Fermentability coefficient of tropical forages and fermentative profile and in vitro degradability of silages

Abstract

Our objective was to determine the fermentability coefficient (FC) of different tropical forages and the fermentative profile, chemical composition, and in vitro degradability of their silages. A randomised block design was used, with five treatments and four repetitions. The forage species studied were whole-plant maize (Zea mays), Cenchrus purpureus cv. BRS Capiaçu, Urochloa brizantha cv. BRS Piatã, U. decumbens cv. Basilisk, and Megathyrsus maximus cv. BRS Zuri. The FC was highest (P < 0.001) in whole-plant maize, followed by BRS Capiaçu, and lowest for BRS Piatã, Basilisk, and BRS Zuri. The fibrous fraction of the plant was higher (P < 0.001) for BRS Capiaçu and lower for whole-plant maize. The pH and ammonia content were lower (P < 0.001) for maize and BRS Capiaçu silages, and higher for BRS Zuri silage. Maize silage showed the highest effective DM digestibility (P < 0.001). The BRS Capiaçu silage exhibited a higher (P = 0.009) indigestible NDF fraction. In conclusion, if whole-plant maize cannot be produced for silage, ensiling perennial grasses, such as BRS Capiaçu, BRS Piatã, and Basilisk, is recommended due to their higher ensilability compared to BRS Zuri.

Introduction

Different forage species are used for silage production worldwide, exhibiting various fermentation patterns due to their chemical and microbiological characteristics. Among the chemical characteristics, the levels of dry matter (DM), water-soluble carbohydrates (WSC), and buffering capacity (BC) are the main factors affecting the ensilability of forage^1,2. These three factors interact to generate an index that predicts the risk of clostridial fermentation in forages, known as the fermentability coefficient (FC)^2,3.

Whole-plant maize is considered the standard species for producing high-quality silage. In contrast, tropical perennial grass silages face challenges in reducing pH due to excess moisture and a low WSC content, resulting in a low FC and increasing the likelihood of clostridial fermentation^3,4. Nevertheless, because of their high forage yield per hectare, tropical perennial grasses are widely used in tropical regions for silage production, particularly species from the genera Cenchrus sp., Urochloa sp., and Megathyrsus sp⁵. Another factor influencing fermentation is the indigenous microbial population, which differs among forage species and is affected by the vegetative stage, climatic conditions, and geographic region^6–8. Thus, various factors interact during the fermentation process, which is also influenced by harvest management, water activity, and nitrate concentration⁹.

Due to the interaction of various factors, the use of the FC index may not accurately predict the ensilability of a forage species. Recently, Carvalho et al.³ demonstrated that FC developed using data from temperate grasses could be applied to tropical grasses. However, Heinritz et al.¹⁰ observed that in tropical legumes, FC was not able to predict the extent of butyric acid formation, serving only as an indicator of silage mass acidification. Thus, a better understanding of both factors related to ensilability and the nutritional value of these forages can aid in selecting forage species and in developing strategies to optimise fermentation and reduce DM losses.

Our hypothesis is that different tropical forages exhibit distinct ensilability, as well as differences in the nutritional value of their silages, even when they have similar fermentability coefficients. Our objective was to evaluate the ensilability of whole-plant maize (Zea mays) and tropical perennial grasses, namely Cenchrus purpureus cv. BRS Capiaçu, Urochloa brizantha cv. BRS Piatã, U. decumbens cv. Basilisk, and Megathyrsus maximus cv. BRS Zuri, managed under the same soil and climatic conditions, as well as comparing the fermentative profile, chemical composition, and in vitro degradability of DM and neutral detergent fiber of their respective silages.

Materials and methods

Experimental location and management of tropical forages

The experiments were conducted at the Unit of Teaching, Research, and Extension in Forage and Pasture Science and at the Forage and Silage Microbiology Laboratory of the Department of Animal Science at the Federal University of Viçosa, located in Viçosa, Minas Gerais, Brazil. The municipality has an altitude of approximately 649 m, latitude south 20°45’14”, and longitude west 42°52’54”. The climate is classified as Cwa (subtropical climate)¹¹with annual precipitation and temperature averages of 1200 mm and 21 °C, respectively.

The experiment followed a randomised block design with five treatments and four replications. The treatments consisted of the following forage species: whole-plant maize (Zea mays L.), Cenchrus purpureus (Schumach.) cv. BRS Capiaçu, Urochloa brizantha [(Hochst. ex A. Rich.) R.D. Webster] cv. BRS Piatã, U. decumbens [(Stapf) R.D. Webster] cv. Basilisk, and Megathyrsus maximus [(Jacq.) B.K. Simon & S.W.L. Jacobs] cv. BRS Zuri. The four blocks were distributed over an area of ​​600 m² and each block contained five plots (3 × 5 m), one for each forage species. The current study complies with Brazilian ethical regulations. All methods were performed in accordance with relevant guidelines and regulations for plants.

Soil correction and fertilisation were carried out according to the recommendations of Ribeiro et al.¹²based on the following soil characteristics: pH (water) = 5.0; P = 2.5 mg/dm^+ 2; K⁺ = 56 mg/dm^+ 2; Ca^+ 2 = 1.18 cmolc/dm²; Mg^+ 2 = 0.45 cmolc/dm²; H⁺ + Al^+ 3 = 6.27 cmolc/dm²; sum of base = 1.77 cmolc/dm² effective cation exchange capacity = 2.17 cmolc/dm² cation exchange capacity at pH 7 = 8.04 cmolc/dm³; and base saturation = 22%. Base saturation was increased to 60% by the prior application of 3.60 t/ha of dolomitic limestone (equivalent neutralising power of 85%). All forage species were planted on the same day, with a planting fertilisation of 90 kg/ha of P₂O₅, using a commercial 08-28-16 (N-P-K; Fertilizantes Heringer^®, Manhuaçu, MG, Brazil) mixture.

