Effects of Plantain (Plantago lanceolata) and Chicory (Cichorium intybus) on Ruminal Fermentation and Enteric Methane Production by the Rumen Simulation Technique (RUSITEC)

Abstract

Alternative forages, such as plantain (Plantago lanceolata) and chicory (Cichorium intybus), show potential to improve ruminant nutrition while reducing environmental impacts; however, their influence on rumen function remains insufficiently defined. Using eight RUSITEC fermenters (two vessels per treatment per run, across two independent runs; n = 4 fermenter replicates per treatment overall), plantain (PL), chicory (CHI), a plantain–chicory mixture (PCH), and a grass-based sward (GBS) were incubated for a 10-day adaptation period followed by a 5-day sampling period to assess nutrient digestibility, fermentation, methane (CH₄) production, and microbial composition. PL resulted in the lowest CH₄ emissions (4.13%) and ammonia nitrogen (NH₃–N; 5.92 mg/day), whereas CHI achieved the highest dry matter digestibility (77.2%) but also the highest CH₄ emissions (5.55%). PCH presented intermediate CH₄ levels (4.52%), showing efficient nutrient use, while GBS and CHI achieved the highest NDF disappearance (NDFomd: 49.0% and 50.9%, respectively). Bacterial diversity (Shannon index) was lower in PL (1.71) than in CHI (4.25) and GBS (4.23), with PCH being intermediate (3.47). These findings support PL and PCH as promising dietary options to enhance rumen efficiency and reduce methane and nitrogen emissions. In vivo validation is recommended.

1. Introduction

Modern grazing systems predominantly rely on temperate grasslands, such as those dominated by ryegrass (Lolium perenne) and white clover (Trifolium repens), to meet livestock nutritional demands [1]. However, climate change has reduced forage quality and growth rates in these pastures, challenging their ability to sustain livestock nutrition [2]. Alternative forages, such as ribwort plantain (Plantago lanceolata) and chicory (Cichorium intybus), offer resilience to water deficits, maintaining high dry matter (DM) yields and nutritional value during warmer seasons [3,4,5,6]. These species not only enhance resource use efficiency but also reduce the environmental impact of ruminant production through improved forage quality and the presence of bioactive secondary compounds, such as aucubin, catalpol, and acteoside in plantain and sesquiterpene lactones in chicory [7,8,9]. These compounds exhibit antimicrobial, antioxidant, and anti-inflammatory properties that have been associated in the literature with reduced methane (CH₄) emissions and altered nitrogen metabolism in the rumen through mechanisms that are not limited to—and in several cases are distinct from—inhibition of biohydrogenation, which chiefly affects ruminal fatty acid metabolism [7,9,10,11,12,13].

Recent studies suggest that incorporating plantain and chicory into ruminant diets can modify ruminal fermentation, improve nutrient uptake, and decrease CH₄ emissions [14,15,16]. These forages are also associated with increased polyunsaturated fatty acid content in milk and meat compared to ryegrass-based diets [16,17,18,19]. Chicory enhances DM and organic matter (OM) digestibility, while plantain reduces ammonia nitrogen (NH₃-N) production, indicating improved protein metabolism and reduced nitrogen excretion [3,11,15,20,21,22]. While the individual nutritional and secondary metabolite profiles of plantain and chicory and their separate effects on digestibility, nitrogen metabolism, or methane emissions have each been documented to some extent in isolation, data directly comparing their combined effects as a mixed sward—together with the associated shifts in the rumen microbial community measured concurrently with fermentation and CH₄ output—remain scarce. The rumen simulation technique (RUSITEC) offers a powerful tool to address these knowledge gaps, allowing controlled and reproducible evaluation of fermentation processes, nutrient degradation, and gas production while maintaining a stable ruminal microbial ecosystem. RUSITEC was selected because it allows simultaneous, replicated, and reproducible assessment of fermentation kinetics, gas/CH₄ production, and microbial community composition under standardized conditions, enabling direct, controlled comparison of individual forages and their mixture that is difficult to achieve given the greater inter-animal variability inherent to in vivo trials. A grass-based sward was selected as the reference treatment because perennial ryegrass is the most widely used species in Chilean pastures [23], providing a practical benchmark against which the nutritional and fermentation performance of the alternative forages can be evaluated under comparable management. Unlike plantain and chicory, grass-based swards are not characterized by the bioactive secondary compounds described above [7,8,9], which provides a basis for the contrasting fermentation and methane responses evaluated here. To our knowledge, few studies have evaluated plantain and chicory both individually and as a binary mixture against a conventional grass-based sward under controlled in vitro conditions, and fewer still have combined fermentation and CH₄ output with rumen bacterial community composition. Therefore, this study aimed to evaluate whether plantain, chicory, and a plantain–chicory mixture differ from a grass-based pasture in nutrient digestibility, ruminal fermentation parameters, enteric CH₄ production, and rumen bacterial community composition using the RUSITEC system.

2. Materials and Methods

This experiment was conducted at the Livestock Systems and Nutrition Laboratory, Universidad de Concepción, Chillán, Chile. All procedures involving donor cows were certified by the Ethics and Animal Welfare Committee of the Universidad de Concepción (N° CBE28-2019, approved on 29 May 2019).

Data collection and analysis for this study were conducted between November 2018 and June 2019.

