Soil and pasture health underlie improved beef nutrient density determined by untargeted metabolomics in Southern US grass finished beef systems

Abstract

As concerns regarding beef production on human and environmental health increase, interest in sustainable practices has grown. This study compared soil, plant, and meat samples from three Southern US grass-fed beef systems to a paired grain-fed beef system to assess soil health, forage phytochemical richness, and meat nutritional composition. Soil samples from pasturelands had 1.4 times higher organic matter and 1.7- to 3.0-fold higher levels of minerals like potassium, phosphorus, and calcium compared to paired feed croplands. Grass-fed beef contained 3.1-fold higher phytochemical antioxidants than grain-fed beef, resulting from a 118.2-fold higher phytochemical content in forage. Vitamins A and E in grass-fed beef were also 2.9- and 4.2-fold higher, respectively. Urate levels were 2.0-fold higher in grass-fed samples, while homocysteine and 4-hydroxynonenal glutathione, associated with reduced metabolic health, were elevated in grain-fed samples. The study provides evidence of the beneficial effects of grass-fed beef systems along the soil-plant-animal-human nutrition continuum.

Beef is a staple food in the American diet providing important nutrients including essential amino acids, vitamins and minerals, and other bioactive compounds important for human health¹. However, concerns about environmental impacts of beef production and potential effects of red meat consumption on human health are growing². In response, more producers are seeking to improve environmental stewardship of production and improve the nutritional qualities of beef³.

In the United States, the dominant approach to maximizing cattle growth and carcass quality is the use of corn (and other feed grains) in feedlot systems during the last several months of the animal’s life. While efficient, this practice is suggested to have several drawbacks on soil health and environmental sustainability in the long run⁴. In particular, current prevalent animal feed crop production methods often rely substantially on monocultures, synthetic fertilizers, and irrigation, which may deplete soil nutrients, degrade soil structure, and contaminate water sources⁵. Moreover, crop monocultures reduce plant diversity and biodiversity, which can negatively affect soil fertility and ecosystem function^6,7.

A growing number of ranchers in the US are moving towards finishing animals on pasture using adaptive grazing practices both for broader environmental, nutritional, and economic reasons^8,9. Rearing and finishing of cattle on pasture using adaptive and rotational grazing practices have been found to improve plant diversity^10,11, soil carbon^12,13, and general ecosystem function¹⁴; however, planned strategic approaches that are tailored to local environments appear key to achieve positive outcomes^15,16. Grass-finished beef has also garnered increasing consumer attention in recent years due to its perceived health benefits^17,18 and environmental considerations^19,20.

Animals grazing fresh forages tend to consume higher and more diverse polyphenol pools compared to animals consuming conserved forages and/or total mixed rations (TMR)²¹, as the polyphenol content in hay or silage is reduced by 30–60% reduced compared to fresh forage^22,23. Unlike other micronutrients, e.g., minerals and vitamins, less attention is currently paid to phytochemical abundance and related metabolites found in animal-sourced foods. However, enrichment of these compounds livestock diets can improve animal metabolic health²⁴, in addition to increasing the broader nutrient density of meat and milk²¹.

While grass-fed beef is recognized for its potential enhancement of omega-3 fatty acids, the capacity of grass-fed systems to enhance soil health and supply an increased quantity of health-promoting nutrients requires further exploration. Leheska et al.²⁵ previously studied intricate relationships within soil-plant-animal nutrition, suggesting a potential link between improved soil mineral content and the mineral composition of grass-fed meat. Furthermore, a recent review by van Vliet et al.²¹ documented that livestock raised on a diverse array of forages yields meat and milk concentrated in additional health-promoting phytochemicals. Additionally, long-standing research suggests that increasing plant diversity can lead to improved pasture productivity and soil health²⁶. This highlights the potential of pasture-based systems to concentrate additional, potential health-promoting compounds in meat, governed by grazing management strategies that may also improve soil health and ecosystem function. However, there is a need for more extensive research to establish potential linkages between metrics of soil, plant, animal, and human health.

Advancements in omics-based techniques have allowed for a deeper profiling of the nutritional composition of foods²⁷, which extends the field beyond macronutrients, vitamins, and minerals to a broader array of metabolites with potential health value. This project aimed to discern the potential differences in the metabolite profiles of grass-fed and grain-fed beef, and linking them soil health and the abundance of nutrients in respective feeding systems²⁸. Additionally, the study investigated health biomarkers in animals, providing insights into the potential impacts of varying nutritional statuses on their health and its potential relationship with meat nutritional composition.

Results

Physicochemical analysis of pastureland and cornfield soils

The soil physicochemical analysis revealed potentially important trends related to soil health in pastures and paired cornfields, as presented in Table 1. On average, soils from pasturelands exhibited significantly higher soil organic matter (SOM), electrical conductivity (EC), estimated nitrogen release (ENR), zinc, calcium, and potassium compared to soils from paired corn fields (all, p ≤ 0.05). Soil pH and iron concentration tended to be greater under pastures than corn fields (p ≤ 0.10). However, an opposite trend was observed for soil pH and CEC in the Blackland Prairie (BLP) and Southern Piedmont Highlands (SPH) soils, respectively. The pastureland of BLP was ~9% more acidic than its paired cornfield, and in the paired corn field of SPH, CEC was ~11% higher than its pastureland.

Table 1.

