applied at a rate of 106 cfu/g of fresh matter (FM). Each LAB was separately diluted in sterilized water, and sprayed uniformly onto the forage using a hand sprayer, which was constantly hand mixed and yield applying amount of 4 L inoculant-diluted solution/t of fresh forages. The same amount of water was added to the CK treatment. And, the treated forage from each pile was packed in a 20 L plastic silo equipped with a lid that only enabled gas release. The density of all silages was about 500 ± 25 kg/m3 on FM basis. Three of silage silos with same treatment were sampled after 30, 60 and 90 days of ensiling at ambient temperature of < 20 ℃. Samples from the fresh forage and the silages were subject to analysis of chemical composition, aerobic stability, microbial population and/or bacterial community.
Chemical analysis
Samples were dried at 65 ℃ for a constant weight to determine dry matter (DM) content, and then ground through 0.20 mm sieve for water soluble carbohydrate (WSC) analysis by the method of McDonald [24]. The DM loss was calculated by formula as follows: DM loss (%) = 100 × [1 - (pre-ensiled forage weight / silage weight at opening)].
Fresh sample of 20 g was mixed with 180 mL ultrapure water for 3 min in a stomacher blender. The pH of filtrate was determined by pH meter. Filtrate of about 10 mL was subjected to centrifugation (4500×g, 15 min, 4 °C), and the supernatant was analyzed for lactic acid, acetic acid, propionic acid and butyric acid using high performance liquid chromatography [25]. Identification and quantification of phenyllactic acid was determined by the method of Jung [5]. Ammonia nitrogen was determined by methods of Broderick & Kang [26].
Microbial population analysis
Microbial population on fresh samples was determined by the method of Cai [27]. Ten grams of each fresh sample were put into a sterile glass bottle, suspended in 90 mL of sterile water, and homogenized for 2 h in a laboratory blender (LB20ES, Shanghai Prime Science Co., Ltd, Shanghai, China). Serial dilutions were made. The number of LAB were counted on MRS agar (GCM188, Land Bridge Technology Co., Ltd, Beijing, China), incubated at 37 °C for 48 h. Yeasts were counted on malt extract agar with 1.5 mg/L Tetracycline (CM173, Land Bridge Technology Co., Ltd, Beijing, China), incubated at 30 °C for 48 h. Yeasts were distinguished from molds by colony appearance and observation of cell morphology.
Bacterial community analysis
The extraction of bacterial DNA from fresh sample was determined by the method of Li [25]. In brief, Phusion®High-Fidelity PCR Master Mix (New England Biolabs) was used to carry out PCR reactions, following the manufacturer’s instructions. The primer 515 F and 907 R was chosen to amplify the V4-V5 region of 16S rRNA gene. The PCR amplicons were then sequenced by using an Illumina MiSeq PE2500 platform at Novogene Company (Beijing, China). After sequencing, paired reads were merged using FLASH (V 1.2.7), and filtered by QIIME. The uparse method was employed to assign operational taxonomic units (OTUs) to the 16S rRNA at a cutoff level of 3% on the Usearch software platform (version 7.1). Based on OTUs results, the alpha indices were calculated with QIIME (Version 1.7.0) and displayed with R software (Version 2.15.3).
Aerobic stability
The aerobic stability of silage was measured by the method of Kung [13]. At silo opening, approximately 5.0 kg of silage from each silo was returned to clean buckets without packing under air-controlled temperature of 25 ± 0.5 °C. A thermocouple probe was placed in the geometric center of each silage mass, and temperatures were recorded by a data logger (YA204R, YADU Electronic Technology Co., Ltd, Shanghai, China) every 30 minutes. Silos were covered with 2 layers of cheesecloth and exposed to air. Aerobic stability was determined as the number of hours before the temperature of the silage mass increased 2 °C above ambient temperature of each silage mass.
Statistical analysis
Data was analyzed as a 4 × 3 factorial arrangement in a completely randomized design. The model included the fixed effects of additive (A), storage period (S), and their interaction (A × S). Data were analyzed using the Fit Model procedure of JMP (SAS Institute Inc., Cary, NC), and differences are reported as significant when P ≤ 0.05. Means were separated by Tukey’s test (P ≤ 0.05).