Whole corn grain addition in sugarcane silage avoids fermentative losses and improves in situ degradation of silage La adición de granos enteros de maíz a ensilaje de caña de azúcar reduce pérdidas por fermentación y mejora la degradación in situ del ensilaje

Sugarcane silage (SS) is generally susceptible to yeast action, resulting in dry matter losses due to high soluble carbohydrate concentration. We evaluated the effects of adding corn grain and microbial inoculant at ensiling on fermentative profile, losses, chemical composition and degradation of silages. Forty experimental silos (PVC tubing) were assigned at random to a 5 × 2 factorial arrangement with: (1) 5 corn additions at ensiling: CONT straight sugarcane silage; GC2 sugarcane with ground corn (processed through a 2 mm sieve) added at ensiling; GC8 sugarcane with ground corn (processed through an 8 mm sieve) added at ensiling; WC sugarcane with whole corn grain added at ensiling; and RCS rehydrated corn ensiled without sugarcane; and (2) 2 microbial inoculant additions at ensiling: 0 and 8 mg of commercial inoculant per kg of feed. Corn grain was added at the rate of 100 g per kg of fresh sugarcane. Adding corn grain to sugarcane at ensiling improved SS fermentation and silage chemical composition. There was no benefit from grinding the grain before adding it to sugarcane. Microbial inoculant had little effect on SS fermentation. Studies comparing corn grain with other energy sources, e.g. molasses or cassava, for addition at ensiling sugarcane seem warranted along with feeding studies with livestock to assess intake and subsequent performance. The overall benefits of adding the energy sources at ensiling versus feeding them directly to animals with untreated sugarcane silage should be determined.


Introduction
In subtropical conditions, sugarcane generally produces higher dry matter (DM) yields per unit area and energy value at maturity than other tropical forages (Daniel et al. 2013) and fresh sugarcane is traditionally fed to cattle during periods of low pasture availability (Santos et al. 2010). Conserving sugarcane as silage would allow greater flexibility in feeding strategies. However, sugarcane silage (SS) is generally susceptible to the action of yeast fungi owing to high soluble carbohydrate concentration, producing a typical alcoholic fermentation of soluble carbohydrates into ethanol, CO2 and water (Sá Neto et al. 2013) and, consequently, increased DM losses (Pedroso et al. 2008). Soluble carbohydrate concentration in the final product is lower and the level of fibrous components is higher than in the raw material, while ruminal degradation of the ensiled material is lower than that of fresh sugarcane.
One alternative to counteract the undesirable outcomes from natural fermentation in SS is the inclusion of corn grain and other additives at ensiling. Incorporating corn grain with fresh sugarcane when making SS could reduce ethanol production and DM losses (Gómez-Vázquez et al. 2011), while the use of other additives, e.g. inoculants, should help to inhibit epiphytic yeast populations and mitigate nutrient losses (Ávila et al. 2014).
Maize cultivars produced in Brazil have a vitreous endosperm, which limits starch digestibility in the gastrointestinal tract of animals. Rehydrated corn grain silage (RCS) has been used to improve starch digestibility of corn kernels in Brazilian production systems (Silva et al. 2018). Ethanol produced in SS can also solubilize proteins of the grain's endosperm (Zhang et al. 2011), improving starch availability. In addition, grinding of corn grain could affect starch digestibility by animals and starch solubilization in the silos. As suggested by Junges et al. (2017), bacterial activity is the most critical determinant of protein degradation and could be enhanced by increasing the soluble carbohydrate concentration in SS.
Among silage additives, microbial inoculants have been used to reduce the undesirable effects of SS fermentation (Carvalho et al. 2014;Santos et al. 2015;Jacovaci et al. 2017). Inoculants containing homolactic bacteria improve lactic acid production and reduce silage pH, without positively affecting alcohol fermentation (Pedroso et al. 2008;Santos et al. 2015). Pediococcus acidilactici establishes a low silage pH (Fitzsimons et al. 1992) and Propionibacterium spp. produce propionic acid from lactic acid (McDonald et al. 1991) with potential negative effects on yeast growth.
The present study aimed to evaluate any benefits from the inclusion of corn grain, processed at different particle sizes, and bacterial inoculants at ensiling on fermentation, chemical composition and in situ degradation of DM and neutral detergent fiber (NDF) of SS. We hypothesized that the addition of corn grain when ensiling sugarcane, regardless of the microbial inoculant supply, would reduce fermentative losses in SS and improve chemical composition and DM degradation relative to straight SS without corn or rehydrated corn grain silage (RCS).

