Introduction
Although annual legumes are traditionally used for grain production, whole-plant legumes show high dry matter (DM) productively (Mustafa and Seguin, Reference Mustafa and Seguin2003) and could be used to meet ruminant nutritional requirements. Among annual legumes, soybean plants have been highlighted in Brazilian conditions due to the high availability of cultivars, management knowledge, and other technologies. However, soybean harvest is seasonal, necessitating the conservation to use in animal feeding.
Whole-plant soybean has high buffering capacity and low water-soluble carbohydrates content (Ni et al., Reference Ni, Wang, Zhu, Yang, Zhou, Pan, Tao and Zhong2017), impairing its silage fermentation profile. According to Weinberg and Muck (Reference Weinberg and Muck1996), homofermentative lactic acid bacteria improve lactic acid production and inhibit ethanol and ammonia-N (NH3-N) production, increasing the dry matter recovery by 12%. Especially in tropical conditions, these inoculants may decrease aerobic stability because of insufficient production of short-chain fatty acids that can inhibit yeasts and moulds (Weinberg et al., Reference Weinberg, Ashbell, Hen and Azrieli1993; Schmidt et al., Reference Schmidt, Hu, Mills and Kung2009). More recently, heterofermentative lactic acid bacteria (LAB) inoculants containing Lactobacillus buchneri have been used to increase acetic acid production, reduce mould and yeast count, and increase aerobic stability of silages (Weinberg et al., Reference Weinberg, Ashbell and Hen1999; Filya, Reference Filya2003).
Silage treatment with chemical additives has been used to modulate fermentation. Chitosan (CHI) is a biopolymer derived from chitin deacetylation and has antimicrobial activity against bacteria and fungi (Kong et al., Reference Kong, Chen, Xing and Park2010). Other studies of our research group showed positive effects of CHI addition to sugarcane on silage fermentation, fermentative losses, aerobic stability, and nutritional value of silage (Gandra et al., Reference Gandra, Oliveira, Takiya, Goes, Paiva, Oliveira, Gandra, Orbach and Haraki2016; Del Valle et al., Reference Del Valle, Zenatti, Antonio, Campana, Gandra, Zilio, Mattos and Morais2018, Reference Del Valle, Antonio, Zilio, Dias, Gandra, Castro, Campana and Morais2020). These effects have been associated with the direct inhibition of undesirable microorganisms, such as mould and yeast. Gandra et al. (Reference Gandra, Takiya, Del Valle, Oliveira, de Goes, Gandra, Batista and Araki2018) evaluated the association of CHI and a homofermentative LAB inoculant containing Lactobacillus plantarum and Propionibacterium acidipropionici during the whole plant soybean silage (WPSS) ensiling. Although those authors reported that CHI had a positive impact on lactate content, DM recovery, and in vitro degradation of WPSS, there was no associative effect. Casquete et al. (Reference Casquete, Castro and Teixeira2016) observed that chitosan utilization in food conservation inhibits aerobic and pathogenic bacteria growth. They also reported a synergistic effect with LAB inoculation, which potentially inhibits the growth of deteriorating microorganisms during the aerobic stage of ensiling and improves silage fermentation.
Therefore, we established the hypothesis that CHI or homofermentative LAB inoculant would increase lactic acid count, reduce silage pH and DM losses, while improving the nutritional value and aerobic stability of WPSS in relation to heterofermentative LAB inoculant and control-treated silos. This study was conducted to evaluate the effects of CHI, homofermentative, and heterofermentative microbial inoculant on silage fermentation, count of LAB and mould, fermentation losses, chemical composition, in vitro degradation, and aerobic stability of WPSS.
Materials and methods
The trial was performed between March and July 2019, at the Agrarian Sciences Center of São Carlos Federal University (UFSCar), Araras, Brazil.
