**Methane Production of Fresh Sainfoin, with or without PEG, and Fresh Alfalfa at Di**ff**erent Stages of Maturity is Similar but the Fermentation End Products Vary**

### **Pablo José Rufino-Moya, Mireia Blanco, Juan Ramón Bertolín and Margalida Joy \***

Centro de Investigación y Tecnología Agroalimentaria de Aragón (CITA), Instituto Agroalimentario de Aragón–IA2 (CITA-Universidad de Zaragoza), Avda, Montañana 930, 50059 Zaragoza, Spain; pjrufino@cita-aragon.es (P.J.R.-M.); mblanco@aragon.es (M.B.); jrbertolin@cita-aragon.es (J.R.B.) **\*** Correspondence: mjoy@aragon.es; Tel.: +34-976-716-442

Received: 14 March 2019; Accepted: 18 April 2019; Published: 26 April 2019

**Simple Summary:** In the last years, there has been increasing interest in the use of forages containing condensed tannins (CT) in ruminant nutrition. Condensed tannins can reduce the methane emissions and the ruminal degradation of protein, improving the animal performances to different extents depending on the source and dose of CT. In vitro fermentation of sainfoin has not been studied in fresh forage. The effect of CT can be studied in comparison with a similar CT-free forage or using polyethylene glycol (PEG), which is a tannin-blocking agent. The maturity stage influences the chemical composition to a different degree depending on the legume species, and can affect the content and fractions of CT. The aims of this trial were to compare the fermentation parameters of sainfoin with or without PEG, to detect the differences due to CT, at different maturity stages (vegetative, start-flowering, and end-flowering) and compare them with the fermentation parameters of alfalfa. The main results were that sainfoin had greater in vitro organic matter degradability (IVOMD) and lower ammonia and acetic:propionic ratio than alfalfa. Sainfoin CT had effect on ammonia and individual fatty acid proportions. In conclusion, fermentation end products were affected both by the chemical composition and CT contents.

**Abstract:** Alfalfa and sainfoin are high-quality forages with different condensed tannins (CT) content, which can be affected by the stage of maturity. To study the effects of CT on fermentation parameters, three substrates (alfalfa, sainfoin, and sainfoin+PEG) at three stages of maturity were in vitro incubated for 72 h. Sainfoin had greater total polyphenol and CT contents than alfalfa. As maturity advanced, CT contents in sainfoin decreased (*p* < 0.05), except for the protein-bound CT fraction (*p* > 0.05). The total gas and methane production was affected neither by the substrate nor by the stage of maturity (*p* > 0.05). Overall, sainfoin and sainfoin+PEG had greater in vitro organic matter degradability (IVOMD) than alfalfa (*p* < 0.05). Alfalfa and sainfoin+PEG presented higher ammonia content than sainfoin (*p* < 0.001). Total volatile fatty acid (VFA) production was only affected by the stage of maturity (*p* < 0.05), and the individual VFA proportions were affected by the substrate and the stage of maturity (*p* < 0.001). In conclusion, alfalfa and sainfoin only differed in the IVOMD and the fermentation end products. Moreover, CT reduced ammonia production and the ratio methane: VFA, but the IVOMD was reduced only in the vegetative stage.

**Keywords:** polyethylene glycol; gas production; in vitro organic matter degradability; condensed tannins; ammonia; volatile fatty acid; in vitro assay

#### **1. Introduction**

There is increasing interest in legume-based forage production systems because of their low reliance on fertilizer nitrogen, potential environmental benefits, and high protein content that contribute to low-input and sustainable livestock production systems [1]. Alfalfa (*Medicago sativa L*.) and sainfoin (*Onobrychis viciifolia Scop*.) are two pluriannual legumes widely grown in the Mediterranean area, presenting high forage productive capacity, high nutritional value, and restorative action for soil fertility [2]. However, alfalfa has a low content of polyphenols and is considered virtually free of condensed tannins (CTs), whereas sainfoin presents a high content of polyphenols and a medium to high content of CTs [3–5].

Alfalfa is usually fed as hay to ruminants mainly to avoid bloat, although continuous grazing is possible without bloat incidence both in sheep [6] and growing cattle [7]. Thus, the majority of studies that compared the ruminal fermentation of alfalfa and sainfoin have been performed using hays [3,4,8]. The differences between alfalfa and sainfoin have been ascribed to differences in the chemical compositions [4], but also to the presence of CTs [8]. To clarify whether the differences between alfalfa and sainfoin are only due to CTs, polyethylene glycol (PEG) must be added as a tannin-blocking agent [8]. To the best of our knowledge, the ruminal fermentation of both species was compared in fresh forages only by McMahon et al. [9] and Chung et al. [10]. Depending on their content, characteristics, and properties [11], CTs from sainfoin alter both the breakdown of protein in the rumen to ammonia and methane, gas and the production of total volatile fatty acids (VFAs) [3,4,12]. These, in turn, are associated with the variety, stage of maturity, and environmental factors [5,13,14].

The objectives of this trial were to compare the in vitro fermentation of alfalfa and sainfoin at three stages of maturity and to clarify whether the differences between both legumes were due to the CTs of sainfoin using PEG.

#### **2. Materials and Methods**

#### *2.1. Experimental Design*

Three substrates (alfalfa, sainfoin, and sainfoin+PEG) at three stages of maturity (vegetative, start-flowering, and end-flowering) were used to evaluate the effect of sainfoin CTs on in vitro fermentation. Alfalfa was used as a tannin-free legume.

#### *2.2. Animal and Diets*

#### 2.2.1. Forages, Crop Management, and Harvest

The experiment was performed at the CITA Research Institute at Zaragoza (41◦42 N, 0◦49 W), altitude 216 m a.s.l., located in Ebro Valley, north-eastern Spain. The silt loam soil at the site had pH 8.1 and 1.81% organic matter and contained 16% clay, 53.5% silt, and 30.5% sand. Alfalfa and sainfoin were cultivated and managed as described by Lobón et al. [15]. Samples of forage were collected fortnightly from 14 April to 22 September 2015. The stage of maturity of the sainfoin and alfalfa was classified into vegetative, start-flowering, and end-flowering according to Borreani et al. [16] and Kalu and Fick [17], respectively. In each sampling, 10 samples per legume were obtained from 0.25 m2 areas randomly allocated in the plot. These samples were mixed homogenously, and a part of the sample was separated manually into stems, leaves, and flowers to study their respective percentages. Another part of the samples was dried at 60 ◦C for 48 h for chemical analyses, and the rest of the sample was freeze-dried in a Genesis Freeze Dryer 25 (Hucoa Erlöss, SA/Thermo Fisher Scientific, Madrid, Spain) for polyphenol and CT analyses and in vitro fermentation assays. Samples for the chemical analyses were ground and sieved through a 1-mm screen (Rotary Mill, ZM200 Retsch, Hann, Germany), and a part of the samples was ground and sieved through a 0.2-mm screen for crude protein (CP), polyphenol, and CT determination. All the samples were stored at −20 ◦C in total darkness.

The number of samples of vegetative, start-flowering and end-flowering stages of each forage to study the in vitro fermentation were chosen in concordance with the representativeness of the plant development. The vegetative stage was the most frequent stage (55%), followed by end-flowering (27%) and start-flowering (18%) stages. Three samples of the vegetative stage and two samples of the start- and end-flowering of each legume species were studied.

#### 2.2.2. Animals and Sampling of Ruminal Digesta

The procedures used in the trial followed the Spanish guidelines for experimental animal protection (RD 53/2013) and were approved by the Institutional Animal Care and Use Committee of the Research Centre (Procedure number 2011-05). Rasa Aragonesa wethers (n = 4; Live weight: 65 ± 2.1 kg) were used as donors of ruminal content. The management of the rumen-cannulated wethers and the sampling of the ruminal digesta was made as reported in Rufino-Moya et al. [18]. Briefly, wethers were fed a diet composed by alfalfa hay (70%) and barley grain (30%) at energy maintenance level. Before morning feeding, ruminal digesta from each wether was collected into a prewarmed insulated thermo, individually strained through four layers of cheesecloth and homogenized. Rumen fluid from the four wethers was mixed (pH: 6.76 ± 0.099), and a buffer solution was added in a proportion of 1:2 (*v*/*v*) based on the protocol of Menke and Steingass [19]

#### 2.2.3. In Vitro Gas Production Technique and Sampling

The production of gas was measured with the Ankom system (Ankom Technology Corporation, Fairport, NY, USA), which had 310 mL capacity bottles fitted with pressure and temperature sensors. The valve open time was one second, the threshold for gas release was 5 PSI and the bottles were not shaken. Five hundred mg of freeze-dried substrate (alfalfa, sainfoin, or sainfoin+PEG) were incubated with 60 mL of buffered solution:rumen fluid (2:1 *v*/*v*) in a water bath (at 39 ◦C) for 72 h. To make the sainfoin+PEG samples, PEG-4000 (Merck, Darmstadt, Germany) was added to the buffered rumen fluid at a concentration of 2.3 g/L [12]. Four runs were conducted on four separate days, and each sample was incubated in duplicate in each run. Gas production and corrected with the blanks (two bottles without substrate were included in each run).