Maize (hybrid LG 6030 PRO2, Limagrain, Curitiba, PR, Brazil) was sown with 0.8 m spacing between rows at a rate of 6 seeds per linear metre. The BRS Piatã, Basilisk, and BRS Zuri (Semensol Sementes, Tupaciguara, MG, Brazil) were sown in rows spaced 0.4 m apart, with a seeding rate of 4.8 kg/ha of pure viable seeds for the Urochloa cultivars and 3.1 kg/ha for BRS Zuri. BRS Capiaçu was planted with a spacing of 0.8 m using a density of 8 buds per linear metre. All crops were planted on November 16, 2020.

At 30 days after sowing or planting, all cultivars received a dose of 40 kg/ha of nitrogen in the form of urea, and the whole-plant maize and BRS Capiaçu crops received an additional dose of 40 kg/ha of K₂O in the form of potassium chloride. Weed control was done manually. At 60 days after sowing, BRS Piatã, Basilisk, and BRS Zuri were cut at 10 cm above the ground (uniformization cutting), followed by nitrogen fertilisation with 40 kg/ha of N and 40 kg/ha of K₂O, using the commercial formulation 20-00-20 (Fertilizantes Heringer^®, Manhuaçu, MG, Brazil). Uniformization cutting and nitrogen fertilisation were carried out to ensure that the regrowth of these grasses was synchronised with the optimal ensiling maturity of BRS Capiaçu and whole-plant maize, thereby enabling all forages to be ensiled on the same day.

Harvesting and ensiling

The BRS Piatã, Basilisk, and BRS Zuri were harvested for silage 60 days after uniformization cutting, following the recommendations of Santos et al.^13,14. BRS Capiaçu and whole-plant maize were harvested 120 days after planting/sowing. At the time of harvesting for ensiling (in the same day), BRS Capiaçu, BRS Piatã, Basilisk, and BRS Zuri were 278, 99, 79, and 166 cm tall, respectively.

The forages were manually harvested, excluding the border rows. The material was chopped into a theoretical particle size of 1.5 cm using a stationary forage harvester (PN Plus 2000, Nogueira^®, São João da Boa Vista, SP, Brazil) and divided into individual 10 kg piles (1 pile per plot). No additives were applied, and no wilting was done for the perennial tropical grasses. Before ensiling, 300 g of the fresh forages were sampled and partially dried in an oven with forced air circulation at 55 °C for 72 h. Subsequently, the material was ensiled in 12-L plastic buckets (1 bucket per plot) and manually compacted with an average density of 578 kg/m³ (± 35.77) based on fresh matter (FM). The buckets were sealed with lids, secured with 6 layers of adhesive tape, and stored in a covered shed at an ambient temperature (average 19.7 °C) until the time of opening.

Analysis of chemical composition, microbial population, and fermentation profile

After 60 days of storage, the buckets were opened, and the silage from the top (± 5 cm) and bottom (± 5 cm) of the bucket was discarded. The remaining silage was homogenised, and 300 g samples from each bucket were collected for partial drying in an oven with forced air circulation at 55 °C for 72 h. All partially dried samples (fresh forage and silage) were ground in a Willey mill (Tecnal^®, Piracicaba, SP, Brazil) using a 1-mm sieve for subsequent chemical composition analysis and in vitro degradability assessment (only for silages).

To quantify the microbial population in the fresh forages and silages, an aqueous extract was obtained by homogenising 25 g of the material for 1 min in an industrial blender with 225 mL of sterile Ringer‘s solution (Oxoid™, Hampshire, UK). The aqueous extract was filtered through a double layer of sterile gauze and subjected to serial dilutions ranging from 10⁻¹ to 10⁻⁷. Plating was performed using the pour-plate technique in sterile Petri dishes. The population of lactic acid bacteria (LAB) was determined on MRS agar (Difco™ Lactobacilli MRS Agar, Le Pont de Claix, France) and incubated at 37 °C for 48 h. Enterobacteria was cultivated on VRB agar (CM0107 Violet Red Bile Agar, Oxoid™, Hampshire, UK) and incubated at 37 °C for 24 h. Yeast and filamentous fungi (FUN) were cultured on DRBC agar (Dicloran Rose Bengal Chloramphenicol, Oxoid™, Hampshire, UK) at 25 °C for 72 h for yeast and 120 h for FUN. Plates containing between 25 and 250 colony-forming units (cfu) were considered countable.

A second aliquot of the aqueous extract was used to measure the pH with a digital pH meter (Tecnal^®, Piracicaba, SP, Brazil). Subsequently, 10 mL of the aqueous extract was collected and placed in tubes containing 1 mL of sulphuric acid (50% v/v) and frozen at − 20 °C for later analyses of ammonia (g/kg N-total) according to Okuda et al.¹⁵, WSC content in fresh forages and silages according to the method proposed by Nelson et al.¹⁶and organic acids in the silages. To quantify the organic acids, the samples were treated with calcium hydroxide and copper sulphate and analysed using HPLC following Siegfried et al.¹⁷. The HPLC device (SPD-10 AVP, Shimadzu™, Tokyo, Japan) was equipped with a refractive index detector, and an Aminex HPX-87 H column (BIO-RAD™, CA, USA) was used, with a mobile phase containing 0.005 M H₂SO₄ and a flow rate of 0.6 mL/min at 50 °C.

