Microbial inoculation of a grass-dominant pasture improves rumen function in vitro

Abstract

Introduction:

Interventions applied to pastures that modify nutritional attributes of the pasture may increase ruminant productivity and reduce the environmental footprint of production. We evaluated bovine in vitro fermentation responses to a microbial inoculant applied to a mixed-sward pasture, along with nutritional attributes of the treated pasture.

Methods:

A 4.2 ha heterogeneous pasture sward was randomly assigned to either a single intervention (inoculant) area or a control (no inoculant) area. Graze2Rumen^® (Agrovive Biologicals) inoculant was applied to the intervention area. The pastures were potentially non-equivalent across the field. A fermentation study with a single-factor design was then performed using two treatment groups. Inoculated and control pastures were chopped daily and fed to five fistulated Holstein cows per treatment group for 35 days. At weekly intervals for five weeks, a rumen sample was collected from each cow, and the following in vitro measures were determined: apparent digested organic matter (aOMD), hydrogen, carbon dioxide, methane, ammonia, total gas, apparent partitioning factor (aPF), microbial biomass production (aMBP), and pH after 24 hours of fermentation. Pasture samples were collected weekly for wet chemistry analysis from one week before (-7 days) to two weeks after the fermentation study (49 days).

Results and discussion:

Pasture samples obtained from potentially non-equivalent pastures, one inoculated and one untreated, showed that the inoculated pasture had a higher neutral detergent fiber content than the control for weeks 5, 6, and 8, and differences in phosphorus, potassium, and starch or resistant sugar content were associated with inoculation. An increase in aMBP of fermented rumen fluid was a key indicator of the potential for increased milk, wool, or meat production resulting from inoculation. This finding is consistent with approximately an 8.7% reduction in methane production and an 11.1% improvement in organic matter digestibility with inoculation. Similarly, a 15% improvement in the apparent partitioning factor with inoculation indicates the potential for increased feed intake if these results are reflected in vivo. These early results indicate the potential for aerial application of a microbial inoculation to affect pasture characteristics and possibly provide a scalable intervention to improve production and reduce the environmental footprint of ruminant production.

1 Introduction

This pilot study was designed to test the efficacy of a microbial inoculation program outlined in US Patent Number US20250312384A1 (Hagen, 2022) on established heterogeneous pasture, with the intent of evaluating effects on the nutritional attributes of the pasture, rumen fermentability, and methane production. The microbial Graze2Rumen^® program (Agrovive Biologicals, Tea, SD, USA) contains a mixture of microbes designed to promote endogenous colonization of beneficial microbial populations during crop development, resulting in plants with altered sap chemistry and microbial residues that influence rumen ecology (Hagen, 2022). The recommended timing of application for the product is during pasture establishment, allowing integration of the microbes into plant tissue and expression of microbial characteristics that may improve the nutritional value of inoculated plants.

Many ruminal intervention strategies demonstrate efficacy in high-concentrate or low neutral detergent fiber (NDF) diets (Weimer et al., 2010; Popova et al., 2022), potentially limiting their application in pasture-based production environments. Typically, temperate pastures are higher in protein, sugars, and NDF and lower in starch than total mixed rations (TMRs) fed to dairy cattle. Temperate pastures are considerably higher in protein and lower in starch than feedlot beef diets. This study evaluated whether an established heterogeneous pasture inoculated with the microbial intervention Graze2Rumen^® (Agrovive Biologicals) and harvested to mimic a grazing system would improve the nutritional attributes of the forage, as determined by chemical analysis, and its fermentability in an in vitro rumen fluid system. It was hypothesized that microbial inoculation of the established heterogeneous pasture sward would have no effect on nutritional attributes or ruminal fermentation characteristics, including methane production.

2 Materials and methods

Animal use was approved by the Fermentrics Animal Care Committee at a facility registered under the OMAFRA Animals for Research Act (license 0191-01).