After 72 h of incubation, the bottles were placed in ice to stop fermentation (5–10 min), and then tempered (at room temperature for 10–15 min). A sample of the gas produced was transferred (at atmospheric pressure) with a syringe attached to a manometer into a Vacutainer®tube to determine CH4 and conserved at 4 ◦C until analysis. At the end of gas sampling, the pH was measured immediately with a microPH 2002 (Crison Instruments S.A., Barcelona, Spain). The sampling to determine ammonia (NH3-N) content and VFA were carried out as reported in Rufino-Moya et al. [18]. Briefly, to determine the ammonia content, 2.5 mL of liquid was mixed with 2.5 mL HCl 0.1 N. For VFA determination, 0.5 mL of the liquid was added to 0.5 mL of deproteinizing solution and 1 mL of distilled water. Tubes with samples for determination of ammonia and VFAs were stored at −20 ◦C until future analyses. The entire incubated sample was filtered through a preweighed bag (50 μm; Ankom) to estimate the in vitro organic matter degradability (IVOMD). Briefly, the bags were sealed, washed, dried at 103 ◦C for 48 h, and finally, placed in a muffle at 550 ◦C to obtain the ashes. The organic matter of the bag content was obtained as DM-ashes, and the IVOMD was calculated.

#### *2.3. Analytical Methods*

#### 2.3.1. Chemical Composition

All the analyses of the chemical composition were analyzed as reported in Rufino-Moya et al. [18] according to official methods. Briefly, AOAC methods were used to determine the contents of dry matter (index no. 934.01), ash (index no. 942.05), and CP (index no. 968.06) [20]. Neutral detergent fiber (NDFom), acid detergent fiber (ADFom), and lignin (sa) contents were determined according to the method described by Van Soest et al. [21] using the Ankom 200/220 fiber analyzer (Ankom Technology Corporation). The NDFom was assayed with a heat stable amylase. The lignin (sa) was analyzed in ADF residues by the solubilization of cellulose with sulfuric acid. All the values were corrected for ash-free content. The ether extract (EE) was determined following the approved procedure Am 5-04 [22] using an XT10 Ankom extractor (Ankom Technology Corporation). The nonstructural carbohydrates (NSC) were calculated as *NSC* = 1000 − *CP* − *EE* − *NDF* − *ash*, according to Guglielmelli et al. [3].

The content of total polyphenol (TP) and the fractions of CT were determined as described in Rufino-Moya et al. [18]. Briefly, the TP were extracted using the method of Makkar [23] and were quantified following the method of Julkunen-Tiitto [24]. The extractable CT (ECT), protein-bound CT (PBCT), and fiber-bound CT (FBCT) were extracted and fractioned following the method of Terrill et al. [25] and quantified by the colorimetric HCl-butanol method described by Grabber et al. [26]. The standard used for the quantification was extracted and purified from sainfoin using the method described by Wolfe et al. [27]. An Heλios β spectrophotometer was used to measure the samples and standard calibration at 550 nm.

2.3.2. Determination of Parameters of the In Vitro Gas Production Technique

Gas production, recorded hourly for 72 h, was used to estimate the parameters of the kinetics of fermentation adjusting the gas produced to the model described by France et al. [28]:

$$P = A \times \left(1 - e^{-ct}\right)$$

where *P* is the cumulative gas production (mL) at time *t* (h), *A* is the potential gas production (mL), and *c* is the rate of gas production (h<sup>−</sup>1).

An HP-4890 (Agilent, St. Clara, CA, USA) gas chromatograph (GC) equipped with a flame ionization detector (FID) and a TG-BOND Q+ capillary column (30 m × 0.32 mm id × 10 μm film thickness, Thermo Scientific, Waltham, MA, USA) was used to determine CH4. Helium was used as the carrier gas at a flow rate of 1 mL/min. The oven temperature was maintained at 100 ◦C (isothermal program). The splitless injection volume was 200 μL. Methane identification was based on the retention time compared with the standard. The methane concentration was calculated from the peak concentration:area ratio using the peak area generated from standard gas as the reference (CH4; 99.995% purity [C45], Carburos Metálicos, Barcelona, Spain).

The content of ammonia was measured in Epoch microplate Spectrophotometer (BioTek Instruments, Inc., Winooski, VT, USA) using a colorimetric method described by Chaney and Marbach [29].

A Bruker Scion 460 GC (Bruker, Billerica, MA, USA) equipped with CP-8400 autosampler, FID and a BR-SWax capillary column (30 m × 0.25 mm i.d. × 0.25 μm film thickness, Bruker) was used to determine the concentration of acetic, propionic, iso-butyric, butyric, iso-valeric, and valeric acids. Helium was the carrier gas (flow rate of 1 mL/min). The oven temperature program was 100 ◦C, followed by a 6 ◦C/min increase to 160 ◦C. The injection volume was 1 μL at a split ratio of 1:50. The individual VFAs were identified based on retention time comparisons with commercially available standards of acetic, propionic, iso-butyric, butyric, iso-valeric, valeric, and 4-methyl-valeric acids at ≥99% purity (Sigma-Aldrich, St. Louis, MO, USA).

#### *2.4. Statistical Analyses*

Data were analyzed using SAS v. 9.3 (SAS Inst. Inc., Cary, NC, USA). The fermentation kinetic parameters (*A* and *c*) were estimated using a nonlinear regression model (NLIN procedure). The contents of secondary compounds were analyzed using the GLM procedure with the substrate, the stage of maturity and its interaction as fixed effects. Total gas and CH4 production, *A*, *c*, IVOMD and the fermentation end products were analyzed using mixed models considering the substrate, the stage of maturity and its interaction as fixed effects and the run as random effect. If the interaction was not significant, it was deleted from the model and the analysis was repeated. The least square means, their

associated standard errors and the differences were obtained. Pearson correlation coefficients between variables were calculated using the CORR procedure. For the entire test, the effects were considered a significant probability at a value of *p* < 0.05 or a trend at *p* = 0.05.

#### **3. Results**

#### *3.1. Chemical Composition*

The chemical composition and the percentage of stems, leaves, and flowers of both legume species at the three stages of maturity are shown in Table 1. On average, alfalfa and sainfoin had similar ADFom (231 g/kg DM), CP (198 g/kg DM), EE (15 g/kg DM), and NSC (335 g/kg DM) contents. Alfalfa, however, had higher ash and NDFom contents and a lower lignin (sa) content than sainfoin. For both forages, NDFom and ADFom content increased as the stage of maturity progressed, whereas the CP content decreased. As the forage matured, the lignin (sa) content increased only in alfalfa whereas the contents of EE and NSC decreased only in sainfoin.

**Table 1.** Chemical composition and plant components of alfalfa and sainfoin at three stages of maturity.


<sup>1</sup> crude protein; <sup>2</sup> Neutral detergent fiber; <sup>3</sup> acid detergent fiber; <sup>4</sup> nonstructural carbohydrates.

Alfalfa and sainfoin had similar proportions of leaves (51.9 vs. 53.6%). However, alfalfa had a greater proportion of stems (45.2 vs. 39.7%) and a lower proportion of flowers (2.8 vs. 6.7%) than sainfoin. Regarding the stage of maturity, the proportion of stems and flowers increased, whereas the proportion of leaves decreased as the stage of maturity advanced.

#### *3.2. Contents of Total Polyphenols and Condensed Tannins*

The content of total polyphenols and the total (TCT) and fractions of CT were affected by the interaction between the legume species and the stage of maturity (*p* < 0.05) (Figure 1). Alfalfa presented steady contents of total polyphenols, TCTs, ECTs, PBCTs, and FBCTs, which were lower than those of sainfoin (*p* < 0.001) regardless of the stage of maturity. The contents of polyphenols, TCTs, ECTs, and FBCTs decreased as maturity advanced (*p* < 0.05).

**Figure 1.** Effect of the species and the stage of maturity on the contents of total polyphenols (TP), total condensed tannins (TCT), extractable CT (ECT), protein-bound CT (PBCT), and fiber-bound (FBCT). Within a parameter, means with different letter differ at *p* < 0.05. (Within each species: n = 3 for the vegetative stage, n = 2 for the start-flowering, and n = 2 for the end-flowering stages.)