2.1. Forage Collection and Treatments

Four pasture treatments were evaluated: plantain (Plantago lanceolata, cv. Ceres Tonic, PL), chicory (Cichorium intybus, cv. Choice, CHI), a mixture of plantain and chicory (30% PL and 70% CHI, PCH), and a grass-based permanent sward (GBS) comprising perennial ryegrass (Lolium perenne, cv. Rohan, 39.7%), cocksfoot (Dactylis glomerata, cv. Safin, 26.8%), white clover (Trifolium repens, cv. Weka, 16.7%), brome (Bromus valdivianus, cv. Póker, 11.2%), and tall fescue (Festuca arundinacea, cv. Kora, 5.34%). The 30:70 proportion in PCH corresponds to the natural proportion of the two species in the sward at the time of harvest, when chicory had become the dominant species. The establishment and management of the field trial in which the plantain–chicory mixture and the grass-based sward were grown are described in [24]. Plantain and chicory were obtained from the plots containing the plantain–chicory mixture by collecting each species separately. Four forage samples were collected for each pasture evaluated in November 2018, at a vegetative to pre-flowering stage, at the Sheep Unit of the Austral Agricultural Experimental Station, Universidad Austral de Chile, Valdivia, Chile, at a cutting height of 5 cm. The swards had been grazed by lambs and allowed to regrow before cutting so the harvested material was regrowth and not a first cut. Samples were freeze-dried within 24 h of harvest and ground through a 4 mm screen using a Wiley mill (Arthur H. Thomas, Philadelphia, PA, USA). Subsequently, the four forage samples from each pasture were pooled and stored at room temperature in sealed bags until their use in the RUSITEC incubation trial, at which time 10 g of each forage sample was placed in a nylon bag (100 mm × 200 mm, 50 μm pore size; ANKOM Technology Corporation, New York, NY, USA). The chemical composition of the forages is presented in Table 1.

Table 1. Chemical composition (mean ± standard deviation) of the different forage treatments used in RUSITEC.

Component 2	Forage Treatments 1
	PL	CHI	PCH	GBS
Dry matter, % as incubated 3	89.7 ± 0.25	89.4 ± 0.31	90.4 ± 0.18	91.3 ± 0.22
Ash, % of DM	11.9 ± 0.35	11.3 ± 0.28	11.1 ± 0.42	8.10 ± 0.15
Crude protein, % of DM	17.5 ± 0.54	20.0 ± 0.62	18.8 ± 0.48	15.5 ± 0.38
Neutral detergent fiber, % of DM	30.1 ± 1.12	23.1 ± 0.95	27.1 ± 1.05	47.3 ± 1.45
Acid detergent fiber, % of DM	19.6 ± 0.78	12.2 ± 0.52	16.0 ± 0.67	25.8 ± 0.92
Hemicellulose, % of DM	10.4 ± 0.45	10.9 ± 0.38	11.1 ± 0.51	21.5 ± 0.84
Metabolizable energy, MJ/kg DM	11.3 ± 0.21	13.0 ± 0.33	11.8 ± 0.25	11.6 ± 0.19

¹ PL: plantain, CHI: chicory, PCH: plantain and chicory mixture, GBS: grass-based sward. ² DM: dry matter. ³ Dry matter of the freeze-dried, ground (4 mm) forage incubated in RUSITEC.

2.2. Experimental Design and RUSITEC Apparatus

A randomized complete block design was used, with four pasture treatments tested in duplicate across two independent runs using the RUSITEC system [25]. The RUSITEC system (Universidad de Concepción, Chile) consisted of eight 1 L polycarbonate fermenter vessels maintained at 39 °C in a water bath with constant vertical agitation. Treatments were randomly assigned to vessels within each run, with two vessels assigned per treatment in each run, giving four fermenter replicates per treatment in total across the two runs; the fermenter, nested within run, served as the experimental unit for all treatment comparisons. Each run included a 10-day adaptation period followed by a 5-day sampling period. Artificial saliva [26] was infused through an input port, fermentation effluent was collected in 1 L Erlenmeyer flasks, and gases were collected in 2 L vinyl bags.

2.3. Incubation Procedure

Rumen fluid and digesta were collected 2 h post-feeding through the ruminal cannula of two Aberdeen Angus cows fed a diet of 25% dry matter commercial concentrate (14% crude protein; COPEVAL, Libertador Bernardo O’Higgins, Chile), 75% DM clover/ryegrass hay, and a vitamin/mineral block (Veterquimica, Metropolitan Region, Chile). Ruminal digesta was filtered through four layers of cheesecloth, pooled and transported to the laboratory in pre-warmed (39 °C) containers. Each fermenter was filled with 800 mL of ruminal fluid and 200 mL of modified McDougall’s buffer (containing 1 g/L (NH₄)₂ SO₄) [27] and was flushed with CO₂ to maintain anaerobiosis. Initially, each fermenter received one nylon bag with 40 g of rumen solid digesta and another with 10 g of forage. After 24 h, the digesta bag was replaced with a new 10 g forage bag. Subsequently, one forage bag was replaced daily, ensuring a 48 h incubation period per bag. Artificial saliva was infused at a 3% per hour dilution rate using a peristaltic pump (Gilson, Middleton, WI, USA). During daily bag exchanges, fermenters were flushed with CO₂ to maintain an anaerobic environment, and effluent was collected in flasks containing sodium azide (1 g/L) to halt fermentation.