2.1 Pasture experimental design

A heterogeneous mixed pasture sward, 4.2 ha in area and located approximately 0.5 km from the cattle used in the fermentation study, was randomly assigned to either a single intervention (inoculant) area or a control (no inoculant) area, each 2.1 ha in size. The established sward consisted of timothy grass (Phleum pratense; C3), orchard grass (Dactylis glomerata; C3), brome (Bromus spp.; C3), ryegrass (Lolium perenne; C3), and fescue (Lolium arundinaceum; C3). The pastures were potentially non-equivalent across the field. The field, located in Ontario, Canada, had not been fertilized for at least 15 years and had not received chemical amendments during that period. The field had been oversown evenly with 70% timothy grass and 30% orchard grass five years prior to the study. Topography was similar between both areas and very flat, with natural drainage. There were no trees or other vegetation, aside from the mixed pasture, within or surrounding the field, and environmental conditions were assessed to be even across the field. The soil type was Renfrew clay throughout, with a pH of 6.8 and organic matter content of 3.6%. The intervention was applied to one of the two pasture areas to facilitate practical implementation of feeding 10 large cows each day for 35 days. Five pasture samples were collected from each designated control and inoculant area two days prior to commencement of the study (inoculation), at the midpoints of each plot and equidistant from the respective plot edges. Samples were dried in a forced-air oven at 62.5°C for 24 h and ground according to National Forage Testing Association (NFTA) guidelines. The 1 mm sample was analyzed using a Phoenix 5000 Top Loader near-infrared (NIR) (Blue Sun Scientific, Jessup, MD, USA) with a resolution of 2 nm and a spectral range of 1100–2500 nm at Fermentrics (McNab-Braeside, Ontario, Canada).

The inoculant, Graze2Rumen^® (Agrovive Biologicals), was applied at a rate of 0.42 L/ha, diluted in 4 L of carrier water, at a speed of 14.5 km/h using a 36 m boom set at a pressure of 6.89 kPa with flat-fan nozzles (Case IH Patriot 4440, Racine, WI, USA) to the intervention area on 12 June 2023. No surfactants or other additives were applied to either area. The composition of the inoculant is provided in US Patent Number US20250312384A1 (Hagen, 2022). However, Clostridium spp., Pseudomonas spp., Delftia spp., Bacillus spp., Prosthecobacter spp., Chryseobacterium spp., Brevundimonas kwangchunensis, Burkholderia spp., and Lactiplantibacillus plantarum (formerly Lactobacillus plantarum) are present at concentrations of 1.0 x 10⁵/mL for all species except Chryseobacterium spp. and Lactiplantibacillus plantarum, which are present at 1.0 x 10²/mL.

2.2 Pasture harvest and feed preparation

Pasture was harvested in sequential strips daily beginning 26 days after inoculation using a flail chopper (New Holland Agriculture, New Holland, PA, USA) to simulate the pasture available for intensive grazing. No strips were harvested more than once. The inoculant area was harvested first into a designated inoculant feed bin, followed by passes over an adjacent stubble area to minimize cross-contamination of the harvest of the subsequent control area into a designated control feed bin. Pasture was harvested within one hour prior to the morning feeding of the cattle in the fermentation study. The quantity of chopped inoculated grass required for the morning feeding was unloaded from the feed bin using a designated inoculant tractor (loader) and mixed immediately with a supplement (0.9 kg/cow per feed) using a Rissler TMR feeder (I.H Rissler Manufacturing, Mohnton, PA, USA). The TMR was dispensed into the respective individual feeders within one hour of the grass harvest. The TMR feeder was then completely emptied, and the process was repeated for the control grass. The remaining chopped grass from the control and inoculant harvests was heaped in separate shaded feed bunks and fed in the afternoon, approximately nine hours after harvest, following mixing with the supplement using the same procedure as the morning feeding. Unused grass was discarded following the afternoon feeding. The supplement nutrient composition was 14.5% crude protein, 3.2% crude fat, 13% crude fiber, 1.5% calcium, 0.5% phosphorous, 0.7% sodium, 0.25 mg/kg selenium, 15,000 IU/kg vitamin A, 2,000 IU/kg Vitamin D₃, and 80 IU/kg Vitamin E (New Life Mills, Hanover, Ontario, Canada). The supplement consisted of the following ingredients: wheat shorts, ground corn, soybean hulls, gluten feed, dried distillers grain, limestone, salt, magnesium oxide, liquid fat, dairy premix, vitamin E, vitamin A, dry krave (Adisseo, Alpharetta, GA, USA), manganese sulfate, copper sulfate, SelSaf 3,000 (selenium yeast; Phileo by Lesaffre, Marcq-en-Barœul, France), and vitamin D₃.