#### *3.3. In Vitro Fermentation*

The pH was affected by the interaction between the substrate and the stage of maturity (*p* < 0.01; Table 2). Alfalfa had greater pH than sainfoin and sainfoin+PEG (*p* < 0.05). Sainfoin and sainfoin+PEG were affected similarly by the stage of maturity (*p* < 0.001), with the greatest pH value in the vegetative

stage (*p* < 0.05) Total gas and CH4 production were affected neither by the substrate nor by the stage of maturity (*p* > 0.05; Table 2). However, the interaction between the substrate and the stage of maturity affected *A* (*p* < 0.001) and c (p = 0.05). Alfalfa showed lower *A* values at start-and end-flowering stages (*p* < 0.001) and greater *c* in the vegetative and start-flowering stages (*p* < 0.05) than sainfoin and sainfoin+PEG (*p* < 0.05). Regarding the effect stage of maturity, sainfoin and sainfoin+PEG had the lowest *A* values in the vegetative stage (*p* < 0.05). As the stage of maturity progressed, *c* increased in sainfoin and sainfoin+PEG substrates (*p* < 0.05).

The IVOMD was also affected by the interaction between the substrate and the stage of maturity (*p* < 0.001; Table 2). Alfalfa had lower IVOMD than both sainfoin substrates in the start-flowering and end-flowering stages (*p* < 0.05), whereas sainfoin+PEG had greater IVOMD than alfalfa and sainfoin in the vegetative stage (*p* < 0.05). In alfalfa, the IVOMD decreased as the stage of maturity advanced (*p* < 0.05). The sainfoin and sainfoin+PEG showed the greatest IVOMD in the start-flowering stage (*p* < 0.05). The IVOMD was correlated with *A* (r = 0.60; *P* < 0.01) and with the total VFA production (r = 0.51; *p* < 0.05).

The NH3-N content was only affected by the substrate (*p* < 0.001); sainfoin produced a lower NH3-N concentration than alfalfa and sainfoin+PEG (Table 2). In contrast, total VFA production was only affected by the stage of maturity (*p* < 0.05), the start-flowering stage presented greater VFA production than the rest of the stages (105, 99, and 100 mmol/L for start-flowering, vegetative, and end-flowering, respectively). Regarding the individual VFAs, alfalfa had a lower acetic acid proportion and greater proportions of the rest of the individual VFAs than sainfoin (*p* < 0.001). When comparing both sainfoin substrates, sainfoin had a greater acetic acid proportion and lower proportions of the rest of the VFAs than sainfoin+PEG (*p* < 0.001). Sainfoin presented the greatest C2:C3 ratio, followed by sainfoin+PEG and alfalfa, which had the lowest ratio (*p* < 0.001). Regarding the effect of the stage of maturity, the vegetative stage had a lower proportion of acetic acid and greater proportions of propionic, iso-butyric, and iso-valeric acid than the rest of stages of maturity (*p* < 0.001). The vegetative stage had a lower C2:C3 ratio than the other stages (*p* < 0.001). The CH4:VFA ratio was only affected by the substrate (*p* = 0.01); sainfoin+PEG presented a greater CH4:VFA ratio than the other substrates (*p* < 0.05).



#### **4. Discussion**

In Mediterranean areas, there is an increasing interest to reintroduce forage-based systems in ovine production to ensure the viability and sustainability of the farms. Legumes are especially advisable due to their nutritional quality for ruminants and environmental benefits [1]. Moreover, the presence of CT in some legumes may decrease CH4 production and improve the performance of ovine to different extents depending on the source and dose of CT. As these legumes are usually fed conserved, either as silage or as hay, there is scarce information on the fermentation parameters when alfalfa and sainfoin are offered fresh. Previous experiments showed differences between the fermentation parameters of alfalfa and sainfoin hay and silage [4,30], however, it is not clear if the differences were due to the chemical composition, the presence of CT in sainfoin or both. With the present study, the fermentation parameters of alfalfa and sainfoin, with or without PEG, in fresh at different stages of maturity were compared to try to clarify the origin of the differences in fermentation. The use of gas production technique is a good tool to evaluate the effect of CT on fermentation parameters, but the fermentation is influenced by the time of incubation, species of the animal donor and the diet [31,32]. Furthermore, the effects of sainfoin CT vary according the variety, harvest time, and cultivation site [13,14], making it difficult to compare the results with other studies. In the present study, the content of total CT and their fractions were analyzed, however, the chemical characteristics of CT (molecular weight, degree of polymerization, prodelphinidin/procyanidin ratio, cis/trans ratio, etc.) were not evaluated.

#### *4.1. E*ff*ect of the Substrate*

References showed noticeable variability in the chemical composition of alfalfa and sainfoin among studies, which is related to the stage at harvest, leaf:stem ratio, soil characteristics, weather conditions, and the cultivars utilized [5,13,14]. In the present study, the similar CP and ADF contents and different NDF contents of alfalfa and sainfoin agree partially with the results reported by Chung et al. [10], who observed similar NDF, ADF and CP contents in fresh alfalfa and sainfoin at the late vegetative stage. However, at the early vegetative stage, the same authors observed greater NDF and ADF contents in alfalfa than sainfoin and similar CP contents.

Alfalfa has a low content of total polyphenols and is considered a CT-free legume [5], although it may present very low CT content in the seed coats. Therefore, the present results related to the presence of CTs agree with previous studies that analyzed both legumes offered fresh [33,34] or as preserved forages [4,30].

The pH did not negatively affect the fermentation environment because the values were within the range of 6.2 to 6.8 and these values ensure a favorable environment for the activity of cellulolytic bacteria [35]. The inclusion of PEG in the current experiment did not affect pH, as reported in fresh sainfoin [12] and sainfoin hay [8]. Regarding gas production, the similar production of alfalfa and sainfoin agrees with the similar gas production observed in alfalfa and sainfoin leaves incubated in Rusitec units [9] and alfalfa and sainfoin silages [30]. However, when alfalfa and sainfoin hays were studied, differences in gas production were reported [3,4,8]. The inconsistency of the results of the type of substrate on gas production might be related to the differences in chemical composition, the different characteristics of CTs and of type and settings of the in vitro assay [14]. In that sense, the similar gas production between sainfoin and sainfoin+PEG was unexpected because previous experiments reported increases in gas production when PEG was added to fresh sainfoin [12,14]. According to Azuhnwi et al. [13], the inclusion of PEG increased gas production by 2.7 to 9.6%, depending on the sainfoin variety, site, and stage at harvest. In the current experiment, the inclusion of PEG slightly increased the gas production, although not statistically significantly.

The similar CH4 production of alfalfa and sainfoin recorded in the present study agrees with results reported using fresh forages [10] and silages [30]. However, the inclusion of extracts from sainfoin accessions in alfalfa decreased CH4 production, but the effect was greatly dependent on the accession and the dose of inclusion [11]. Moderate CT content may have beneficial effects reducing rumen CH4 emission production [36]. The action of CT on methanogenesis can be attributed to indirect effects via reduced hydrogen production (and presumably reduced forage digestibility) and via direct inhibitory effects on methanogens [34]. Regarding the effect of PEG, the inclusion of PEG in sainfoin did not affect CH4 production in previous studies [12,37] as in the current experiment. The structural features of condensed tannins affect in vitro CH4 production, which may be linked to the interaction of CTs with dietary substrate or microbial cells [11,38]. Therefore, the type of CT and dose present in the current experiment might not be sufficient to modify CH4 production.

The reduced *A* in alfalfa compared with sainfoin and sainfoin+PEG in the start- and end-flowering stages can be related to the higher fiber fraction, as reported by Guglielmelli et al. [3]. In the current experiment, the presence of CTs in sainfoin had no effect on *A*, as reported by Calabrò et al. [8]. The higher *c* in alfalfa when compared to sainfoin agrees with the results reported by Hatew et al. [11], although the effect on *c* depends on the types and concentrations of sainfoin CTs.

The lower IVOMD of alfalfa, when compared to sainfoin and sainfoin+PEG, was also reported using fresh forage estimated in situ [10] and in vitro [39] and could be related to the greater fiber fraction. The increased IVOMD in sainfoin+PEG with respect to sainfoin at the vegetative stage could be related to the blockage of CTs by the PEG. However, Theodoridou et al. [12] reported no effect of the inclusion of PEG in sainfoin on IVOMD studied at 24 h, regardless of the stage of maturity. The discrepancy between studies could be related to the content, characteristics and structures of CTs, which depends on the botanical species and variety of the source [13,14].

The NH3-N contents recorded in the present study are in line with most similar studies that compared alfalfa and sainfoin [3,4,12]. The reduced NH3-N concentration in sainfoin with respect to alfalfa and sainfoin+PEG confirmed the inhibition elicited by CTs in the ruminal degradation of dietary proteins due to the formation of complexes CT-protein at ruminal pH [40]. In contrast, the effect of CTs on total VFA production is not clear. Some studies reported a lower total VFA production in sainfoin silage than in alfalfa silage [30], and the inclusion of different doses of extracted accession of sainfoin in alfalfa decreased or maintained the total VFA production, depending on the accession [11]. In the current experiment, the total VFA production was not affected by the legume species, as observed by other authors [4,10]. The inclusion of PEG did not affect total VFA production in the current experiment as reported for sainfoin hay [8,37], which is contrary to the increase of total VFA production observed by Hatew et al. [14]. The differences between the studies could be due to the time of incubation, species of the animal donor, chemical structure, and biological activity of CTs [14,31].