Soil health metrics of paired pasturelands and corn fields

	Southern Piedmont Highlands (SPH)		Southern Blue Ridge (SBR)		Blackland Prairie (BLP)		T-test
	Pasture (n = 2)	Corn (n = 2)	Pasture (n = 2)	Corn (n = 2)	Pasture (n = 2)	Corn (n = 2)	Pasture vs Corn (all samples)
SOM (g kg –1)	68 ± 0.6	52 ± 2.9	69 ± 7.3	49 ± 3	53 ± 0.1	34 ± 0.2	≤0.01
pH	6.5 ± 0.0	4.9 ± 0.0	7.3 ± 0.7	7.1 ± 0.0	6.5 ± 0.1	7.1 ± 0.1	0.07
CEC (meq 100 g–1)	8.3 ± 1.1	9.2 ± 8.5	37.6 ± 2.1	27.9 ± 3.2	8.2 ± 0.1	6.4 ± 0.2	0.12
EC (mmhos cm–1)	0.2 ± 0.1	0.2 ± 0.1	0.3 ± 0.1	0.2 ± 0.1	0.2 ± 0.1	0.1 ± 0.1	0.02
ENR (ppm)	109.0 ± 0.0	101.0 ± 1.5	109.5 ± 3.6	98.5 ± 2.2	102.0 ± 0.0	84.0 ± 0.0	≤0.01
Iron (mg kg–1)	170.5 ± 3.6	100.5 ± 23.4	91.5 ± 33.3	139.5 ± 17.7	210.5 ± 5.0	99.0 ± 2.9	0.10
Zinc (mg kg–1)	14.8 ± 1.7	2.4 ± 0.1	3.2 ± 1.6	2.7 ± 0.4	4.9 ± 1.5	2.7 ± 0.3	0.05
Calcium (mg kg–1)	1002.5 ± 123.8	230.5 ± 17.7	6672.5 ± 621.6	3863.0 ± 449.8	898.0 ± 17.0	919 ± 39.6	0.05
Phosphorus (mg kg–1)	54.5 ± 10.7	64.0 ± 28.3	30.0 ± 1.5	18.5 ± 0.8	173.0 ± 83.5	53.5 ± 14.9	0.11
Cobalt (mg kg–1)	0.1 ± 0.1	0.1 ± 0.1	0.3 ± 0.1	0.5 ± 0.1	0.2 ± 0.1	0.6 ± 0.1	0.03
Sulfur (mg kg–1)	14.0 ± 0.0	17.0 ± 1.5	12.0 ± 1.5	13.5 ± 0.8	13.5 ± 0.8	11.5 ± 0.8	0.23
Potassium (mg kg–1)	131.5 ± 40.4	52.5 ± 2.2	208.0 ± 41.1	189.0 ± 8.5	245.5 ± 33.3	103.0 ± 24.1	0.01

Values means ± standard daviation of three analytical replicates sampled from a composite sample of 4 subsamples.

SOM soil organic matter, CEC Cation exchange capacity, ENR estimated nitrogen release.

Nutritive value of pasture and total mixed rations

Routine nutritional analysis of pasture forages and feedlot TMR samples indicated numerical differences but did not reach statistical signficance (Supplemental Table 2). The TMR samples averaged 11.2% crude protein, while the forage samples averaged 14.6% crude protein. Likewise, Acid Detergent Fiber and Neutral Detergent Fiber levels were lower in the Feedlot TMR than in the forages (24.6 vs 12.1 for ADF, and 49.7 vs 20.6 for NDF, respectively). There were differences between farms for several major nutrients. In many cases, general indicators of forage quality including crude protein, ADF, NDF and mineral levels were highest for the SBR site, intermediate for the BLP site, and lowest at the SPH site. For example, Crude Protein was 20.6, 12.8 and 10.6%, and ADF was 24.6, 32.7, and 34.0% for SBR, BLP and SPH, respectively.

Metabolomics profiling of pasture vs total mixed rations

The untargeted metabolomics analysis of the pasture forages and TMR samples identified 846 metabolites. A total of 69% of metabolites were significantly different (p ≤ 0.05) between pasture and TMR samples (Table 2). Among the 581 significantly (p ≤ 0.05) different metabolites, 319 were higher in pasture, while 262 were more abundant in TMR. Additionally, among the 6% metabolites with 0.05 > p ≤ 0.10, 18 were higher in pasture and 35 were higher in TMR. As a result, clear separation (component 1, 62.8%) was achieved by Partial Least Squares Discriminant Analysis (PLS-DA) between pasture and TMR (Fig. 1a). Pasture forage samples displayed broader variation, indicating differences among the pastures of each farm, which is attributed to differences in plant species composition between grass-fed farms. As only one grain-fed operation was included, TMR samples showed less variation. The volcano plot (Fig. 1b) highlighted the top 20 metabolites that differentiate pasture and TMR samples, while the heatmap (Fig. 2) displays the top 50 metabolites selected by the PLS-DA VIP score. Key metabolites that appeared in the volcano plot are omega-3 galactolipids (e.g., 1-palmitoyl-2-linolenoyl-digalactosylglycerol 16:0/18:3, 1-linoleoyl-2-linolenoyl-galactosylglycerol 18:2/18:3, and 1,2-dilinolenoyl-digalactosylglycerol 18:3/18:3), phytochemicals (e.g., chlorogenate, coumarquinate, isorhamnetin 3-rutinoside, and luteolin 7-O-glucoside), dipeptides (e.g., pyroglutamylvaline) and glycerolipids (e.g., glycerol). Phytochemicals were also dominantly featured in the heatmap displaying elevated levels of various phenolics and polyphenols (e.g., coumarquinates, feruloylquniates, quercetin 3-glucoside, cryptochlorogenic acid, rutin, and genetic acid-5-glucoside) while higher concentrations of various dipeptides, including alanylhistidine, glycylisoleucine, leucylalanine, and alanylproline, were observed in TMR (Supplemental Table 3).