Pasture samples were collected weekly from one week before (-7 days) to two weeks after the fermentation study (49 days) and submitted to A & L Laboratories (London, Ontario, Canada) for wet chemistry analyses. The samples collected before and after the fermentation study were pooled from samples collected at grazing height from six locations within the control and inoculant areas, respectively. The post-fermentation study collections were obtained from unharvested areas of the field. During the fermentation study, six subsamples of harvested grass were collected and pooled per treatment group as the forage bins were unloaded into the TMR feeder on the morning prior to rumen fluid collection days (days 6, 12, 20, 27, and 34). The samples were dried in a forced-air oven at 62.5°C for 24 h.

The following AOAC methods were used by A & L Laboratories: dry matter by forced-air oven drying at 60°C for 12 h (934.01), protein (990.03), acid detergent fiber (ADF) and lignin (973.14), NDF (2002.4), ash (942.05), starch (996.11), and fat (920.39). Protein fractions, including soluble protein, acid detergent insoluble protein, and neutral detergent insoluble protein were, were determined according to Roe et al. (1990). Minerals were assayed using inductively coupled plasma atomic emission spectroscopy according to the U.S. Environmental Protection Agency (EPA) method 6010.

2.3 Fermentation study

At 26 days post-inoculation, a fermentation study was conducted using a single-factor design with two replicates per treatment (inoculant vs control).

Ten fistulated, non-lactating Holstein cows housed in a free-stall barn were divided into two groups (n = 5/group): control (no inoculant) or inoculant. The cattle were fed 5.75 kg of dry matter (DM) of their respective treatment diets twice daily at 09:30 and 17:00 using a rail headlock system for 35 days. Divisions between feed boxes prevented crossfeeding. The cattle were not released from the headlocks until the respective feeds had been consumed, resulting in no feed refusals. At weekly intervals for five weeks (days 7, 14, 21, 28, and 35), 1 L of rumen fluid/mat was collected from the anterior position of the mid-rumen of each cow via the fistula into prewarmed insulated thermos containers at 09:45 and used to measure fermentation metrics. The proximity of the research barn allowed rumen fluid/mat samples to be processed within 15 min of collection.

To provide substrate for the in vitro rumen fluid samples, a subsample of the TMR samples collected one day prior to rumen fluid collection (days 6, 12, 20, 27, and 34) and dried for wet chemistry analysis was ground to 6 mm.

In addition, 24 h prior to the collection of the rumen fluid samples in the morning, 80 mL of K-State buffer was aliquoted into 20 fermentation flasks (two replicates per treatment). The flasks were then bathed in carbon dioxide and preheated at 39.5°C for 16 h. The samples of rumen fluid/mat were strained through two layers of cheesecloth and quickly added to the preheated fermentation flasks using an 80:20 ratio of buffer to filtered rumen fluid. Substrate (400 mg per flask) from the respective treatment group was then added, and the lids were applied. The flasks were further bathed in carbon dioxide, returned to the heated water bath, connected to infrared gas sensors (Johnson et al., 2026) with a collection data rate of 100x/s, and left to ferment for 24 h.

Following fermentation, the samples were dried, weighed, and evaluated with an NDF assay (ANKOM NDF method 15), in which neither alpha amylase nor sodium sulfite was added. Upon completion of the assay, the samples were further dried, and the results were used to determine microbial biomass production (aMBP). The metrics apparent digested organic matter (aOMD), hydrogen, carbon dioxide, methane, ammonia, total gas, apparent partitioning factor (aPF), and pH were also measured. Apparent partitioning factor (aPF) is defined as the ratio of truly degraded substrate to gas volume produced (Blümmel et al., 1997). An increase in this value reflects the potential for increased dry matter intake of forages.

