Addition of Organic Acids and Lactobacillus acidophilus to the Leguminous Forage Chamaecrista rotundifolia Improved the Quality and Decreased Harmful Bacteria of the Silage

Simple Summary The leguminous forage Chamaecrista rotundifolia contains a high crude protein content and can be an important feed source for livestock. Forage is preserved mainly as silage, and microbial fermentation is key to high-quality silage. The fermentation quality of silage depends on the microbial community structure and the metabolites produced during silage fermentation. The current study provides evidence that the addition of malic or citric acid and/or Lactobacillus acidophilus (L) improves the quality of silage and inhibits the growth of harmful microorganisms. Abstract This study aimed to investigate the effects of citric acid, malic acid, and Lactobacillus acidophilus (L) on fermentation parameters and the microbial community of leguminous Chamaecrista rotundifolia silage. Fresh C. rotundifolia was treated without any additive (CK), or with L (106 CFU/g fresh weight), different levels (0.1, 0.3, 0.5, and 1% fresh weight) of organic acid (malic or citric acid), and the combinations of L and the different levels of organic acids for 30, 45, and 60 days of ensiling. The effects of malic acid and citric acid were similar during the ensiling process. Treatment with either citric or malic acid and also when combined with L inhibited crude protein degradation, lowered pH and ammonia nitrogen, and increased lactic acid concentration and dry matter content (p < 0.05). The neutral detergent fiber and acid detergent fiber increased initially and then decreased with fermentation time in all treatments (p < 0.05). Increasing the level of organic acid positively affected the chemical composition of C. rotundifolia silage. In addition, the addition of 1% organic acid increased the relative abundance of Lactobacillus, while the relative abundances of Clostridium and Enterobacter decreased at 60 days (p < 0.05). Moreover, both organic acids and combined additives increased (p < 0.05) the relative abundance of Cyanobacteria at 60 days of fermentation. We concluded that adding malic acid, citric acid, and L combined with an organic acid could improve the quality of C. rotundifolia silage and increase the relative abundance of beneficial bacteria. The addition of organic acid at a level of 1% was the most effective.


Introduction
Chamaecrista rotundifolia, an important leguminous forage grass, is sown for livestock fodder in tropical and subtropical regions due to its nutritive value, nitrogen fixation activity, and resistance to diseases and pests [1]. It was reported that cattle grazing native

Silage Preparation
In September 2020, C. rotundifolia was harvested from Wanan Farm (Zhangpu County, Zhangzhou, China) at 8-10 cm above ground. The forage was spread out evenly indoors and dried until the dry matter (DM) reached 25.0% fresh weight (FW), which took approximately 4 h.
C. rotundifolia was cut into 2-3 cm lengths with a grass cutter, and treated with different additives as follows: distilled water (CK), Lactobacillus acidophilus (L, provided by Fujian Academy of Agricultural Sciences, Fuijan, China) at 1 × 10 6 cfu/g fresh weight, different levels (0.1, 0.3, 0.5, and 1% FW) of malic acid (M1-M4) and citric acid (C1-C4) (analytical reagent, Shanghai Aladdin Biochemical Technology, Shanghai, China) and L combined with the different levels of malic acid (ML1-ML4) or citric acid (CL1-CL4). The additives, based on the FW of the wilted raw material, were dissolved in 10 mL of sterile water and sprayed evenly on the surface of the C. rotundifolia with a small spray bottle (an equal amount of distilled water was sprayed on CK). Then the C. rotundifolia was packed into vacuum sealed bags (24 cm × 35 cm) of 400 g each. The bags were sealed with an automatic vacuum Animals 2022, 12, 2260 3 of 17 compressor, and the C. rotundifolia was fermented anaerobically at a room temperature of 25 ± 3 • C. At 30,45, and 60 days of ensiling, three bags were randomly selected from each treatment to determine nutritional components and fermentation parameters. In total, there were 9 bags (3 fermentation times × 3 replicates) for each treatment.