2.4. Nutrient Disappearance and Chemical Analysis

From days 11 to 15, one forage bag was removed daily from each fermenter after 48 h of incubation. Bags were individually rinsed with cold distilled water until the rinse water was clear and then oven-dried at 55 °C for 48 h (Universal Oven UN55, Memmert, Schwabach, Germany). Residual dry matter was determined separately for each bag and sampling day. For chemical characterization of the residual material, residues from three sampling days were pooled within each fermenter and experimental run, ground through a 1 mm screen (Wiley Mill, Philadelphia, PA, USA), and analyzed for DM, crude protein (CP) and ash according to AOAC methods 930.15, 990.03, and 942.05, respectively [28]. Organic matter (OM) was calculated as 100 minus ash. Neutral detergent fiber (NDF) was determined according to Van Soest et al. [29], and acid detergent fiber (ADF) according to AOAC method 973.18 [28]. Hemicellulose (HE) was calculated as NDF minus ADF. Metabolizable energy (ME) was estimated using local regression based on digestible organic matter content (DOMD) from the two-stage rumen fluid–pepsin method [30,31]. The chemical composition obtained from each fermenter-specific composite residue was used together with the daily amount of residual dry matter to estimate the residual amount of OM, CP, NDF, ADF, and hemicellulose for each sampling day. Nutrient disappearance was calculated daily as the difference between the amount of each nutrient initially incubated and the corresponding amount remaining after 48 h of incubation, expressed relative to the amount initially incubated.

2.5. Gas and CH₄ Sample Collection

Total gas production was measured daily via water displacement [32]. A 15 mL gas sample was collected from each bag’s septum using a 21G needle syringe and stored in vacuum-sealed exetainers (Labco Ltd., Ceredigion, UK) for methane analysis; this aliquot was added back to the total gas volume recorded by water displacement so that daily total gas production corresponded to the sum of the sampled aliquot and the remaining gas volume, and sampling did not reduce the total gas volume used for methane calculations. Methane concentration in the collected samples was determined via gas chromatography (Agilent Technologies 7890, GS-CarbonPLOT column, 30 m × 0.32 mm × 3 μm; helium carrier gas, 27 cm/s; oven 35 °C; injector 185 °C; detector 150 °C). Methane production was calculated from the CH₄ proportion determined by gas chromatography and the total daily gas volume as follows:

CH₄ (mg/d) = [V gas (L/d) × (CH₄/100)] × (16.04 g/mol/22.4 L/mol) × 1000

where 16.04 g/mol is the molecular weight of methane and 22.4 L/mol is the molar gas volume at standard temperature and pressure (used as the conversion factor). Gas volumes were not additionally corrected for the actual ambient temperature or atmospheric pressure at the time of measurement; because all fermenters were measured under the same laboratory conditions and the same conversion factor was applied consistently across treatments, this does not affect the relative comparison among treatments, although it may slightly influence the absolute methane values expressed in mg/day. Effluent pH was measured daily during bag exchanges (Orion Star A121, Thermo Scientific, Massachusetts, MA, USA).

2.6. Ruminal Fermentation Parameters

For volatile fatty acid (VFA) analysis, 1.5 mL effluent subsamples were preserved with 200 µL of 10% sulfuric acid and stored at −20 °C. For ammonia nitrogen (NH₃-N), 1.5 mL subsamples were preserved with 150 µL of trichloroacetic acid. VFA concentrations were quantified via gas chromatography (Agilent Technologies 7890, Santa Clara, CA, USA), and NH₃-N was analyzed spectrophotometrically (Merck Spectroquant Pharo 300, Darmstadt, Germany) [27]. VFA and NH₃-N production were calculated by multiplying concentrations (mmol/L and mg/L, respectively) by daily effluent volume (L/d).

2.7. DNA Isolation and Sequencing

Fermentation fluid for DNA extraction was collected directly from the fermenter vessel contents on day 15 (final sampling day), immediately after the morning bag exchange, and was frozen at −80 °C until DNA extraction. DNA was extracted from fermentation fluid samples using a QIAamp DNA Mini Kit (QIAGEN, Hilden, Germany) and submitted to the Integrated Microbiome Resource (IMR), Dalhousie University, Canada. The V6–V8 region of the 16S rRNA gene was amplified using primers B969F (5′-ACGCGHNRAACCTTACC-3′) and BA1406R (5′-ACGGGCRGTGWGTRCAA-3′). Amplicons were sequenced on an Illumina MiSeq platform (2 × 300 bp paired-end reads).

2.8. Bioinformatic and Microbiome Analysis

Raw paired-end reads targeting the bacterial 16S rRNA gene V6–V8 region were processed using QIIME2 (v2023.9; [33]). Sequence quality control and denoising were performed with the DADA2 plugin (v2023.9; [34]), which included removing the first 30 bases to remove primer sequences and low-quality 5′ ends, and truncating forward and reverse reads at 280 and 250 bases, respectively. Subsequently, paired-end reads were merged, denoised, and filtered to remove chimeric and PhiX sequences, leading to the inference of amplicon sequence variants (ASVs). Taxonomic classification of the resulting ASVs was conducted using a naïve Bayes classifier trained on full-length 16S rRNA gene sequences from the SILVA reference database (v138; [35]). For phylogenetic analyses, a tree was constructed by aligning ASV sequences with MAFFT (v7.520; [36]) and inferring an approximate maximum-likelihood tree with FastTree (v2.1.11; [37]).

Before diversity analyses, ASVs with a total frequency below 10 reads across all samples were removed. Alpha diversity was assessed using observed ASVs, Shannon’s diversity index, Pielou’s evenness, and Faith’s phylogenetic diversity via QIIME2’s core-metrics-phylogenetic pipeline. For alpha diversity calculations only, samples were rarefied to 23,716 reads, corresponding to the smallest library in the filtered table, so that all 16 samples (n = 4 per treatment) were retained. Differences in alpha diversity indices among treatments were tested with the Kruskal–Wallis test followed by Dunn’s post hoc test with Benjamini–Hochberg false discovery rate (FDR) correction [38,39]. Beta diversity was evaluated on non-rarefied data by principal coordinate analysis (PCoA) of robust Aitchison distances, which account for the compositional nature of microbiome data by applying a robust centered log-ratio (rCLR) transformation [40,41]. Differences in microbial community composition among treatments were tested using permutational multivariate analysis of variance (PERMANOVA, 999 permutations) applied to the robust Aitchison distance matrix using the adonis2 function in the vegan R package (v2.6-6; [42]) within R software (v4.3.2; [43]).

