Species and Harvest Time of Fresh Tropical Grasses Affect Rumen Fermentation as Determined by In Sacco and In Vitro Incubations

Abstract

The consumption of forages high in slow-degradable carbohydrates by dairy cattle leads to greater ruminal acetate production, which benefits milk fat content. Although tropical grasses are typically rich in fibrous materials, the milk fat content of dairy cows in Southeast Asia is low. Here, we investigate the effects of the species and harvest time of three common tropical grasses (Guinea, King, and Mulato II) harvested at three instances (early, normal, late) on in sacco degradation kinetics and in vitro gas production (IVGP) characteristics. Grass samples were subjected to (1) chemical analysis, (2) a fully automated in vitro gas recording system using rumen fluid to measure fermentation characteristics over 72 h, and (3) in sacco degradation using the nylon bag technique, employing seven incubation times up to 336 h. Forage quality decreased with maturity, as reflected in changes to digestibility and fiber content. Overall, early harvested grasses yielded the highest total gas production (311 ± 12.5 mL/g OM) followed by normal (300 ± 45.7 mL/g OM) and late (273 ± 19.5 mL/g OM) harvested grasses. The in vitro fermentable fraction (A1 + A2) was the highest for early harvested grasses, with the A2 parameter, relevant for milk fat content, being the highest for Guinea (81.6% A2/(A1 + A2)) and the lowest for King grass (71.0% A2/(A1 + A2)). Consequently, early harvested Guinea had the longest incubation times (10.5 h) and lowest fermentation rates (Rmax2 = 12.8 mL/g OM/h). Regression analysis showed relationships between NDF content and degradability. Harvesting tropical grass earlier than customarily practiced enhanced forage quality and ruminal degradability. Of the three grasses studied, each at three levels of maturity, early harvested Guinea grass was the most promising candidate for improving milk fat content in Southeast Asian dairy cows. This grass showed a high fermentable OM content, with a large proportion of slow-degradable carbohydrates.

1. Introduction

Dietary neutral detergent fiber (NDF) content has important implications for dairy production, as the fermentation of NDF primarily yields acetate and therefore (potentially) increases milk fat content [1]. Conversely, the fermentation of water-soluble carbohydrates (WSCs) promotes propionate formation, thereby stimulating milk yield [2,3]. Thus, greater proportions of NDF vs. WCS in dairy cow rations may enhance the fat content of milk. Tropical grasses typically contain higher NDF contents than grasses from temperate regions [4,5,6]. While the higher NDF content of tropical grasses might suggest enhanced milk fat content, empirical evidence indicates otherwise. In Southeast Asia, for instance, dairy cows consistently have relatively low milk fat percentages (below 4%, [7]). This (apparent) contradiction highlights the nuanced interactions between grass composition, ruminal degradation, and cow milk production. Unfortunately, detailed knowledge on the NDF degradation of tropical grasses is scant, which underscores the need for region-specific nutritional research.

The organic matter degradability (OMD) of selected tropical grasses was studied in an in vitro gas production technique (IVGPT) experiment [8]. In this study, the total gas production or proportions of produced volatile fatty acid (VFA) fermentation end products were found to be relatively similar to what is reported for temperate grasses in the literature. Obviously, in vitro studies do not fully represent the conditions of the rumen. To determine the rate and extent of the rumen degradability of grasses, the in sacco (nylon bag) technique has been widely used. While in sacco analysis provides information on the kinetics of substrate degradation, the IVGPT provides information on the kinetics of gas production which reflects the production of VFAs in rumen fermentation [9]. Combining data obtained by the IVGPT with the in sacco ruminal degradation of these grasses provides more comprehensive information on in vivo rumen processes.

Although the IVGPT has been applied to some tropical grasses [10,11,12], very few studies determined the effects of harvest time on gas production kinetics [13], and it is unclear how degradability may differ for tropical grasses compared to temperate grasses. Furthermore, dried substrates have been generally utilized in the in sacco method and IVGPT, whereas fresh materials as consumed by cows have rarely been used. While fresh samples provide a more realistic view on ruminal degradation, it is true that by using dry samples, risks in terms of variability due to nutrient leaching are avoided. Therefore, samples that were not incubated (time point 0 h) were subjected to a washing step.

The objective of the current study was to determine the effects of grass species and harvest time on in sacco degradation kinetics and in vitro gas production characteristics utilizing fresh samples of three tropical grasses (Guinea, King, and Mulato II) commonly used in Southeast Asia. Additionally, we aimed to determine correlations between the nutritive values of the grasses on the one hand and in sacco degradation parameters on the other hand. Based on a previous evaluation of various feedstuffs commonly used in Vietnam [8], we hypothesized that tropical grass species differ significantly in their NDF degradation kinetics and that harvest time influences both degradability parameters and fermentation end-product profiles. These differences in degradation patterns could help in understanding the paradoxical relationship between the high NDF content and low milk fat percentages observed in Southeast Asian dairy production systems.