Generally, the presence of CT from sainfoin leads to an increase in the propionic acid proportion and a reduction in the C2:C3 ratio [10,11,38]. However, the effect on each individual VFA proportion is variable due to the type of substrate, types, and contents of CTs and length of incubation period or the donor animal [11,31]. In the present study, sainfoin had a greater acetic acid proportion than alfalfa, which was similar to results from Guglielmelli et al. [3] using hays and Grosse-Brinkhaus et al. [30] using silages. The C2:C3 ratio recorded in the present study was greater in sainfoin than in alfalfa and sainfoin+PEG, which is in contrast with other studies that did not observe effects of the type of substrate or the addition of PEG [3,12,37]. Sainfoin had lower valeric acid and branched-chain VFA proportions than alfalfa and sainfoin+PEG because of the presence of CTs [4,30,38]. Condensed tannins reduce the proportions of branched-chain VFAs due to reduce protein degradation in the rumen because these VFA are products of the breakdown of the carbon skeleton of amino acids during rumen fermentation [41].

#### *4.2. E*ff*ect of the Stage of Maturity*

The decrease in CP content and the concomitant increase in the cell wall (NDFom and ADFom) content as the stage of maturity progressed in both forages is a result of the decrease of the proportion of leaves to stems and the increase of lignified tissues [10]. The steady lignin (sa) content in sainfoin during the development of plants can be due to some interference between this compound and CTs during analysis, as reported by Guglielmelli et al. [3].

In the current experiment, the TP and TCT contents were affected by the stage of maturity, as reported in previous studies [3,5,42], although the magnitude of the effect varied among the studies. Regarding the CT fractions, there is little information about the influence of the stage of maturity. As in the present study, Chung et al. [10] observed a reduction of the ECT fraction in the end-flowering stage with respect to the vegetative stage in sainfoin. However, Jin et al. [43] observed greater TCT, ECT, and PBCT contents in *Dalea purpurea* at flowering than at the vegetative stage due to the high percentage of flowers, which are very rich in CTs [44]. From a physiological point of view, the reduction in the secondary compound concentrations as maturity advances could be due to a sort of dilution as a consequence of the growth and expansion of plant cells [45] and to the decrease in the leaf: stem ratio as a consequence of the reduction of leaves, which are rich in CTs [44].

The lack of an effect of the stage of maturity on pH values in alfalfa agrees with the results reported by Chung et al. [10], but the reduction of pH in sainfoin at both flowering stages disagrees with the abovementioned study. In this sense, the stage of maturity had no effect on pH when sainfoin hay was incubated [3,8]. More studies considering the stage of maturity, the chemical composition and the presence of secondary compounds in alfalfa and sainfoin must be performed to discern the importance of these factors on the ruminal pH.

Previous studies reported a reduction of in vitro gas and CH4 production as the stage of maturity advanced, associated with the chemical composition and the CT content [12,46]. However, the stage of maturity had no effect on gas and CH4 production in the current experiment, which is in agreement with in vitro [3] and in vivo [10] experiments. The similar chemical composition observed between stages of maturity, the low biological activity of CT and the interactions between nutritive components and antinutritional factors could be responsible for the similar gas and CH4 production [3].

As expected, the IVOMD of alfalfa decreased as the fiber fraction increased with maturity [47]. In contrast, sainfoin, with or without PEG, presented considerably high IVOMD at the start-flowering stage with respect to the rest of the stages in agreement with Theodoridou et al. [12]. These results could be due to the different biological activity of CTs at vegetative and start-flowering stages [5], because the chemical composition and CT contents were similar in both stages.

The advancing of the stage of maturity tended to reduce NH3-N production as a consequence of the decrease of CP content in concordance with several studies carried out in vitro [3,12] and in vivo [10,42]. Studies concerning the effect of the stage of maturity on the total production of VFAs and their proportions show discrepancies. The chemical composition of the substrates, the length of the incubation period, and the inoculum donor animal are determinant factors that can influence VFA production and proportions [31]. In the current experiment, the start-flowering stage presented the highest total VFA production, in concordance with the highest IVOMD observed. However, Theodoridou et al. [12] studied the effect of the stage of maturity of fresh sainfoin with a similar CT content in a 24 h in vitro assay and did not find an effect on the total VFA production.

In relation to the proportion of individual VFAs, the effect of the stage of maturity on these parameters has been reported in vitro and in vivo in previous studies [3,10], but the results are not consistent. In the current experiment, as maturity advances, there is an increase in acetic acid and a decrease in propionic acid proportions, thus increasing the C2:C3 ratio due probably to increase of fiber and reduction of CT content [42]. The reduction of the proportion of iso-butyric and iso-valeric acids in the start- and end-flowering stages and valeric acid at the start-flowering stage in comparison with the vegetative stage might be explained by the decrease in CP content, because they are products of the breakdown of the carbon skeleton of amino acids during rumen fermentation, as the maturity of the forage advanced [10,47].

#### **5. Conclusions**

In conclusion, sainfoin might be an alternative to alfalfa due to the high IVOMD and the potential protection against ruminal protein degradation, according to the results of ammonia content, branched-chain VFAs, and valeric acid proportion from sainfoin in vitro. The effect of the stage of

maturity was less than expected, probably due to the high quality of the forages. It is required to study the effects of the type of substrate and stage of maturity on animal performance to recommend the best stage of maturity to cut sainfoin and alfalfa.

**Author Contributions:** Conceptualization, M.B.; Data curation, M.B. and M.J.; Formal analysis, P.J.R.-M. and M.B.; Funding acquisition, M.J.; Investigation, P.J.R.-M. and J.R.B.; Methodology, M.B. and M.J.; Project administration, M.J.; Resources, M.B. and M.J.; Supervision, M.B. and M.J.; Visualization, P.J.R.-M.; Writing—original draft, P.J.R.-M.; Writing—review & editing, M.B. and M.J..

**Funding:** This research was funded by the Ministry of Economy and Competitiveness of Spain, the European Union Regional Development Funds (INIA RTA2012-080-00, INIA RZP2017-00001) and the Research Group Funds of the Aragón Government (A14\_17R). P.J. Rufino's contract is supported by a doctoral grant from the INIA-EFS, M. Blanco has a contract supported by INIA-EFS.

**Acknowledgments:** Appreciation is expressed to the staff of CITA de Aragón for their help in data collection. Special thanks to the lab team and to the crop team for their assistance.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **In Vitro Evaluation of Di**ff**erent Dietary Methane Mitigation Strategies**

#### **Juana C. Chagas, Mohammad Ramin and Sophie J. Krizsan \***

Department of Agricultural Research for Northern Sweden, Swedish University of Agricultural Sciences (SLU), Skogsmarksgränd, 90183 Umeå, Sweden; juana.chagas@slu.se (J.C.C.); mohammad.ramin@slu.se (M.R.)

**\*** Correspondence: sophie.krizsan@slu.se; Tel.: +46-90-7868748

Received: 15 October 2019; Accepted: 5 December 2019; Published: 11 December 2019

**Simple Summary:** Dietary methane mitigation strategies do not necessarily make food production from ruminants more energy-efficient, but reducing methane (CH4) in the atmosphere immediately slows down global warming, helping to keep it within 2 ◦C above the pre-industrial baseline. There is no single most efficient strategy for mitigating enteric CH4 production from domestic ruminants on forage-based diets. This study assessed a wide variety of dietary CH4 mitigation strategies in the laboratory, to provide background for future studies with live animals on the efficiency and feasibility of dietary manipulation strategies to reduce CH4 production. Among different chemical and plant-derived inhibitors and potential CH4-reducing diets assessed, inclusion of the natural antimethanogenic macroalga *Asparagopsis taxiformis* showed the strongest, and dose-dependent, CH4 mitigating effect, with the least impact on rumen fermentation parameters. Thus, applying *Asparagopsis taxiformis* at a low daily dose was the best potential dietary mitigation strategy tested, with promising long-term effects, and should be further studied in diets for lactating dairy cows.