Materials and Methods
The experiment was carried out at the Federal University of São Carlos, Araras, São Paulo, Brazil. Sugarcane (variety RB83-5054), at 8 months of growth (first cut) and 17.5% Brix, was used. Sugarcane from 5 locations/plots was manually harvested and chopped in a stationary cutter (Dedini®, Piracicaba, Brazil) to an ideal cut length of 10 mm.
Forty experimental silos (PVC tubes -28 cm diameter and 25 cm long) were randomly assigned to a 5 × 2 factorial arrangement to evaluate: (A) 5 levels of corn addition: 1) control (CONT) -sugarcane ensiled without corn addition; 2) GC2 -SS with 100 g ground corn (2 mm sieve-processed)/kg fresh sugarcane added at ensiling; 3) GC8 -SS with 100 g 8 mm sieve-processed corn/kg fresh sugarcane; 4) WC -SS with 100 g whole corn grain/kg fresh sugarcane; and 5) RCS -rehydrated corn silage (whole grain ensiled without sugarcane); and (B) 2 levels of microbial inoculant addition at ensiling: 0 and 8 mg commercial inoculant/kg total ensiled material. Each kg of inoculant contained 3.9 × 10 10 colony-forming units (CFU)/g of Pediococcus acidilactici and 3.75 × 10 10 CFU/g of Propionibacterium acidicipropionici. Samples of fresh sugarcane and corn grain were collected for chemical analyses (Table 1). Ensiling was performed using PVC tubes equipped with Bunsen valves. Sand (2 kg) was placed in the bottom of the tubes and separated from the ensiled material by a nylon mesh screen to drain effluent. Inoculant and corn were added individually to the sugarcane and the total thoroughly mixed manually before being assigned to a tube. Microbial inoculant was diluted in water and sprayed onto the fresh sugarcane. Ensiled material was compacted manually (650 kg/m 3 for SS and 1,000 kg/m 3 for RCS) and tubes were sealed, weighed and stored at room temperature (about 25 °C) in a shed for 60 days. Immediately before opening, the silos were reweighed to determine DM and gas losses, expressed as a proportion of the DM ensiled.
Samples (500 g) from the center of the mass of those silos treated with whole corn grain (WC) were used for whole corn grain selection and recuperation calculation. Subsamples of corn grain from the WC treatment were called 'recovered' and used for chemical analysis and in situ degradation assay.
Fresh forage and silages were analyzed for DM concentration in a forced-air oven at 60 °C for 72 h, then ground through a 2 mm screen (SL-31, Solab Científica, Piracicaba, Brazil) and analyzed for ash, crude protein (CP) and ether extract (EE) according to AOAC (2000). Neutral detergent fiber (NDF, with heat-stable amylase, without sodium, and expressed inclusive of residual ash) and acid detergent fiber (ADF) were determined according to Van Soest et al. (1991). Two cannulated dairy cows previously adapted to a diet with 60:40 forage:concentrate ratio were used for in situ degradation assays. Ruminal incubation was performed for 96 hours, using 5 × 5 cm non-woven tissue bags (Casali et al. 2008). After removal, bags were washed in running water and evaluated for NDF concentration.
To identify and quantify yeasts and molds, 10 g silage was diluted with 90 mL sterilized peptone water (1%, w/v). Serial dilutions were pour-plated on Dichloran Rose Bengal Chloramphenicol Agar. Agar plates were incubated aerobically at 28 °C for 7 days. Colony-forming units were transformed into log10/g values (Downes and Ito 2001). Water activity (WA) was assessed using a benchtop water activity meter (Aqualab 4T, Decagon Devices Inc., Pullman, USA).
Another sample (500 g) of SS was used for silage juice extraction with a hydraulic press. The extract was filtered through cheesecloth and pH was determined immediately. The silage juice sample was centrifuged (500 × g for 15 min) and the supernatant was used for NH3-N and organic acid evaluation. Ammonia nitrogen was determined by the colorimetric phenol-hypochlorite method (Broderick and Kang 1980). Concentrations of ethanol plus acetic, propionic and butyric acids were determined by gas chromatography (GC-2010 Plus chromatograph, Shimadzu, Barueri, Brazil), fitted with a flame-ionization detector and automatic sample injection. Lactic acid concentration in silage samples was assessed using the spectrophotometric method (Pryce 1969).
Gas and effluent losses were calculated according to the following 3 equations: where: GL is gaseous loss; SWE and SWO are the silo weights at ensiling and opening, respectively; and EDM is the ensiled dry matter.
where: EL is effluent loss; ESWO and ESWE are the empty (but including the sand plus effluent) silo weights at opening and ensiling, respectively.
where: DMR is dry matter recovery; DMO is dry matter at silo opening; and DME is dry matter at ensiling.

Statistical analysis
For fermentative profile, losses and chemical composition, data from RCS were removed for statistical analysis. The PROC MIXED of SAS 9.3. (SAS Institute Inc., Cary, USA) was used, according to the following statistical model: where: Yijk is the value of the dependent variable; µ is the overall mean; Ci is the fixed effect of corn (i = 1 to 4); Ij is the fixed effect of microbial inoculant (j = 1, 2); C×Iij is an interaction term; eijk is the residual error; and N stands for Gaussian distribution.