2. Materials and Methods

2.1. Grass Sample Collection

Three tropical forage grasses were evaluated in this study: Guinea grass (Panicum maximum Jacq.), King grass (Pennisetum purpureum × Pennisetum glaucum), and Mulato II grass (Brachiaria ruziziensis (B. ruziziensis × B. decumbens × B. brizantha)). Guinea grass is characterized by moderate growth rates, reaching heights of 40–75 cm with fresh matter yields of 3.97–8.53 tons/ha/cut within 60 days of growth. King grass demonstrates vigorous growth capacity, attaining heights exceeding 120 cm and the highest biomass production (4.47–15.80 tons/ha/cut) during the same period. Mulato II exhibits a more compact growth habit with heights ranging from 30 to 47 cm yet reaching high yields (5.40–9.80 tons/ha/cut), particularly during early growth phases [14].

The grasses, planted at the Animal Husbandry Research and Development Centre for Mountainous Zone, Song Cong town, Thai Nguyen province, Vietnam, were harvested at different regrowth stages between June and August 2018. The center is located at 21°29′14″ N 105°48′47″ E and experiences an annual rainfall of 2168 mm with an average temperature of 23 °C. An initial fertilizer dressing of N:P:K with 160:80:80 kg/ha/yr was applied at sowing, with further annual applications at the same rate. Annually, an amount of 20 tons/ha/yr of cattle manure was applied manually. The cattle manure used was not chemically analyzed, but manure from Vietnamese dairy cattle typically contains N:P:K with a ratio of 4.0:1.9:1.6.

All grasses were cut and harvested according to predetermined harvesting time points, which were based on standard local farming practices. Harvest times were specifically selected for each grass species to reflect authentic practices, rather than standardized experimental conditions. This ensures comparable physiological stages across species despite their inherent growth differences, allowing for representative inter-species comparisons and ecological validity. For each grass species, three harvesting times, relative to local farming practices, were selected: early (−2 weeks), normal (0 weeks), and late (+2 weeks). Specifically, Guinea grass was harvested at weeks 2, 4, and 6; King grass at weeks 5, 7, and 9; and Mulato II at weeks 4, 6, and 8. For each grass species at each harvest time, a 10 m × 10 m area was marked out for sampling, and by walking in a ‘W’ pattern, 20 evenly spaced cores were manually cut and collected using a sickle. The grass was harvested, leaving around 10 cm of stubble above the ground level. After harvesting, each selected species and harvesting time grass sample was manually cut to 3 cm and mixed thoroughly before collecting a 5 kg representative sample which was then divided equally across two bags (one for analysis and one for reserve) and stored at −20 °C in Vietnam. Subsequently, all the frozen grass samples were transported in sealed plastic bags maintaining −20 °C conditions to Wageningen University & Research (WUR), the Netherlands, for analyses.

2.2. Chemical Analyses

Upon arrival at WUR, the frozen grass samples were thawed, finely cut to a length of ~1 cm using a paper slicer (JAC Duro BEL 450, ABO, Leek, The Netherlands), and homogenized. Representative subsamples (~500 g fresh weight) were dried for 16 h at 70 °C before being ground (1 mm screen) using a cross beater mill (Peppink 100 AN, Deventer, The Netherlands) and analyzed in duplicate for dry matter (DM) and ash, following AOAC [15] method no. 930.15 and no. 942.05. The organic matter (OM) content was calculated as the difference between DM and ash contents. Ether extract (EE) was determined by the Soxhlet method with petroleum ether as a solvent following AOAC [15] method no. 963.15. Crude protein (CP) was calculated from nitrogen (N × 6.25) obtained via Kjeldahl method no. 5983-2 [16]. The neutral detergent fiber (NDF, with heat-stable α-amylase) content was analyzed according to Van Soest et al. [17], while acid detergent fiber (ADF) and acid detergent lignin (ADL) contents were determined according to Van Soest [18].

2.3. In Sacco Determination of Rumen Degradability

2.3.1. Sampling

After thawing and finely cutting the grass samples to a length of ~1 cm, samples were immediately re-frozen and stored at −20 °C pending further treatment. To be used for an in sacco study, the grass samples were defrosted and homogenized again, and representative samples (4.87 ± 0.236 g of DM) were accurately weighed into N-free 10 cm × 19 cm nylon bags (porosity 30%, pore size 40 μm; PA 40/30, Nybolt, Switzerland). With increasing incubation times, more bags were used to ensure sufficient residue for later evaluation. Specifically, 4 bags were used for 0, 3, and 6 h of incubation; 5 bags were used for 9 and 12 h of incubation; 7 bags for 24 h; and 10 bags for 48 h of incubation. To ensure the correct determination of the undegradable fraction, 10–12 bags were subjected to long-term incubation (336 h), following a standardized protocol as described by Tas et al. [19] and Ali et al. [20]. In total, 1243 bags were kept frozen until the start of rumen incubations. Occasionally, a bag was inadmissible, usually due to the rupture of the bag. In the present study, the number of bags per incubation that were inadmissible was never more than two.