The BC was measured according to Playne and McDonald¹⁸using 15 g of fresh forage macerated with 250 mL of distilled water. The aqueous extract was titrated to pH 3 with 0.1 N HCl to remove carbon dioxide, then titrated with 0.1 N NaOH to pH 6, noting the volume of 0.1 N NaOH used to increase the pH from 4 to 6. The BC was converted to g of lactic acid/kg DM using the equation proposed by O’Kiely and Pahlow¹⁹: BC = 0.0154 × BC (mEq/kg DM) – 0.2115 (R² = 0.95). The FC was calculated according to the equation proposed by Weissbach et al.². : FC = DM (g/kg) + 80 × WSC (g/kg DM) / BC (g lactic acid/kg DM). The minimum DM content to inhibit clostridial fermentation was determined according to Weissbach et al.². : DMmin_w = 450–80 × (WSC/BC), and by the equation proposed by Carvalho et al.³. for tropical grasses, DMmin_c = 386–79 × (WSC/BC).

The fresh forage and silage samples, ground to 1 mm, were analysed for their DM (method 934.01), ash (method 942.05), crude protein (CP; method 984.13), acid detergent fibre (ADF), and lignin contents (method 973.18), according to AOAC²⁰. The neutral detergent fibre (NDF) concentrations were determined with the addition of thermostable α-amylase without the use of sodium sulphite, following Van Soest et al.²¹. and modified by Senger et al.²². Residues from the ADF and NDF analyses were subjected to ash content determination²³ and nitrogen compound analysis²⁴. The NDF and ADF contents were expressed excluding residual ash and protein (NDFap and ADFap, respectively). Hemicellulose (HEM) was calculated by the difference between NDFap and ADFap, and cellulose (CEL) was calculated by the difference between ADFap and lignin, all expressed in g/kg DM. These calculations were performed through sequential analyses of the same sample.

In vitro degradability assay

To estimate the in vitro degradability, two heifers (1/2 Nelore × Red Angus) with an average weight of 330 kg, fitted with rumen cannulas, were housed in individual pens as donors of ruminal inoculum. The animals were adapted for 14 days²⁵ to a diet with 12% CP and a forage-to-concentrate ratio of 80:20 based on DM. The procedures for the use and handling of animals in this study were previously approved by the Animal Experimentation Ethics Committee of UFV (protocol 019/2021). All methods were carried out in accordance with relevant guidelines and regulations. The methods were also in accordance with Animal Research Reporting In Vivo Experiments (ARRIVE) guidelines for the reporting of animal experiments.

The silage samples, ground to 1 mm, were weighed (500 mg) and placed in F57 bags (Ankom Technology Corp™, Macedon, NY, USA). The ruminal fermentation simulation process was conducted in a Daisy incubator (Ankom Technology Corp™, Macedon, NY, USA) following the method proposed by Tilley and Terry²⁶ and adapted by Holden²⁷. Samples from only three of the four replicates per treatment were evaluated in this analysis, with one treatment being excluded (cultivar Basilisk) due to the limitation of the number of jars in the Daisy incubator (4). The ruminal fluid was collected 1 h after feeding, and three incubation runs (replicates) were performed. The bags were incubated at 0, 3, 6, 12, 24, 48, 72, and 96 h. Two bags were added at each time point for each forage species, and samples of each forage species were placed individually in separate jars to avoid associative effects in the degradation estimates²⁸. To maintain anaerobic conditions in the jar, CO₂ was infused every time the jars were opened. The incubations were performed in reverse order of the times so that all bags were removed simultaneously, allowing for uniform washing.

At the end of the incubations, all bags were manually washed in running water until the water became clear. The bags for time zero were not incubated but were washed with the others. After washing, the bags were partially dried in an oven with forced air ventilation at 55 °C for 72 h and then analysed for DM and NDF contents (without ash and protein correction) following the previously described methodology.

The parameters of the in vitro degradation of DM were estimated using the equation proposed by Orskov and McDonald²⁹: Y(t) = a + b × (1 – exp ^{(−kd × t)}), where Y(t) = degraded fraction of DM (g/kg); “a” = readily soluble fraction (g/kg); “b” = potentially degradable fraction in the rumen (g/kg); “kd” = rate constant for degradation of “b” per h (g/kg per h); t = time (h). After estimating the DM parameters, they were used to estimate effective digestibility and potential degradability according to the equation proposed by Orskov and McDonald²⁹: ED = a + (b × kd / kd + kp) and PD = a + b, where ED, where DE (g/kg) = ruminal effective digestibility of DM; PD (g/kg) = ruminal potential degradability; kp = ruminal passage rate (ruminal passage rates of 2, 5, and 8% per h were used).

The parameters of NDF degradation were obtained according to the equation proposed by Van Milgen et al.³⁰. : RNDF(t) = b × [1 + (λ × t)] × exp ^{(−λ × t)} + Ind, where RNDF(t) = undegraded NDF at time “t” (g/kg); “b” = potentially degradable fraction in the rumen (g/kg); λ = joint fractional rate of latency and degradation (h^− 1); t = time (h); Ind = indigestible fraction (g/kg). The degradation rate of NDF was calculated based on λ, using the properties of Γ(2) distribution³¹: kd = (0.59635 × λ), where kd = constant rate of degradation of fraction “b” (g/kg per h); λ = joint fractional rate of latency and degradation (h^− 1).

The degradation parameters a, b, kd, λ, and Ind were estimated using PROC NLIN procedures (version 9.4, SAS Institute Inc., Cary, NC, USA), assuming the Gauss–Newton algorithm for convergence.

Statistical analysis

The microbial count data were converted to a logarithmic base (log10 cfu/g). All data were analysed using the GLM procedure in SAS 9.4 according to the following model: Yijk = µ + Ei + ßj + ɛijk; where Yijk is the dependent variable; µ is the overall mean; Ei represents the fixed effect of the forage species (i = 1, 2, 3, 4, and 5); ßj represents the random effect of the block (j = 1, 2, 3, and 4); ɛijk is the random error, assuming a normal distribution of the data (NID), (0; σ2ε).