2.4 Statistical analysis

All analyses were completed using Stata (Version 18.0, StataCorp, TX, USA), and analyses were conducted blinded to treatment group assignment. Baseline nutritional components of pasture samples collected from the designated control and inoculant areas two days prior to study commencement (inoculation) were analyzed using a t-test (Stata ttest). Evaluation of wet chemistry-determined nutritional components of pasture samples collected from one week before (-7 days) to 2 weeks after the fermentation study (49 days) was conducted using Stata anova, with a model containing week, treatment, and their interaction for eight samples from each treatment group. Marginal means (Stata margins) were calculated and plotted (Stata marginsplot).

After initial evaluation of the in vitro fermentation data through visualization and assessment of normality, all models included the two-way interaction between treatment and day, with day treated as a repeated measure and replicate within cow as random effects using Stata mixed. Margins were calculated using Stata margins, and group differences were determined using pwcompare (groups). Statistical significance was defined as P < 0.05.

An estimate of additional dry matter intake potential was derived from Blümmel and Becker (1997) using the equation in their Table 2 to estimate dry matter intake (DMI) (g/kg liveweight/day) from the gas production associated with NDF fractions. An estimate of metabolizable energy (MJ/kg of ME) was based on day 35 data results and the methods described by Tagliapietra et al. (2011) in their Table 4.

3 Results

3.1 Pasture analysis

The baseline pre-inoculation samples differed in only one tested component (neutral detergent insoluble crude protein percentage), indicating consistency between the designated plot areas before study commencement (Table 1). Differences (P <0.05) associated with treatment were observed for dry matter, ash, starch, phosphorus, and potassium (Table 2). These samples were obtained from areas within two large plots in which the pastures are potentially non-equivalent. Week-by-treatment effects differed only for dry matter percentage. However, NDF was initially lower at weeks 1 and 2 for the inoculant treatment, similar at weeks 3, 4, and 7, and higher in the inoculant treatment at all other weeks; however, the overall treatment-by-week interaction was not significant (P = 0.112; Table 2; Figure 1A). Week effects were observed for dry matter, magnesium, and starch (P ≤0.05; Table 2).

Figure 1

3.2 In vitro fermentation study

The interaction between treatment and day was significant for aOMD, with the inoculant treatment considerably increasing aOMD on days 7 and 14 compared with the control treatment (Table 3; Figure 1B). The aOMD increased by 11.1% with inoculation relative to the control, and marked differences in responses across days were observed except on days 21 and 28 (P <0.001; Table 3).

Methane effects differed significantly by treatment, with 8.7% less methane produced by the inoculant treatment than by the control treatment. However, there was no interaction between treatment and day or effect of day (Table 3; Figure 1C). Hydrogen production did not differ by treatment, and no treatment-by-day interaction was observed (Table 3). Hydrogen production was lower on days 7 and 35 compared with the other days in the study. Carbon dioxide production did not differ overall by treatment (P = 0.233). However, the controls produced 16.6% more carbon dioxide on day 14 than the inoculant treatment, whereas on day 21 inoculant treatment produced 7.9% more carbon dioxide than the controls (P <0.001). Overall, carbon dioxide production was lower only on day 35 (P<0.001; Table 3). Total gas production was 4.5% lower in the inoculant group than in the control group (P = 0.002; Table 3), driven by a 16.3% decrease in carbon dioxide at day 14 (Figure 1D). Overall, total gas production was lowest on day 35 (Table 3; Figure 1D).

Apparent partitioning factor increased by 15% with inoculation and was influenced by day. Values were considerably higher on days 7 and 14 in the inoculant group than in the control group (Table 3; Figure 1E). Microbial biomass production was 10.2 mg/g greater (5.2%) in the inoculant treatment than in the control treatment, with higher aMBP values observed for the inoculant group at every time point except day 14 (Table 3; Figure 1F). Ammonia production was not influenced overall by treatment (Table 3), although the controls produced more ammonia on day 14 than the inoculant treatment, whereas the inoculant treatment produced more ammonia on day 28 than the controls (data not shown). There was no main effect of treatment on pH. However, groups differed at each time point, with higher pH values on days 7 and 28 in the inoculant treatment and higher pH values on days 14, 21, and 35 for the controls (data not shown).