Soybean, treatments and experimental design
Soybean (cultivars M6410IPRO®, Monsoy – Bayer Crop Science, São Paulo, Brazil) was seeded on 19 November 2018, in ten different plots (almost 1000 m2 each one). 110 days after the seeding, almost 40 kg of whole-plant soybean were manually harvested from each area at the R6 stage (Coffey et al., Reference Coffey, Granade, Moyer, Anderson and Bush1995) and chopped in a stationary hammer mill (TRF300®, Trapp, Jaguará do Sul, Brazil) to produce four experimental silos (one for each treatment) from each area (plot). The experimental design was a completely randomized block to evaluate the following treatments: (1) CON: WPSS without additives; (2) CHI: WPSS treated with 6 g/kg DM of chitosan; (3) LBB: WPSS treated with 5.0 × 107 colony-formins units (CFU) of Lactobacillus buchneri (NCIM 40788, Lasil Cana®, Lallemand Animal Nutrition, Montreal, Canada) per kg of fresh matter; and (4) LPP: WPSS treated with 1.6 × 108 CFU of Lactobacillus plantarum and 1.6 × 108 CFU of Pediococcus acidilactici (Kera SIL®, Kera Nutrição Animal, Bento Gonçalves, Brazil) per kg of fresh matter. Chitosan had a density of 640 g/L, 883 g/kg DM, 20.0 g/kg of ash, pH of 7.0–9.0, viscosity <200 cPS, and 70 g/kg nitrogen (Polymar Indústria, Fortaleza, Brazil). Chitosan level was based on a previous study of our research group (Del Valle et al., Reference Del Valle, Antonio, Zilio, Dias, Gandra, Castro, Campana and Morais2020), and inoculant inoculation rates were defined according to the manufacturer's recommendations. Each silo material was individually weighted, manually mixed, and randomly allocated to one silo (PVC tubing with 28 cm i.d, 25 cm height, and equipped with Bunsen valve to avoid gas penetration).
Procedures and sampling
Before the treatments were applied, one sample from each area (n = 10) was sampled to evaluate chemical composition. The particle size of ensiled WPSS was analysed using the Penn State Particle Separator (Maulfair et al., Reference Maulfair, Fustini and Heinrichs2011). Buffering capacity was assessed using Playne and McDonald (Reference Playne and McDonald1966) method. In addition, 5 kg of dried sand was placed at the bottom of silos to collect effluent losses, and a nylon screen was placed to avoid silage sampling contamination. Before and after filling, silos were weighed using a 5-g sensitivity scale (Mettler Toledo, Barueri, Brazil). Additionally, whole silo weight was recorded at 5, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, and 120 days after the ensiling to estimate gas losses (GL) throughout the ensiling period. Silos were opened 120 days after ensiling: topmost and bottom phase (5 cm) was discarded, and the remaining silage was homogenized for sampling.
A silage sample (200 g) was pressed in a hydraulic press (PHE-45®, Engehidro, Piracicaba, Brazil) to obtain silage fluid. Silage pH was evaluated using a digital potentiometer (LUCA-210®, Lucadema, Sao José do Rio Preto, Brazil) and the remained fluid sample was frozen for NH3-N, ethanol, and organic acids evaluation. Another sample (500 g) was frozen for chemical analysis and in vitro degradation assay. 10-g of fresh silage was diluted in 90 ml of physiological solution (NaCl, 9 g/l). For microbiological enumeration, 1 ml was diluted in 9 ml of physiological solution. Five subsequent dilutions were performed. A pour plating method in a 10-fold serial dilution on MRS Agar® (Kasvi, São José dos Pinhais, Brazil) incubated at 30°C for 48 h for LAB colony counts (Briceño and Martínez, Reference Briceño and Martínez1995) and on potato dextrose agar (Kasvi) incubated at 26°C for 5 d for yeast and mould colony counts (Rabie et al., Reference Rabie, Liibben, Marais and Jansen van Vuuren1997). Countable CFU results were transformed for log10 CFU/g, and the average was considered to express the result for each silo. To determine aerobic stability, 3 kg of silage was placed without compaction in a plastic bucket (one for each silo; n = 40) and stored in a controlled temperature room (17.0°C ± 1.05; mean ± s.d.) for 7 d. Silage pH was evaluated every 24 h and temperature was measured every eight hours, using spit thermometers (K29-5030®, Kasvi Produtos Laboratoriais, Pinhais, Brazil).
