Introduction

Paper mulberry (Broussonetia papyrifera) has been considered as a pioneer source to cope with the shortage of feed and the development of livestock [1]. The crude protein content of whole-plant paper mulberry could be equivalent to alfalfa [2]. Besides, paper mulberry contains abundant biologically active compounds, amino acids, vitamins, digestible crude fiber and mineral elements [3, 4]. Due to these excellent features, paper mulberry can be a new type of high-quality protein feedstuff for ruminants.

Paper mulberry usually grows in the high humidity and rainy area in south of China, which is not conducive to production and storage of hay. Ensiling is a reliable way to effectively preserve fresh forage due to its good palatability and low nutrient loss [5, 6]. Ensiling has been proved to be an efficient method for preserving paper mulberry [7]. However, it has been found that high temperature and humidity conditions lead to the propagation of harmful microorganisms, which subsequently leads to the deterioration of silage. In addition, the features of high moisture and low water-soluble carbohydrate (WSC) concentration in fresh paper mulberry make it difficult to ensile, revealed by unpleasant smelling and low fermentation quality [8,9,10,11]. Therefore, it is critical to find the optimal way to improve the paper mulberry silage quality.

In China, more than 20 million tons of wheat bran (WB) are produced annually [12]. As a traditional animal feed, WB has a high dry matter content, rich in carbohydrates and other nutrients [13]. Adding WB is an effective method to reduce feed moisture content. During ensiling, WB could provide fermentation substrate for the growth of lactic acid bacteria (LAB) and accelerate the succession process to become dominant bacteria. The exogenous LAB has the characteristics of high utilization rate of sugar, fast acid production and pH reduction [14], which can speed up the ensiling process and inhibit the growth of harmful microorganisms [15]. In this regard, adding LAB inoculant and/or WSC substrates might be critical for enhancing paper mulberry silage quality.

The purpose of our research was to evaluate the application of LAB and WB on the fermentation quality of paper mulberry silage. We investigated the fermentation characteristics, chemical composition and microbial community of paper mulberry silage.

Materials and methods

Silage preparation

The experiment was conducted on 16 September, 2020 at Shandong hengda coal industry co. LTD (E116.87°, N35.88°) in Taian, Shandong, China. The industry is located in a monsoon climate of medium latitudes, with a mean annual temperature of 15.5 °C, relative humidity of 81% and annual rainfall of 1250–1440 mm.

Paper mulberry (Broussonetia papyrifera L.-Zhongke No.1), cultivated at the experimental station, was approximately 1.2 m in height and harvested from May to October in a year. Paper mulberry was harvested in the third cutting and grew approximately 4 months. The whole paper mulberry was obtained at the second round of cutting, leaving stubble of 15 cm. Before ensiling, the harvested paper mulberry was directly chopped into 1–2 cm pieces and then mixed manually.

The L. plantarum strain was isolated from paper mulberry silage in our laboratory, named of Forage Production and Processing Lab. The bacterial inoculants were added at a level of 106 cfu g−1 of fresh material (FM). Wheat bran (WB, applied at 30% rate of FM) was purchased from China Oil and Foodstuffs Corporation Organization, Zhengzhou Haijia Food Co. Ltd.

A single factor completely randomized design was used in this experiment. The silage treatments were as followed: (i) no additives (control group, CK); (ii) application of L. plantarum (LP silage); (iii) application of wheat bran additives (WB silage); (iv) combination of L. plantarum and wheat bran (LP + WB silage). Then, 500 g of materials with different additives were mixed homogenously and packed into polyethylene bags (30 cm × 20 cm). The silage bags were vacuum-sealed and stored at room temperature (20–30 °C) for 60d. After 60d of ensiling, three bags for each treatment were opened to analyze the fermentation characteristics, chemical composition and bacterial communities [16].