2.3.2. In Sacco Procedure

The in sacco incubation procedure was performed as described by Tas et al. [19] and de Jonge et al. [21]. Three rumen-fistulated Holstein-Friesian cows that were in similar lactation stages (~248 ± 184 days in milk with 26.1 ± 9.81 kg milk/d) at Utrecht University (Utrecht, The Netherlands) experimental facilities were used for the 0–48 h incubations. As the final incubation required 10–12 bags for each sample type, four cows were needed for the 336 h incubations to avoid overfilling the rumen, which can hinder the in sacco degradation process. Therefore, for the 336 h incubations, four cows at the experimental facilities of WUR were used. According to Ali et al. [22], there is no variation between cows in long-term incubation, such as one lasting 336 h. All these cows were fed twice per day at the maintenance level with an ad libitum water supply. The roughage contained 40% maize silage and 60% grass silage, with the addition of concentrates.

Nylon bags containing samples were incubated for the seven different time points (3, 6, 9, 12, 24, 48, and 336 h) over a seven-week period. For each time point, the all-in all-out procedure was used for the incubation, where bags of the same time point were all placed into and taken out of the rumen at the same time. The bags were randomly distributed across different animals to control for and minimize potential cow-specific effects on the disappearance measurements. In order to decrease variation, all incubations started at the same time relative to the morning feeding. For each incubation time point, each cow had a maximum of 35 bags containing different sample types. These were placed in a ‘larger mesh’ net attached to a stainless steel weight (1.5 kg) to ensure the bags remained submerged in the ventral rumen sac.

After incubation, the bags were retrieved from the rumen and placed immediately into ice water to cool down in order to terminate fermentation. Then, the bags were washed using a domestic washing machine following the standard ‘wool wash program’ without centrifuging. In addition, unincubated bags (0 h) were cooled and washed as well to determine the washable (W) fraction [19]. After the washing procedure, the contents of all bags were oven-dried at 70 °C until stable weight and stored in sealable plastic bags. Nylon bags containing grass samples from the same incubation time were acclimatized for 30 min, weighed, pooled, ground to pass a 1 mm sieve (Retsch ZM200, Retsch, Wuppertal, Germany), and then stored in plastic pottles, pending chemical analyses of DM, ash, and NDF.

2.3.3. In Sacco Experiment: Curve Fitting and Calculations

Incubated residue weights for the OM and NDF fractions were divided by their respective initial weight for every nylon bag to calculate the amount degraded. Subsequently, time points containing outliers were identified using the coefficient of variation (>10%) of DM disappearance. Outliers within the identified time points were selected and removed based on visual inspection. A total of 34 outliers were removed from the entire data set (n = 424). The parameters calculated were the washable fraction (W), potentially degradable fraction (D), and the undegradable fraction (U), as well as the rate of degradation of D (Kd). W and U were measured as the residue after washing without incubating or after 336 h incubation in the rumen, respectively. Using the PROC NLIN version 2023.03 procedure of SAS Institute Inc. (SAS version 9.4, 2012, Cary, NC, USA), the OM and NDF residue percentages were fitted to the following equation [23] to derive degradation parameters (D and Kd) for each grass:

$$P=W+D(1-e^{-Kd\times t})$$

where P is the percentage of OM or NDF that disappeared at time t; W is the intercept representing the washable fraction (%); D is the potentially degradable fraction (%); Kd is the fractional degradation rate of fraction D (%/h); and t is the time of incubation.

Using the above estimated parameters, the effective degradability (ED) was calculated by the equation provided by Dhanoa [24]:

$$ED=W+(D\times {\displaystyle \frac{Kd}{Kd+Kp}})$$

where ED is the effective degradability (%); W, D, and Kd are as explained above; and Kp is the fractional passage rate (the outflow rate of the rumen) of D (%/h) and assumed to be 0.025 as the chemical compositions of the grasses used in our research were close to those of the grasses used in the research of Warner et al. [25].

2.4. In Vitro Determination of Gas and Methane (CH₄) Production

2.4.1. In Vitro Gas and CH₄ Measurement and Donor Animals

At WUR, fresh grass samples (~1 cm) were defrosted, and ~1 g of DM was accurately weighed into 250 mL fermentation bottles (Schott, Mainz, Germany) with six replicated bottles per sample. Total cumulative IVGP and CH₄ production were measured using a fully automated gas production system [26] over 72 h. Rumen fluid was collected into pre-warmed thermos flasks flushed with CO₂ from four fistulated Holstein-Friesian lactating dairy cows before the morning feeding and transported to the laboratory of the Animal Nutrition Group at WUR, the Netherlands. Cows were fed a grass silage and maize silage-based diet with concentrates added depending on the lactation stage. Equal volumes of rumen fluid from two cows were combined to create two independent mixtures. Each mixture of rumen fluid was filtered over two layers of cheesecloth and combined with a pre-warmed anaerobic buffer solution (39 °C) at a ratio of 1:2 (v/v) [27]. For each mixture of rumen fluid, three bottles were used as analytical replicates. The buffered rumen fluid mixture (120 mL) was added to each sample-containing fermentation bottle (Schott, Germany) with three replicate bottles containing only 120 mL of buffered rumen fluid per rumen fluid mixture (blank bottles). All bottles were equally distributed over six shaking water baths kept at 39 °C and pre-flushed with CO₂. Gas production was measured for 72 h with a fully automated in vitro gas recording apparatus [27].