For the in vitro degradability assay, data were analysed also using the GLM procedure in SAS 9.4 according to the following model: Yijk = µ + Ei + ßj + ɛijk; where Yijk is the dependent variable; µ is the overall mean; Ei represents the fixed effect of the forage species (i = 1, 2, 3, and 4); ßj represents the random effect of the block/run (j = 1, 2, and 3); ɛijk is the random error, assuming a normal distribution of the data (NID), (0; σ2ε). Means were compared using Tukey’s test with a significance level of 0.05 to control for the probability of Type I error.

Results

Fermentability coefficient and microbial population of forages before ensiling

Tropical forage species affected all studied variables. The DM content was higher (P < 0.001) in whole-plant maize, followed by BRS Capiaçu, which did not differ from BRS Piatã (Fig. 1). The WSC concentration was higher (P < 0.001) in whole-plant maize, followed by BRS Capiaçu, which had a higher WSC concentration than BRS Piatã, Basilisk, and BRS Zuri (Fig. 2). The BC was higher (P < 0.001) for Basilisk and lower for BRS Capiaçu, which did not differ from whole-plant maize (Fig. 3).

Fig. 1.

Dry matter (DM) content of tropical forages at the time of ensiling. Means followed by different letters are significantly different according to Tukey’s test (P < 0.001, SEM = 14.245).

Fig. 2.

Water-soluble carbohydrates (WSC) content of tropical forages at the time of ensiling. Means followed by different letters are significantly different according to Tukey’s test (P < 0.001, SEM = 7.096).

Fig. 3.

Buffering capacity (BC) of tropical forages at the time of ensiling. Means followed by different letters are significantly different according to Tukey’s test (P < 0.001, SEM = 2.101).

The WSC/BC ratio was higher (P < 0.001) for whole-plant maize and BRS Capiaçu compared to the other grasses (Fig. 4), while FC was higher (P < 0.001) in whole-plant maize, followed by BRS Capiaçu, being superior to BRS Piatã, Basilisk, and BRS Zuri (Fig. 5). The minimum DM content to inhibit clostridial fermentation was similar among perennial grasses (Fig. 6); however, the levels estimated by the equation proposed by Carvalho et al.³ were much lower than those estimated by Weissbach et al.².

Fig. 4.

Water-soluble carbohydrates: buffering capacity ration (WSC/BC) of tropical forages at the time of ensiling. Means followed by different letters are significantly different according to Tukey’s test (P < 0.001, SEM = 0.505).

Fig. 5.

Fermentability coefficient (FC) of tropical forages at the time of ensiling. Means followed by different letters are significantly different according to Tukey’s test (P < 0.001, SEM = 52.356).

Fig. 6.

Minimum dry matter content of tropical forages to prevent butyric fermentation estimated using the equation proposed by Weissbach et al.². (DMmin_w; black bar) and Carvalho et al.³. (DMmin_c; gray bar) (descriptive statistics).

The initial LAB population was higher (P < 0.001) in whole-plant maize compared to BRS Capiaçu, Basilisk, and BRS Zuri but did not differ from BRS Piatã (Fig. 7). The yeast population was lower (P < 0.001) in BRS Capiaçu and BRS Zuri compared to the other cultivars. The FUN (P < 0.001) and ENT (P = 0.032) population was lower in BRS Capiaçu compared to the other grasses. The initial pH was higher (P < 0.001) for BRS Capiaçu and BRS Zuri compared to the other grasses (Fig. 7).

Fig. 7.

Microbial population and pH of tropical forages at the time of ensiling. Means followed by different letters are significantly different according to Tukey’s test. Lactic acid bacteria (LAB; P < 0.001, SEM = 0.101), Yeasts (P < 0.001, SEM = 0.140), Filamentous fungi (FUN; P < 0.001, SEM = 0.173), Enterobacteria (ENT; P = 0.032, SEM = 0.260), and pH (P < 0.001, SEM = 0.049).

Chemical composition of the forage before ensiling

There was an effect of the tropical forage species on all evaluated chemical characteristics (P ≤ 0.022; Table 1). The ash concentration was higher (P < 0.001) in BRS Zuri, which did not differ from Basilisk. The concentrations of NDFap and ADFap were lower (P < 0.001) in whole-plant maize and higher in BRS Capiaçu. The HEM content was higher (P < 0.001) in BRS Piatã and Basilisk, intermediate in BRS Capiaçu and BRS Zuri, and lower in whole-plant maize (Table 1). The lignin content was higher (P = 0.022) in BRS Capiaçu and lower in whole-plant maize. The CEL content was higher (P < 0.001) in BRS Capiaçu, followed by BRS Piatã, Basilisk, and BRS Zuri. Whole-plant maize showed the lowest CEL content (Table 1). The CP levels were higher (P < 0.001) in BRS Piatã, Basilisk, and BRS Zuri grass, and lower in whole-plant maize. NDIP was higher (P < 0.001) in BRS Zuri and lower in Basilisk grass, which did not differ from BRS Piatã or whole-plant maize. ADIP was higher (P < 0.001) in BRS Capiaçu compared to the other cultivars (Table 1).

Table 1.

Chemical composition (g/kg DM) of different tropical forages species before ensiling¹.