Sampling and Chemical Analysis
After 30, 45, and 60 days of fermentation, C. rotundifolia was removed from the bags, dried at 65 • C for 48 h for DM determination, and ground into powder to determine the chemical composition. Briefly, WSC content was measured by the anthrone-sulfuric acid colorimetric method [15]; total nitrogen (TN) content was measured by an automatic nitrogen determinator (KDN-103F, Shanghai Fiber Inspection Instrument Co., Ltd., Shanghai, China); crude protein (CP) content was measured by TN × 6.25; acid detergent fiber (ADF) and neutral detergent fiber (NDF) according to Van Soest et al. [16]; and ash content according to Monti et al. [17].
To determine fermentation parameters, 10 g samples were homogenized with 90 mL ultrapure water, sealed with sealing film, refrigerated at 4 • C for 24 h, and then filtered through four layers of medical gauze. The pH was measured by a pH meter (FiveEasy Plus, Mettler Toledo International Trading Co., Ltd., Shanghai, China). Ammonia nitrogen (NH 3 -N) content was determined by the phenol-sodium hypochlorite colorimetric method [18], and lactic, acetic, and propionic acids were determined following Wang et al. [19], using a high-performance liquid chromatograph (HPLC, WondaSil C18 Superb column, Shimadzu, Kyoto, Japan; SPD 210 nm; flow rate 1 mL/min; temperature 30 • C).

DNA Extraction, Amplicon Library Preparation, and Sequencing
After 60 d of fermentation, a portion of each silage sample was snap-frozen (−80 • C) for further analysis. To analyze the microbial diversity of 60 d silage, CK, L, and organic acid treatments with higher fermentation quality (C4, CL4, M4, ML4) were selected for high-throughput sequencing. Total genome DNA was extracted using the CTAB/SDS method, and DNA concentration and purity were monitored on 1 % agarose gels. The PCR was performed using diluted genomic DNA as a template, specific primers with Barcode, Phusion ® High-Fidelity PCR Master Mix with GC Buffer (New England Biolabs (Beijing) LTD., Beijing, China), and high performance and fidelity enzymes based on the selection of sequencing regions to ensure amplification efficiency and accuracy. The hypervariable regions V3-V4 of the bacterial 16S rDNA were obtained using 315F (CCTAYGGGRBGCASCAG) and 806R (GGACTACNNGGGTATCTAAT) as primers. The PCR products were detected by electrophoresis using 2% concentration of agarose gel, purified by magnetic beads, quantified by enzyme labeling, mixed thoroughly in equal amounts according to the concentration of PCR products, and the products were recovered for the target bands using a gel recovery kit (Qiagen, Hilden, Germany). The PCR amplification products were identified by agarose gel electrophoresis, and a MiSeq library was constructed and sequenced using TruSeq ® DNA PCR-Free. The library quality was assessed on the Qubit@2.0 Fluorometer (Invitrogen, Carlsbad, CA, USA) and the Agilent Bioanalyzer 2100 System (Agilent, Santa Clara, CA, USA). After the library was qualified, it was double-ended sequenced based on the Illumina NovaSeq sequencing platform to obtain a complete microbial community.

Sequence Processing and Analysis
The reads of each sample were spliced using FLASH after truncating the barcode and primer sequences to obtain raw tags [20]. The raw tags were processed according to the QI-IME quality control process to obtain high-quality clean tags [21]. The tag sequences were compared with the species annotation database (https://github.com/torognes/vsearch/ accessed on 6 January 2021) to detect chimeric sequences, and the final effective tags were obtained by removing the chimeric sequences. Sequences were clustered into operational taxonomic units (OTUs) with 97% consistency by default for all effective tags using Uparse [22]. MUSCLE was used for a fast multiple sequence comparison to obtain the phylogenetic relationships of all OTU-representative sequences [23]. Finally, the data of each sample were homogenized, the least amount of data in the sample was used as the criterion for homogenization, and the subsequent alpha and beta diversity analyses were based on the homogenized data. QIIME (version 1.9.1) was used to calculate Observedspecies, Chao1 richness estimate, Shannon diversity index, Simpson diversity index, ACE richness estimate, and Goods-coverage [24]. The R (version 2.15.3) for Windows was used to determine alpha diversity index inter-group variance, which was done separately with parametric and non-parametric tests: the T-test and Wilcox-test for two groups and the Tukey-test and Wilcox-test for more than two groups (Agricolae Package). To analyze Beta diversity, ImageGP was used for analyzing principal coordinate analysis (PCoA) based on a Bray-Curtis dissimilarity matrix [25]. Spearman's correlation analysis tested the correlations between the main genera and silage fermentation quality.