Table 2.

Untargeted metabolomics profiling comparisons of beef and feed samples

Wilcoxon rank-sum t-test	Feed	Meat
	Pasture TMR	Bf Grass Bf_Grain
Total biochemicals identified	846	745
Total metabolites, p ≤ 0.05	581 (69%)	281 (38%)
Biochemicals	319↑ │ 262˄	128↑ │ 153˄
Total metabolites, 0.05 > p ≤ 0.10	53 (6%)	63 (8%)
Biochemicals	18↑ │ 35˄	34↑ │ 29˄

The p values stand for t-test with unequal variance. (^↑) indicates higher in plant pasture and grass-fed beef (Bf-Grass) samples; (^˄) indicates higher in total mixed rations (TMR) and grain-fed beef (Bf-Grain) samples.

Fig. 1. Multi- and univariate analysis of metabolite profiles in pasture and total mixed ration samples.

a Partial Least Squares Discriminant Analysis and b Volcano plot of differential metabolites in both pasture (n = 13) and total mixed ration (TMR; n = 4) samples fed to grass-fed and grain-fed beef cattle, respectively. Names of the samples correspond to the location of the grass-fed ranches (Southern Piedmont Highlands [SPH], Southern Blue Ridge [SBR], Blackland Prairie [BLP]) and a Midwest Feedlot (MWF) in the USA.

Fig. 2. Heatmap of the top 50 metabolites detected in feed samples ranked based on PLS-DA VIP values.

The names below the heatmap correspond to individual samples: n = 13 pasture Samples and n = 4 total mixed ration (TMR) samples. Names of the samples correspond to the location of the grass-fed ranches (Southern Piedmont Highlands [SPH], Southern Blue Ridge [SBR], Blackland Prairie [BLP]) and Midwest Feedlot (MWF) in the USA. Supplemental Table 1 provides the full names and abbreviations of the compounds.

The ChemRich analysis revealed that 41 out of 45 identified metabolic pathways were significantly (p ≤ 0.01) different between the pasture and TMR samples (Table 3). Metabolic variation was dominated by phenolic and glutamate metabolism followed by aspartate and phospholipid (omega-3) metabolism. In particular, pathways clusters for omega-3 fatty acids (e.g., lyso-phospholipids: 1-linolenoyl-GPG [18:3], galactolipids: 1-palmitoyl-2-linolenoyl-digalactosylglycerol [16:0/18:3], and lyso-galactolipids: 2-linolenoyl-digalactosylglycerol [18:3]) were higher in pasture samples while omega-6 clusters {e.g., free fatty acid: linoleate [18:2n6]) were higher in TMR samples.

Table 3.

Differences in metabolite abundance across sub-biochemical pathways between pasture forage and total mixed rations (TMR) samples