Item1	Tropical forages					SEM2	P-value3
	Maize	BRS Capiaçu	BRS Piatã	Basilisk	BRS Zuri
Ash	32.0d	54.3c	74.5b	86.1ab	98.3a	5.563	< 0.001
NDFap	476c	733a	672b	678b	672b	20.433	< 0.001
ADFap	241c	462a	381b	382b	404b	16.911	< 0.001
Lignin	42.5b	70.1a	52.1ab	57.3ab	53ab	1.974	0.022
HEM	235c	272b	291a	296a	268b	5.065	< 0.001
CEL	198c	392a	329b	324b	351b	15.233	< 0.001
CP	59.3c	77.2b	102a	106a	101a	4.473	< 0.001
NDIP	279cb	311b	286cb	251c	453a	16.012	< 0.001
ADIP	55.0b	72.1a	43.3b	43.4b	56.6b	2.830	< 0.001

NDFap - neutral detergent fiber expressed exclusive of residual Ash and protein; ADFap - acid detergent fiber expressed exclusive of residual Ash and protein; HEM - hemicellulose; CEL - cellulose; CP - crude protein; NDIP - acid detergent insoluble protein (g/kg CP); ADIP - neutral detergent insoluble protein (g/kg CP)². SEM - standard error of the mean³. Means with different letters differ by tukey’s test (P < 0.05).

Fermentation profile and microbial population of silages

The population of LAB was higher (P < 0.001) in the silage of BRS Piatã, Basilisk, and BRS zuri, which did not differ from BRS Capiaçu silage, and was lower in maize silage. The yeast (P = 0.173) population did not differ among the silages. Enterobacteria were detected only in the BRS Zuri silage (3.73 log cfu/g). The FUN population was higher (P = 0.008) in the BRS Capiaçu silage, which did not differ from the maize silage (Table 2). Maize and BRS Capiaçu silages had the lowest (P < 0.001) pH values, while BRS Zuri silage had the highest pH value. Maize silage had a higher (P < 0.001) lactic acid content, followed by BRS Capiaçu silage, whereas BRS Zuri silage had the lowest lactic acid content, not differing from Basilisk silage (Table 2).

Table 2.

Microbial population (log cfu/g FM), pH, and fermentation end-products (g/kg DM) of silages from different tropical forages species after 60 days of storage¹.

Item1	Tropical forages silage					SEM2	P-value3
	Maize	BRS Capiaçu	BRS Piatã	Basilisk	BRS Zuri
LAB	6.11b	7.08ab	7.98a	7.97a	8.14a	0.211	< 0.001
ENT	nd	nd	nd	nd	3.73	-	-
Yeast	4.25	3.74	3.76	3.62	4.17	0.104	0.173
FUN	2.98ab	3.51a	2.45b	2.44b	2.61b	0.128	0.008
pH	3.69c	3.63c	4.49b	4.61b	5.50a	0.161	< 0.001
LA	85.8a	45.3b	18.6c	14.9 cd	1.1d	7.474	< 0.001
AA	16.7ab	7.0c	13.9b	14.6b	23.2a	1.345	< 0.001
PA	3.2a	1.97b	2.65ab	2.07b	1.93b	0.138	< 0.001
BA	nd	nd	nd	nd	16.1	-	-
Ethanol	30.8a	12.8b	13.0b	9.75b	10.5b	1.930	< 0.001
Ammonia	45.7c	62.2c	105bc	158b	394a	30.035	< 0.001

LAB - lactic acid bacteria; ENT – enterobacteria; nd - not detected; FUN - filamentous fungi; LA - lactic acid; AA - acetic acid; PA - propionic acid; BA – butyric acid; ammonia (g/kg total nitrogen)². SEM - standard error of the mean³. Means with different letters differ by tukey’s test (P < 0.05).

The acetic acid concentration was higher (P < 0.001) in BRS Zuri silage, which did not differ from maize silage, and was lower in BRS Capiaçu silage (Table 2). Propionic acid was higher (P < 0.001) in maize silage, which did not differ from BRS Piatã silage. BRS Zuri silage had a high concentration of butyric acid (16.1 g/kg), which was not detected in the other silages. The ethanol concentration was higher (P < 0.001) in maize silage compared to the others. The ammonia concentration was higher (P < 0.001) in BRS Zuri silage and lower in maize and BRS Capiaçu silages, which did not differ from BRS Piatã silage (Table 2).

Chemical composition of silages

Maize silage had the highest (P < 0.001) DM content, while Basilisk silage had the lowest, not differing from BRS Zuri silage. The ash content was higher (P < 0.001) in BRS Zuri silage, which did not differ from Basilisk silage (Table 3). The NDFap content was lower (P < 0.001) in maize silage compared to the other silages. The ADFap and lignin contents were lower (P < 0.001) in maize silage and higher in BRS Capiaçu silage. HEM and CEL were lower (P < 0.001) in maize silage compared to the other silages (Table 3). The CP content was higher (P = 0.005) in BRS Piatã silage, which did not differ from BRS Capiaçu and Basilisk silages, and was lower in maize silage, which did not differ from BRS Zuri silage. The NDIP content was higher (P < 0.001) in BRS Zuri silage compared to the other silages. There was no effect of forage species on the ADIP content (P = 0.488), with a mean value of 47.6 g/kg of PB. Maize silage had a higher (P < 0.001) WSCr content (Table 3).

Table 3.

Chemical composition (g/kg DM) of silages from different tropical forage species after 60 days of storage¹.