Sampling and Chemical Analysis
The data were subjected to a two-way analysis of variance (SPSS software, version 26.0, Chicago, IL, USA). The fixed effects were the ensiling days, additive treatments, and the interactions between ensiling days and additive treatments. Differences among means were compared by Duncan's multiple range test, and p < 0.05 was accepted as significant.

Effect of Additives on the Nutrient Composition of Chamaecrista Rotundifolia Silage
Low DM and WSC contents tend to cause spoilage of silage, and organic acids and LAB are often added to limit spoilage and improve silage quality. After 30 days of fermentation, the DM content of C. rotundifolia silage was higher with any of the additives than with CK silage (p < 0.05). The DM and WSC contents increased with increasing levels of organic acids; however, fermentation time had no significant effect on the DM content of the silage. The CP contents in the L, M2-M4, ML2-ML4, C2-C4, and CL2-CL4 silages were greater than in the CK silage (p < 0.05) at all fermentation times, with CL4 the greatest. The WSC content of M4 silage was higher (p < 0.05), and that of L silage was lower (p < 0.05) than that of CK silage. The CP and DM contents of the silages with all additives were higher than those in the CK silage. Similar changes were reported by Tao et al. [26]. Citric acid and malic acid, as intermediates in the tricarboxylic acid cycle, rapidly lower the pH, thereby inhibiting the growth and activity of harmful microbiota, which reduces the loss of DM and the degradation of CP [27]. Under anaerobic conditions, a low pH suppresses unwanted microbial activity if LAB ferments WSC into sufficient lactic acid [28], and can then preserve most of the CP in silage [29]. M4 had a greater concentration of WSC at 30 days of fermentation, and the lactic acid content was greater than that of CK, indicating that WSC content can promote the production of lactic acid during fermentation [30]. In the present study, both LAB and organic acid additions increased the lactic acid concentration and reduced the concentrations of acetic and propionic acids.
The NDF content increased initially and then decreased with time of fermentation, and after 60 days, the NDF content was lower than at 30 and 45 days (p < 0.05) in all treatments. The NDF content at 30 days of fermentation was lowest in the M4 silage (p < 0.05), and at 60 days was lowest (p < 0.05) in the L silage among treatments. The ADF contents of the L, M4, ML4, CL2-CL4, and C2-C4 silages were generally lower than those of CK silage (p < 0.05) at all fermentation times. After 60 days of fermentation, the ADF content in each treatment decreased significantly, with 1% malic acid being the lowest. At 45 days of fermentation, ML4 and CL4 treatments had lower (p < 0.05) ash contents than CK treatment, while at 60 days of fermentation, the ash content in all silages was lower than in CK (p < 0.05) (Tables 1 and 2). The NDF and ADF contents of forages affect the digestibility of feed in ruminants. In this study, the L, M4, ML4, C4, and CL4 silages reduced the NDF and ADF contents in the silage due, at least in part, to the hydrolysis of digestible parts of cell walls by organic acids during fermentation [31]. In addition, the NDF and ADF contents in the LAB and LAB with organic acid silages were lower after 60 days than after 30 and 45 days of fermentation, which may be due to the continued hydrolysis of structural carbohydrates [32]. In general, the acid hydrolysis of structural carbohydrates is accompanied by the release of WSC, which was evident in this study [29]. The L in this study belonged to homotypic fermentative LAB, which uses more WSC to ferment into lactic acid than heterofermentative LAB. Therefore, the addition of a combination of L and citric acid or malic acid was effective in reducing fiber content and thereby improving the digestion and utilization of C. rotundifolia silage by herbivores. The level of 1% citric or malic acid gave the best results.