Sub Pathway	Cluster size	t-test	FDR	Key compound	Altered metabolites	↑Pasture	↑TMR
Phenolic Metabolism	114	≤0.01	≤0.01	Chlorogenate	60	44	16
Glutamate family	54	≤0.01	≤0.01	Ornithine	37	5	32
Aspartate family (OAA derived)	45	≤0.01	≤0.01	Fructosyllysine	33	4	29
Phospholipids	39	0.000	≤0.01	1-palmitoyl-2-linoleoyl-GPG (16:0/18:2)	33	33	0
Dipeptide	28	≤0.01	≤0.01	Alanylhistidine	26	0	26
Aromatic amino acid metabolism (PEP derived)	38	≤0.01	≤0.01	N-succinyl-phenylalanine	22	3	19
Lyso-phospholipids	30	≤0.01	≤0.01	1-linolenoyl-GPG (18:3)	21	20	1
Pyrimidine metabolism	28	≤0.01	≤0.01	2’-o-methyluridine	20	4	16
Purine metabolism	28	≤0.01	≤0.01	Adenosine	20	7	13
Sucrose, glucose, fructose metabolism	27	≤0.01	≤0.01	Erythrulose	20	15	5
Branched Chain Amino Acids (pyruvate derived)	23	≤0.01	≤0.01	N-succinyl-leucine	16	2	14
Galactolipids	14	≤0.01	≤0.01	1-palmitoyl-2-linolenoyl-digalactosylglycerol (16:0/18:3)	14	14	0
Amino sugar and nucleotide sugar	21	≤0.01	≤0.01	Glucosamine-6-phosphate	13	8	5
Fatty acid, hydroxy	20	≤0.01	≤0.01	2-hydroxyoleate	13	2	11
Glutathione metabolism	25	≤0.01	≤0.01	5-oxoproline	12	2	10
Free fatty acid	17	≤0.01	≤0.01	Linoleate (18:2n6)	11	1	10
Fatty acid, Dicarboxylate	15	≤0.01	≤0.01	Dodecenedioate (C12:1-DC)	11	3	8
Serine family	17	≤0.01	≤0.01	N-acetylglycine	10	2	8
Lyso-galactolipids	10	≤0.01	≤0.01	2-;inolenoyl-digalactosylglycerol (18:3)	10	10	0
Glycerolipids - Diacyl	15	≤0.01	≤0.01	Palmitoyl-linolenoyl-glycerol (16:0/18:3) [2]	9	3	6
TCA cycle	9	≤0.01	≤0.01	Tricarballylate	7	4	3
Dipeptide Derivative	8	≤0.01	≤0.01	Pyroglutamylvaline	7	0	7
Nicotinate and nicotinamide metabolism	9	≤0.01	≤0.01	Nicotinamide ribonucleotide (NMN)	6	5	1
Phospholipid Metabolism	7	≤0.01	≤0.01	Glycerophosphoethanolamine	6	3	3
Sphingolipid	6	≤0.01	≤0.01	Sphingadienine	6	0	6
Amines and polyamines	11	≤0.01	≤0.01	N1,N10-dicoumaroylspermidine	5	0	5
Glycolysis	6	≤0.01	≤0.01	Lactate	5	2	3
Sulfolipids	6	≤0.01	≤0.01	SQDG (18:3/16:0)	5	5	0
Photorespiration	5	≤0.01	≤0.01	Tartronate (hydroxymalonate)	4	4	0
Ascorbate metabolism	5	≤0.01	≤0.01	Threonate	4	4	0
Vitamin B metabolism (B6 or B12)	5	≤0.01	≤0.01	Pyridoxate	4	1	3
Tocopherol metabolism	5	≤0.01	≤0.01	Alpha-tocopherol	4	2	2
Glycerolipids - Monoacyl	11	≤0.01	≤0.01	1-linolenoylglycerol (18:3)	3	3	0
Fatty acid ester	4	≤0.01	≤0.01	Oleoylcholine	3	0	3
Oxylipins	3	≤0.01	≤0.01	12,13-DiHOME	3	0	3
Riboflavin and FAD metabolism	3	≤0.01	≤0.01	Flavin mononucleotide (FMN)	3	2	1
Calvin cycle and pentose phosphate	3	≤0.01	≤0.01	Sedoheptulose-7-phosphate	3	3	0
Chlorophyll and heme metabolism	3	≤0.01	≤0.01	Pheophytin A	3	3	0
Auxin metabolism	4	≤0.01	≤0.01	Indoleacetate	2	0	2

Metabolomics profiling of grass-fed vs grain-fed beef

The PLS-DA plot of grass-fed (Bf_Grass) and grain-fed (Bf_Grain) beef samples show clear separation (component 1, 35%) among the two finishing systems (Fig. 3a). Within grass-fed beef samples, the BLP samples showed a separation compared to the other two farms (SPH and SPB).The t-test found that 38% of metabolites (281 out of 745) significantly differed (p ≤ 0.05) between the grass-fed beef (Bf_Grass) and grain-fed beef samples (Bf_Grain) (Table 2). Among the 281 significantly varied metabolites, 128 compounds were higher in grass-fed beef, while 153 were higher in grain-fed beef. Additionally, among 63 of metabolites with 0.05 > p ≤ 0.10, 34 were higher in grass-fed beef and 29 were higher in grain-fed beef. The volcano plot (Fig. 3b) highlights the top 20 metabolites that differentiate between the grass-fed and grain-fed samples, while the heatmap (Fig. 7) displays the top 50 metabolites selected by PLS-DA VIP score. Key metabolites that appeared in the volcano plot are phenolic metabolites such as benzoylcarnitne, homostachydrine, stachydrine, tryptophan betaine, N-methylpipecolate (N-MethylPip), and amino acid metabolites such as cadaverine and homocysteine. The heatmap (Fig. 7) further highlights amino acid metabolites (e.g., leucylalanine and tryptophylglycine), and phenolic metabolites (e.g., catechol sulfate) as dominant metabolites that differ between grass-fed and grain-fed beef.

Fig. 3. Multi- and univariate analysis of metabolite profiles in grass-fed and grain-fed beef samples.

a Partial Least Squares Discriminant Analysis (PLS-DA) and b Volcano plot of differential metabolites in grass-fed (n = 8) and grain-fed (n = 8) beef samples. Names of the samples correspond to the location of the grass-fed ranches (Southern Piedmont Highlands [SPH], Southern Blue Ridge [SBR], Blackland Prairie [BLP]) and Midwest Feedlot (MWF) in the USA.

The ChemRich analysis revealed that 41 out of 102 identified metabolic pathways differed (FDR ≤ 0.01) between the grass-fed and grain-fed beef samples (Table 4), further highlighting distinct phenotypes as a result of the finishing system. The metabolic variation altered in response to finishing mode, was dominant in dipeptides, which were found to be elevated in grain-fed beef; followed by phenolic metabolism and fatty acid (acyl carnitine, polyunsaturated) metabolism, which were predominantly elevated in the grass-fed beef samples (Supplemental Table 4).

Table 4.