Chemical analysis and in vitro assay
Silage fluid was centrifuged at 500 × g for 15 min. Ammonia-N was evaluated by the Kjeldahl method (method 984.13; AOAC, 2000) without acid digestion. The supernatant of the previously mentioned centrifugation was acidified using formic acid (9:1 ratio, v/v). Organic acids were determined using gas chromatography (GC-2010 Plus chromatograph, Shimadzu, Barueri, Brazil), equipped with an AOC-20i auto-sampler, Stabilwax-DA™ capillary column (30 m, 0.25 mm i.d., 0,25 μm df, Restek©) and a flame ionization detector. Sample (1 μL) was injected with a split ratio of 40:1, using Helium as the carrier gas at a linear velocity of 42 cm/s. The injector and detector temperatures were 250 and 300°C, respectively, and the column's initial temperature was 40°C. The method was calibrated using the WSFA-2 standard (Ref. 47056, Supelco©) and ethanol (Ref. 459828, Sigma-Aldrich©) solutions. The chromatogram was analysed using the GCsolution v. 2.42.00 software (Shimadzu©). The lactic acid concentration was analysed using a spectrophotometric method (Pryce, Reference Pryce1969).
Samples frozen for chemical analysis were thawed at room temperature, dried at 60°C for 72 h, and ground in a knife mill to pass through a 1-mm sieve (SL-31, Solab Científica, Piracicaba, Brazil). It was analysed for DM (method 950.15), ash (method 942.05), ether extract (EE; method 920.39), crude protein (CP; 6,25 × N – method 984.13), acid detergent fibre (ADF), and lignin (method 973.18) as described in AOAC (2000) contents. The neutral detergent fibre (NDF) was analysed without alpha-amylase, and sodium sulphite (Van Soest et al., Reference Van Soest, Robertson and Lewis1991). Non-fibre carbohydrate (NFC) was calculated as follows: NFC (g/kg DM) = 1000 – (NDF + CP + ash + EE). In vitro degradation of DM and NDF was determined according to Tilley and Terry (Reference Tilley and Terry1963) modified by Holden (Reference Holden1999). Samples were processed in a knife mill using a 2-mm sieve and placed in non-woven fabric tissue (5 × 5 cm and 100 g DM/m2; Casali et al., Reference Casali, Detmann, Valadares Filho, Pereira, Henriques, de Freitas and Paulino2008) bags. The bags (three per sample) were incubated for 48 h at 39°C in an in vitro incubator (NL162®, New Lab, Piracicaba, Brazil). Each vial received 1.6 l of McDougall (Reference McDougall1948) buffer and 0.4 l of fresh ruminal fluid. It was sampled from two Holstein heifers (400 kg of body weight) maintained in a pasture without concentrate. The inoculum was CO2 saturated before sample introduction, and 40 bags were incubated in each vial. After removal, samples were washed in running tap water and analysed for NDF content, as previously described.
Protein characterization was performed as described in Cornell Net Carbohydrate and Protein System (CNCPS; Sniffen et al., Reference Sniffen, O'Connor, Van Soest, Fox and Russell1992). Non-protein nitrogen (A-fraction) and soluble protein fraction (B1) were determined after buffer solubilization (Roe et al., Reference Roe, Sniffen and Chase1990). True protein fraction was determined after trichloroacetic acid precipitation (Van Soest et al., Reference Van Soest, Sniffen, Mertens, Fox, Robinson and Krishnamoorthy1981). Unavailable protein (C-fraction) was defined as acid detergent insoluble crude protein (ADIP). Slowly degradable protein fraction (B3) was obtained by the difference amount of the neutral detergent insoluble protein (NDIP) and ADIP. Rumen fermentable fraction of protein (B2) was calculated by the difference of buffer insoluble protein and NDIP (Sniffen et al., Reference Sniffen, O'Connor, Van Soest, Fox and Russell1992). All the protein fractions were expressed in g/kg of CP.