Effect of Additives on the Fermentation Parameters of Chamaecrista rotundifolia
Acid production and pH reduction during fermentation are spontaneous results of microbial activity and are used as important indicators in evaluating the quality of silage [33]. Grass silage pH is generally below 4.2, which is designated as excellent quality, and legume silage with high CP content has a higher pH, generally between 4.0 and 5.6 [34]. In the present study, the pH of the silage was lower (p < 0.05) with the addition of organic acids or LAB than the CK silage (Tables 3 and 4) and decreased with an increase in the level of citric acid or malic acid (p < 0.05). The lowest pH occurred in silage with 1% citric or malic acid and with these acids plus LAB. In addition to the lowered pH, the NH 3 -N:TN ratio was below 0.1 with all additives during the 60 days of fermentation, indicating good silage quality. However, overall, fermentation time did not affect the pH or the NH 3 -N:TN ratio. The rapid reduction of NH 3 -N:TN ratios in the C4 and M4 silages may be due to the accumulation of lactic acid and the rapid acidification of silage. At 30 and 60 days of fermentation, all silages with additives had greater lactic acid concentrations than the CK silage. Similarly, Li et al. [35] reported that the addition of organic acids rapidly reduced the pH of alfalfa silage, thus inhibiting the activity of undesirable microorganisms and protease activity and resulting in lower non-protein nitrogen and NH 3 -N contents in alfalfa silage. The NH 3 -N:TN ratio decreased with increasing citric acid addition at 30 and 60 days of fermentation (p < 0.05), and at 45 days of fermentation, the NH 3 -N:TN ratio was lower, except for M1, in all silages than the CK silage (p < 0.05). The acetic acid concentration was lower in all silages than in CK, but only C3 remained lower throughout (p < 0.05). Citric acid and malic acid displayed similar effects in reducing acetic acid concentrations. Propionic acid concentrations in all silages were lower (p < 0.05) than in the CK silage at 30 and 60 ensiling days.
The addition of 0.5% and 1% citric or malic acids during 60 days of fermentation reduced the pH, NH 3 -N:TN ratios and acetic acid concentration of C. rotundifolia silage, increased the lactic acid content, reduced DM loss, and improved the quality of C. rotundifolia silage. Similarly, Ke et al. [36] reported that with the addition of L. plantarum, malic acid and citric acid, pH, and the NH 3 -N:TN ratio of alfalfa silage were reduced when compared to the control silage. The addition of two organic acids further improved silage quality, but Lactobacillus plantarum and an organic acid displayed a superimposed effect. In the present study, L and L combined with an organic acid produced small amounts of lactic acid, with no effect on days of fermentation, which is consistent with the study of Ni et al. [37]. Moreover, Muck et al. [34] reported that low levels of WSC in raw materials may reduce the improvement of silage quality by the addition of LAB. This could explain why, in the current study, silages with LAB alone or when combined with an organic acid did not produce substantial amounts of lactic acid.       SEM, standard error of means; NH 3 -N:TN, ammoniacal nitrogen as a percentage of total nitrogen. Chamaecrista rotundifolia was treated with the following: distilled water (CK), Lactobacillus acidophilus (L), different levels (0.1, 0.3, 0.5, and 1%) of malic acid (ML1-ML4) and citric acid (CL1-CL4). D, ensilage days effect; T, treatment effect; D × T, the interaction between ensilage days and treatment. Means of additive treatment within a row (a-h) followed by different lowercase superscripts differ (p < 0.05). Means of ensiling time treatment within a column (A-C), followed by different uppercase superscripts, differ (p < 0.05).