For the measurement of CH₄, 10 µL of gas was collected with a gas-tight syringe (Hamilton 1701N, point style 5 needles, 51 mm; Hamilton, Bonaduz, Switzerland) from the headspace of each bottle and injected directly into a gas chromatograph as described by Pellikaan et al. [28,29]. The gas chromatograph was fitted to a flame ionization detector and contained a packed column (Porapak, 6 m × 1/8 in., 50–80 mesh, Grace/Alltech, Lexington, KY, USA) with N as the carrier gas (100 kPa) and an oven temperature of 60 °C. Methane in the headspace of bottles was measured at 0, 3, 6, 9, 12, 24, 30, 36, 48, 60, and 72 h with a gas chromatograph (GC8000 Top, CE Instruments, Milan, Italy).

2.4.2. Gas and Methane Production: Curve Fitting and Calculations

Gas and CH₄ production from all samples were corrected for the corresponding production by blank bottles at each time point. Using the nonlinear least squares regression procedure of SAS Institute Inc., the following equation [30] was fitted to the corrected cumulative data:

$$\mathrm{G}\mathrm{P}=\sum _{\mathrm{i}=1}^{\mathrm{n}}{\displaystyle \frac{\mathrm{A}\mathrm{i}}{{1+(\mathrm{C}\mathrm{i}/\mathrm{t})}^{\mathrm{B}\mathrm{i}}}}$$

where gas production (GP-72 h) (mL/g OM) is the cumulative produced gas or CH₄; n = the total number of phases; i = the number of phases; Ai (mL/g OM) is the estimated asymptotic gas or CH₄ production in phase i; Bi is a constant determining the switching characteristic of the curve in phase i; Ci (h) is the time at which half of the asymptotic gas or CH₄ production was reached in phase i; and t (h) is the time of incubation.

A tri-phasic model (n = 3) was fitted to the cumulative gas production following the procedure as described by Groot et al. [30], where phases 1 and 2 are assumed to relate to the fermentation of the soluble and non-soluble fraction, respectively; phase 3 is assumed to be related to microbial turnover. The time windows related to the asymptotes of GP for phases 1, 2, and 3 (A1, A2, and A3, respectively) were pre-set from 0 to 3, 3 to 20, and 20 to 72 h after the start of the incubation of the substrate, respectively, to enable the estimation of the various parameters (Bi and Ci, respectively). The aforementioned time points were empirically determined by van Gelder et al. [31] based on the work of Cone et al. [32]. Data on CH₄ production were fitted according to the abovementioned model. The maximum rate of gas production (Rmax) and time to reach Rmax (TRmax) were calculated as described by Groot et al. [30].

2.5. Statistical Analyses

The effects of grass species (G) and harvest time (H) on in vitro gas production parameters (model 1) and in sacco degradation characteristics (model 2) were subjected to an analysis of variance (ANOVA) using the PROC MIXED procedure in SAS with the following models:

Yijk = μ + Gi + Hj + (G × H)ij + Rk + eijk

(1)

Yij = μ + Gi + Hj + eij

(2)

where Yij/Yijk are the response variables (i.e., GP-72, fermentation or degradation kinetic parameters); μ is the overall mean; Gi is the effect of grass species (i = 1 to 3); Hj is the effect of grass harvest time (j = 1 to 3); (G × H)ij is the interaction effect of experimental grass i and harvesting week j; Rk is the random effect of rumen fluid mixture (l = 1 to 2); and eijk/eij are the residual error terms. Differences were determined using the least square means procedure and Tukey’s multiple comparisons. Throughout, the level of statistical significance was pre-set at p < 0.05, while a trend was declared at 0.05 ≤ p < 0.10.

To explain the variation in the in sacco parameters of the tropical grasses, simple linear regressions were performed using the PROC REG procedure in SAS with in sacco kinetic parameters as dependent variables and the chemical components of grasses as independent variables. Pearson’s correlation coefficients between dependent and independent variables were determined if data were normally distributed as tested using Kolmogorov–Smirnov. If data were not normally distributed, Spearman’s correlation coefficients were determined.

3. Results

3.1. Chemical Composition of Tropical Grasses at Different Maturity Stages

The chemical composition of the selected tropical grasses at different harvest times (g/kg DM except for DM, which was expressed as % of fresh matter) (

Table 1. Chemical composition (g/kg DM unless otherwise indicated) of three fresh grasses commonly used in dairy cattle nutrition in Vietnam harvested early, normal, or late.