Results
Corn addition decreased (P<0.01) acetic, propionic and butyric acid concentrations in SS and tended to decrease (P=0.07) ethanol concentration, while decreasing (P≤0.01) losses and increasing (P<0.01) DM recovery (Table 2). Microbial inoculant addition increased (P≤0.04) silage pH plus ethanol and butyric acid concentrations, but decreased (P≤0.04) acetic acid concentration, mold and yeast counts and DM recovery, and tended (P≤0.07) to increase lactic acid concentration and effluent losses. Compared with ground corn, WC tended (P≤0.07) to decrease mold and yeast counts and to increase DM recovery. The corn × microbial inoculant interaction was significant (P≤0.02) for NH3-N and water activity (WA) plus gaseous and total losses. Corn addition decreased (P<0.01) NH3-N only in silages treated with microbial inoculant (Figure 1). Ground corn tended to increase (P≤0.09) NH3-N in silages relative to whole corn, regardless of microbial inoculant application (Figure 1) and the effect was stronger with finer grinding (P≤0.06).
Regardless of inoculation, corn addition decreased (P<0.01) gaseous and total fermentative losses of silage (Table 2; Figure 2). Whole corn decreased (P=0.02) Table 2. Effects of addition of ground or whole corn grain and microbial inoculant at ensiling on fermentation, mold counts and fermentative losses in sugarcane silage.   gaseous and total losses relative to ground corn only in silage treated with inoculant. Finer grinding reduced gaseous and total fermentative losses relative to coarser grinding (P≤0.01) only without addition of microbial inoculant (Table 2; Figure 2). Adding corn decreased water activity (WA) in inoculated silages (P=0.01). In addition, silage with WC showed lower WA if treated with microbial inoculant (P<0.01; Table 2) and a tendency for lower WA in noninoculated silage (P=0.09). There was no corn addition × INO interaction effect on chemical composition and DM degradation of silage (P≥0.22; Table 3). Corn addition increased (P≤0.01) silage DM, OM, NFC, CP and EE concentrations and DM degradation and decreased (P<0.01) NDF and ADF concentrations.
Microbial inoculant addition decreased (P=0.03) silage OM concentration and increased (P=0.04) CP concentration (Table 3). Ground corn tended to decrease NFC and to increase NDF concentrations, in comparison with whole corn (P≤0.09). Particle size of ground corn had no effect (P≥0.13) on SS composition and DM degradation (Table 3).
Recovered corn from WC and RCS showed similar OM, NDF, ADF and NFC concentrations (P≥0.12; Table  4). However, recovered corn had lower (P≤0.01) DM and CP concentrations and higher (P<0.01) EE concentration than RCS. Additionally, recovered corn tended to have higher (P=0.09) DM degradation than RCS. Microbial inoculant had no effect on any aspects of corn grain chemical composition and DM degradation (P≥0.15).  Probabilities: PROC -processing effect; INO -microbial inoculant effect; PROC × INO -interaction between processing and inoculant effects P≥0.14.