Putative microbial biomarkers and taxa associated with treatments were identified using LEfSe (v1.1.2; [44]), which combines non-parametric statistical testing with linear discriminant analysis (Kruskal–Wallis and pairwise Wilcoxon tests; p < 0.05; LDA score > 2.0). In parallel, differential abundance analysis was conducted using a zero-inflated Gaussian mixture model (ZIGMM) implemented in the metagenomeSeq R package (v1.42.0: [45]), with cumulative-sum scaling (CSS) normalization, to identify genera exhibiting statistically significant abundance differences among treatments. Log₂ fold changes (log₂FC) are reported as the focal treatment relative to each comparison treatment so that positive values indicate higher abundance in the focal treatment.

2.9. Statistical Analysis

Data normality and homoscedasticity were assessed with the Shapiro–Wilk and Levene tests, respectively. In all analyses, the fermenter was the experimental unit. Daily nutrient disappearance (DM, OM, CP, NDF, ADF, and hemicellulose), pH, NH₃–N, individual and total VFAs, total gas, and CH₄, all determined for each of the five sampling days, were analyzed with a mixed model in Stata (version 16; statistical software, College Station, TX, USA), including treatment, sampling day, and their interaction as fixed effects, and run and fermenter(run) as random effects. Sampling day was modeled as a repeated measure, and for each variable, the covariance structure with the lowest Akaike’s information criterion (AIC) was selected among seven candidate structures (autoregressive, banded, compound symmetry, heterogeneous autoregressive, heterogeneous compound symmetry, Toeplitz, and unstructured [46]; denominator degrees of freedom were estimated by the Kenward–Roger method). No significant treatment × sampling day interaction was found for any variable. Treatment least-squares means were separated using Tukey’s adjustment. Significance was declared at p < 0.05 and a tendency at 0.05 ≤ p < 0.10.

3. Results

3.1. Nutrient Disappearance

Forage type significantly influenced nutrient disappearance in the RUSITEC system (Table 2). CHI exhibited the highest in vitro DM digestibility (IVDMd; 77.2%) and organic matter disappearance (OMd; 75.8%), followed by PL (71.3% and 69.9%), PCH (72.0% and 70.7%), and GBS (67.7% and 66.5%) (p = 0.001). Crude protein disappearance (CPd) was highest in GBS (75.7%) and CHI (68.9%) and lowest in PL (54.7%), with PCH being intermediate (64.4%); CHI was not significantly different from GBS or PCH (p < 0.001). Neutral detergent fiber disappearance (NDFomd) was highest in CHI (50.9%) and GBS (49.0%), intermediate in PCH (46.3%, which was not significantly different from either CHI/GBS or PL), and lowest in PL (40.0%; p = 0.003). Acid detergent fiber disappearance (ADFomd) was highest in GBS (42.6%) and lowest in PL (34.0%), with CHI and PCH being intermediate and not significantly different from either extreme (p = 0.032). Hemicellulose disappearance (HEd) was significantly greater in CHI (61.8%) than in PL (51.2%), with PCH (57.2%) and GBS (56.6%) showing intermediate values (p = 0.002).

Table 2. Effects of different forage types on nutrient disappearance in the rumen simulation technique.

Item (%) 2	Forage Treatments 1				SEM	p-Value
	PL	CHI	PCH	GBS
IVDMd	71.3 b	77.2 a	72.0 b	67.7 b	1.96	0.001
OMd	69.9 b	75.8 a	70.7 b	66.5 b	2.01	0.001
CPd	54.7 c	68.9 ab	64.4 b	75.7 a	3.32	<0.001
NDFomd	40.0 b	50.9 a	46.3 ab	49.0 a	3.00	0.003
ADFomd	34.0 b	41.3 ab	38.9 ab	42.6 a	2.70	0.032
HEd	51.2 b	61.8 a	57.2 ab	56.6 ab	3.31	0.002

¹ PL: plantain, CHI: chicory, PCH: plantain and chicory mixture, GBS: grass-based sward. ² IVDMd: in vitro dry matter disappearance, OMd: organic matter disappearance, CPd: crude protein disappearance, NDFomd: neutral detergent fiber disappearance, ADFomd: acid detergent fiber disappearance, HEd: hemicellulose disappearance. Means within a row with different letters differ (p < 0.05).

3.2. Fermentation Characteristics

Forage type did not affect effluent volume or pH in the RUSITEC fermenters (p > 0.05), with pH values ranging from 6.93 to 7.00 (Table 3). However, forage type significantly influenced NH₃-N production (p < 0.01). PL exhibited the lowest NH₃-N production (5.92 mg/d), and GBS showed the highest (25.0 mg/d). CHI (18.9 mg/d) was intermediate and significantly lower than GBS, while PCH (22.5 mg/d) was intermediate and not significantly different from either CHI or GBS. Total VFA production was unaffected by forage type (p = 0.737), averaging 28.7–32.2 mmol/d across treatments. Significant differences were observed in the VFA profile among treatments. The PCH treatment resulted in the highest acetate proportion (60.2%; p < 0.05), whereas propionate proportions were highest in PL (34.5%) and lowest in CHI (26.0%; p < 0.001). The acetate-to-propionate (A:P) ratio was lowest in PL (1.60) and highest in CHI (2.27; p < 0.001). No significant differences were observed in isobutyrate, valerate, or caproate proportions (p > 0.05); the butyrate proportion tended to differ among treatments (p = 0.054), with the numerically highest value in CHI, followed by GBS, PCH, and PL. Notably, 2- and 3-methylbutyrate proportions were higher in CHI (0.55%) compared to those in other treatments (0.11–0.15%; p < 0.001).