Microbial Diversity in Chamaecrista rotundifolia Silage
The sequences were grouped in OTUs to an agreement level of 97%, and the degree of overlap of bacterial OTUs in each treatment is presented in Figure 1A. There were 1305 OTUs after 60 days of fermentation, with 138 common OTUs and 36, 83, 75, 104, 171, and 152 OTUs unique to CK, L, M4, C4, ML4, and CL4, respectively. Rarefaction curves were used to estimate species richness as a function of the sampling results. In the present study, the curves plateaued, indicating that the sequencings were saturated and that all microorganisms were identified ( Figure 1B).

269
The sequences were grouped in OTUs to an agreement level of 97%, and the degree 270 of overlap of bacterial OTUs in each treatment is presented in Figure 1A. There were 1305 271 OTUs after 60 days of fermentation, with 138 common OTUs and 36, 83, 75, 104, 171, and 272 152 OTUs unique to CK, L, M4, C4, ML4, and CL4, respectively. Rarefaction curves were 273 used to estimate species richness as a function of the sampling results. In the present 274 study, the curves plateaued, indicating that the sequencings were saturated and that all 275 microorganisms were identified ( Figure 1B). 276 277 Figure 1. Petal diagram illustrating the degree of overlap of bacterial OTUs among the 6 groups. 278 Each petal represents a group, the middle CORE number represents the number of OTUs common 279 to all groups, and the number on the petal represents the number of OTUs specific to that group 280 (A). Rarefaction curves of the observed species index. The horizontal coordinate is the number of 281 sequencing strips selected randomly from a sample, and the vertical coordinate is the number of 282 OTUs that can be constructed based on the number of sequencing strips to reflect the sequencing 283 depth. Different samples are represented by different color curves (B). 284 The alpha diversity of the microbial community in each silage treatment is presented 285 in Figure 2. Chao1, an estimator of species richness based on the number of rare species, 286 and ACE were used to estimate the number of OTUs in the community [38]. In the current 287 study, the Shannon, Chao1, and ACE indices were lower in the L than in the CK silage, 288 indicating that microbial diversity was reduced in the L treatment. Yang et al. [39] re-289 ported that LAB proliferated rapidly as the dominant bacteria after inoculation. In addi-290 tion, a reduced pH during fermentation inhibits the activity of bacterial groups, which 291 leads to a reduction in microbial diversity. Compared with the CK silage, the Shannon 292 index was higher in the ML4 silage, the Chao1 and ACE indices were higher in the C4 and 293 ML4 silages, and the ACE index was higher in the CL4 silage. This indicates that the ad-294 dition of an organic acid and mixed additives could increase the bacterial community 295 richness, which was similar to the results of Wang et al. [40] The ACE and Chao1 indices 296 were higher (p < 0.05) in M4, C4, ML4, and CL4 silages than in the CK silage, and the 297 Chao1 index was higher (p < 0.05) in the C4 and ML4 silages than in the CK silage. The 298 The alpha diversity of the microbial community in each silage treatment is presented in Figure 2. Chao1, an estimator of species richness based on the number of rare species, and ACE were used to estimate the number of OTUs in the community [38]. In the current study, the Shannon, Chao1, and ACE indices were lower in the L than in the CK silage, indicating that microbial diversity was reduced in the L treatment. Yang et al. [39] reported that LAB proliferated rapidly as the dominant bacteria after inoculation. In addition, a reduced pH during fermentation inhibits the activity of bacterial groups, which leads to a reduction in microbial diversity. Compared with the CK silage, the Shannon index was higher in the ML4 silage, the Chao1 and ACE indices were higher in the C4 and ML4 silages, and the ACE index was higher in the CL4 silage. This indicates that the addition of an organic acid and mixed additives could increase the bacterial community richness, which was similar to the results of Wang et al. [40]. The ACE and Chao1 indices were higher (p < 0.05) in M4, C4, ML4, and CL4 silages than in the CK silage, and the Chao1 index was higher (p < 0.05) in the C4 and ML4 silages than in the CK silage. The Shannon index was higher in the ML4 silage than the CK silage (p < 0.01) and was lower in the L silage than the C4 and ML4 silages (p < 0.05); the Simpson index was higher in the ML4 than the L silage (p < 0.05), and was higher in the ML4 (p < 0.01) and CK (p < 0.05) silages than the CL4 silage.