Item1	Tropical forages silage					SEM2	P-value3
	Maize	BRS Capiaçu	BRS Piatã	Basilisk	BRS Zuri
DM	322a	201b	195cb	161d	176 cd	13.332	< 0.001
Ash	33.5d	56.3c	77.6b	94.3ab	104a	6.056	< 0.001
NDFap	444b	693a	646a	644a	673a	21.401	< 0.001
ADFap	226c	426a	379b	381ab	417ab	17.129	< 0.001
Lignin	31.6c	61.0a	46.8b	51.6ab	48.6b	2.353	< 0.001
HEM	218b	267a	267a	264a	257a	4.725	< 0.001
CEL	203b	365a	332a	329a	361a	14.037	< 0.001
CP	62.6c	84.0abc	95.6a	88.8ab	69.8bc	3.585	0.005
NDIP	130b	113b	104b	164b	300a	18.235	< 0.001
ADIP	39.7	45.6	38.6	54.1	60	4.038	0.488
WSCr	9.0a	5.1b	4.6b	5.8b	5.0b	0.396	< 0.001

DM - dry matter (g/kg fresh matter); NDFap - neutral detergent fiber expressed exclusive of residual Ash and protein; ADFap - acid detergent fiber expressed exclusive of residual Ash and protein; HEM - hemicellulose; CEL – cellulose; CP - crude protein; NDIP - acid detergent insoluble protein (g/kg CP); ADIP - neutral detergent insoluble protein (g/kg CP); WSCr – water-soluble carbohydrates residual². SEM - standard error of the mean³. Means with different letters differ by tukey’s test (P < 0.05).

In vitro degradability of silages

Maize silage had the highest soluble fraction “a” (P < 0.001) and degradation rate “kd” of DM (P < 0.001), while BRS Zuri silage had the lowest values for these variables (Table 4). The potentially degradable fraction in the rumen “b” showed the opposite trend, with a higher (P < 0.001) content in BRS Zuri silage and a lower content in maize silage (Table 4). The disappearance behaviour of DM over incubation times is shown in Fig. 8.

Table 4.

In vitro degradability of DM and NDF of silage from different species of tropical forage¹.

Item1	Tropical forages silage				SEM2	P-value3
	Maize	BRS Capiaçu	BRS Piatã	BRS Zuri
DM (g/kg)
a	316a	134b	142b	95.4c	25.734	< 0.001
b	496c	579bc	666b	827a	38.413	< 0.001
kd	0.0344a	0.0245b	0.0229b	0.0141c	0.002	0.001
PD	812b	714b	808b	922a	24.403	0.003
ED 2	629a	453c	496b	435c	22.963	< 0.001
ED 5	517a	325b	350b	276c	27.458	< 0.001
ED 8	464a	270b	289b	219c	27.993	< 0.001
NDF (g/kg DM)
b	644ab	578b	674a	664a	13.188	0.012
ʎ	0.0512ab	0.0614a	0.0594ab	0.0494b	0.00189	0.044
kd	0.0306ab	0.0366a	0.0354ab	0.0294b	0.00113	0.044
Ind	333b	439a	344b	354b	14.452	0.009

DM – dry matter; NDF - neutral detergent fiber; a - readily soluble fraction; b - potentially degradable fraction in the rumen; kd - rate constant for degradation of “b” (per h); PD - potential degradation; ED - effective degradability at different passage rates (2. 5. and 8%); λ - joint fractional rate of latency and degradation (per h); Ind - indigestible fraction of NDF². SEM - standard error of the mean³. Means with different letters differ by tukey’s test (P < 0.05).

Fig. 8.

In vitro DM disappearance curve as a function of incubation time of silages from different tropical forages (descriptive statistics). The bars represent the standard error of the mean.

The potential degradability “PD” (P = 0.003) of DM was higher for BRS Zuri silage compared to the other silages. However, the effective DM degradability was higher in maize silage, regardless of passage rate. The ED at 2% was lower (P < 0.001) in BRS Capiaçu and BRS Zuri silages. However, with an increase in the passage rate to 5 and 8%, BRS Zuri silage showed the lowest (P < 0.001) ED compared to the others. The ED of BRS Piatã and BRS Capiaçu silages were similar at 5 and 8% passage rates (Table 4).

The potentially degradable fraction “b” of NDF was higher (P = 0.012) in BRS Piatã and BRS Zuri silages compared to BRS Capiaçu silage, without differing from maize silage. The joint fractional latency and degradation rate (λ) were higher (P = 0.044) for BRS Capiaçu silage compared to BRS Zuri silage, without differing from corn and BRS Piatã silage (Table 4). The indigestible fraction of NDF was higher (P = 0.009) in the BRS Capiaçu silage compared to the other silages (Table 4). The disappearance behaviour of NDF over incubation times is shown in Fig. 9.

Fig. 9.

In vitro NDF disappearance curve as a function of incubation time of silages from different tropical forages (descriptive statistics). The bars represent the standard error of the mean.

Discussion

The ensilability of a tropical forage can be estimated based on the levels of DM, WSC, and BC, generating an index that predicts the risk of inadequate fermentation, mainly by clostridia². According to Carvalho et al.³. tropical grasses require a minimum FC of 400 to ensure butyric acid-free fermentation without the addition of additives. However, this minimum FC can be reduced depending on the DM content, the initial population of LAB, and nitrate concentration.

Whole-plant maize and BRS Capiaçu exhibited an FC above 400, indicating high ensilability. Indeed, these two species produced silages with a pH below 4.2, low ammonia concentrations, and no detectable butyric acid, indicating a proper fermentation profile³². The FC for whole-plant maize found in our study is close to the FC reported by Wang et al.³³. However, the FC values reported in the literature for C. purpureus cv. Napier³⁴ and other clones of this species³⁵ are significantly lower than our findings, which can be attributed to differences in the cultivar or clone. In our study, we used cultivar BRS Capiaçu, which has been reported to have an adequate fermentation pattern when harvested with a DM content close to 200 g/kg of FM^36,37.

However, the other perennial forage species showed an FC < 350, indicating low ensilability and a higher risk of Clostridium fermentation. Nevertheless, only BRS Zuri silage exhibited typical clostridial fermentation, according to Kung et al.³². Despite having similar DM and WSC contents and WSC/BC ratios, the fermentation profiles of the tropical perennial grasses differed. Therefore, FC should not be used as the sole indicator of ensilability potential, as the success of fermentation in the silo also depends on other management factors.