Grass	Harvest Time *	DM, %	OM	CP	EE	NDF	ADF	ADL
Guinea	Early (−2)	16.4	890	218	28.7	573	315	20.2
	Normal	15.7	883	169	26.3	663	410	31.6
	Late (+2)	15.5	877	143	28.5	661	397	30.6
King	Early (−2)	13.8	881	139	33.4	557	297	19.4
	Normal	13.3	909	88.7	27.9	628	360	23.6
	Late (+2)	18.5	931	89.9	23.7	642	399	47.3
Mulato II	Early (−2)	19.9	864	145	25.7	441	223	19.3
	Normal	17.5	875	107	19.2	586	320	24.5
	Late (+2)	17.4	879	136	18.1	600	326	30.9

* The normal harvest time for Guinea at 4 weeks, King at 7 weeks, and Mulato II at 6 weeks. The values between brackets are the number of weeks compared to normal. DM = dry matter; OM = organic matter; CP = crude protein; EE = ether extract; NDF = neutral detergent fiber; ADF = acid detergent fiber; ADL = acid detergent lignin.

) was within previously reported ranges [33,34,35].

The DM content of Guinea and Mulato II grass decreased with harvest time. Furthermore, for all grass species, NDF content (mean ± SD) was on average 595 ± 68.9 g/kg DM and increased with harvest time. Across harvest times, the NDF content of Mulato II was 8.8% less than average. With increasing harvest times, a decrease was observed for CP content (r = −0.84; p = 0.004; n = 9) with the CP content of King grass (106 ± 29 g/kg DM) being 22.6% lower than the average CP content of all grasses (137 ± 41 g/kg DM).

3.2. Effects of Tropical Grasses at Different Maturity Stages on In Vitro Gas Production

Both grass species and harvest time (

Table 2. In vitro 72 h cumulative gas (GP-72 h) and methane (CH₄-72 h) production and associated model parameters of organic matter (OM) of three fresh grasses (G) commonly used in dairy cattle nutrition in Vietnam, harvested at different times (H).

Parameter	Guinea			King			Mulato II			SEM	p-Value
	Early (−2)	Normal	Late (+2)	Early (−2)	Normal	Late (+2)	Early (−2)	Normal	Late (+2)		G	H	G × H
GP-72 h, mL/g OM	312 b	260 cde	295 bc	298 b	350 a	264 de	323 ab	291 bcd	259 e	7.5	0.049	<0.001	<0.001
A1, mL/g OM	41 c	27 e	37 cde	54 b	71 a	54 b	54 b	40 cd	29 de	2.5	<0.001	0.001	<0.001
A2, mL/g OM	178 a	130 de	156 bc	150 c	171 ab	120 e	172 ab	144 cd	119 e	4.2	0.022	<0.001	<0.001
A1 + A2, mL/g OM	219 abc	156 fg	193 cde	204 bcd	241 a	174 efg	226 ab	184 def	147 g	6.2	<0.001	<0.001	<0.001
C1, h	1.0	1.1	1.1	1.1	1.0	1.0	1.1	1.1	1.1	0.04	0.023	0.938	0.089
C2, h	10.5 ab	10.9 a	9.9 c	9.5 d	9.1 de	9.0 e	10.3 bc	10.2 bc	10.2 bc	0.09	<0.001	<0.001	<0.001
Rmax1, mL/g OM/h	24.6 cd	15.5 d	21.2 cd	31.6 bc	45.2 a	38.0 ab	31.1 bc	23.0 cd	15.8 d	2.56	<0.001	0.208	<0.001
Rmax2, mL/g OM/h	12.8 a	9.6 def	11.7 ab	11.0 bcd	12.7 a	9.0 ef	11.6 abc	10.2 cde	8.4 f	0.34	<0.001	<0.001	<0.001
CH4-72 h
mL/g OM	41 b	34 b	41 b	41 b	47 a	35 b	41 b	35 c	31 c	1.2	<0.001	<0.001	<0.001
% of GP-72	13.0	12.9	13.9	13.8	13.5	13.7	12.8	11.9	12.2	0.22	<0.001	0.226	0.226

^{a,b,c,d,e,f,g} Values within rows with different superscripts differ (p < 0.05). Ai = asymptote of gas or CH₄ production in phase i (i = 1 and 2 for gas); Ci = incubation time at which half of maximum gas or CH₄ production was formed in phase i; Rmaxi = maximum rate of gas production in phase i.

) affected total gas production (p = 0.049 and p < 0.001, respectively). All parameters were affected by interactions between grass species and harvest time (p < 0.001), except for incubation time C1 (p = 0.089). Across harvesting times, King grass yielded 304 ± 43.3 mL/g OM, which was 5.2 and 4.5% more gas than Mulato II and Guinea, respectively. Generally, increasing maturity was associated with lower gas production, but in King grass, GP-72 h was the highest when harvested at the normal harvest time (350 mL/g OM). Gas production was found to be negatively correlated with lignin content (r = −0.69; p = 0.04; n = 9).

The fermentable fraction (A1 + A2) was the highest for early harvested grasses but not in King grass when harvested at the normal time (p < 0.001). The A2 fraction (asymptote of gas production associated with the insoluble fraction) of the fermentable fraction was the highest for Guinea (81.6%) and the lowest for King grass (71.0%). Consequently, Guinea had the longest incubation times and reached the lowest maximum gas production rates (Rmax2).