Principal Coordinates Analysis (PCoA) determines and visualizes the similarities or dissimilarities of bacterial communities [41]. In this study, the PCoA plot revealed a separation and difference in bacterial communities in each ensiled treatment, indicating that the microbiota was altered with different additives (Figure 3). This difference in silage quality may be due to changes in the microbial community [42]. Therefore, based on alpha and beta diversity analyses, we concluded that malic acid, citric acid, and LAB treatments could affect the microbial diversity and community structure of C. rotundifolia silage.
Shannon index was higher in the ML4 silage than the CK silage (p < 0.01) and was lower 299 in the L silage than the C4 and ML4 silages (p < 0.05); the Simpson index was higher in the 300 ML4 than the L silage (p < 0.05), and was higher in the ML4 (p < 0.01) and CK (p < 0.05) 301 silages than the CL4 silage. Principal Coordinates Analysis (PCoA) determines and visualizes the similarities or 310 dissimilarities of bacterial communities [41]. In this study, the PCoA plot revealed a sep-311 aration and difference in bacterial communities in each ensiled treatment, indicating that 312 the microbiota was altered with different additives (Figure 3). This difference in silage 313 quality may be due to changes in the microbial community [42]. Therefore, based on alpha 314 and beta diversity analyses, we concluded that malic acid, citric acid, and LAB treatments 315 could affect the microbial diversity and community structure of C. rotundifolia silage.   (Figure 4). There were 5 biomarkers in the CK silage, which showed Clostridiaceae at the family level and Clostridium at the genus level; 4 biomarkers in the L silage, with Lactobacillus at the family and genus levels; 4 biomarkers in the M4 silage, with Cyanobacteria at the phylum level and chloroplasts not defined at the genus level; 4 biomarkers in the M4 silage with Cyanobacteria at the phylum level and chloroplasts not defined at the genus level; 6 biomarkers in the ML4 silage, with Rhizobiaceae and Lachnospiraceae at the family level; one biomarker in the CL4 silage, with Lactobacillus plantarum at the species scale; and no biomarker in the C4 silage. Li et al. [43] analyzed differences in silage microbiome using the LEfSe method and found a significant correlation with silage fermentation.  Figure 5A). Cyanobacteria was a dominant phylum in M4, C4, ML4, and CL4, with 24.6%, 339 14.8%, 14.9%, and 20.4%, respectively, but was only 2.01% and 2.80% in L and CK silages. 340 The highest relative abundances of the phylum Aspergillus in CK, M4, C4, ML4, CL4, and 341 L silages were 23.5%, 9.70%, 12.8%, 21.1%, 10.5%, and 8.08%, respectively. In addition, 342 Firmicutes had the highest relative abundance of bacteria at the phylum level in CK, L, M4, C4, ML4, and CL4, with 73.2%, 89.4%, 65.4%, 70.1%, 61.4%, and 68.1%, respectively ( Figure 5A). Cyanobacteria was a dominant phylum in M4, C4, ML4, and CL4, with 24.6%, 14.8%, 14.9%, and 20.4%, respectively, but was only 2.01% and 2.80% in L and CK silages. The highest relative abundances of the phylum Aspergillus in CK, M4, C4, ML4, CL4, and L silages were 23.5%, 9.70%, 12.8%, 21.1%, 10.5%, and 8.08%, respectively. In addition, ML4 had low relative abundances of Bacteroidota (1.13%) and Actinobacteriota (1.03%). Fermentation is a process of microbial colony