Differences in metabolite abundance across sub-biochemical pathways in grass-fed (Bf_Grass) and grain-fed (Bf_Grain) beef samples

Sub Pathway	Cluster size	t-test	FDR	Key compound	Altered metabolites	↑Bf_Grass	↑Bf_Grain
Dipeptides	26	≤0.01	≤0.01	Leucylglycine	21	0	21
Phenolic metabolism	35	≤0.01	≤0.01	Benzoylcarnitine	15	11	4
Methionine, Cysteine, SAM and Taurine Metabolism	21	≤0.01	≤0.01	Homocysteine	9	2	7
Monoacylglycerol	16	≤0.01	≤0.01	1-dihom-linolenylgly (20:3)	9	9	0
Leucine, Isoleucine and Valine Metabolism	22	≤0.01	≤0.01	Isoleucine	9	0	9
Lysophospholipid	38	≤0.01	≤0.01	1-linolenoyl-gpc (18:3)	9	6	3
Phosphatidylcholine (PC)	34	≤0.01	≤0.01	1-linoleoyl-2-linolenoyl-GPC (18:2/18:3)	9	7	2
Acyl Carnitine, Polyunsaturated	10	≤0.01	≤0.01	Arachidonoylcarnitine (20:4)	8	8	0
Histidine Metabolism	18	≤0.01	≤0.01	Histidine methyl ester	8	4	4
Tyrosine Metabolism	12	≤0.01	≤0.01	3-methoxytyrosine	7	2	5
Acyl Carnitine, Monounsaturated	10	≤0.01	≤0.01	Undecenoylcarnitine(11:1)	6	3	3
Fatty Acid, Dicarboxylate	12	≤0.01	≤0.01	2-hydroxysebacate	6	2	4
Tryptophan Metabolism	11	≤0.01	≤0.01	Tryptophan	6	3	3
Urea cycle; Arginine and Proline Metabolism	20	≤0.01	≤0.01	N-delta-acetylornithine	6	2	4
Glycine, Serine and Threonine Metabolism	9	≤0.01	≤0.01	Threonine	5	1	4
Polyamine Metabolism	7	≤0.01	≤0.01	5-methylthioadenosine (MTA)	5	2	3
Long Chain Polyunsaturated Fatty Acid (n3 and n6)	15	≤0.01	≤0.01	Linolenate [alpha or gamma; (18:3n3 or 6)]	5	5	0
Glutamate Metabolism	11	≤0.01	≤0.01	Beta-citrylglutamate	5	4	1
Glycolysis, Gluconeogenesis, and Pyruvate Metabolism	10	≤0.01	≤0.01	3-phosphoglycerate	5	3	2
Phenylalanine Metabolism	5	≤0.01	≤0.01	Phenyllactate (PLA)	4	0	4
Long Chain Saturated, Acyl Carnitine	7	≤0.01	≤0.01	Palmitoylcarnitine (C16)	4	4	0
Fatty Acid, Branched	3	≤0.01	≤0.01	(12 or 13-methylmyristate (a15:0 or i15:0))	3	3	0
Glycerolipid Metabolism	3	≤0.01	≤0.01	Glycerophosphoglycerol	2	0	2
Modified Peptides	3	≤0.01	≤0.01	Pyroglutamylvaline	2	0	2

Relationship between phytochemicals in feed and their transfer to meat

The comparison of the metabolites in the two types of beef and their respective feed type provides an initial insight into the transfer of phytochemicals from feed to animal and their downstream metabolism (Fig. 4). A total of 11 out of 33 phytochemicals detected in meat samples were also found in the feed samples. In seven of these phytochemicals, their abundance in meat was directly related to feed. For example, ergothioneine was 11.4-fold higher in pasture compared to TMR feed, resulting in 2.8-fold higher levels of ergothioneine in grass-fed beef compared to grain-fed beef. On the other hand, catechol sulfate, 4-ethylphenol sulfate, dimethyl sulfone, 3-hydroxyhippurate, cinnamoylglycine were significantly higher in grass fed beef but not reported in the feed samples, as these compounds represent metabolized versions of upstream phenolics and polyphenols. Piperidine, an alkaloid metabolite, and 4-methylcatechol sulfate, a phenylsulfate, were significantly (all p ≤ 0.05) higher in grain fed beef, and likely represent metabolized versions of alkaloids and benzene-related compounds in the TMR. In conclusion, pasture samples had 118.2 times more phenolics than TMR, resulting in a 3.1-fold increase in phenolics and their metabolites in grass-fed beef compared to grain-fed beef.

Fig. 4. Conceptual model of relationship between phytochemicals found in pasture and total mixed ration (TMR) and the appearance of metabolized versions in grass-fed (Bf_Grass) and grain-fed (Bf_Grain) beef.

The results are presented as fold changes, comparing pasture to TMR and grass-fed (Bf_Grass) to grain-fed (Bf_Grain) beef. Red shaded cells indicate significantly higher (p ≤ 0.05) mean values in TMR or Bf_Grain, while green shaded cells indicate significantly higher (p ≤ 0.05) mean values in pasture or Bf_Grass. Light green shading indicates trends (p ≤ 0.1). The predicted pathway of phenolics from feed to ribeye is also illustrated, with green text indicating higher abundance in pasture compared to TMR, and yellow text indicating the reverse. The text within rectangles filled with green shows higher levels in Bf_Grass, while yellow indicates higher levels in Bf_Grain.