Total CH₄ production (CH₄-72 h, Table 2) was affected by grass species (p < 0.001) and maturity (p < 0.001). Across harvest times, the fermentation of King grass resulted in the highest CH₄ production with 41 ± 6 mL/g OM, compared to Guinea (39 ± 4 mL/g OM) or Mulato II (36 ± 5 mL/g OM). CH₄ production, expressed as % of GP-72 h, was on average 13.1 ± 0.7% and was only different between grass species, as Mulato II had a lower relative CH₄ production (12.3 ± 0.5%).

3.3. In Sacco Degradation of Tropical Grasses at Different Maturity Stages

In sacco degradation parameters (

Table 3. Effects of grass species and harvest time on in sacco rumen degradation parameters (expressed as %, unless otherwise indicated) of organic matter and neutral detergent fiber of three fresh grasses commonly used in dairy cattle nutrition in Vietnam.

Parameter	Grass Species (G)			Harvest Time (H)			SEM	p-Value
	Guinea	King	Mulato II	Early	Normal	Late		G	H	Linear
Organic matter
W	6.9	19.4	10.4	15.9	7.4	13.4	2.77	0.073	0.197	0.562
U	37.5	33.3	35.1	27.3 b	38.5 a	40.1 a	1.30	0.179	0.004	0.002
D	55.6	47.4	54.5	56.8	54.2	46.5	2.92	0.213	0.138	0.067
W + D	62.5	66.8	64.9	72.7 a	61.5 b	59.9 b	1.30	0.179	0.004	0.002
Kd (%/h)	2.54	2.54	2.79	3.09	2.40	2.38	0.159	0.508	0.057	0.034
ED	34.9 b	43.4 a	38.9 ab	47.2 a	33.9 b	36.1 b	1.09	0.014	0.002	0.002
Neutral detergent fiber
W	0.15	6.27	0.52	1.80	0.52	4.62	1.911	0.146	0.389	0.356
U	34.8	35.3	33.9	26.6	37.5	40.0	3.23	0.953	0.083	0.042
D	65.0	58.4	65.6	71.7	62.0	55.3	4.20	0.475	0.118	0.052
W + D	65.2	64.7	66.1	73.5	62.5	60.0	3.23	0.953	0.083	0.042
LT (h)	0.88	4.54	2.77	1.51	4.44	2.24	2.320	0.581	0.676	0.835
Kd (%/h)	2.44	2.69	2.45	2.86	2.50	2.21	0.292	0.797	0.379	0.190
ED	32.3	36.6	32.7	40.0 a	31.0 b	30.7 b	1.59	0.221	0.023	0.014

^a,b Values within rows for grasses or harvest time with different superscripts differ (p < 0.05). W = washable fraction; U = undegradable fraction; D = potentially degradable and insoluble fraction; LT = lag time; Kd = fractional degradation rate constant of D fraction; ED = effective degradability, using 2.5%/h passage rate.

) and disappearance patterns (Figure 1) were assessed for both OM and NDF. The three selected species generally displayed different trends for in sacco OM and NDF disappearance (OMD and NDFD, respectively).

The effective degradability of OM differed between grass species (p = 0.014) and between harvest times (p = 0.002). After two weeks of incubation, King grass demonstrated the highest ED_OM (43.4 ± 7.4%), followed by Mulato II grass (38.9 ± 6.6%) and Guinea grass (34.9 ± 7.7%). Comparisons between grasses of different maturity showed that early harvested grass also scored significantly higher in OM degradability, with lower U fractions (p = 0.004) and higher potentially degradable fractions (W + D; p = 0.004) compared to grasses that were harvested later. NDF degradation was not affected by grass species. Harvest time, on the other hand, did cause a difference (p = 0.023), as early harvested grasses were consistently associated with higher ED_NDF (Table 3; Supplementary Table S1). At the initial incubation time, a small amount of OM (Figure 1B,C) and NDF (Figure 1E) was already lost in some cases, especially for King grass. The lost fraction was mostly due to the washable fractions.

3.4. Relationships Between Chemical Composition and Ruminal Degradation

Regression analyses were performed to detect correlations between the grasses’ chemical composition and in sacco ruminal degradation. Linear regression equations are shown in

Table 4. Pearson’s correlation coefficients and linear regression equations of relationship between various indices of cell wall fractions of tropical grasses and rumen degradation characteristics.