Table 3. Effects of different forage types on fermentation characteristics measured in the rumen simulation technique.

Response	Forage Treatments 1				SEM	p-Value
	PL	CHI	PCH	GBS
Effluent, mL	725	707	709	716	21.0	0.733
pH	6.95	6.93	7.00	6.97	0.018	0.108
NH3-N, mg/d	5.92 c	18.9 b	22.5 ab	25.0 a	1.130	<0.001
Total VFA, mmol/d	32.2	31.9	29.6	28.7	2.57	0.737
VFA proportions, mmol/100 mmol total VFA
Acetate	54.2 b	57.6 ab	60.2 a	56.5 ab	2.95	0.007
Propionate	34.5 a	26.0 c	27.7 bc	30.3 b	2.70	<0.001
Butyrate	8.16	11.6	8.88	9.69	4.35	0.054
Isobutyrate	0.46	0.52	0.59	0.61	0.065	0.183
Valerate	2.21	3.06	2.14	2.51	1.076	0.136
2- and 3-methylbutyrate	0.15 b	0.55 a	0.12 b	0.11 b	0.046	<0.001
Caproate	0.56	0.60	0.40	0.46	0.087	0.219
A:P	1.60 c	2.27 a	2.22 ab	1.89 bc	0.112	0.001

¹ PL: plantain, CHI: chicory, PCH: plantain and chicory mixture, GBS: grass-based sward, A:P: ratio of acetate to propionate. Means within a row with different letters differ (p < 0.05).

3.3. Total Gas and Methane Production

The effects of forage type on total gas and CH₄ production are presented in Table 4. Total gas production did not differ significantly among treatments (p = 0.135), ranging from 1.58 L (PCH) to 1.78 L (CHI). CH₄ content, expressed as a percentage of total gas, varied significantly (p = 0.022), with PL exhibiting the lowest proportion at 4.13% and CHI the highest at 5.55%. Absolute CH₄ production (mg) was greater in CHI (70.7 mg), followed by GBS (52.1 mg), whereas PCH (49.1 mg) and PL (47.9 mg) exhibited lower values. Similarly, CH₄ yield per gram of OM incubated was highest in CHI (8.93 mg/g OM), while the remaining forages produced lower amounts. CH₄ yield per gram of OM digested (mg CH₄/g OMd) showed a tendency toward significance (p = 0.066), with CHI exhibiting the highest value (11.7 mg/g OMd).

Table 4. Effects of different forage types on gas and methane production in a RUSITEC.

Item	Forage Treatments 1				SEM	p-Value
	PL	CHI	PCH	GBS
Total gas, L/d	1.59	1.78	1.58	1.59	0.064	0.135
CH4, %	4.13 b	5.55 a	4.52 ab	4.75 ab	0.409	0.022
CH4, mg/d	47.9 b	70.7 a	49.1 b	52.1 ab	5.62	0.028
CH4, mg/g OM	6.06 b	8.93 a	6.10 b	6.21 b	0.705	0.020
CH4, mg/g OMd	8.83	11.7	8.75	9.35	0.791	0.066

¹ PL: plantain, CHI: chicory, PCH: plantain and chicory mixture, GBS: grass-based sward. mg/g OM = CH₄ yield per gram of organic matter incubated; mg/g OMd = CH₄ yield per gram of organic matter apparently disappeared. Means within a row with different letters differ (p < 0.05).

3.4. Metabarcoding Analyses

Metabarcoding of fermentation fluid samples (n = 4 per treatment) generated 1.07 million raw reads. After quality filtering and denoising with the DADA2 algorithm, 632,773 high-quality reads were retained, with an average of 39,548 reads per sample (range: 23,806–55,555; Supplementary Table S1). After removal of low-frequency ASVs (<10 reads in total), 762 ASVs and 630,513 reads were retained, and the smallest library (23,716 reads) was used as the rarefaction depth for alpha diversity analyses (Supplementary Table S2). Shannon diversity and Pielou’s evenness differed among treatments (Kruskal–Wallis, H = 11.32, p = 0.010, and H = 10.48, p = 0.015, respectively; Supplementary Table S2). Both indices were lowest in PL (Shannon: 1.71 ± 0.41; evenness: 0.235 ± 0.051). Dunn’s post hoc test with Benjamini–Hochberg FDR correction showed that Shannon diversity was significantly lower in PL than in CHI (4.25 ± 0.03) and GBS (4.23 ± 0.87) (FDR-adjusted p = 0.018 for both), whereas evenness was significantly lower in PL than in CHI (0.567 ± 0.026; FDR-adjusted p = 0.011), with only a tendency for GBS (0.513 ± 0.095; FDR-adjusted p = 0.064). PCH (Shannon: 3.47 ± 0.15; evenness: 0.448 ± 0.008) was intermediate and did not differ significantly from any other treatment. Observed ASVs (H = 10.61, p = 0.014) and Faith’s phylogenetic diversity (H = 12.79, p = 0.005) also differed among treatments, being highest in GBS (299.8 ± 38.4 ASVs; 21.39 ± 0.78) and lowest in PL (151.8 ± 22.4 ASVs; 15.98 ± 0.40), with GBS being significantly higher than PL for both indices (FDR-adjusted p = 0.010 and 0.004, respectively) and CHI and PCH being intermediate (Figure 1; Supplementary Table S2).