Fermentation quality, microbial population and chemical composition analysis

Twenty grams of samples were homogenized in 180 mL distilled water and stored at 4 °C for at least 4 h and then filtered by qualitative filter paper [17]. The pH, lactic acid (LA), acetic acid (AA), propionic acid (PA), butyric acid (BA) and NH3-N were immediately measured with filtrate. The pH was measured with a glass pH meter (PHS-3C, INESA Scientific Instrument, Shanghai, China). The NH3-N content was accessed by the phenol–sodium hypochlorite method, as described by Guan et al. [18]. The contents of organic acid (LA, AA, PA and BA) were measured by high-performance liquid chromatography method (column: Shodex RS Pak KC-811; detector: DAD, 210 nm, SPD-20A; Shimadzu Co., Ltd., Kyoto, Japan; 3 mmol L−1 HClO4) [19]. The microbial count of all samples was determined by the plate culture method. The filtrate was serially diluted from 10–1 to 10–6 and each diluted suspension was spread. The populations of LAB, coliform bacteria, yeasts and molds were determined on de Man Rogosa Sharpe (MRS) agar (AOBOX Biotechnology Co. Ltd., Beijing China), Blue Light (BL) agar and Rose Bengal (RB) agar, respectively, after incubation at 30 °C for 48–72 h [20].

On days 0 and 60, a total of 42 fresh and silage of paper mulberry samples were dried to determine the DM concentration in an air oven at 65 °C for at least 48 h. All samples were milled to pass through a 1.0 mm size of the sieve for chemical composition analysis. Ether extract (EE) were estimate by AOAC [21]. The content of total nitrogen (TN) was determined by the Kjeldahl procedure and the crude protein (CP) content was calculated by multiplying TN by 6.25 [21]. Neutral detergent fiber (NDF) and acid detergent fiber (ADF) were measured by the method of Van Soest et al. [22]. The content of WSC was analyzed by the methodology of Murphy et al.[23].

Microbial diversity analysis

The method of DNA extraction of each sample was according to Liu et al. [24] with slight modifications. After monitored the quality and concentration of DNA, the full-length 16S ribosomal RNA (rRNA) gene was amplified with the polymerase chain reaction (PCR), with the primer 27F (5′-AGRGTTTGATYNTGGCTCAG-3′) and the reverse primer 1492R (5′-TASGGHTACCTTGTTASGACTT-3′). The PCRs program was: 95 °C for 2 min, 25 cycles at 98 °C for 10 s, 55 °C for 30 s, 72 °C for 90 s and a final extension at 72 °C for 2 min [25].

After extraction, purification and qualification, the PCR products were sequenced on PacBio Sequel platform (Biomarker Technologies, Beijing, China). The raw sequences were selected by Single Molecule Real Time (SMRT) according to Li et al. [26]. We clustered the unique sequence set into operational taxonomic units (OTUs) based on a 0.97 threshold identity using UCLUST [27]. The high-quality sequences were annotated based on Silva132 database [28].

Statistical analysis

Analysis of variance (ANOVA) was performed using generalized linear modeling in the Statistical Package for the Social Sciences (SPSS version 21.0, SPSS Inc., Chicago, IL, USA) to examine the chemical composition, fermentation quality and microbial counts data among samples. Duncan’s multiple range methods were used to determine the significant difference between means with three replicates per treatment. The significance was declared at P < 0.05.

After the OTUs analysis, QIIME software (version 2) [29] was operate to calculated alpha diversity indices. Principal component analysis (PCA) was performed to access the differences in community composition and structure of microbiota, basing on the beta-diversity analysis. Heatmap analysis was operated to analysis the relationship between pH, AA, NH3-N, CP, WSC, LA. LA/AA and microbial community of paper mulberry silage using R-based statistics tool. Linear discriminant analysis effect size (LEfSe) analyses were conducted using a free online platform [30].

Accession number

The raw sequencing data including all samples have been deposited into the NCBI Sequence Read Archive (SRA) database under the Accession Number of PRJNA972621.

Results

Chemical and microbial composition of raw materials

The chemical characteristics and microbial composition of fresh paper mulberry and wheat bran are listed in Table 1. The DM contents of paper mulberry and wheat bran were 259.29 and 917.70 g kg−1 FM, respectively. In addition, the CP, NDF and ADF of raw paper mulberry were 183.84 g kg−1 DM, 442.52 g kg−1 DM and 251.03 g kg−1 DM, respectively. Paper mulberry had low WSC (25.79 g kg−1 DM) and LAB (5.09 log10 cfu g−1 FM).