4 Discussion

Limitations of this study include the possibility that variation in treatment responses was confounded by pasture area allocation, despite a random selection of area within a consistent soil type. Therefore, it is possible that the swards selected for fermentation differed because of factors unrelated to treatment that may have influenced nutritional value. However, baseline NIR results indicated consistency between areas prior to intervention. Study strengths include the use of multiple sites with heterogeneous, established mixed pasture across the two pasture areas. Although the heterogeneity of the pasture sward within each strip of pasture harvested was not quantified, the diversity of species created a greater challenge for the intervention, demonstrating the potential for wider application in the field if subsequent outcomes are positive. In addition, the replicated random allocation of rumen samples obtained from 10 cows to dynamic fermentation evaluation provided outcomes, including aOMD, aPF, and aMBP, which evaluate the potential of feed to provide estimates on animal production and methane production. The in vitro assessments used in this study provided high-resolution, real-time gas kinetics and carbohydrate degradation rates, enabling pool-specific modeling of substrate fermentation (France et al., 2000; Krizsan et al., 2012).

As the pasture samples were harvested over several weeks, it was not surprising that dry matter, NDF (P = 0.112), starch, and magnesium concentrations differed over time, reflecting expected changes in pasture swards in the summer period. It is likely that the influence of days on apparent organic matter digestibility reflects the maturity of the pasture harvested. As the pasture matured, the inoculated pasture had a higher NDF content than the control pasture (Figure 1A), following initially numerically lower NDF values. However, over the fermentation test period (weeks 3 to 4), NDF concentrations were similar (Figure 1A). Because protein and non-structural carbohydrate contents were similar between the control and inoculated pasture, the observed increase in aMBP (Figure 1F), particularly when pastures had higher NDF concentrations (Figure 1A), was surprising. However, this finding is consistent with the reduction in methane production and differences in apparent organic matter digestibility associated with inoculation. The increased yield of aMBP from the inoculated pasture indicates a highly positive response to intervention that, if matched by similar in vivo responses, would result in increased weight gains and/or milk production. Microbial biomass production from substrates is a key determinant of rumen efficiency. The production of carbon dioxide is strongly associated with short-chain fatty acid and methane production (Opatpatanakit et al., 1994; Blümmel et al., 1997); however, treatment differences were observed only for methane production. The difference in total gas production between treatments appears to be strongly influenced by a large peak in carbon dioxide in the controls on day 14. Total gas production is strongly associated with true substrate degradability (Blümmel et al., 1997). Consequently, while there were differences in fermentation outcomes, such as aMBP and aPF, between treatment groups, differences in the true substrate degradability were not generally evident throughout the study. In addition, differences in phosphorus, potassium, and starch or possibly resistant sugar content between the treatment groups indicate differences in substrate associated with these groups. The effects of treatment-by-day interaction observed for ammonia production reflect that ammonia production is influenced by the availability of substrate, which enhances the potential to incorporate nitrogen in the plant, whether in the form of protein, nitrate, nitrite, or other forms, into microbial protein.

Blümmel et al. (1997) proposed the concept of an apparent partitioning factor (aPF), defined as the ratio of truly degraded substrate to gas volume produced. Forages with a high aPF, meaning low gas production per unit of truly degraded substrate in vitro, were associated with higher DMI across 55 forage types in vivo (Blümmel et al., 1997; Blümmel and Becker, 1997). Later in the season, NDF increased in the pasture, and effects on aPF were not evident (Figure 1E). If the aPF benefits observed with inoculation were reflected with in vivo responses, feed intake for livestock fed inoculated pasture would increase, and the metabolizable energy content of the inoculated pasture would be higher than that of the control pasture (Blümmel et al., 1997; Tagliapietra et al., 2011). These differences in energy, when combined with the increase in aMBP production, could translate into substantial production differences for ruminant livestock if the in vitro responses are reflected in in vivo production.