**Abstract:** We assessed and ranked different dietary strategies for mitigating methane (CH4) emissions and other fermentation parameters, using an automated gas system in two in vitro experiments. In experiment 1, a wide range of dietary CH4 mitigation strategies was tested. In experiment 2, the two most promising CH4 inhibitory compounds from experiment 1 were tested in a dose-response study. In experiment 1, the chemical compounds 2-nitroethanol, nitrate, propynoic acid, p-coumaric acid, bromoform, and *Asparagopsis taxiformis* (AT) decreased predicted in vivo CH4 production (1.30, 21.3, 13.9, 24.2, 2.00, and 0.20 mL/g DM, respectively) compared with the control diet (38.7 mL/g DM). The 2-nitroethanol and AT treatments had lower molar proportions of acetate and higher molar proportions of propionate and butyrate compared with the control diet. In experiment 2, predicted in vivo CH4 production decreased curvilinearly, molar proportions of acetate decreased, and propionate and butyrate proportions increased curvilinearly with increased levels of AT and 2-nitroethanol. Thus 2-nitroethanol and AT were the most efficient strategies to reduce CH4 emissions in vitro, and AT inclusion additionally showed a strong dose-dependent CH4 mitigating effect, with the least impact on rumen fermentation parameters.

**Keywords:** antimethanogenic; chemical inhibition; global warming; halogenated compound; macroalgae; methane production; methanogenic inhibitor; plant inhibitory compound

#### **1. Introduction**

The global population is growing and, although there is enough food in the world today, there are major differences in how people live. Meat and milk from ruminants are high-quality foods and a large proportion of their production is based on grass, but production is still resource-intensive. Future intensification of agriculture can reinforce negative effects such as greenhouse gas (GHG) emissions, the main contributor to climate change through global warming [1,2].

Methane (CH4) is a powerful GHG that plays a key part in global climate change and concentrations have been rising rapidly in the atmosphere over the past decade. Recently published data based on radioactive carbon (C14) content in CH4 indicate that anthropogenic emissions of CH4 in recent decades have been higher than previously estimated [3]. Satellite data [4] suggest that the increased global CH4 emissions in the period 2005–2015 were mostly due to increased extraction of shale gas, and that the natural gas and oil industry contributes twice as much CH4 emissions as animal agriculture.

Methanogenesis in the rumen is an essential metabolic process required to remove molecular hydrogen generated during fermentation. The production of CH4 is influenced by animal species, age, management, and diet. The Rumen Census project sequenced a wide variety of rumen and camelid foregut microbial communities in many samples from a wide variety of animal species and countries, to identify factors such as diet, host species, or geography causing the greatest variation in CH4 emissions [5]. The results showed that rumen archaeal diversity was similar irrespective of host or diet, and a core rumen bacterial population of 67% of the community occurred irrespective of host or diet. The main diversity changes in other bacteria present were caused by diet, and not host genetics [6]. Although dietary strategies for mitigating enteric CH4 production in ruminants have been intensively studied, no single most efficient dietary strategy has been identified for dairy cows on forage-based diets.

Methane losses from typical dairy cow diets are 6–7% of gross energy intake, but losses are approximately 3% in feedlot situations, indicating that feeding high-concentrate diets can reduce CH4 production [7]. However, recent data indicate that the effect on CH4 production of including more grain in dairy cows diets is small [8] and use of this strategy can therefore be questioned. Use of antimethanogenics or plant inhibitory compounds in ruminant diets can also reduce GHG emissions, and has been suggested as an effective and feasible strategy in the livestock sector [9]. Dietary mitigation strategies do not necessarily make food production from ruminants more energy-efficient, but they reduce CH4 emissions to the atmosphere and thus immediately slow down global warming [10], contributing to keep the planet within 2 ◦C of the pre-industrial baseline [11]. The use of CH4 inhibitors might be the most immediate and efficient strategy to reduce CH4 emissions from dairy cows. The most successful inhibitor suggested in vivo so far is 3-nitroxypropanol that has showed CH4 reducing effects when provided to dairy cows in a low dose [12]. The tropical macroalgae *Asparagopsis Taxiformis* is a recent and natural supplement that has shown very promising CH4 inhibitory effects in vitro [13].

In vitro gas production technique has been developed to evaluate factors influencing digestibility and fermentation kinetics from feeds. The technique has been used to estimate CH4 emission with the advantage of screening large number of samples, providing large amount of data points, and allowing accurate predictions of in vivo CH4 production [14].

This study assessed and ranked a wide variety of dietary CH4 mitigation strategies using an automated gas in vitro system, in order to provide background for future in vivo evaluations of dietary manipulation strategies for efficiently reducing CH4 production from domestic ruminants.

#### **2. Materials and Methods**

Two in vitro experiments were conducted to assess different dietary antimethanogenic compounds. In experiment 1, the dietary CH4 mitigating strategies tested comprised six chemical inhibitory compounds at two levels, three plant-derived inhibitory treatments at two levels, five different potentially CH4-reducing diets with the active ingredients in two levels except for one of the diets, and two typical grass silage fermentation acids at two levels to mimic different silage fermentation qualities. In experiment 2, the two most promising CH4 inhibitory treatments from experiment 1 were tested in a dose-response experiment designed to represent a wide range of treatment levels.

#### *2.1. Experimental Treatments*

#### 2.1.1. Experiment 1

All experimental diets were composed from a control diet that consisted of timothy grass (*Phleum pratense*), rolled barley (*Hordeum vulgare*), and rapeseed (*Brassica napus*) meal in a ratio of 545:363:92 g/kg diet dry matter (DM). The grass and rolled barley originated from Röbäcksdalens research farm in Umeå (63◦45' N, 20◦17' E), Sweden. The rapeseed meal was a commercial solvent-extracted and heat-moisture-treated protein supplement ExPro-00SF (Aarhus Karlshamn AB, Malmö, Sweden). All potential dietary CH4 mitigating strategies tested in experiment 1 are listed in Table 1. The chemical compounds 2-nitroethanol (2-NE), propynoic acid, ferulic acid, p-coumaric acid, and bromoform (Sigma-Aldrich Sweden AB, Stockholm, Sweden) were added without replacing any DM of the control diet. Nitrate was added to the control diet to represent one level of 21 g NO3/kg DM or 0.0890 g Ca(NO3)2 × 4H2O/g DM (Sigma-Aldrich Sweden AB, Stockholm, Sweden). The nitrate treatment was compared with a zero-nitrate treatment in which 0.0350 g urea/g DM and 0.051 g CaCO3/g DM (J.T. Baker BV, Deventer, The Netherland) were added to the control diet to achieve an isonitrogenous and equivalent diet (159 g crude protein (CP)/kg DM). The plant-derived compounds rowan (*Sorbus aucuparia*) berries and the forb fireweed (*Chamerion angustifolium*) were added to replace grass and barley in the control diet, such that the ratio of forage:concentrate was kept constant relative to all other diets. These ingredients were collected in Umeå (63◦ N, 20◦ E), Sweden in October and July 2018, respectively. The red seaweed *Asparagopsis taxiformis* (AT) was added in such a small dose in both levels of the treatment that no replacement of control dietary ingredients was made. The seaweed was harvested in the Azores (38.6◦ N, 28◦ W), Portugal, in October 2018. Replacements in the potentially CH4-reducing diets were also made so that the forage:concentrate ratio was kept constant relative to all other diets and to contain 160 g CP/kg diet DM. Rapeseed oil and oats (*Avena sativa*) were added to replace grass and barley on a DM basis. These ingredients were also collected in Umeå in July 2018. Dried distiller's grains (Agrodrank 90, Agroetanol, Östergötland, Sweden) replaced rapeseed meal in the control diet and was added to represent an increment of 20 g/kg DM in CP between the levels (CP concentration 160 g/kg DM and 180 g/kg DM, respectively). The CP concentration was made iso-nitrogenous to the control diet for the lowest level when dried distiller's grain replaced rapeseed meal.


**Table 1.** Experimental treatments evaluated in vitro in experiment 1 for methane (CH4) mitigation potential.

DM = dry matter; <sup>1</sup> 0.035 g of urea + 0.051 g of CaCO3 on DM basis included in control diet in comparison with nitrate treatment; <sup>2</sup> 0.089% Ca(NO3)2 × 4H2O on DM basis; <sup>3</sup> Urea was added to correct CP at 160 g/kg DM; <sup>4</sup> Urea was added to correct CP at 160 g/kg DM.

In the treatments where maize (*Zea mays*) silage replaced grass silage, urea was added to make diets isonitrogenous to the control diet. No correction of CP concentration was made in the diet when red clover (*Trifolium pratense*) replaced grass.

#### 2.1.2. Experiment 2

In experiment 2, AT (0, 0.06, 0.13, 0.25, 0.5, and 1.0 g/kg of diet organic matter (OM)) and 2-NE (0, 0.3, 0.7, 1.3, 2.6, and 5.1 m*M*) were tested in a dose-response experiment comprising six different treatment levels. The same control diet as in experiment 1, of timothy, rolled barley, and rapeseed meal, was used in experiment 2.

#### *2.2. In Vitro Incubations*

The handling of animals in this experiment was approved by the Swedish Ethics Committee on Animal Research (Dnr A 32-16), represented by the Court of Appeal for Northern Norrland in Umeå, and the experiment was carried out in accordance with laws and regulations governing experiments performed with live animals in Sweden.