The DM content is the variable that most significantly impacts fermentation³. Spoelstra³⁸ reported that silages with more than 250 g of DM/kg of FM did not experience clostridial fermentation. In our study, only whole-plant maize had a DM content exceeding 250 g/kg. However, despite having a lower DM content, BRS Capiaçu silage exhibited a fermentation pattern similar to maize silage, with low pH and ammonia values, indicating that the WSC content in this crop was sufficient for rapid acidification and inhibition of undesirable microorganisms. Indeed, the WSC content of BRS Capiaçu was close to the minimum (60 g/kg) recommended by McDonald et al.³⁹. Additionally, the WSC/BC ratio, representing the acidification capacity^2,3was greater than 3, indicating that the amount of fermentable carbohydrates was sufficient for lactic acid bacteria to acidify the silage mass³.

In tropical grasses, the greatest challenge for silage production is the inhibition of clostridial fermentation, primarily due to the low DM content found in these plants³. The minimum DM content required to inhibit clostridial fermentation was 387 g/kg, according to the equation proposed by Weissbach et al.². This value is quite high and difficult to achieve under practical conditions when ensiling forage plants with good nutritional value without using wilting or moisture-sequestering additives. However, using the equation proposed by Carvalho et al.³. with data from tropical grasses, the minimum DM content to prevent butyric acid formation is 324 g/kg of FM, which is more easily achievable and requires less moisture-sequestering additive or wilting time. Additionally, these authors observed that the minimum DM content decreased linearly as the WSC/BC ratio increased. This demonstrates that producing high-quality silages from tropical perennial grasses is not unfeasible, although it is challenging in field conditions.

The silages of BRS Piatã and Basilisk exhibited a typical fermentation pattern for tropical grasses⁴⁰. Despite these crops having low DM and WSC contents, the pH of their silages was reduced and stabilised at 4.49 and 4.61, respectively, within an acceptable range for grass silages, according to Kung et al.³². This reduction was sufficient to suppress the presence of enterobacteria and the activity of clostridia, thereby reducing proteolysis. These silages had a higher ammonia concentration compared to maize silage (reference silage), but the levels remained within an acceptable limit, according to Kung et al.³². Additionally, no butyric acid was detected in the silages.

Although BRS Piatã, Basilisk, and BRS Zuri cultivars initially had similar levels of WSC, some of the fibre can be solubilised during fermentation, providing soluble carbohydrates for LAB^1,41. In the present study, BRS Piatã and Basilisk showed higher HEM concentrations at the time of ensiling, and over the storage period, there was a more pronounced reduction (2.65% points) in this fibre component in their silages. However, little difference was observed in the HEM content between the ensiled material and the Zuri silage (reduction of 0.4% points). The greater HEM solubilisation in the BRS Piatã and Basilisk cultivars may have provided WSC, which was not quantified in the plant because it was in an insoluble form. The release of WSC possibly allowed the acidification of the ensiled mass⁴²thus improving the fermentation profile of BRS Piatã and Basilisk silages compared to BRS Zuri silage, even though they had similar CF levels. Furthermore, Carvalho et al.³. reported that the minimum nitrate content required to inhibit the action of clostridia is higher in Aries II (M. maximus) silage than in marandu grass (U. brizantha) silage. The causes of this are not fully understood, but it partially explains the differences in the fermentation profiles between the BRS Zuri, BRS Piatã, and Basilisk silages observed in this study. The nitrate content was not measured in our study; however, it is mainly influenced by nitrogen fertilisation⁴³which was similar among the grasses. Therefore, nitrate levels are expected to be close, reinforcing the findings of Carvalho et al.³.

Some studies have shown that the chemical composition of forage influences the fermentation profile more than the initial autochthonous population^8,44as it affects microbial succession throughout the fermentation process due to differences in the pH reduction rate, a factor responsible for significant variations in the microbial population⁴⁵. In our study, we observed this specifically with cultivar BRS Capiaçu, which, despite having a low initial LAB population, was efficient in dominating the process and providing good fermentation, inhibiting the growth of enterobacteria, which can be attributed to its higher WSC content.

BRS Zuri silage exhibited inadequate fermentation despite having a high initial autochthonous LAB population and a low enterobacteria population, highlighting the importance of faster acidification and favourable microbial succession. After 60 days of storage, enterobacteria was suppressed in all silages except in BRS Zuri silage, indicating high competition with LAB, which, despite being in a high population, was insufficient to conduct adequate fermentation. The reduction in undesirable microbial populations in silage reflects the combined presence of good ensilability conditions, including nutrient availability and water, efficient conversion of these nutrients into fermentation products, and pH reduction⁹. Furthermore, the presence of enterobacteria in BRS Zuri silage could also explain the high ammonia concentration. According to Li et al.⁴⁶. a high presence of genes encoding proteolytic enzymes was detected in alfalfa silage, primarily attributed to the Enterobacteriaceae family.

According to Gomes et al.⁴⁷. a high LAB population is not sufficient to prevent clostridial fermentation in grass silage with a low WSC, as observed in BRS Zuri silage, which had a low WSC/BC ratio at ensiling. Therefore, FC could be used as a guideline in choosing additives for silage, emphasising the importance of characterising forages used for silage production³. Grasses with an FC below 350 require additives to increase the DM content (moisture sequestrants or wilting) or to provide soluble sugars (e.g., molasses or fibrolytic enzymes), as these components directly impact FC. However, for an FC between 350 and 450, the use of inoculants containing homofermentative LAB would optimise WSC utilisation, leading to greater lactic acid formation and a rapid pH reduction. For grasses with an FC above 450, naturally adequate fermentation is expected but attention should be paid to post-silo opening issues due to the lower formation of antifungal acids under these conditions⁴⁸. Additionally, the combined adoption of the two types of additives could generate a synergistic effect under conditions of low ensilability^10,49,50.