Calculations and statistical analysis
Gas losses (GL), effluent losses (EL), and DM recovery (DMR) were calculated according to Jobim et al. (Reference Jobim, Nussio, Reis and Schmidt2007):
WSWB and WSWA are the whole silo weight before and after the storage, respectively, and EM is the ensiled matter (fresh or dried).
ESWA and ESWB is the empty silo weight after and before the storage, respectively.
where ODM is the DM at the opening, and EDM is the ensiled dry matter.
Calculations and statistical analysis
Data were analysed using the PROC MIXED of SAS (version 9.4, SAS Inst. Inc., Cary, NC, USA) and the following model:
with $b_j\approx N( {0, \;\sigma_b^2 } )$
and $e_{ij}\approx N( {0, \;\sigma_e^2 } )$
, where: Y ij is the observed value of the dependent variable; μ is the overall mean; T i is the fixed effect of treatment (i = 1−4); b j is the random effect of block (area; j = 1−10); e ij is the random residual error; N stands for Gaussian distribution; $\sigma _b^2$
and $\sigma _e^2$
are the variances associated with the random effects of blocks and residue, respectively.
Gas losses throughout the ensiling period and evaluations of pH and temperature after aerobic exposure were evaluated using the following model:
with $b_j\approx N( {0, \;\;\sigma_b^2 } ) , \;\omega _{ij}\approx N( {0, \;\;\sigma_\omega^2 } )$
, and e ijk ≈ MVN(0, R) where: Y ijk is the observed value; μ, T i, and b j were previously defined; ω ijk is the error associated with experimental units (silos); P k is the fixed effect of the period/time (k = 1−13 for gas losses throughout ensiling; 1−21 for temperature; and 1−7 for pH after aerobic exposure); T × P ik is the interaction between treatment and period effects; e ijk is the experimental error; N stands for Gaussian distribution; $\sigma _b^2$
and $\sigma _\omega ^2$
are variances associated with blocks and silos, respectively; MVN stands for multivariance normal distribution; R is a variance and covariance matrix due to repeated measures. Matrixes (CS, CSH, AR, ARH, TOEP, TOEPH, FA, UN, ANTE) were evaluated using the Bayesian method. Treatment effects were studied using a protected Fisher's means test (LSD) at 5% of probability.
Results
Fresh whole-plant soybeans used in the present study averaged 250 g/kg DM, 472 g/kg NDF, 57.2 g/kg of ether extract, 618 g/kg of DM in vitro degradation, and 784 g/kg of particles higher than 8 mm (Table 1). The addition of LPP during the ensiling reduced (P ≤ 0.05) NH3-N, ethanol, butyric, acetic, propionic, and branched-chain fatty acid concentration relative to other treatments evaluated in the present study (Table 2). Both microbial inoculated silos (LBB and LPP) had lower (P ≤ 0.05) pH values than CON and CHI-treated silages. In addition, LBB decreased (P ≤ 0.05) ethanol concentration when compared to CHI and CON. Silages from CON and CHI treatments showed a similar (P > 0.05) fermentation profile.
Table 1. Composition and buffer capacity of whole-plant soybeans (n = 10) at ensiling (g/kg DM, unless stated)
s.d., standard deviation; DM, dry matter.
Table 2. Fermentation profile of whole-plant soybean silage treated with chitosan or microbial inoculants
a–cFisher's means test at 5% of probability.
1 Treatments: CON: WPSS without additive; CHI: WPSS with 6 g/kg DM of chitosan; LBB: WPSS with 5.0 × 107 CFU/kg fresh matter of Lactobacillus buchneri; LPP: WPSS with 1.6 × 108 CFU of Lactobacillus plantarum and 1.6 × 108 CFU of Pediococcus acidilactici per kg of natural matter.