Discussion
Fresh sugarcane used in the present study averaged 17.5% Brix and 257 g DM/kg, which is lower than the recommended level for ensiling (300 g/kg; McDonald et al. 1991). However, these values are similar to those reported by other authors, e.g. Sá Neto et al. (2013) for fresh sugarcane. We chose this material to provide a higher challenge for evaluation of the treatments, as ensiling of sugarcane with low DM concentration could benefit more from corn addition. As expected, corn addition to sugarcane at ensiling increased DM concentration of ensiled material and consequently improved fermentation conditions, decreasing the production of acids and WA, especially in inoculant-treated silos. According to Greenhill (1964), after the breakdown of the cell walls, WA depends mainly on the moisture content of the ensiled material. Material with high DM concentration shows decreased WA, lower bacterial counts and delayed growth of lactic acid bacteria (LAB) (Castro et al. 2006). Bernardes et al. (2007) showed that WA is directly associated with counts of mold and yeast. While corn addition decreased acid concentration and ethanol production in the present study, ground corn increased WA relative to whole corn.
Although concentrations of acids in corn-treated silages were lower than in straight sugarcane silage, in general, corn addition had no effect on pH of the silage. This result supports the findings of Andrade et al. (2001), who added ground corn ears to sugarcane with urea at ensiling and found no differences in silage pH. In the present study, the production of lactic acid, the primary acidogenic acid evaluated in the current trial, was not affected by corn addition at ensiling. Similarly, Bernardes et al. (2007) evaluated the addition of dehydrated corn grain with cob and straw to sugarcane at ensiling and also found no effects on silage pH, but chemical composition was improved.
Water may be produced during bacterial fermentation of sugars, mainly during the conversion of substrate to ethanol by yeasts (Pedroso et al. 2008). This water accumulation results in increased effluent losses and decreased silage DM concentration and DM recovery in silos without corn treatment. Additionally, WC decreased total losses and improved DM recovery, relative to GC, especially in inoculant-treated silos. Besides higher WA, GC2 addition resulted in lower effluent and gaseous losses, without affecting DM recovery, relative to GC8. Finely ground corn with smaller particle size can make compaction of material and expulsion of air from the ensiled material easier than whole or coarsely ground corn.
Furthermore, corn processing increases microbial adhesion to the endosperm and consequently increases NH3-N concentration (Lee et al. 2002). The critical interaction between corn particle size and action of microbial inoculant was highlighted by the more evident corn effect in inoculated silos. As NH3-N is produced from proteolysis (Albrecht and Muck 1991), decreased microbial action could negatively affect the NH3-N level, as observed in corn-treated silos. Although Oliveira et al. (2017) found decreased NH3-N concentration in silages treated with homofermentative LAB regardless of forage type, inoculant had no effect in the present study. Despite the presence of propionic acid bacteria in the evaluated inoculant, the addition of inoculants showed no effect on propionic acid concentration but increased ethanol concentration of the silage. Borreani et al. (2018) observed that yeast activity converted glucose to ethanol, resulting in high fermentative losses in silage. Both homofermentative and heterofermentative lactic acid bacteria potentially improve silage fermentation, but heterofermentative bacteria are more effective in inhibiting fungal growth (Filya 2003). Santos et al. (2015) observed that adding homofermentative LAB (Propionibacterium acidicipropionici, Lactobacillus plantarum and Enterococcus faecium) to sugarcane at ensiling increased ethanol production and DM losses, even when heterofermentative LAB were included, whereas silage inoculated exclusively with heterofermentative LAB (L. buchneri) had reduced DM losses and alcoholic fermentation. In the present study, microbial inoculant decreased DM recovery and concentrations of DM and OM in silage.
In general, corn addition to SS reduced fermentative losses and, consequently, improved chemical composition of silage. Traditionally, in SS, high yeast populations (Ávila et al. 2010) convert most of the water-soluble carbohydrate, which is in the NFC component, into fermentation endproducts, such as volatile organic compounds and mainly ethanol (Daniel et al. 2013). In keeping with these outcomes, corn addition reduced fermentative losses in SS and increased the concentration of NFC in our study, resulting in decreased fiber concentration via a dilution effect and improved DM degradation. The different responses in NDF and NFC as a result of the addition of ground and whole corn seem related to lower fermentative losses in silos containing WC grain.
In evaluating recovered whole corn grain (RWCG) relative to rehydrated corn silage (RCS), we found similar OM, NDF, ADF and NFC concentrations for both forms of corn, while RWCG had lower DM and CP concentrations and higher EE concentration than RCS. Although Junges et al. (2017) attributed 60% of protein degradation in RCS to bacterial activity, solubilization in fermentation end-products could increase protein degradation (Lawton 2002) and improve silage DM degradation. Lower CP concentration and increased DM degradation observed in RWCG relative to RCS are reflected in increased acid levels and bacterial activity in the silos. Higher ethanol concentration in SS could solubilize kernel protein (Zhang et al. 2011), which was not recovered in the silage samples.
At ensiling rehydrated corn contained 660 g DM/kg and DM recovery after ensiling was 926 g DM/kg silage, which is in accordance with average recovery levels reported by Kung Jr et al. (2004). On the other hand, RWCG showed a recovery rate of 966 g DM/kg (P>0.05). However, owing to the importance of this variable in silage making, more studies are necessary to evaluate the role of adding corn in improving DM recovery in sugarcane silage. Our results suggest that there is little merit in grinding the grain before adding it to the sugarcane at ensiling. In some situations, other energy sources like molasses or cassava may be a cheaper source than corn grain and studies are needed to test their efficacy relative to corn. In addition, feeding studies to evaluate feed intakes by livestock and subsequent performance using whole corn grain and other energy additives in SS at ensiling or fed directly with sugarcane silage are warranted.

Conclusions
Microbial inoculant containing mainly homolactic bacteria had little effect on fermentation of sugarcane silage and there appears little merit in adding it to fresh sugarcane at ensiling. The addition of corn grain at ensiling improved SS fermentation and silage composition but there seems little value in grinding the grain before adding it to the sugarcane. Further studies comparing other energy sources with corn as additives at the ensiling of sugarcane seem warranted as well as feeding studies to compare intakes of the various products by livestock and subsequent performance. The alternative of feeding the energy sources with straight sugarcane silage as opposed to adding them at ensiling should be assessed.