The chemical composition varied among the studied species, especially concerning the fibrous fraction and protein content. Regarding the fibrous fraction, BRS Capiaçu exhibited higher contents of NDFap, ADFap, and lignin at ensiling, which can be attributed to the growth pattern of this species, requiring greater deposition of structural components for support. However, in whole-plant maize, the deposition of NDFap and ADFap differs from that of perennial grasses due to starch accumulation during grain filling, leading to a dilution effect of the fibrous fraction. This results in maize silage having lower NDFap and ADFap contents compared to perennial tropical grasses.

The higher protein content in BRS Piatã, Basilisk, and BRS Zuri compared to BRS Capiaçu and whole-plant maize can be attributed to the higher proportion of leaves in these crops at ensiling. In the silages, the protein content mainly impacted BRS Zuri silage, which showed fermentation with high ammonia formation, indicating intense proteolysis. This led to a reduction in protein content from 101 g/kg (forage) to 69.8 g/kg (silage). As a portion of the easily accessible protein for microorganisms was lost, the fraction associated with fibre became concentrated, resulting in higher NDF-bound protein (NDIP) levels. The ADIP content differed only in fresh forages, with this difference disappearing in silages. According to Muhandiram et al.⁵¹. protein losses through proteolysis during fermentation pose a barrier to the development of sustainable ruminant production systems. Therefore, characterising commonly used forage species in silage production can facilitate decision-making in choosing additives, increasing their efficiency, and making the system more productive and sustainable.

Maize silage exhibited the highest levels of soluble fraction ‘a‘ of DM, which is readily degraded in the rumen. This was mainly due to the presence of higher WSCr and starch concentrations, which, after undergoing fermentation, had a higher degradation rate due to the breakdown of the protein matrix covering starch granules^52,53. Furthermore, soluble proteins are contained in fraction ‘a‘ of DM. Maize silage showed a lower NDF-bound protein (NDIP) content compared to BRS Zuri silage. This fibre-associated protein has a slower degradation rate, thereby reducing the soluble or readily available fraction in the silage. BRS Zuri silage exhibited a lower soluble fraction ‘a’, which may be attributed to its inadequate fermentation profile; however, we observed a higher potentially degradable fraction ‘b’ in this silage. According to McCuistion et al.⁵⁴silages with a higher soluble fraction ‘a’ of DM tend to have a lower potentially degradable fraction and vice versa, which aligns with our findings, in which the potentially degradable fraction ‘b’ of DM was lower for maize silage and higher for BRS Zuri silage.

PD is related to the capacity of material to degrade under an infinite rumen residence time. However, it is recognised that the rumen is dynamic and exhibits different passage rates. As observed, effective degradation using passage rates of 2, 5, and 8% differed significantly from PD. With higher passage rates, the residence time of silage in the rumen decreases, meaning silages with higher degradation rates (‘kd’) are more efficient in providing nutrients and avoiding limitations in ruminal filling^55,56. Thus, regardless of the passage rate, maize silage is more efficient in terms of nutrient availability to animals, especially those with higher passage rates. When we analysed the PD, we observed a higher value for BRS Zuri silage. However, due to its fermentation profile, there may be a lower voluntary intake of this silage when provided to animals⁵⁷. Additionally, in silages with clostridial fermentation, the formation of butyric acid and biogenic amines can reduce acceptance by animals and increase the incidence of ketosis in dairy cows during the transition period^57,58.

A factor that directly influences animal productivity, especially in dairy cows, is the NDF digestibility⁵⁹. The potentially degradable fraction of NDF was higher for perennial grass silages, except for the BRS Capiaçu silage. The lower potentially degradable fraction of NDF in BRS Capiaçu silage can be explained by its high lignin content, which forms bonds with CEL and HEM, thus reducing its degradability⁶⁰. Moreover, the higher lignin content increases the indigestible fraction of NDF, causing a greater physical effect in the rumen and directly impacting animal intake⁵⁶.

In conclusion, tropical forages with similar fermentability coefficients exhibit different fermentation profiles; therefore, this isolated indicator does not determine the ensilability or the fermentation profile of tropical grasses, as other factors influence ensilability.

Whole-plant maize and BRS Capiaçu produce silages with better fermentation profiles due to higher fermentation coefficients. However, the silage of the BRS Piata and Basilisk can be considered of acceptable quality, despite their lower fermentation coefficient. The nutritive value varies among grasses and the choice of which specie to ensile should be based on the nutritional requirements of the animal category.

Author contributions

Author’s contributionW.S.A. Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing - original draft, Writing -review & editing. G.F.L.C. Formal analysis, Investigation, Validation, Visualization. R.L.F. Formal analysis, Investigation, Validation, Visualization. T.C. S. Investigation, Visualization. D.N.C. Investigation, Visualization. J.P.S.R. Investigation, Visualization. T.N.T. Investigation, Visualization. J.P.S.R. Investigation, Methodology, Validation, Visualization O.G.P. Methodology, Resources, Validation, Visualization. K.G.R. Conceptualization, Data curation, Funding acquisition, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing - review & editing.All authors reviewed the manuscritp.

Funding

Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG-APQ-01527-21) and Coordination of Improvement of Higher Education Personnel (CAPES-PROEX-88887.844747/2023-00).

Data availability

All data generated or analyzed during this study are included in this published article.

Declarations

Competing interests

The authors declare no competing interests.