2 Standard error of mean.
3 Branched-chain fatty acids.
Between evaluated treatments, LBB inoculation reduced (P ≤ 0.05) LAB counts, whereas LPP decreased (P ≤ 0.05) effluent and gas losses as well as improved (P ≤ 0.05) DM recovery (Table 3). The lowest GL throughout the ensiling process was in LPP-treated silages (72 g/kg), followed by LBB (109 g/kg), which decreased losses in relation to CON and CHI-silos (116 g/kg; Figure 1). Treatments did not affect (P = 0.627) mould and yeast counts. LPP-treated silages showed the highest (P ≤ 0.05) DM and OM concentration, as well as DM and NDF degradation (Table 4). Chitosan increased (P ≤ 0.05) CP and NDIP relative to other treatments. There was no treatment effect (P ≥ 0.170) on WPSS NDF, ADF, NFC, and EE, which averaged 510, 341, 171, and 77.5 g/kg, respectively.
Fig. 1. Gas losses after ensiling of whole-plant soybean silage treated with chitosan and microbial inoculants. Treatments: CON (
): WPSS without additive; CHI (
): WPSS with 6 g/kg DM of chitosan; LBB (
): WPSS with 5.0 × 107 CFU/kg fresh matter of Lactobacillus buchneri; LPP(
): WPSS with 1.6 × 108 CFU of Lactobacillus plantarum and 1.6 × 108 CFU of Pediococcus acidilactici per kg of natural matter.
Table 3. Fermentation losses and microbial counts of whole-plant soybean silage treated with chitosan or microbial inoculants
a,bFisher's means test at 5% of probability.
1 Treatments: CON : WPSS without additive; CHI: WPSS with 6 g/kg DM of chitosan; LBB: WPSS with 5.0 × 107 CFU/kg fresh matter of Lactobacillus buchneri; LPP: WPSS with 1.6 × 108 CFU of Lactobacillus plantarum and 1.6 × 108 CFU of Pediococcus acidilactici per kg of natural matter.
2 Standard error of mean.
Table 4. Chemical composition and in vitro degradation of whole-plant soybean silage treated with chitosan or microbial inoculants
a–cFisher's means test at 5% of probability.
1 Treatments: CON : WPSS without additive; CHI: WPSS with 6 g/kg DM of chitosan; LBB: WPSS with 5.0 × 107 CFU/kg fresh matter of Lactobacillus buchneri; LPP: WPSS with 1.6 × 108 CFU of Lactobacillus plantarum and 1.6 × 108 CFU of Pediococcus acidilactici per kg of natural matter.
2 Standard error of mean.
CHI-treatment increased (P ≤ 0.05) C-fraction of protein in relation to microbial inoculant treatments (Table 5). Additionally, LPP reduced C-fraction relative to the control. However, treatments did not affect (P ≥ 0.069) other protein fractions of silage evaluated in the present study. There was no treatment and time interaction effect (P ≥ 0.560) on silage pH and temperature after aerobic exposure (Figs 2 and 3, respectively). Average silage pH was higher (P ≤ 0.05) for CON (5.57) and CHI-treated (5.58) silos in relation to LBB (5.45) and LPP-treated (5.46) silages. Additionally, treatments did not affect (P = 0.682) the temperature of silage after aerobic exposure, although the difference relates to environment temperature decreased (P < 0.001) across the time up to 168 h.
Table 5. Protein fractions of whole-plant soybean silage treated with chitosan or microbial inoculants
a–bFisher's means test at 5% of probability.
1 Treatments: CON: WPSS without additive; CHI: WPSS with 6 g/kg DM of chitosan; LBB: WPSS with 5.0 × 107 CFU/kg fresh matter of Lactobacillus buchneri; LPP: WPSS with 1.6 × 108 CFU of Lactobacillus plantarum and 1.6 × 108 CFU of Pediococcus acidilactici per kg of natural matter.