Two lactating Swedish Red cows, fed ad libitum on a diet of 600 g/kg grass silage and 400 g/kg concentrate on a DM basis (presenting chemical composition as 509 g/kg of DM, 425 g/kg NDF, and 171 g/kg CP), were used as donor animals of rumen inoculum for all incubations. The rumen fluid from each cow was filtered separately using a double layer of cheesecloth into Thermos flasks that were pre-warmed and flushed with carbon dioxide (CO2) prior to collection. Rumen fluid was transported to the laboratory within 15 min. Equal amounts from each cow were immediately blended, strained through four layers of cheesecloth, and added to buffered mineral solution [15] including PeptoneTM (pancreatic digested casein; Merck, Darmstadt, Germany) at 39 ◦C under constant mixing and CO2 flushing, to give a buffered rumen fluid solution with a rumen fluid:buffer ratio of 1:4 by volume.

Prior to each in vitro incubation, dietary ingredients were dried at 60 ◦C for 48 h and milled in a Retsch SM 2000 cutting mill (Retsch GmbH, Haan, Germany) to pass through a 1-mm screen. Then 1003 ± 38 mg of DM substrate were weighed into serum bottles flushed with CO2, and 60 mL of the previously prepared buffered rumen fluid were added. All bottles were placed in a water bath and gently and continuously agitated at 39 ◦C during an incubation period of 48 h.

These procedures were repeated for six runs in total and all samples were incubated, with three replicates of each sample. All runs included triplicate bottles with blanks (i.e., bottles with 60 mL of buffered rumen fluid with no sample or treatment in), and samples were randomly allocated to the in vitro incubation bottles and never incubated in the same bottle in more than one run.

#### *2.3. In Vitro Gas Production Measurements and Sampling*

Gas production was measured using a fully automated system (Gas Production Recorder, GPR-2, Version 1.0 2015, Wageningen UR), with readings made every 12 min and corrected to the normal air pressure (101.3 kPa) [16].

Measurement of CH4 in vitro was performed according to Ramin and Huhtanen [14] on gas samples withdrawn during the incubation period (0.2 mL) from each bottle at 2, 4, 8, 24, 32, and 48 h. Concentration of CH4 was determined with a Varian Star 3400 CX gas chromatograph (Varian Analytical Instruments, Walnut Creek, CA, USA) equipped with a thermal conductivity detector.

Liquid samples of 0.6 mL were collected from the bottles at 8, 24, and 48 h of incubation and immediately stored at −20 ◦C until analysis of volatile fatty acids (VFA). Liquid samples for ammonia-nitrogen (NH3-N) analysis were taken at 8 and 24 h of incubation, and also stored at −20 ◦C before further analysis. Liquid samples from the replicate treatments between runs were pooled before NH3-N and VFA analysis.

After 48 h of incubation, all bottles were removed from the water bath and placed on ice to stop fermentation. The residue was used for in vitro determination of true organic matter digestibility (TOMD).

#### *2.4. Chemical Analysis*

The concentrations of DM and OM in the individual dietary ingredients were quantified by AOAC [17] method 930.15 and method 942.05, respectively. Concentrations of nitrogen were determined by Kjeldahl digestion of 1000 mg sample in 12 M sulfuric acid using Foss Tecator Kjeltabs Cu (Höganäs, Sweden) in a Block Digestion 28 system (SEAL Analytical Ltd., Mequon, WI, USA), followed by determination of total nitrogen by continuous flow analysis using an Auto Analyzer 3 (SEAL Analytical Ltd., Mequon, WI, USA). The samples were analyzed for neutral detergent fiber (NDF) using a heat-stable α-amylase [18] in an ANKOM200 Fiber Analyzer (Ankom Technology Corp., Macedon, NY, USA).

In vitro TOMD was determined for all samples in all runs by analyzing ash-free NDF concentrations in the residues using 07-11/5 Sefar Petex (Sefar AG, Heiden, Switzerland) in situ bags according to Krizsan et al. [19].

Individual VFA concentrations in rumen fluid samples were determined using a Waters Alliance 2795 UPLC system as described by Puhakka et al. [20], and NH3-N concentration according to the method provided by SEAL Analytical (Method no. G-102-93 multitest MT7) using AutoAnalyzer 3.

Bromoform concentration in AT was analyzed according to Roque et al. [21] using an Agilent 7890B GC applied to Agilent 7000C triple quad Mass Spectrometer equipped with a ZB-5ms column (Agilent Technologies, Inc. Santa Clara, CA, USA).

#### *2.5. Calculations*

Mean blank gas production within run was subtracted from sample gas production. In vivo predicted CH4 production was calculated as described by Ramin and Huhtanen [14] as:

CH4 = 265 × CH4 concentration + total gas production × CH4 concentration × 0.55

where total gas production is in mL/g sample, 265 is the total headspace volume (mL), and 0.55 is the ratio of CH4 emissions in the outflow gas from the in vitro system. A mean retention time of 50 h (20 h in the first compartment and 30 h in the second compartment), corresponding to the maintenance level of feed intake, was used in model simulations.

Total VFA (TVFA) production was calculated as: the molar proportion of individual VFA were calculated related to TVFA.

> TVFA (mmol) = ( individual VFA concentration − mean of blank VFA) × 0.06 (amount of buffered rumen fluid)

The molar proportion of individual VFA were calculated related to TVFA. The in vitro TOMD was calculated as:

$$\text{TOTD} \left(\text{g/kg}\right) = \frac{\text{incubated OM} \left(\text{g}\right) - \text{NDF residue corrected for ash andblank} \left(\text{g}\right)}{1000 \times \text{incubated OM} \left(\text{g}\right)}$$

#### *2.6. Statistical Analysis*

Data on in vivo predicted CH4 production and in vitro TOMD from Experiment 1 were analyzed using the MIXED procedure in SAS (SAS Institute Inc., Cary, NC, version 9.4), by a model correcting for random effect of bottle and fixed effect of run and treatment:

$$\mathbf{Y}\_{\rm ijk} = \mu + \mathbf{T}\_{\rm i} + \mathbf{R}\_{\rm j} + \mathbf{B}\_{\rm k} + \mathbf{e}\_{\rm ijk}$$

where Yijk is dependent variable ijk, μ is overall mean, Ti is treatment i, Rj is run j, Bk is bottle k, and eijk ~ N(0,σ<sup>2</sup> *<sup>e</sup>*) is the random residual error. Orthogonal contrasts were included for evaluation of control diet vs. treatment and of linear responses to level of treatment.

Data on measured VFA and NH3-N concentrations from Experiment 1 were evaluated in a repeated measurements model using the Toeplitz function in the MIXED procedure in SAS (SAS Institute Inc., Cary, NC, USA, version 9.4) (level within treatment was used as subject). The model accounted for effects of treatment and time, and interactions between treatment and time:

$$\mathbf{y}\_{\mathrm{i}\mathbf{j}} = \boldsymbol{\mu} + \mathbf{T}\_{\mathrm{i}} + \mathbf{A}\_{\mathrm{j}} + (\mathbf{TA})\_{\mathrm{i}\mathbf{j}} + \mathbf{e}\_{\mathrm{i}\mathbf{j}}$$

where yij is the dependent variable ij, μ is overall mean, Ti is treatment i, Aj is time j, (TA)ij is interaction between treatment i and time j, and eij ~ N(0,σ<sup>2</sup> *<sup>e</sup>*) is the random residual error.

Data on predicted in vivo CH4 production, in vitro TOMD, total VFA (TVFA), and molar proportions of individual VFA and NH3-N from experiment 2 were subjected to linear and quadratic regression analysis using the REG procedure in SAS (SAS Institute Inc., Cary, NC, USA, version 9.4). Best fit was judged from lowest root mean square error and highest adjusted R2.

Effects were considered statistically significant at *p*-value ≤ 0.05.

#### **3. Results**

The chemical composition of the control diet and the potential CH4 reducing diets is shown in Table 2. The AT bromoform concentration was 6.84 mg/g DM.



NDF = neutral detergent fibre.