Equation	Pearson’s r	R2adj	Pconstant	Pslope
OM degradation
Kd = 5.93 (±0.546) − 0.006 (±0.0001) × NDF	−0.917	0.819	<0.001	<0.001
ED = 82.4 (±17.24) − 0.073 (±0.0288) × NDF	−0.691	0.403	0.002	0.039
NDF degradation
U = −89.8 (±64.72) + 0.88 (±0.189) × ADF	0.871	0.723	0.208	0.002
D (%) = −89.8 (±4.32) − 0.98 (±0.150) × ADL	−0.926	0.837	<0.001	<0.001
D (g/kg DM) = −155.8 (±73.65) + 14.9 (±2.07) × hemicellulose	0.938	0.864	0.072	<0.001
Kd = 3.39 (±0.458) − 0.032 (±0.0159) × ADL	−0.601	0.269	<0.001	0.087
ED = 52.3 (±9.31) − 0.054 (±0.0271) × ADF	−0.604	0.274	0.001	0.085

Equations are expressed as coefficients of intercept (±SE) and X variable 1 (±SE). U = undegradable fraction (g/kg DM); D = potentially degradable and insoluble fraction; Kd = fractional degradation rate constant of D; ED = effective degradability (%), using 2.5%/h passage rate; NDF = neutral detergent fiber (g/kg DM); ADF = acid detergent fiber (g/kg DM); ADL = acid detergent lignin (g/kg DM); hemicellulose (i.e., NDF-ADF, % of total carbohydrates).

. The values of ED_OM (Table 4) and Kd_OM can be explained by NDF content for 40.3% (Figure 2A) and 81.9%, respectively. The NDF content was further analyzed, which revealed how ADF content is predominantly determinative for the undigestible NDF fraction (72.3%; p = 0.002). For ED_NDF, there was a tendency towards significance (27.4%; p = 0.085) (Figure 2B,C). The digestible fraction tended to be lower when ADL content was high. The hemicellulose content (Figure 2D) of tropical grasses was found to be positively correlated with NDF degradation; the variation in the hemicellulose content explained 86.4% of the variation in NDF degradation (p< 0.001).

4. Discussion

The present study provides a comprehensive examination of the chemical composition, in vitro gas production, and in sacco degradation kinetics of three tropical grass species (Guinea, King, and Mulato II) that were harvested at different times. Overall, the fermentation of younger grasses led to increased gas production, often with larger proportions of gas resulting from the fermentation of slow-digestible carbohydrates. This suggests that early harvesting could benefit milk fat content. Particularly young Guinea grass seems to be, in this respect, a suitable forage type.

To optimally utilize feed, the fermentable fraction of a forage should be as large as possible. Total gas production reflects the magnitude of this fraction. The first 20 h (A1 + A2) is especially relevant; thereafter, gas production is considered a result of microbial turnover [30,31]. During the first 20 h, King normal and Guinea early showed the highest gas production, implying that these grasses have the highest amount of fermentable OM. Gas production, however, is the result of both fast- and slow-fermentable carbohydrates. Slow-fermentable carbohydrates are important for acetate production, a precursor for milk fat [1]. Fast-fermentable carbohydrates, on the other hand, are water-soluble carbohydrates, and their fermentation is typically associated with propionate production, which benefits milk yield but counteracts the effects of acetate [36]. Therefore, hypothetically, the implementation of more complex carbohydrates could benefit milk fat content. If so, desired forages are the grasses that predominantly consist of slow-degradable carbohydrates: grasses of which fermentation resulted in lower A1 values (gas production from fast fermentation) compared to A2 values (gas production from slow fermentation). In the present study, this was true for early harvested Guinea grass. King grass had (relatively) more soluble carbohydrates, which resulted in a shorter incubation time (C2) and faster maximum gas production rate (Rmax2) compared to Guinea early. Of all nine grasses, the fermentation of Guinea early took the most time.

The fermentation of King normal and Guinea early also resulted in higher CH₄ productions. This makes sense, as CH₄ production is positively correlated with the acetate-to-propionate ratio as well, since propionate competes with CH₄ for metabolic hydrogen, and acetate does not [37]. Moreover, in the formation of acetate, metabolic hydrogen is released, thus enhancing methane formation.

Methane production was also affected by maturity, which could imply that early harvesting benefits acetate production (and consequently milk fat content). Mulato II produced less CH₄, which is linked to lower acetate production [1]. As lower acetate production is associated with less milk fat, Mulato II is most likely an unsuitable forage for improving milk fat content. In view of the latter, based on gas production parameters alone, it appears that King grass has promising aspects regarding total gas production, but, due to the distribution of A1 vs. A2, early harvested Guinea grass seems the most suitable forage for improving milk fat. Overall, the IVGP results of Mulato II were on all fronts inferior to those of Guinea and King grasses. Gas production was lower, but the fermentation rate was higher (Table 2). This finding can be interpreted for lower fermentable content and a lower acetate-to-propionate ratio. Therefore, in the context of increasing milk fat content, Mulato II is the least ideal grass species.