Table 1 Chemical composition and microbial population of raw materials prior to ensiling
Full size table

Chemical composition of paper mulberry silage

The chemical composition of paper mulberry silage is shown in Table 2. There was no difference in NDF contents between control and treatments (P > 0.05). The WB and LP + WB treatments significantly increased (P < 0.05) DM and WSC contents and decreased (P < 0.05) ADF and EE contents compared with control and LP treatments. The CP content of treatments was significantly higher than control while the CP in LP and WB treatments was highest (Table 2).

Table 2 Chemical composition of paper mulberry silage
Full size table
Fig. 1
figure 1

The OTU information and Shannon index of paper mulberry added with lactic acid bacteria and wheat bran after 60 days of ensiling. a OTU; b Shannon index. CK, no additives; LP, Lactobacillus plantarum additives; WB, 30% wheat bran additives; LP + WB: Lactobacillus plantarum additives plus 30% wheat bran; p < 0.01 means obviously difference significant, 0.01 < p < 0.05 means difference significant; p > 0.05 means no significance

Full size image

Dynamic changes of before and after ensiling

The dynamic change of chemical composition paper mulberry silage at the before and after ensiling process is shown in Table 3. The dynamic change of ADF content before and after ensiling in WB and LP + WB treatments was significantly higher (P < 0.05) than control and LP treatments. The WB and LP + WB treatments had lower (P < 0.05) DM, CP and WSC losses compared with control. No difference in NDF content was observed among treatments (P > 0.05).

Table 3 Dynamic changes (%) of chemical composition of paper mulberry silage at the before and after ensiling processes
Full size table

Fermentation characteristics and microbial population of silage

As shown in Table 4, the treatments had a significant influence on pH, NH3-N, LA, AA, PA, BA contents and LAB counts (P < 0.05). The WB and LP + WB treatments significantly increased (P < 0.05) LA content and decreased pH, the contents of NH3-N, AA, BA and the amount of LAB compared with control and LP treatments. There was no difference in yeast, molds and coliform between control and treatments (P > 0.05).

Table 4 Fermentation quality and microbial population of paper mulberry silage
Full size table

Bacterial community indices of paper mulberry silage

The coverage values of all samples were beyond 0.995, indicating the credible analysis for sequencing depth of the microbial composition of paper mulberry silage. A total of 155 OTUs were clustered at 97% sequenced similarity. Shannon index ranged from 1.11 to 2.43. Compared with control, inoculations with treatments significantly increased (P < 0.05) Shannon, while the highest was 2.43 of LP + WB treatment (Fig.1).

As shown in Fig. 2, the variance of bacterial community was observed by principal component analysis (PCA) on the OTUs levels. The bacterial community in control were clearly separated from the other groups, which suggested that bacterial community changed significantly after the addition of LP and WB and combinations of LP and WB.

Fig. 2
figure 2

The principal component analysis (PCA) by operational taxonomic units (OTUs) for paper mulberry silage with different inoculant treatments. CK, no additives; LP, Lactobacillus plantarum additives; WB, 30% wheat bran additives; LP + WB: Lactobacillus plantarum additives plus 30% wheat bran

Full size image

The bacterial community at genus (A) and species (B) levels are shown in Fig. 3, using a threshold of > 0.1% of total. At the genus levels, Lactobacillus was the dominant genus in all silages, and the proportion of Lactobacillus in LP, WB and LP + WB were 84.45%, 89.27% and 86.10%, respectively, which were lower than control (P < 0.05). In addition, the proportion of Weissella in LP was higher than other treatments.