Figure 1. Alpha diversity indices of rumen microbial communities across different forage treatments in the RUSITEC system. Boxplots represent Shannon’s diversity index, Pielou’s evenness, observed ASVs, and Faith’s phylogenetic diversity. Each point represents one fermenter (n = 4 per treatment); indices were calculated after rarefaction to 23,716 reads per sample. PL: plantain; CHI: chicory; PCH: plantain–chicory mixture; GBS: grass-based sward.

At the genus level, Streptococcus dominated in PL (81.6 ± 7.7%), followed by lower abundances in PCH (53.1 ± 5.9%), CHI (28.5 ± 8.7%), and GBS (49.1 ± 12.8%). Lactobacillus was abundant in CHI (32.4 ± 1.7%) but represented less than 10% in other treatments. Kandleria was present at similar levels in CHI (12.4 ± 8.5%), PCH (11.8 ± 8.6%), and GBS (11.4 ± 4.1%), while Olsenella showed a higher relative abundance in GBS (6.2 ± 0.9%) compared with that in other forages. Other genera detected at lower relative abundances included Megasphaera (1.1–2.9%) and Bifidobacterium (1–2%), showing a variable distribution among treatments (Figure 2).

Figure 2. Relative abundance (%) of the main bacterial genera detected in the fermentation fluid across different forage treatments in the RUSITEC system. Bars show treatment means (n = 4 fermenters per treatment), with less abundant genera grouped together; variability (mean ± SD) of the differentially abundant genera is reported in Supplementary Table S3. PL: plantain; CHI: chicory; PCH: plantain–chicory mixture; GBS: grass-based sward.

Beta diversity analysis, visualized by principal coordinate analysis (PCoA) of robust Aitchison distances, revealed distinct microbial community structures across treatments (Figure 3). The first two principal coordinates explained 24.12% and 15.04% of the total variance, respectively, indicating separation among treatment groups. PERMANOVA confirmed significant differences in microbial composition among treatments (p < 0.001, F = 3.74, R² = 0.48).

Figure 3. Principal coordinate analysis (PCoA) of rumen microbial communities across different forage treatments based on robust Aitchison distances. Each point represents one fermenter (n = 4 per treatment). PL: plantain; CHI: chicory; PCH: plantainchicory mixture; GBS: grass-based sward.

The LEfSe analysis identified distinct microbial communities among the four forage treatments in the RUSITEC experiment (Supplementary Table S3). Streptococcus (LDA: 5.43) and Selenomonas (LDA: 3.55) were enriched in the PL treatment. The CHI treatment exhibited higher relative abundances of Lactobacillus (LDA: 5.16), Kandleria (LDA: 4.77), and Desulfovibrio (LDA: 4.09). In the PCH treatment, Sharpea (LDA: 4.25) and Bifidobacterium (LDA: 4.19) were the most differentially abundant taxa. The GBS treatment was enriched in fibrolytic genera, including Olsenella (LDA: 4.46), Butyrivibrio (LDA: 3.74), and Ruminococcus (LDA: 3.51).

The zero-inflated Gaussian mixture model (ZIGMM) analysis, applied to genus-level microbial data using the metagenomeSeq package, also detected significant differences in taxa abundance among forage treatments. In the PL treatment, Streptococcus (log₂FC: 1.11 vs. PCH, 2.18 vs. CHI, 1.86 vs. GBS; padj < 0.05) and Selenomonas (log₂FC: 1.13 vs. PCH, 2.73 vs. CHI, 1.52 vs. GBS; padj < 0.05) were significantly enriched. The CHI treatment showed higher relative abundances of Lactobacillus (log₂FC: 1.72 vs. PL, 3.65 vs. GBS, 1.91 vs. PCH; padj < 0.01) and Desulfovibrio (log₂FC: 2.21 vs. PL, 1.59 vs. GBS; padj < 0.05). In PCH, Sharpea (log₂FC: 4.32 vs. PL, 2.60 vs. GBS; padj < 0.001) and Bifidobacterium (log₂FC: 1.48 vs. PL, 8.96 vs. GBS; padj < 0.05) were significantly enriched. The GBS treatment exhibited higher abundances of Olsenella (log₂FC: 2.39 vs. PL; padj < 0.01) and Rikenellaceae [RC9 gut group] (log₂FC: 1.40 vs. PCH, 4.31 vs. CHI; padj < 0.05).

4. Discussion

The present study shows that forage type was associated with differences in in vitro nutrient disappearance, fermentation patterns, and ruminal bacterial community composition, indicating that plantain (Plantago lanceolata) and chicory (Cichorium intybus) elicited distinct responses in rumen function relative to a grass-based sward. These differences are consistent with the contrasting chemical composition and fiber degradability of the forages and may also involve secondary metabolites reported for these species, which were not analyzed here.