Lipid and energy metabolism in grass-fed and grain-fed beef

Fatty acid metabolites of meat samples revealed notable differences among the two finishing systems, with very long-chain polyunsaturated fatty acids (PUFAs), saturated fatty acids (SFAs), and long-chain acyl carnitines being amongst the commonly altered metabolite classes (Supplemental Table 4). Among PUFAs, eicosapentaenoate (EPA; 20:5n3), docosahexaenoate (DHA; 22:6n3), and linolenate [alpha or gamma; (18:3n3 or 6)] were respectively 3.4-, 1.8- and 5.4-fold higher in grass-fed beef (all p ≤ 0.05), whereas linoleate (18:2n6), an omega-6 fatty acid dominant in grains, was 0.82-fold lower in grass-fed beef (p = 0.189). Similar to long-chain chain PUFAs, long-chain SFAs were more abundant in grass-fed beef, with 2.7- and 1.5-fold higher levels of nonadecanoate (19:0) and arachidate (20:0) compared to grain-fed beef (all p ≤ 0.05). Likewise, long-chain polyunsaturated and saturated acyl carnitines were elevated in grass fed beef with 16.1-, 11.0-, and 11.8-fold higher levels of linolenoylcarnitine (C18:3, docosapentaenoylcarnitine (C22:5n3), arachidonoylcarnitine (C20:4) respectively, compared to grain-fed beef (all, p ≤ 0.001).

On average, tricarboxylic acid (TCA) cycle intermediates were upregulated by 1.3-fold in grass-fed beef compared to grain-fed beef; however, only alpha-ketoglutarate and malate reached statistical significance with 1.8- and 2.7-fold higher levels (all, p ≤ 0.05; Supplemental Table 4). Whereas glucose, ribose, and lactate were 1.3-, 1.4 and 1.1-fold higher in grain-fed beef (all p ≤ 0.05), respectively.

Endogenous oxidative stress markers and antioxidants levels in beef

The analysis of endogenous stress markers and antioxidant levels revealed higher levels of oxidative stress and lower antioxidant status in grain-fed animals finished in a feedlot compared to grass-fed animals finished on pasture (Fig. 5). Levels of homocysteine and 4-hydroxynonenal glutathione, two common oxidative stress markers, were 2.5-fold and 2.3-fold elevated in grain-fed beef compared to grass-fed beef, respectively (all, p ≤ 0.05). Conversely, levels of two common antioxidants, urate and oxidized glutathione, were 2.1-fold and 1.3-fold higher in grass-fed beef compared to grain-fed beef, respectively (all, p ≤ 0.05).

Fig. 5. Oxidative stress and antioxidant markers in grass-fed (Bf_Grass) and grain-fed (Bf_Grain) beef.

The bar graph shows the abundance status of oxidative stress markers and antioxidants levels across two types of finished cattle (Bf_Grass and Bf_Grain): a Homocysteine, b 4-HNE-glutathione, c Glutathione (oxidized), and d Urate. The groups were compared using a Wilcoxon rank sum t-test (p ≤ 0.05); (*) indicates p ≤ 0.05, and (**) indicates p ≤ 0.01.

Vitamins levels in the grass-fed vs grain-fed production system

Grass-fed beef contained 4.2-fold higher levels of alpha-tocopherol (Vitamin E), which was directly related to the 113.1-fold higher levels in forage samples (all, p ≤ 0.05) (Fig. 6). Additionally, 17.9-fold higher levels of β-carotenes in forages resulted in 4.9-fold, non-significantly, higher (p = 0.14) levels of β-carotene in grass-finished beef and 2.9-foldhigher levels of its metabolite retinol (Vitamin A) (p = 0.03) compared to grain-fed beef. Regarding B-vitamins, 4.0- and 1.4-fold, non-significantly, higher levels of thiamin (vitamin B₁; p = 0.38) and pantothenate (vitamin B5; p = 0.11) in TMR samples compared to forage samples led to 1.8- and 1.9-fold higher levels of thiamin and pantothenate in grain-fed beef compared to grass-fed beef, respectively (all, p ≤ 0.001). Similarly, 3.0- and 7.2-fold higher levels of pyridoxine and pyridoxamine (vitamin B₆ metabolites) in TMR compared to pasture samples led to 1.5-fold higher levels of pyridoxamine phosphate levels in grain-fed vs grass-fed beef (p = 0.006). On the other hand, 5.6-fold higher levels of flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN) (vitamin B₂ metabolites), and 94.2-fold higher levels of ascorbate and dehydroascorbate (vitamin C metabolites) in pasture compared to TMR samples did not translate to differences between grass-fed and grain-fed beef, with riboflavin (a vitamin B₂ metabolite upstream of FAD and FMN) being 1.7-fold elevated in grain-fed beef compared to grass-fed beef (p = 0.001). No differences between forages and meat samples were found for other vitamins.

Fig. 6. Comparison of vitamin metabolism in two types of beef cattle finishing diets {pasture and total mixed rations (TMR)} and their respective impact on beef (Bf_Grass and Bf_Grain) samples.

a Comparison of alpha-tocopherol abundance and b gamma/beta-tocopherol levels between two types of beef cattle finishing diets and their corresponding meat, and c Sub pathway analysis of vitamin metabolites, withgreen shaded cells indicating significantly higher (p ≤ 0.05) mean values in pasture samples and grass-fed beef, respectively and red shaded cells indicating significantly higher (p ≤ 0.05) mean values in grain-fed beef, respectively. Green shaded cells indicate significantly higher (p ≤ 0.05) mean values in grass-fed beef. Light red and light green shaded cells (0.05 ≥ p ≤ 0.10) indicate trends.