Fig. 3
figure 3

Bacterial community and relative abundances by genus (a) and species (b) for paper mulberry silage after ensiling. CK, no additives; LP, Lactobacillus plantarum additives; WB, 30% wheat bran additives; LP + WB: Lactobacillus plantarum additives plus 30% wheat bran

Full size image

At the species levels, L. farciminis and L. ginsenosidimutans were the dominant species in control. In LP treatment, L. farciminis, and W. cibaria were the representative bacterium. Interestingly, treatments with wheat bran shifted the dominant bacteria to L. brevis and L. farciminis. The relative abundance of L. brevis and L. farciminis in WB treatment were 57.55% and 30.03%, while the relative abundance in LP + WB treatment were 47.75% and 36.97%, respectively. L. brevis can be used as a biomarker in WB treatment by the LEfSe algorithm (Fig. 4). These findings clearly indicated lactic acid bacteria and wheat bran influenced bacterial community significantly.

Fig. 4
figure 4

Linear discriminant analysis effect analysis of bacterial community (LDA > 3.0) at species level in paper mulberry silages compared by different additives. CK, no additives; LP, Lactobacillus plantarum additives; WB, 30% wheat bran additives; LP + WB: Lactobacillus plantarum additives plus 30% wheat bran

Full size image

Correlation between bacterial community and fermentation products

Lactobacillus brevis was negatively correlated to pH (r = − 0.91), AA content (r = − 0.86) and NH3-N (r = − 0.69) (P < 0.05), but positively correlated to CP content (r = 0.68), LA content (r = 0.89), WSC content (r = 0.64) and ratio of LA and AA (r = 0.86), respectively (P < 0.05). In contract, L. farciminis, L. ginsenosidimutans, L. maninotivorans, and L. dextrinicus were positively correlated to pH, AA content and AN/TN, but were negatively correlated to LA, WSC content and ratio of LA and AA (P < 0.05).

Discussion

Previous studies have confirmed that paper mulberry is difficult to be well ensiled due to its low WSC and high moisture content [31]. Additives, such as LAB and bran, are often widely utilized to effectively improve the silage fermentation quality [32,33,34,35]. WB and LP + WB treatments contained lower AA and NH3-N contents (P < 0.05), indicating less protein degradation in silage with WB and LP + WB. In addition, L. brevis, with high acid production capacity, is the dominant in WB and LP + WB treatment. Therefore, these results verified that ensiling with WB and LP + WB could enhance paper mulberry silage quality (Table 4, Fig. 3B and Fig. 5).

Fig. 5
figure 5

Dendro-heatmap visualized interrelationship between bacterial community and fermentation products at species levels in paper mulberry silage. Corresponding value of middle heat map is Spearman correlation coefficient r, which ranges between 1 and − 1, r < 0 indicates a negative correlation (blue) and r > 0 indicates a positive correlation (red). * significance at P < 0.05; ** significance at P < 0.01; *** significance at P < 0.001. CK, no additives; LP, Lactobacillus plantarum additives; WB, 30% wheat bran additives; LP + WB: Lactobacillus plantarum additives plus 30% wheat bran

Full size image

In this experiment, the CP content of fresh paper mulberry was much lower than that determined by Dong et al. [36], but much higher than that in our previous study [37]. ADF and NDF contents of fresh paper mulberry were lower than that reported by Li et al. [38], but comparable to those reported by Li et al. [19]. These discrepancies might be due to the factors such as growth period, harvest time, fertilization and climate conditions [39, 40]. The DM content of paper mulberry was 259.29 g kg−1 FM, which was not conducive to well-quality fermentation [41]. Higher WSC (> 50 g kg−1 DM) and LAB (> 5 log10 cfu g−1 FM) of raw materials are necessary to obtain well-preserved silage [42]. In this study, WSC content and LAB number of fresh paper mulberry were below the minimal requirement, which was not conducive to silage fermentation [10].

All treatments showed higher CP content and lower CP loss compared with control, which could be due to LAB and WB limiting the growth of undesirable microbes [43]. The WB and LP + WB treatments had higher ADF content changes rate than control and LP treatment, which could be due to acidolysis during ensiling process [44]. The WB and LP + WB treatments showed lower ADF content than control and LP treatments, which was due to lower ADF in wheat bran. After 60 days of ensiling, the WSC contents in silages were lower than that in fresh paper mulberry, which was possible due to LAB converting WSC into organic acids and ethanol during ensiling process. The lower WSC content changes in WB and LP + WB treatment than that control and LP treatment could be due to the different fermentation pattern, but the exact reason needs to be further classified in the future [45].