4.1. Nutrient Disappearance and Fermentation Dynamics

Chicory exhibited the greatest in vitro dry matter disappearance (IVDMd; 77.2%) and organic matter disappearance (OMd; 75.8%), consistent with its low structural fiber content (NDF 23.1% and ADF 12.2% of DM, Table 1) and with the higher soluble sugar content reported for vegetative chicory relative to perennial ryegrass (111 vs. 74 g/kgDM) [3,9] These compositional traits may enhance microbial accessibility and degradation efficiency, as previously demonstrated in both in vitro and in vivo trials [3,47]. The elevated CP disappearance observed in CHI and GBS suggests active proteolysis and microbial assimilation of nitrogenous substrates under these conditions, and the high NH₃-N observed in GBS likely reflects the rapid degradation of soluble nitrogen in ryegrass, leading to ammonia accumulation [48,49]. In contrast, the lower CP disappearance and NH₃-N concentration detected in PL (54.7% and 5.92 mg/d, respectively) point to a reduced rate of protein degradation and ammonia formation. Plantain is known to contain phenolic glycosides, such as aucubin and acteoside, previously reported to suppress proteolytic and deaminating bacteria [7,11,50]. Although these compounds were not quantified in the forage batches used in this study, their documented effects offer a plausible, though unconfirmed, explanation for the lower NH₃-N observed in PL. Aucubin may bind free amino acids or inhibit microbial deamination, whereas acteoside could enhance microbial protein synthesis by providing fermentable carbon substrates [7,11]. These combined effects could reduce ruminal ammonia accumulation, which may in turn limit subsequent urinary nitrogen losses, a trend consistently reported in grazing studies with Plantago species [15,51,52]. Because microbial protein synthesis and total nitrogen flows were not measured here, whether this reflects genuinely improved nitrogen use efficiency or simply reduced proteolytic activity cannot be determined from the present data.

The low NDFomd observed in PL (40.0%) could also reflect an influence of plantain secondary metabolites on fibrolytic microbes, as acteoside- and aucubin-containing extracts have been reported to reduce fiber-degrading bacteria in vitro [50]; this possibility was not tested in the present study. This contrasts with the higher fiber disappearance in CHI and GBS, which is consistent with more extensive fibrolytic fermentation. Collectively, these results suggest that CHI and GBS promote greater nutrient degradability and microbial fermentation intensity.

Differences in volatile fatty acid (VFA) profiles further illustrate the metabolic specialization induced by each forage type. The dominance of acetate in PCH (60.2%) and the higher propionate proportion in PL (34.5%) suggest distinct hydrogen utilization pathways. CHI also showed a high acetate proportion (57.6%), consistent with its greater digestibility. The lower acetate-to-propionate (A:P) ratio in PL (1.60) indicates a shift toward glucogenic fermentation, which likely contributes to reduced CH₄ production due to the role of propionate as a competitive hydrogen sink [53]. Conversely, the higher A:P ratio in CHI (2.27) suggests a fermentation pattern favoring acetate production, associated with rapid carbohydrate degradation and greater hydrogen release, which may favor methanogenesis. Despite these compositional and metabolic shifts, total VFA production and pH remained stable across treatments, indicating that forage type primarily modified fermentation end-product proportions rather than total fermentative capacity, consistent with observations by [18].

4.2. Methane Production and Fermentation Efficiency

Methane production differed markedly among forage types. CHI exhibited the highest CH₄ output (70.7 mg, 8.93 mg/g OM; 5.55% of total gas), followed by GBS and PCH, whereas PL showed the lowest values (47.9 mg, 6.06 mg/g OM; 4.13%). These differences reflect both fermentation stoichiometry and substrate composition. The greater CH₄ production in CHI is consistent with its higher digestibility (IVDMd = 77.2%), which increases hydrogen release during carbohydrate fermentation and is expected to increase substrate availability for methanogens [54]. In contrast, the reduced CH₄ emissions from PL correspond to its lower acetate-to-propionate ratio and higher propionate formation, which act as hydrogen sinks and reduce the availability of reducing equivalents for methanogenesis [54,55]. This in vitro response also mirrors in vivo observations, where cows fed plantain produced less CH₄ than those fed ryegrass, an effect largely attributed to the lower dry matter digestibility of PL [13]. The convergence of in vitro and in vivo evidence underscores the central role of substrate degradability in regulating rumen hydrogen dynamics and methanogenic potential.

The methane-mitigating effect of PL also agrees with reports attributing lower CH₄ production in Plantago spp. to iridoid glycosides such as aucubin and acteoside, which have been proposed to alter hydrogen flow through modified bacterial fermentation pathways [11,13], because archaeal community data were not generated in this study, a direct effect on methanogen metabolism cannot be confirmed and remains speculative. Intermediate CH₄ values observed in PCH suggest partial mitigation when both forages are combined, possibly reflecting additive effects between plant secondary metabolites and structural carbohydrates on hydrogen balance. Although CH₄ yield per digested OM only tended to differ (p = 0.066), this supports the concept that substrate degradability and hydrogen disposal pathways jointly determine methanogenic efficiency [56,57].

Interestingly, total gas production did not differ among treatments (p = 0.135), indicating that the variation in CH₄ output is primarily related to differences in gas composition rather than total fermentation intensity. This finding reinforces the importance of microbial routing of reducing power toward alternative sinks such as propionate, microbial biomass, or reductive acetogenesis, processes favored in PL and PCH fermentations [53,58]. Collectively, these results demonstrate the potential of PL as a functional forage for mitigating enteric CH₄ emissions, while the high digestibility of CHI, although nutritionally advantageous, may enhance hydrogen release and counteract its mitigation potential, partially diverging from previous reports [47].

4.3. Rumen Microbial Composition/Ruminal Microbiota Diversity

Alpha diversity patterns indicated that forage type influenced rumen bacterial diversity, evenness and richness. CHI and GBS exhibited greater Shannon diversity than PL, CHI also showed greater evenness, and GBS harbored the richest and most phylogenetically diverse community, whereas PCH was intermediate, suggesting that these forages supported more diverse and evenly structured bacterial consortia than PL (Supplementary Table S2). Greater alpha diversity has been associated with enhanced fermentation stability and metabolic redundancy [59], which may contribute to the observed variation in fermentation profiles among treatments. Beta diversity analysis (PCoA of robust Aitchison distances) further demonstrated distinct community clustering across forages, with PL clearly separated from CHI, PCH, and GBS, as confirmed by PERMANOVA (p < 0.001, R² = 0.48; Figure 3).