2 Standard error of mean.
3 According to Sniffen et al. (Reference Sniffen, O'Connor, Van Soest, Fox and Russell1992).
Fig. 2. Silage pH after aerobic exposure of whole-plant soybean silage treated with chitosan and microbial inoculants. Treatments: CON (
): WPSS without additive; CHI (
): WPSS with 6 g/kg DM of chitosan; LBB(
): WPSS with 5.0 × 107 CFU/kg fresh matter of Lactobacillus buchneri; LPP(
): WPSS with 1.6 × 108 CFU of Lactobacillus plantarum and 1.6 × 108 CFU of Pediococcus acidilactici per kg of natural matter.
Fig. 3. Silage temperature after aerobic exposure of whole-plant soybean silage treated with chitosan and microbial inoculants. Treatments: CON (
): WPSS without additive; CHI (
): WPSS with 6 g/kg DM of chitosan; LBB (
): WPSS with 5.0 × 107 CFU/kg fresh matter of Lactobacillus buchneri; LPP(
): WPSS with 1.6 × 108 CFU of Lactobacillus plantarum and 1.6 × 108 CFU of Pediococcus acidilactici per kg of natural matter.
Discussion
The present study used ten plots to evaluate the effect of additives on WPSS fermentative profile and losses, chemical composition, in vitro digestibility, and aerobic stability. Fresh samples showed a low DM content (250 ± 14.4 g/kg), high concentration of CP (178 ± 17.9 g/kg DM), and buffering capacity (448 ± 26.7 mEq/kg DM), which comprises a challenge for silage preservation (Jatkauskas and Vrotniakiene, Reference Jatkauskas and Vrotniakiene2011). These conditions led to poor silage preservation, as demonstrated by high silage pH, butyric acid, and NH3-N concentrations (Kung Jr et al., Reference Kung, Smith, da Silva, Windle, Silva and Polukis2018). According to Playne and McDonald (Reference Playne and McDonald1966), a high buffering capacity of soybean forage results from the high concentration of proteins and organic acids (malic, citric, nicotinic, malonic and glyceric) and their salts present in the plant tissues. For instance, alfalfa also is a leguminosae with a buffering capacity of 488 meq/kg DM, and that of maize ranges from 200 to 250 mE/kg of DM (McDonald et al., Reference McDonald, Henderson and Heron1991). Considering WPSS characteristics similarity with other leguminoseae silages, it is possible to consider that high buffering capacity is the main challenge to improving the fermentation profile and nutritional value of silage.
Both microbial inoculants (LBB and LPP) reduced silage pH, whereas chitosan showed no effect on this variable. LPP inoculation reduced WPSS pH, which is most related to lactic acid and buffering capacity (Kung et al., Reference Kung, Smith, da Silva, Windle, Silva and Polukis2018) and has an essential effect of inhibiting undesirable microorganisms that consume lactic acid (Ni et al., Reference Ni, Wang, Zhu, Yang, Zhou, Pan, Tao and Zhong2017). The LAB produce lactic acid as the main end-product of carbohydrates fermentation (Muck, Reference Muck2010). However, the concentration of the lactic acid was not affected by additives in the present study. Undesirable microorganisms, such as clostridia, may have been active in transforming protein and sugar into NH3-N and butyric acid in these high moisture silages (Kung et al., Reference Kung, Smith, da Silva, Windle, Silva and Polukis2018). Therefore, NH3-N and butyric acid concentrations were reduced in LPP-treated silages suggesting a suppression effect of homofermentative strains on clostridia activity.
LBB-inoculant reduced silage pH in relation to the control and did not affect organic acid concentration. Reduced pH could be associated with non-statistical higher ethanol and lower NH3-N of LBB-treated silos. L. buchneri-treated silos often show 0.1−0.2 units higher pH than untreated silage (Kleinschmit and Kung, Reference Kleinschmit and Kung2006), because of the conversion of lactic acid to acetic acid, 1,2-propanediol (1,2PD), and ethanol (Oude-Elferink et al., Reference Oude-Elferink, Krooneman, Gottschal, Spoelstra, Faber and Driehuis2001). On the other hand, LBB increased the concentration of ethanol, acetic and propionic acids at the expense of LPP-silos. The greater production of acetic and propionic acids, rather than lactic acid, in silages inoculated with L. buchneri is well documented in the literature (Pahlow et al., Reference Pahlow, Muck, Driehuis, Oude-Elferink, Spoelstra, Buxton, Muck and Harrison2003).