The low pH value around 4.2, a critical criterion for well-preserved silage, was important to inhibit harmful Clostridia [46]. In this experiment, the pH of WB and LP + WB treatment was around 4.2 while the pH of LP treatment was beyond 4.2, indicating that the effect of wheat bran was better than L. plantarum in regulating fermentation quality of paper mulberry silage. This might be because high moisture and low WSC are not conducive to propagation and role of LAB [47]. Generally, lactic acid bacteria are often employed to convert WSC into LA to reduce pH and inhibit harmful microorganisms [48]. In our study, silage treated with LP showed lower LA content and higher NH3-N, AA and BA content, indicating that low WSC in paper mulberry cannot provide a good growth environment for LAB. As a typical indicator of proteolysis [49], silage treated with WB and LP + WB contained lower NH3-N content than LP treatment, indicating less proteolysis occurred with WB and LP + WB. The above results indicated that it is important to regulate the moisture content promote fermentation quality of paper mulberry.

Lactobacillus and Weissella are beneficial to improve silage quality [50]. The dominant genus involved in all treatments were Lactobacillus and Weissella. Lactobacillus is known as a predominant microbial species in high-quality silage [51], which is consistent with our experimental results. LAB plays an important role in producing LA and reducing pH to inhibit harmful bacteria, while as a hetero-fermentative bacteria with weak acid-producing ability [52], Weissella is difficult to become the dominant genus in silage. Interestingly, there was a sharp increase in the relative abundance of Weissella species only in LP treatment. Similar result was also reported by Guo et al. [30]. The high abundance of Weissella in LP treatment would explain its relatively poor fermentation quality compared with WB and LP + WB treatments.

At species level, the addition of L. plantarum did not significantly increase the relative abundance of L. plantarum but increased the relative abundance of L. farciminis and W. cibaria in LP treatment, which might be due to incompatibility between the inoculants and plant materials [53]. L. brevis and L. farciminis were the dominant species in WB and LP + WB treatments while L. farciminis was the dominant species in control and LP treatment. In addition, L. ginsenosidimutans and W. cibaria were the secondary dominant species in control and LP treatment, respectively. Generally, desirable LAB, such as L. plantarum, L. farciminis, L. buchneri and L. brevis, play an essential role in increasing LA content and reducing pH value [54]. In addition, the homo-fermentative L. plantarum and L. farciminis were overtaken by hetero-fermentative L. buchneri and L. brevis in later stage of high-quality silage [55, 56]. At the later of stage of fermentation, lower WSC content was insufficient to support the fermentation of homo-fermentative LAB while hetero-fermentative LAB use lactic acid for growth, resulting in fermentation mode changes [57]. In addition, the anaerobic conditions promoted the growth of LAB that produced organic acids inhibiting other bacteria [58, 59]. However, the role of L. farciminis in silage fermentation remains unclear. In this study, L. brevis was positively relative to LA and negatively to pH, indicating that the production of LA was largely attributable to L. brevis. Similar results have been reported by a previous study [60]. In this study, L. ginsenosidimutans in control could be isolated from fermented foods, such as Korean fermented pickle, and used as probiotic strains to biotransform ginsenosides and improve the taste of functional foods [61]. However, the role of L. ginsenosidimutans in silage needs further study.

Conclusion

Paper mulberry silages with the addition of inoculant had lower NDF, ADF and WSC content than fresh paper mulberry. The addition of LP had no impact in improving the fermentation quality of paper mulberry silage. However, wheat bran (WB) and its combination with L. plantarum (LP + WB) additions could reduce pH, NH3-N and increase LA content. The application of WB and LP + WB shifted the dominant bacteria species to L. brevis. Thus, the addition of wheat bran or combined with lactic acid bacteria were effective ways to enhance paper mulberry silage fermentation.