#### *3.1. Experiment 1*

Predicted in vivo CH4 production derived from analysis of 48 h gas in in vitro incubation of the control diet was 38.7 mL/g DM, in vitro TOMD was 867 g/kg, TVFA was 3.62 mmol, and molar proportion of acetate, butyrate, and propionate was 583, 125, and 237 mmol/mol, respectively. In comparison with the control diet the chemical compounds 2-NE, nitrate, propynoic acid, p-coumaric acid, bromoform, and the plant compound AT, decreased (*p* ≤ 0.01) in vivo CH4 predicted production (Table 3). Addition of 2-NE, bromoform, and AT gave the strongest inhibition (*p* < 0.01) of predicted in vivo CH4 production among all experimental treatments (97%, 95%, and 99% reduction in the value for the control diet). The reduction in predicted in vivo CH4 production achieved by the other compounds ranged between 38% and 64% of the value for the control diet. Surprisingly, none of the potential CH4 reducing diets or lactic acid and acetic acid addition affected CH4 production in this study (*p* ≥ 0.20). In vitro TOMD was negatively affected by the chemical compounds p-coumaric acid and bromoform (*p* < 0.01), while rapeseed oil inclusion in the diet increased in vitro TOMD compared with the control diet (*p* = 0.04). Propynoic acid and bromoform decreased (*p* ≤ 0.01) TVFA

compared with the control diet. Several of the treatments altered the molar proportions of individual VFA. Acetate decreased (*p* ≤ 0.03) on adding 2-NE, propynoic acid, p-coumaric acid, bromoform, AT, or lactic acid to the control diet. For all those treatments except p-coumaric acid and bromoform, there was a concomitant increase (*p* ≤ 0.05) in molar proportions of propionic and butyric acid compared with the control diet. Results of nitrate vs. zero nitrate treatment were: TVFA 2.91 vs. 3.01 mol, acetate 597 vs. 604 mmol/mol propionate 250 vs. 227 mmol/mol and butyrate 87 vs. 123 mmol/mol.

The molar proportion of isobutyrate, isovalerate and valerate, and NH3-N for the control diet and experimental treatments are given in, Table 4. The molar proportions of the branched-chain volatile fatty acids (BCVFA) were altered by many of the CH4 mitigating strategies tested. Compared to the control diet, isobutyrate increased (*p* ≤ 0.01) for p-coumaric acid treatment, while for bromoform treatment the molar proportion decreased (*p* ≤ 0.01). The treatments, 2-NE, propynoic, p-coumaric, ferulic acid, AT, lactic acid, and lactic acid + acetic acid, increased (*p* ≤ 0.04), and bromoform decreased (*p* ≤ 0.01) the molar proportion of isovalerate compared to the control diet. Propynoic acid decreased (*p* ≤ 0.05) while bromoform and AT increased (*p* ≤ 0.05) molar proportion of valerate. Results of nitrate vs. zero nitrate treatment were: isobutyrate 6.63 vs. 9.96 mmol/mol, isovalerate 4.01 vs. 4.94 mmol/mol, valerate 16.4 vs. 16.3 mmol/mol, and NH3-N concentration 436 vs. 557 mg/L.

Tests for linear effects between the two levels according to Table 1 and the control diet (0 here) revealed no significant effect on in vitro TOMD (*p* = 0.148) for all treatments (data not presented). However, there was a significant linear decrease (*p* < 0.01) in predicted in vivo CH4 production for propynoic acid (24 and 0 mL/g DM) and p-coumaric acid (27.1 and 19.8 mL/g DM) when the inclusion level was increased.


**Table 3.** Effect of experimental treatments on predicted in vivo CH4 production (mL/g DM), in vitro true organic matter digestibility (TOMD, g/kg), total volatilefatty acid production (TVFA, mmol), and molar proportions of acetate, propionate, and butyrate (mmol/mol of TVFA) measured in 48 h gas from the in

dietmadebyaddingureaandCaCO3tothecontroldietaccordingtoTable1;numericaldifferencesofTVFAandmolarproportionsofvolatilefattyacidsaregiveninthetext.

#### *Animals* **2019** , *9*, 1120

 vitro


**Table 4.** Effect of experimental treatments on molar proportions of isobutyrate, isovalerate, and valerate (mmol/mol of TVFA), and ammonia concentration (NH3-N,

NA = not analyzed; SEM = standard error mean. 1 Orthogonal contrasts of control diet vs. treatment of the different in vitro traits. 2 Nitrate treatment was compared to the zero nitratediet made by adding urea and CaCO3 to the control diet according to Table 1; numerical differences molar proportions of branched-chain volatile fatty acids and NH3-N are given in thetext.

#### *Animals* **2019** , *9*, 1120

#### *3.2. Experiment 2*

Predicted in vivo CH4 production decreased curvilinearly (*p* < 0.01) with increased levels of both 2-NE (Figure 1).

**Figure 1.** Predicted in vivo methane production based on analysis of 48 h gas from in vitro incubation of a control diet (545:363:92 g/kg of grass silage:barley:rapeseed meal) treated with different levels (three replicates per level) of (**A**) 2-nitroethanol and (**B**) *Asparagopsis taxiformis* in experiment 2.

The TVFA content decreased linearly (*p* < 0.01) from 5.35 to 3.00 mmol at 48 h for 2-NE and from 4.71 to 4.33 mmol at 24 h for AT for the lower to higher level of supplementation (Figure 2). The TVFA content for 2-NE at 8 h (*p* < 0.01; adj R2 = 0.38; RSME = 0.22 mmol) and 24 h (*p* < 0.01; adj R2 = 0.55; RSME = 0.25 mmol) showed curvilinear responses, while for AT the curvilinear pattern was verified at 8h(*p* < 0.01; R<sup>2</sup> = 0.55; RSME = 0.25 mmol) and 48 h (*p* = 0.01; adj R<sup>2</sup> = 0.33; RSME = 0.33 mmol).

**Figure 2.** Total volatile fatty acid (TVFA) content in fluid samples taken at different time points during 48 h in vitro incubation of a control diet (545:363:92 g/kg of grass silage:barley:rapeseed meal) treated with different levels (three replicates per level) of (**A**) 2-nitroethanol and (**B**) *Asparagopsis taxiformis* in experiment 2.

Molar proportion of acetate decreased, while propionate and butyrate proportions increased curvilinearly (*p* < 0.01), at all-time points studied for 2-NE and AT (Figure 3). There were no statistical difference (*p* > 0.05) between the coefficients generated for the equations of the different time points. The best fit equations of molar proportions of VFA were generated for both 2-NE and AT from different sampling time points (Figure 3), but the equations generated were not statistically different (*p* > 0.05) from the other sampling time points (results not presented).

There were no linear or curvilinear relationships between TOMD and level of supplementation for 2-NE (*p* = 0.152) or AT (*p* = 0.142) (results not presented).

The equations of the molar proportions of BCVFA (isobutyrate, isovalerate, and valerate) were statistically different (*p* < 0.05) between the different sampling time points. The best fit equations are presented in Figure 4.

**Figure 3.** Molar proportions of acetate (Ace), propionate (Prop), and butyrate (But) in fluid samples gas samples taken at different time points during 48 h in vitro incubation of a control diet (545:363:92 g/kg of grass silage:barley:rapeseed meal) treated with different levels (three replicates per level) of (**A**) 2-nitroethanol and (**B**) *Asparagopsis taxiformis* in experiment 2.

**Figure 4.** Molar proportions of isobutyrate (Isobut), isovalerate (Isoval), and valerate (Val) in fluid taken at different time points during 48 h in vitro incubation of a control diet (545:363:92 g/kg of grass silage:barley:rapeseed meal) treated with different levels (three replicates per level) of (**A**) 2-nitroethanol and (**B**) *Asparagopsis taxiformis* in experiment 2.

The NH3-N concentration responses decreased linearly (*p* < 0.05) for both 2-NE and AT at both 8 and 24 h, and the best fit equations are presented in Figure 5.

**Figure 5.** Ammonia concentration (NH3-N) in fluid samples taken at different time points during 48 h in vitro incubation of a control diet (545:363:92 g/kg of grass silage:barley:rapeseed meal) treated with different levels (three replicates per level) of (**A**) 2-nitroethanol and (**B**) *Asparagopsis taxiformis* in experiment 2.

#### **4. Discussion**

The CP concentration of the potential CH4 reducing diets varied between 140 and 181 g/kg DM, and reflected the characteristics of the dietary ingredient studied in each diets. Regarding that Peptone™ was included in the buffered rumen fluid, none of the diets supplied an insufficient amount of CP in terms of CP available for rumen microbial growth in comparison with in vivo requirements [22].

Ruminants are valuable food producers world-wide, since they are able to utilize fibrous non-human-edible resources (forages and pasture) through microbial fermentation of feed in the rumen. Recent data indicate that domesticated ruminants are not the major contributor to anthropogenic CH4 emissions. The fermentation of feed and decomposition of manure are the foremost sources of GHG emissions caused by domesticated ruminants [23]. Estimates suggest that livestock are responsible for around 9% and 37% of anthropogenic CO2 and CH4 emissions, respectively [24]. Long-term strategies to improve feed efficiency through targeted breeding [25] and improved longevity or lifetime productivity [26] could reduce CH4 emissions from dairy cows. According to Knapp et al. [9], nutrition and feeding approaches may be able to reduce CH4 emissions per unit of energy-corrected milk by 2.5–15%, while reductions of 15–30% can be achieved by combined genetic and management approaches.