The current results support research on temperate grasses [38,39,40]. Ruminal degradability decreased with grass harvest times due to increasing lignification [41], a pattern consistent with the grass maturation observed in temperate grasses. EDOM was negatively correlated with NDF content (r = 0.69; p = 0.04). Indeed, lignin negatively affects the degradation of fiber since it covalently binds with cellulose [42] and physically shields its surface from enzymes, thereby inhibiting the enzymatic hydrolysis of cellulose [43,44]. An exception was observed for King normal, which caused higher gas production despite its higher lignin content. In view of the chemical composition, high gas production cannot directly be explained by higher OM or NDF, while the A2 fraction in gas production is relatively high. Furthermore, the values for both OMD and NDFD fall within the range reported in previous reported findings of other tropical feedstuffs including tropical roughages/legumes, concentrate feeds, and semi-arid browse forages [45,46,47]. King grass appeared to be a promising grass species, as it demonstrated the highest degradability, in line with the observed total gas production. However, King, especially when harvested at a normal time point, contained a much larger washable fraction, which reflects a higher amount of non-structural carbohydrates, given their water-soluble properties [19]. This explains the larger A1 value, but it is unclear why the washable fraction was lower in the early and late harvested King grasses, but not when harvested at a normal time; usually, more advanced stages of maturity are associated with a greater concentration of cell wall material. Also, the washable fraction of NDF should per definition be absent, yet the NDF of early and late harvested King grasses still contained this fraction. This suggests that in some tropical grasses, NDF may contain additional non-soluble substances.

In line with Huyen et al. [48], the present study confirms that early harvesting can also benefit the forage quality of tropical grasses. Maturity negatively influenced CP content, led to a decrease in the degradability of OM and NDF, and reduced total gas production when fermented in vitro. While a negative correlation between CP content and gas production has been described [49], such effects of CP were not found in the current study. Moreover, the King grass harvested at the normal time had the lowest CP value but had the highest values of GP-72 h, A1, and total A1 + A2 compared to the other grasses. This apparent contradiction might be the result of the existence of an optimum ratio of NDF-to-CP. Such an optimum has been described by Bezabih et al. [11] for Dactyloctenium aegyptium grass, which produced the highest amount of in vitro gas production at an NDF-to-CP ratio of 7.5.

Combining in sacco data with in vitro gas production provided further insights into the ruminal degradability of tropical grasses. The high digestibility values for OM and NDF that were observed for early harvested King grass imply that this type of grass could be used to increase the fermentable OM content of dairy rations in Vietnam. Generally, the nutritional quality of C4 grasses is lower than that of C3 grasses [50]. This is also true for protein content, with ranges from 9.3 to 17.0% DM in C3 grasses and from 7.1 to 11.0% DM in C4 grasses [5,40,51]. The current values of gas production were found to be somewhat greater than previously reported [8]. This may have been caused by differences in sample preparation, biological variations in the rumen fluids that were used, or nutrient retention in fresh samples. Previous research showed that gas production was indeed higher when the same grass species of the same maturity were oven-dried [48]. A similar observation was made by Calabrò et al. [52], who also concluded that fresh samples generated more gas and reached the maximum rate of gas production more quickly than dried samples. Fresh samples retained their original lactic acid and ammonia concentrations, whereas dried samples lost these nutrients during preparation. Calabrò et al. [52] suggested that the difference between fresh and dry samples might be caused by the enhanced colonization of rumen microorganisms on fresh samples, leading to faster fermentation and higher gas production due to greater nutrient availability compared to dried samples.

Knowledge of tropical grasses is limited compared to knowledge of temperate grasses, and certain ‘generally accepted’ implications for the chemical composition of feedstuffs have not been verified for tropical grasses. By performing regression analyses comparing chemical composition with ruminal degradation parameters, we revealed clear relationships between fiber content and digestibility. Forages with high NDF content typically result in a lower ruminal degradation of OM (Figure 1A). Considering that NDF contains hemicellulose, cellulose, and lignin, further analysis showed that the negative effects on ruminal degradation are predominantly caused by lignin and cellulose content (Figure 2B), whereas conversely, hemicellulose content is positively correlated with NDF degradability (Figure 2C). Early harvesting enables farmers to keep the lignin content of tropical grasses low. This does not just benefit the amount of fermentable OM but also enhances effective degradability in the rumen. Thanks to these strong correlations, the chemical composition of these grasses can be used to make a (cautious) estimation of how grass is degraded in the rumen. The same was performed by Ali et al. [20], who found very similar correlations (with a slope of −0.05 compared to −0.07 in the present study) for temperate grasses: in these grasses, NDF is negatively correlated with ED as well.

While the present study presents new findings that can be useful in forage selection for dairy cows, an important note is that the IVGP is tailored for temperate grasses. It is not certain that all parameters are equally relevant or applicable for tropical grasses. For instance, the duration of the three phases of fermentation was determined for temperate grasses. The same (potential) issue may be true for the in sacco experiment. The fact that NDF still had a washable fraction in King grass suggests that the NDF fraction of tropical grasses can contain additional unknown components. Comparative studies could help improve protocols, thereby making sure they are optimally adapted to tropical grasses.

5. Conclusions

The three tropical grasses investigated displayed distinct patterns of in sacco degradation kinetics and in vitro gas production characteristics. Of the nine studied grasses, early harvested Guinea appeared to be the most promising for improving milk fat content. The grasses should additionally be studied when incorporated in a mixed, complete diet to identify possible interactions between different feed compounds, as part of an in vivo trial. Such research should include the evaluation of milk fat contents in Southeast Asian dairy cows and will provide new knowledge on efficiently using tropical grasses in dairy cattle nutrition, without additional impacts on the environment.