Discussion

The comparative soil analysis conducted between pasturelands and paired cornfields indicates that soil health is generally improved in pasturelands compared to neighboring cornfields producing feed grains, in part, destined for use in cattle TMR (Table 1). We found higher levels of SOM, EC, zinc, calcium, and potassium in pastureland compared to paired cornfields. Higher SOM and microminerals can support robust plant growth and improve the nutritional quality of plants^29,30. Since corn is considered a more exhaustive plant compared to most (perennial) forages, its higher nutrient demands can reduce soil organic and mineral reserves if they are not replenished³¹. This generally necessitates consistent use of fertilizers³², which was common practice in the studied cornfields included in this work. While corn fields cycle nutrients from residues (e.g., corn stalks and leaves) left on the field after harvest through decomposition by soil microorganisms, this process may not fully compensate for nutrients removed by the grain³³, which was evidenced by lower levels of SOM and various minerals in cornfields compared to lands managed as pasture. On the other hand, pasturelands typically maintain live plants year-round especially in Southern US, which helps in capturing and recycling nutrients more effectively and reduce erosion³⁴. Grazing animals also contribute to nutrient recycling by depositing manure and urine, which adds organic matter and nutrients back into the soil³⁵. Furthermore, adaptive rotational grazing practices, such as those employed by the grass-fed farmers in this work, have been found to benefit soil carbon levels compared to continuous grazing^10,36. Thus, the grazing management employed by the farmers in this work could have further contributed to the disparities in metrics of soil health and micronutrients in the pasturelands compared to paired corn fields (Supplemental Table 2); however, we did not have a continuous grazed grass-fed control to compare to. On the other hand, higher concentrations of cobalt in cornfields, which was an opposite trend to other minerals in the soil comparisons may potentially indicate the use of sulfate-containing fertilizers in corn crop production³⁷.

Metabolomics analysis of feed (TMR fed to grain-fed cattle) and forage (plants consumed by grass-fed cattle on pasture) also unveiled notable differences (Fig. 1a). Out of 846 compounds detected, 69% metabolites showed significant variation between the two types of finishing systems (Table 2). The substantial differences in the TMR vs pasture metabolome are reflected in the meat metabolomes with 38% of measured metabolites showing significant differences between grass-fed and grain-fed beef (Table 2). A significant shift was observed in the dipeptides sub-pathway, with notable increases in alanylproline, isoleucylglycine, and prolylalanine in grain-fed beef compared to grass-fed beef (Fig. 7). The elevated abundance of dipeptides in grain-fed beef are attributed to their higher levels in the TMR, in which dipeptides also appeared as a dominant metabolite class that separated TMR from forages. Higher levels of dipeptides can contribute and/or are indicative of accelerated weight gain in grain-fed animals³⁸.

Fig. 7. Heatmap of the top 50 metabolites in grass-fed (Bf_Grass; n = 8) and grain-fed (Bf_Grain; n = 8) beef samples, ranked based on PLS-DA VIP p values.

Names of the samples correspond to the location of the grass-fed ranches (Southern Piedmont Highlands [SPH], Southern Blue Ridge [SBR], Blackland Prairie [BLP]) and Midwest Feedlot (MWF) in the USA. Supplemental Table 1 provides the full names and abbreviations of the compounds.

Conversely, the grass-fed beef samples demonstrated significantly higher levels of phenolics and other phytochemical-derived metabolites. Additionally, long-chain PUFAs and acyl carnitines were enriched in the grass-fed beef samples. These findings can be directly attributed to the greater abundance of phenolics, flavonoids, and long-chain omega-3 fatty acids in forages compared to TMR samples. In particular, acyl carnitines in grass-fed meat are primarily enriched in omega-3 fatty acids, such as linolenoylcarnitine (C18:3), docosatrienoylcarnitine (C22:3), dihomo-linolenoylcarnitine (C20:3n3), and docosapentaenoylcarnitine (C22:5n3). Elevated levels of very long-chain fatty acids and acyl carnitines are associated with various health benefits, including reducing low-density lipoprotein (LDL) cholesterol levels, supporting brain function, and decreasing inflammation³⁹. Furthermore, long-chain acyl carnitines transport PUFAs to the mitochondria for β-oxidation⁴⁰. This is potentially observed in the elevated levels of TCA cycle metabolites in the grass-fed beef samples, which indicates increased reliance on oxidative metabolism in these animals⁴¹. In contrast, the grain-fed samples indicate a greater reliance on glycolytic metabolism, which is in line with previous literature⁴¹, and is the result of their high-concentrate diet.

The comparison of phytochemical metabolites in grass-fed and grain-fed beef, and their respective feed type samples (forage vs TMR) also provides an initial insight into the transfer of phytochemicals from feed to animal (Fig. 4); a relationship that is increasingly being recognized but had not studied extensively within the same study^21,42. Seven out of 33 phenolic metabolites detected in meat samples were directly transferred from plant to meat (e.g., n-methylpipecolate, stachydrine, ergothioneine, etc.), thus appearing as the same metabolite in meat as in the TMR and/or forages. This direct transfer suggests digestive and metabolic stability of these compounds in the rumen, aligning with previous studies demonstrating direct transfer of several dietary phytochemicals to animal tissues⁴³. For example, ergothioneine was 11.4-fold higher in forage compared to TMR samples and led to 2.8-fold higher levels in grass-fed beef compared to grain-fed beef. Ergothioneine and its metabolite histidine betaine, which was also 3.1-fold higher in grass-fed beef, are produced by fungi and mycobacteria in soils⁴⁴. Elevated levels of these metabolites provide a potential direct connection with soil health, which we found to be improved in pasturelands compared to paired cornfields.