As inoculation was effective in successive fermentation weeks, despite the intentional heterogeneity of the mixed pasture sward and expected change in nutritional value over summer, the positive results appear more likely attributable to the inoculant intervention than to pasture area effects. Inoculation of an established pasture was effective despite recommendations that inoculation occur during pasture establishment, representing an important finding for producers.

5 Conclusions

These preliminary findings suggest that intervention with the multi-strain microbial inoculant may have altered the chemical composition of pasture samples obtained from potentially non-equivalent pastures in which one pasture was inoculated and the other was untreated. The inoculant also appeared to improve fermentation efficiency, possibly through an enhancement of the microbiome. The potential to deliver these differences through application of a microbial intervention at planting or through aerial foliar application, with subsequent improvements in ruminal fermentation observed some months later, suggests the potential for a scalable intervention that could increase productivity and reduce methane production. Overall, these preliminary findings suggest that there is a pattern of response to inoculation that warrants further study.

References

• 1

  BlümmelM.BeckerK. (1997). The degradability characteristics of fifty-four roughages and roughage neutral-detergent fibres as described by in vitro gas production and their relationship to voluntary feed intake. Brit. J. Nutr.77, 757–768. doi: 10.1079/BJN19970073. PMID:

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 2

  BlümmelM.SteingaßH.BeckerK. (1997). The relationship between in vitro gas production, in vitro microbial biomass yield and 15N incorporation and its implications for the prediction of voluntary feed intake of roughages. Brit. J. Nutr.77, 911–921. doi: 10.1079/BJN19970089. PMID:

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 3

  FranceJ.DijkstraJ.DhanoaM. S.LopezS.BanninkA. (2000). Estimating the extent of degradation of ruminant feeds from a description of their gas production profiles observed in vitro: derivation of models and other mathematical considerations. Brit. J. Nutr.83, 143–150. doi: 10.1017/S0007114500000180. PMID:

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 4

  HagenT. (2022). Microbial compositions and methods for reducing methane emissions. US Patent US20250312384A1. Raison LLP, assignee.

  • Google Scholar
  • View reference in article

• 5

  JohnsonJ.JohnsonR.RakhshanfarM. (2026). dentifying raw materials and gas production data. US Patent US-2026-0037698-A1. Fermentrics Technologies Inc, assignee.

  • Google Scholar
  • View reference in article

• 6

  KrizsanS. J.NyholmL.NousiainenJ.SüdekumK.-H.HuhtanenP. (2012). Comparison of in vitro and in situ methods in evaluation of forage digestibility in ruminants. J. Anim. Sci.90, 3162–3173. doi: 10.2527/jas.2011-4347. PMID:

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 7

  OpatpatanakitY.KellawayR.LeanI.AnnisonG.KirbyA. (1994). Microbial fermentation of cereal grains in vitro. Aust. J. Agric. Res.45, 1247–1263. doi: 10.1071/AR9941247. PMID:

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 8

  PopovaM.FerlayA.BougouinA.EugèneM.MartinC.MorgaviD. P. (2022). Associating changes in the bacterial community of rumen and faeces and milk fatty acid profiles in dairy cows fed high-starch or starch and oil-supplemented diets. J. Dairy Res.89, 249–258. doi: 10.1017/S0022029922000498. PMID:

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 9

  RoeM. B.SniffenC. J.ChaseL. E. (1990). “ Techniques for measuring protein fractions in feedstuffs,” in Cornell Nutrition. Conference for feed manufacturers ( Cornell University, Ithaca. NY).

  • Google Scholar

• 10

  TagliapietraF.CattaniM.HansenH. H.HindrichsenI. K.BailoniL.SchiavonS. (2011). Metabolizable energy content of feeds based on 24 or 48h in situ ndf digestibility and on in vitro 24h gas production methods. Anim. Feed Sci. Technol.170, 182–191. doi: 10.1016/j.anifeedsci.2011.09.008. PMID:

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 11

  WeimerP. J.StevensonD. M.MertensD. R. (2010). Shifts in bacterial community composition in the rumen of lactating dairy cows under milk fat-depressing conditions. J. Dairy Sci.93, 265–278. doi: 10.3168/jds.2009-2206. PMID:

  • Pubmed Abstract
  • CrossRef
  • Google Scholar