According to Senel and McClure (Reference Senel and McClure2004), the chitosan bactericidal effect is dependent on pH: the more significant activity is observed at pH values around 4.5. As soybean silage showed a high pH value (range from 5.64 to 5.80), which resulted in no significant effect of CHI. A similar lack of effect on fermentation parameters was observed by Gandra et al. (Reference Gandra, Takiya, Del Valle, Oliveira, de Goes, Gandra, Batista and Araki2018). Differently, chitosan positively affected sugarcane silage fermentation and nutritional value, for example, which traditionally showed a low pH environment (Del Valle et al., Reference Del Valle, Zenatti, Antonio, Campana, Gandra, Zilio, Mattos and Morais2018, Reference Del Valle, Antonio, Zilio, Dias, Gandra, Castro, Campana and Morais2020). Gandra et al. (Reference Gandra, Takiya, Del Valle, Oliveira, de Goes, Gandra, Batista and Araki2018) also observed improved DM recovery and nutritional value on CHI-treated soybean silages. However, it is essential to highlight that most of the CHI effects observed in that study were dependent of microbial inoculant addition; DM content of fresh soybeans was higher than obtained in the present study (342 vs. 250 g/kg natural matter) and different chitosan levels were evaluated (5 g/kg natural matter vs. 6 g/kg DM). Therefore, Gandra et al. (Reference Gandra, Takiya, Del Valle, Oliveira, de Goes, Gandra, Batista and Araki2018) obtained a lower silage pH, which favours the CHI effect on WPSS compared to the present study.
Ni et al. (Reference Ni, Wang, Zhu, Yang, Zhou, Pan, Tao and Zhong2017) reported a yeast number of 106 CFU/g of fresh matter. Although we had no treatment effect on mould and yeast counts, the treatment average ranged from 6.18 to 6.52 log10/g of fresh matter. High ethanol production has been associated with mould and yeast growth (Muck, Reference Muck2010). According to Kung Jr et al. (Reference Kung, Smith, da Silva, Windle, Silva and Polukis2018) excessively high acetic acid is observed in high moisture silages that show unwanted fermentation, dominated by enterobacteria, clostridia, or heterolactic acid bacteria (McDonald et al., Reference McDonald, Henderson and Heron1991). The critical pH value for controlling Enterobacter growth in silage at 25% DM is 4.35 (Weissbach and Honig, Reference Weissbach and Honig1996). Enterobacteria could ferment lactic to acetic acid and other products, causing loss of nutritive value (Ni et al., Reference Ni, Wang, Zhu, Yang, Zhou, Pan, Tao and Zhong2017). Heterofermentative LAB and species of Enterobacterias produce a mix of fermentation products, in which ethanol may represent 50% of them (Mc Donald et al., Reference McDonald, Henderson and Heron1991). Typically, DM gas losses are linked to ethanol production due to carbon dioxide formed during ethanol fermentation (Driehuis and Wikselaar, Reference Driehuis and Wikselaar2000).
LPP-treated silage showed the highest DM content and a considerably higher DM recovery. Although L. buchneri has been extensively used to produce acetic from lactic acid (Kung et al., Reference Kung, Smith, da Silva, Windle, Silva and Polukis2018), it was not observed in the present study. In clostridial silage, epiphytic heterolactic fermentation prevails, which might minimize the effect of these inoculated bacteria. The production of acids during the aerobic stage of ensiling favours the growth of a more acid-tolerant LAB. When the substrate is not limited, LAB growth reduced the silage pH and produced stable silage. If the substrate is limited, the enterobacteria and clostridia may not be suppressed and may also grow (Rooke and Hatfield, Reference Rooke, Hatfield, Buxton, Muck and Harrison2003). We evaluated oven-dry matter content in the present study. This method could provide biased results (Daniel et al., Reference Daniel, Weiß, Custódio, Neto, Santos, Zopollatto and Nussio2013) because higher DM content observed in LPP-silages was linked with lower organic acid (except lactic) concentration.