#### *4.1. In Vitro Measurements of CH4 Production in Ruminants*

This in vitro study evaluated a wide variety of dietary CH4 inhibitors, which would not have been feasible in an in vivo study. A main advantage of the in vitro gas production system in measuring CH4 emissions is that it provides a large number of data points, allowing accurate estimates of CH4 emissions. However, it is a batch culture approach and has some limitations compared with in vivo studies (e.g., no absorption of VFA over time). The in vitro method used here was developed by Ramin and Huhtanen [14] to overcome this problem and involves a modeling approach based on data obtained from the gas in vitro system. They assumed a gross energy concentration of 18.5 MJ/kg DM, while the predicted proportion of CH4 energy for a sample size of around 1000 mg was calculated to be 0.061. This value is close to observed in vivo values at production levels of intake in dairy cows [27]. Recently, Danielsson et al. [28] evaluated the in vitro technique developed by Ramin and Huhtanen [16] using data (diets) from in vivo studies using a respiration chamber to measure CH4 emissions. The results showed a high correlation (R2 = 0.96) between observed (chamber) data and predicted in vivo CH4 values, confirming that the in vitro system is a useful tool for screening diets and evaluating feed additives.

#### *4.2. Dietary Strategies to Decrease CH4 Production from Ruminants*

In this study, we screened many different dietary strategies with known potential to mitigate CH4 production from ruminants and also a few new potential inhibitors. It is known that improved forage quality, feeding balanced diets to ensure efficient utilization of nutrients, and optimized microbial protein synthesis in the rumen can decrease CH4 production in relation to animal productivity [29]. With respect to improved forage quality, the effects on CH4 production reported in the literature are contradictory. Enteric CH4 production increases with more digestible substrate available for rumen microbes, but overall emissions of CH4 in lactating dairy cows can decrease per kg increase in digestible OM [8]. The mechanism behind this effect is likely that better forage quality improves intake, and thereby increases passage rate. Increased passage rate (i.e., decreased feed retention) and larger animals (i.e., greater body mass) have been associated with reduced CH4 emissions in sheep [30–32].

Contrasted to our results, in measurements in vivo, rapeseed oil added at 50g/kg DM to a grass silage-based diet reduced ruminal CH4 emissions from lactating cows by 22% [33], with the reduction observed being entirely explained by decreases in DM intake and the dilution effect on fermentable OM. Use of dried distiller's grain to replace soybean meal in diets based on grass silage decreased CH4 production in an in vitro study by Franco et al. [34]. The effect was explained by a shift in the ruminal fermentation pattern to decreased acetate and butyrate production and increased propionate production. A similar shift in ruminal fermentation pattern was observed when rapeseed meal replaced soybean meal in vitro in that study [35]. In the present study, dried distiller's grain replaced rapeseed meal in the control diet and the suggested similarities in ruminal fermentation pattern of these protein supplements would explain the lack of effect on predicted in vivo CH4 production. In contrast to our results, Fant et al. [35] observed a significant reduction in predicted in vivo CH4 production of 2.1 mL/g DM when using oats instead of barley as the concentrate carbohydrate source. Inclusion of maize silage is also reported to promote propionate fermentation in the rumen, and thereby decrease CH4 production in dairy cows [36–38]. However, we did not observe this effect with inclusion of maize silage in the diets. Greater molar proportions of acetate and lower proportions of propionate in VFAs when replacing grass with red clover have been reported both in vivo [39] and in vitro [40], suggesting that CH4 production potential is greater when ruminants are fed red clover. Maize silage and red clover diets were only numerically lower respectively higher in in vivo predicted CH4 production compared with the control diet in this study. On the other hand, grasses are generally more likely to accumulate nitrates than legumes, and nitrate inhibits enteric CH4 production by replacing reduction of CO2 to CH4 as a major sink for disposal of H2 in the rumen [41]. Interactions between ruminant physiological responses and diet quality affecting CH4 production might explain the lack of impact on in vivo predicted CH4 production by the potential CH4-reducing diets screened in vitro in this study.

The chemical inhibitors 2-NE and bromoform, and the plant-derived inhibitor AT, gave a very large reduction in predicted in vivo CH4 production in this study. The bromoform concentrations of 1.5 and 3.0 mg/kg DM used in this study were representative of concentrations occurring naturally in AT [42]. Vucko et al. [43] analyzed bromoform concentrations in AT biomass subjected to a wide variety of post-harvesting processes and found a maximum concentration of 4.4 mg/g DM for unrinsed, frozen, and freeze-dried AT. Those authors suggest a bromoform threshold of 1.0 mg/g DM in AT for 100% inhibition of CH4 production in vitro, which corresponds with our results and the levels used in this study.

Machado et al. [44] tested different dosages of AT in vitro and found that production of CH4 was decreased by 84.7% at an inclusion level of 1% (OM basis), while at AT doses greater than 2% (OM basis), CH4 production was decreased by more than 99% compared with the control treatment. In the present study, in vivo predicted CH4 production was inhibited almost completely by AT already at a level of 0.5% on an OM basis. Li et al. [45] added AT to diets fed to sheep and observed a reduction of CH4 production at inclusion levels exceeding 1% of OM intake, but with altered rumen fermentation at all inclusion rates, i.e., at inclusions ≥0.5% of OM intake. On the other hand, in a short-term in vivo experiment by Stefenoni et al. [46], inclusion of AT at 0.5% of DM intake decreased CH4 emission in lactating dairy cows by 80%, with no negative effects on DM intake and milk yield (rumen fermentation parameters were not measured).

An in vitro study by Zhang and Yang [47] indicated high potential of 2-NE to mitigate CH4 production, as also found in the present study. However, they observed a negative effect on in vitro digestibility already at their lowest dose of 5 m*M*, which was not observed in this study. Use of 2-NE in an in vivo trial would not be realistic, considering that the concentration we used in vitro would equate to a daily dose of 0.9 L of 2-NE for a dairy cow with a rumen volume of 200 L.

Except for molar proportion of valerate with increased AT supplementation, all of the BCVFA decreased with increased supplementation in the dose response experiment. The BCVFA are mainly a consequence of the degradation of the amino acids valine, isoleucine, leucine and proline and are used for the biosynthesis of those amino acids and higher branched chain volatile fatty acids. The BCVFA are specific nutrients for the ruminal cellulolytic bacteria, and are believed to have a general positive influence on microbial fermentation [48].

Nitrate, propynoic acid, and p-coumaric acid had much lower inhibitory effects on predicted in vivo CH4 production. Nitrate is reported to be an effective CH4 production mitigating dietary component [49,50]. For example, a 24.8% reduction in CH4 production by lactating cows receiving nitrate at 21.1 g NO3 −/kg of DM was observed by Olijhoek et al. [51]. The dose of nitrate that can

be toxic to ruminants' ranges between 198 and 998 mg/kg live weight and is dependent on diet, administration, and consumption [52]. However, the negative effects of nitrate can be reduced through gradual adaptation of animals to consumption of this nitrogen source, which could contribute to reducing CH4 emissions. The CH4 inhibitory effect of propynoic acid in this study was lower than that observed by Zhou et al. [53] at a comparable inclusion rate (75.7% reduction compared with a control diet with no inhibitor added). Also the lower amount of TVFA compared with the control diet indicated that propynoic acid can potentially affect digestibility. The inhibitory effect of p-coumaric acid on CH4 production by ruminants has not been studied previously and there are no in vitro results with which to compare, but the treatment decreased the dietary TOMD. Lactate in the rumen are metabolized to propionate, which could hypothetical induce changes in ruminal fermentation pattern providing an alternative hydrogen sink to reduce methanogenesis. Likely, the lactic acid preservation has to be more extensive than the levels suggested in this study to have an effect on CH4 production in dairy cows.

#### **5. Conclusions**

This study confirmed that natural bioactives produced by the red seaweed *Asparagopsis taxiformis* can act as a strong natural inhibitor of CH4 production in domesticated ruminants. Use of CH4 inhibitors with high mitigation potential at a reasonable dietary supplementation level could be an important and effective strategy to mitigate CH4 emissions by ruminants. However, *Asparagopsis taxiformis* needs to be further evaluated in vivo to ensure it has no negative effects on animal health, productivity, or product quality. It is also important to establish the long-term CH4 mitigation effect of using this inhibitor.

**Author Contributions:** S.J.K.: conceived and supervised the study and acquired funding. S.J.K., J.C.C.: inputs to data analysis. S.J.K., M.R.: methodology. J.C.C., S.J.K., M.R.: conducted the experiment and wrote the full paper.

**Funding:** This research was funded by FACCE ERA-GAS and FORMAS.

**Acknowledgments:** The authors would like to extend their sincere appreciation to Ann-Sofi Hahlin, Azam Jafari, and Amanda Poppi for their support in laboratory work, and to Carl Tryggers Foundation for providing a postdoc scholarship for J.C.C.

**Conflicts of Interest:** There is no conflict of interest relevant to this publication.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