succession. Aerobic fermentation in the pre-fermentation stage is followed by anaerobic fermentation when beneficial bacteria, mainly LAB, start to dominate. LAB produces lactic acid, which in turn inhibits the growth of undesirable colonies [44]. There are many microbial species in silage, with Firmicutes and Proteobacteria the dominant phyla, and Cyanobacteria to a lesser extent; together, they constitute the main epiphytic bacterial communities [45]. Firmicutes and Proteobacteria play an important role in the degradation of fibers in an anaerobic environment, and changes in DM, NDF, and ADF contents in silage may be related to these phyla [46]. Proteobacteria compete with LAB in the utilization of WSC, and the lower WSC content in the L, C1, C2, M1, and M2 groups may be related to this competition [47]. The relative abundance of Cyanobacteria was higher in the M4 silage than in the CK and L silages, which was attributed to the undefined chloroplasts at the genus level, presumably belonging to the plant sample fraction of C. rotundifolia. At the genus level, Lactobacillus was the dominant bacteria at 60 days of fermentation, followed by Clostridium_sensu_stricto_12, Enterobacter, Methylobacterium-Methylorubrum, Faecalibacterium, and Pectobacterium ( Figure 5B). Enterobacter are non-spore forming, facultative anaerobes that are able to ferment glucose to acetic acid and other metabolites, can cause diseases in animals, such as mastitis, and can destabilize the aerobic stability of the forage [48]. In the present study, the relative abundance of Lactobacillus was greater in the L silage than the other silages (p < 0.05), and in the C4, L4, and CL4 silages was greater than in the CK silage (p < 0.05). The relative abundance of Clostridium_sensu_stricto_12 was lower in all silages than the CK silage (p < 0.05), and of Enterobacter was lower in L and CL4 silages than in CK silage (p < 0.05). Silages with added organic acids had higher (p < 0.05) relative abundances of Unidentified_Chloroplast than CK, and ML4 silage had a higher (p < 0.05) relative abundance of Faecalibacterium than CK silage (p < 0.05). Lactobacillus spp. and Enterococcus spp. dominate the fermentation of feed products and improve silage quality under anaerobic conditions [49]. The dominant genus in the L, M4, C4, ML4, and CL4 silages was Lactobacillus spp., and the relative abundances of Clostridium spp. and Enterococcus spp. were lower than in the CK silage. This may be related to the fact that the M4, C4, ML4, and CL4 silages had increased WSC and CP contents and reduced NH 3 -N:TN ratios. Clostridium perfringens is detrimental to silage quality because it consumes protein and WSC. Ávila and Carvalho [50] reported that Clostridium spp. caused excessive protein degradation, DM loss, and butyric acid production, leading to spoilage and a reduction in silage intake by livestock. However, some Clostridium produce substantial amounts of butyric acid [51]. The decrease in acetic acid content in M4, C4, ML4, and CL4 may be related to the decrease in the relative abundance of Clostridium. The concentrations of lactic acid, acetic acid, and propionic acid and the pH may have a direct effect on bacterial activity and an indirect effect on microbial community structure [44,52]. In the present study, lactic acid and the NH 3 -N:TN ratio were correlated negatively with Clostridium spp. and Enterobacter spp.