Reduced NH3-N and ethanol production are associated with decreased EL in LPP-treated silages due to a more desirable fermentation profile than CON-treated silos. According to Muck (Reference Muck2010), yeasts, moulds, and acetic acid bacteria can grow on silage in aerobic conditions, using fermentation products and residual sugars to produce carbon dioxide, water and heat. Water could be drained from the silage, resulting in increased EL. It is important to highlight three consequences of these effluent losses: (1) effluents are produced from the most nutritive fractions (soluble sugars, protein) of ensiled material (Buxton et al., Reference Buxton, Muck, Harrison, Rooke and Hatfield2003), and reduced effluent production improves the nutritional value of silage, as observed by increased in vitro degradation of LPP-silages in the present study; (2) as effluent production is a physical process that occurred as a consequence of higher moisture content of silage, increased effluent production is linked with depressed silage DM content; (3) effluent has a very high biochemical oxygen demand, being one of the most concentrated farm pollutants (Buxton et al., Reference Buxton, Muck, Harrison, Rooke and Hatfield2003).
Heterofermentative LAB inoculant had no effect on WPSS chemical composition and in vitro degradation. Filya (Reference Filya2003) has already observed no effect of L. buchneri on in situ DM, OM and NDF degradability of maize and sorghum silages. On the other hand, CHI increased CP and NDIP. It is essential to highlight that chitosan has 438 g/kg of CP, mostly insoluble in a neutral condition. In acid conditions, a molecular dissociation and solubilization significantly increase (Goy et al., Reference Goy, Brito and Assis2009). However, microbial inoculants reduced the C-fraction of protein in relation to CON and CHI. We can associate this effect with reduced fermentation losses observed in inoculated-silages, diluting the proportion of low degradable protein fraction.
There was no treatment and time interaction effect on silage pH after aerobic exposure. The differences at the opening were observed throughout the aerobic evaluation. We also observed an unexpected behaviour of silage pH after aerobic exposure. Silage pH started from 5.52 to 5.67 and, after seven days of evaluation, found values between 5.41 and 5.55. According to Parra et al. (Reference Parra, Bolson, Jacovaci, Nussio, Jobim and Daniel2019), reduced soybean silage pH after aerobic exposure could be associated with the degradation of alkalizing substances, such as proteins.
Similarly, there was no treatment effect on silage temperature after aerobic exposure. Although it was expected that LBB could improve the aerobic stability of silage, it was not observed. The buffering capacity of Leguminosae is higher than grasses (Wilkinson, Reference Wilkinson and Wilkinson2005). Evaluating 264 legume silages, Pahlow et al. (Reference Pahlow, Muck, Driehuis, Oude-Elferink, Spoelstra, Buxton, Muck and Harrison2003) observed that 89% of them were stable 156 h after aerobic exposure. Therefore, as previously discussed, undesirable fermentation has a considerable impact on this variable, resulting in no effects of treatments evaluated in the present study.
Conclusion
Homofermentative LAB inoculant containing Lactobacillus plantarum and Pediococcus acidilactici reduces fermentation losses and improves WPSS fermentation profile and nutritional value WPSS, but with no effect on aerobic stability. Heterofermentative LAB inoculation and chitosan have no positive effect on WPSS production.
Acknowledgements
The authors thank Professor Dr Reinaldo G. Bastos and Luiz F. A. Mattos (UFSCar) for providing the physical infrastructure and staff to organic acids evaluations.
Financial support
The authors appreciate the São Paulo Research Foundation (FAPESP, Brazil) for financial support (2017/15457-7).
Conflict of interest
The authors declare that there are no conflicts of interest to the current manuscript.
Ethical standards
Procedures of the present study were previously approved by the UFSCar Ethics Committee (Approval number 1395120219).