Compositional analyses revealed marked taxonomic shifts that reflected forage-specific microbial adaptations. Plantago lanceolata (PL) was dominated by Streptococcus (81.6%) with enrichment of Selenomonas. Selenomonas is associated with the succinate–propionate pathways, a net hydrogen sink consistent with the higher propionate proportion and lower A: P ratio observed in PL [58,60]. Streptococcus, in contrast, is primarily a lactate producer [61]; its overwhelming dominance likely reflects rapid fermentation of soluble substrates rather than a direct role in hydrogen disposal, and the subsequent fate of lactate (e.g., conversion to propionate by lactate-utilizing bacteria) may contribute to the observed VFA profile. CHI was enriched in Lactobacillus and Desulfovibrio (LDA scores: 5.16 and 4.09). Although the higher digestibility and acetate proportion of CHI are consistent with greater hydrogen availability and methane output, the enrichment of Lactobacillus is typical of rapid carbohydrate fermentation [62], whereas the hydrogenotrophic sulfate reducer Desulfovibrio would be expected to compete with methanogens for hydrogen [58]; its role under these conditions remains unclear and warrants targeted functional study. PCH exhibited enrichment in Sharpea and Bifidobacterium, associated with moderate hydrogen fluxes and balanced acetate-to-propionate ratios, while GBS was characterized by fibrolytic taxa such as Olsenella, Butyrivibrio, and Ruminococcus, reflecting its enhanced fiber digestibility [63].

Together, these findings highlight that forage composition shapes rumen microbial community structure and function, with plausible downstream effects on fermentation efficiency and methanogenic potential. The differential enrichment of fermentative and fibrolytic taxa across forages underscores the importance of plant-derived substrates and secondary metabolites as selective forces modulating rumen microbial ecology [64,65].

4.4. Integrative Interpretation and Implications

Collectively, these results demonstrate that Plantago lanceolata and Cichorium intybus were associated with distinct rumen microbial and fermentative responses, including differences in methane output and nitrogen metabolism. In PL, lower CH₄ emissions were observed together with a pattern consistent with reduced proteolytic activity, which may be related to plantain bioactive compounds [13,50], although these were not quantified here, whereas CHI showed extensive substrate degradation and higher digestibility, possibly involving acetogenic fermentation. The intermediate responses observed in PCH may reflect the combined contribution of both forages; however, this study was not designed to test synergistic effects, so whether moderated fermentation intensity and balanced microbial consortia could benefit both nutrient utilization and environmental performance remains to be determined. Because only one plantain-to-chicory ratio (30:70) was evaluated, it is possible that this proportion attenuated the methane-mitigating effect of plantain; testing alternative ratios would be needed to address this.

From a practical perspective, these in vitro findings suggest that P. lanceolata could contribute to lower CH₄ emissions and ruminal ammonia accumulation, which, if confirmed in vivo, could reduce greenhouse gas and potentially nitrogen excretion losses [13,52]. Conversely, C. intybus, with its high digestibility and extensive fermentability, appears suitable for improving nutrient uptake but may not be ideal for methane mitigation when used as the sole forage. The intermediate fermentation and CH₄ yield observed in PCH suggest it may be suitable for mixed diets, in line with a previous in vivo study reporting higher milk protein content in ewes grazing PCH pastures [24], although a direct link cannot be established from our data. Ryegrass (GBS) remains a reliable reference forage for fiber digestion but lacks the functional attributes of the other species. Finally, although the RUSITEC system provides controlled insight into ruminal processes, in vivo validation remains essential, particularly to capture the influence of secondary metabolites, seasonal variation in forage chemistry, and potential long-term microbial adaptations. In addition, the 5-day sampling period, although standard for RUSITEC incubation [66], captures short-term responses and does not reflect longer-term dietary adaptation; the absence of animal performance and product-quality data and the lack of quantification of forage secondary metabolites also limit extrapolation to grazing animals and should be addressed in future in vivo trials. Furthermore, because this study characterized only the bacterial community (16S rRNA V6–V8 region) and did not include archaeal-specific data (e.g., methanogen-targeted qPCR or archaeal 16S sequencing), the mechanistic links proposed here between specific bacterial taxa and methanogenesis remain associative rather than causal and should be confirmed with direct measurements of methanogen abundance and activity in future work.

Future work integrating metatranscriptomic and metabolomic approaches could further elucidate the functional mechanisms underlying these microbial shifts and their persistence under grazing conditions.

5. Conclusions

In this RUSITEC study, Plantago lanceolata and Cichorium intybus differed in ruminal fermentation, CH₄ production, and bacterial community composition. P. lanceolata was associated with lower CH₄ emissions, whereas C. intybus was associated with greater nutrient disappearance and higher CH₄ production. The plantain–chicory mixture (PCH) showed intermediate responses, suggesting a potential complementarity between the two forages that warrants further evaluation. These in vitro findings suggest that plantain and chicory may be of interest as forage options for CH₄ mitigation strategies.

Future in vitro studies should quantify the secondary metabolite profile of the forages and screen alternative plantain–chicory ratios and different concentrations of these compounds and their interactions, while in vivo grazing studies should validate these effects and explore seasonal variability, with the aim of optimizing methane mitigation without compromising nutrient utilization.