butyric acid production, leading to spoilage and a reduction in silage intake by livestock. 376 However, some Clostridium produce substantial amounts of butyric acid [51]. The de-377 crease in acetic acid content in M4, C4, ML4, and CL4 may be related to the decrease in 378 the relative abundance of Clostridium. The concentrations of lactic acid, acetic acid, and 379 propionic acid and the pH may have a direct effect on bacterial activity and an indirect 380 effect on microbial community structure [44,52]. In the present study, lactic acid and the 381 NH3-N:TN ratio were correlated negatively with Clostridium spp. and Enterobacter spp.  Figure 6 presents the relationships between the ensiling fermentation quality indicators (pH, lactic acid, acetic acid, propionic acid, and NH 3 -N:TN) and microbial genera. Li et al. [14] reported that with an accumulation of lactic acid and a decrease in silage pH, Lactobacillus dominated, while the relative abundances of Lactococcus, Enterococcus, Clostridium, and Leuconostoc declined. In the present study. Lactobacillus correlated positively (r = 0.64) with lactic acid content, which is in agreement with Li et al. [45]. In addition, Clostridium_sensu_stricto_12 correlated negatively with lactic acid content (r = −0.74) and positively with amino acids (r = 0.54) and propionic acid (r = 0.58) contents, Enterobacter correlated negatively with lactic acid content (r = −0.55), Ruminocoocus (r = -0.63), Aureimonas (r = −0.70), Roseburia (r = −0.64), and Bacteroides (r = −0.63) correlated negatively with the NH 3 -N:TN ratio, and Pediococcus correlated positively with the NH 3 -N:TN ratio (r = 0.59) and negatively with lactic acid content (r = −0.50). Figure 6 presents the relationships between the ensiling fermentation quality indica-386 tors (pH, lactic acid, acetic acid, propionic acid, and NH3-N:TN) and microbial genera. Li 387 et al. [14] reported that with an accumulation of lactic acid and a decrease in silage pH, 388 Lactobacillus dominated, while the relative abundances of Lactococcus, Enterococcus, Clos-389 tridium, and Leuconostoc declined. In the present study. Lactobacillus correlated positively 390 (r = 0.64) with lactic acid content, which is in agreement with Li et al. [45]. In addition, 391 Clostridium_sensu_stricto_12 correlated negatively with lactic acid content (r = -0.74) and 392 positively with amino acids (r = 0.54) and propionic acid (r = 0.58) contents, Enterobacter 393 correlated negatively with lactic acid content (r = -0.55), Ruminocoocus (r = -0.63), Aureimo-394 nas (r = -0.70), Roseburia (r = -0.64), and Bacteroides (r = -0.63) correlated negatively with 395 the NH3-N:TN ratio, and Pediococcus correlated positively with the NH3-N:TN ratio (r = 396 0.59) and negatively with lactic acid content (r = -0.50). 397 398 Figure 6. Heat map displaying the correlations between the fermentation quality of Chamaecrista 399 rotundifolia silage and the relative abundance of bacterial genera. pH, hydrogen ion concentration; 400 LA, lactic acid; AA, acetic acid; PA, propionic acid; NH3-N, ammonia nitrogen as a percentage of 401 total nitrogen. *p < 0.05; **p < 0.01. 402

403
When malic or citric acid was added separately or combined with L, DM loss of the 404 silage was reduced, pH was lowered, growth, and activity of harmful bacteria were inhib-405 ited, fermentation was promoted, and protein hydrolysis was reduced. After 60 days of 406 fermentation, Firmicutes, which enhances fiber digestion, was the dominant phylum in 407 all treatments. The organic acids L and a combination of both increased the relative abun-408 dance of Lactobacillus and decreased the relative abundance of Clostridium and Enterobac-409 Figure 6. Heat map displaying the correlations between the fermentation quality of Chamaecrista rotundifolia silage and the relative abundance of bacterial genera. pH, hydrogen ion concentration; LA, lactic acid; AA, acetic acid; PA, propionic acid; NH 3 -N, ammonia nitrogen as a percentage of total nitrogen. * p < 0.05; ** p < 0.01.

Conclusions
When malic or citric acid was added separately or combined with L, DM loss of the silage was reduced, pH was lowered, growth, and activity of harmful bacteria were inhib-ited, fermentation was promoted, and protein hydrolysis was reduced. After 60 days of fermentation, Firmicutes, which enhances fiber digestion, was the dominant phylum in all treatments. The organic acids L and a combination of both increased the relative abundance of Lactobacillus and decreased the relative abundance of Clostridium and Enterobacter. We concluded that the addition of malic acid, citric acid, and L improved the quality of C. rotundifolia silage and inhibited the growth of harmful microorganisms. A level of 1% malic acid or citric acid provided the best results.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.