In addition to direct transfer, phenolic metabolites can also undergo metabolic transformation⁴⁵. In particular, various (poly)phenols found in forages are metabolized in the rumen through phase-I metabolism including oxidation, reduction, and hydrolysis, and become conjugated by phase-II enzymes in the liver, to form sulfated, glucuronidated, and hydroxylated derivates, such as catechol sulfate, 4-ethylphenylsulfate, and hydroxyhippurates; mammalian antioxidant compounds that ultimately can become enriched in tissues, such as muscle meat, and biofluids such as blood and milk⁴⁵. To that end, we found a significant elevation phenolic-derived compounds in grass-fed beef such as 4-ethylphenylsulfate (5.4-fold), catechol sulfate (6.0-fold), and 3-hydroxyhippurate (2.4-fold) compared to grain-fed animals. Besides potential direct antioxidant, anti-inflammatory, antimicrobial, and anti-platelet activity of phenolic metabolites⁴⁶, both to livestock⁴⁷ and humans^9,21, phenolic metabolites in plants provide raw materials for rumen gut microbes to produce downstream metabolites, such as catechol sulfate, 4-ethylphenyl, dimethyl sulfone, and cinnamoylglycine. Microbial metabolites of polyphenols can further protect the body against oxidative stress, inflammation, infection, and certain cancers^48–50. Our results are consistent with previous studies, which report higher levels of phenolic-derived antioxidants in grass-fed beef^21,51.

For example, the proposed metabolic pathways leading to the production of 3-hydroxyhippurate in grass-fed beef involve several interconnected biochemical transformations from plant to animal (Fig. 4). First, hippuric acid, a common mammalian antioxidant, is synthesized from benzoic acid, a dominant phenolic acid in plants that serves as a precursor of the biosynthesis of many downstream phytochemicals⁵² and glycine, a common amino acid and secondary metabolite found in structural cell wall proteins of plants⁵³. The conversion of benzoic acid and glycine to hippuric acid is facilitated through intermediary metabolites such as caffeoyl tyrosine and caffeate⁵⁴. This pathway further advances with the hydroxylation of hippuric acid, producing 3-hydroxyhippurate, a key metabolite associated with ingestion of phenol-rich diets⁵⁴. Additionally, feruloylquinic acid plays a crucial role, undergoing conversion to ferulate and 4-hydroxybenzoate, which contributes to the downstream formation of 3-hydroxyhippurate via cinnamoylglycine⁵⁵, which was 2.0-fold higher in grass-fed meat. Another significant pathway involves coumaroylquinic acid, a dominant quinic acid found in plants, which contributes to the production of 4-hydroxycinnamate, a precursor to p-coumaric acid and p-vinylphenol⁵⁶. These compounds are further metabolized into catechol derivatives, including catechol sulfate and 4-ethylphenyl sulfate⁵⁴, both of which were elevated in grass-fed beef. These pathways collectively underscore the complex biochemical processes that are enhanced by a forage-rich diet, resulting in elevated levels of phytochemical-derived antioxidants grass-fed cattle compared to grain-fed cattle (Fig. 4). An average 118.2-fold higher concentration of phytochemical metabolites in pasture samples compared to TMR samples, led to a combined 3.1-fold increase in phytochemical metabolites in grass-fed beef compared to grain-fed beef.

While collectively higher in grass-fed beef, we found four phytochemical metabolites to be elevated in grain-fed beef, namely 4-methylcatechol, and the alkaloids fagomine, piperidine, and its metabolite thioproline. Piperidine is found to have potential antioxidant and anti-cancer effects in vitro^57,58 and fagomine⁵⁹ has been found to improve glucose regulation in vivo; however, some work has implicated piperidine in alkaloid poisoning when livestock are fed conserved forages⁶⁰. Whether observed levels of piperdine are of any concern is not known; however, our findings of higher levels of piperidine in grain-fed beef compared to grass-fed beef are in line with previous work⁶¹.

The elevated oxidative stress markers in grain-fed beef, such as homocysteine and methionine sulfoxide, suggest that these animals potentially experienced greater metabolic stress. This could be due to several factors, including reduced dietary self-selectivity, limited ability to engage in innate behaviors, and/or lower levels of antioxidant intake compared to grass-fed animals^{17,41,62–65}. The increase in 4-hydroxynonenal-glutathione in grain-fed beef compared to grass-fed beef further supports the hypothesis of higher molecular stress in grain-fed cattle, as 4-hydroxynonenal is a byproduct of lipid peroxidation and a biomarker of oxidative stress and cellular damage⁶⁶. Conversely, higher levels of urate and oxidized glutathione in grass-fed beef potentially indicate a more robust antioxidant defense system⁶⁷. Urate is a strong antioxidant that can neutralize reactive oxygen species, while oxidized glutathione is a marker of the body’s ability to recycle antioxidants⁶⁸. Therefore, elevated levels of these compounds in grass-fed beef potentially suggest that these animals are better equipped to handle oxidative stress.

Grass-fed beef exhibited elevated levels of alpha-tocopherol (Vitamin E) and retinol (Vitamin A), corresponding to higher concentrations of alpha-tocopherol and carotene in forages (Fig. 6). Alternatively, the higher levels of alpha-tocopherol and retinol in grass-fed beef, could be because the grass-fed animals were between 28 and 32 months of age at the time of harvest, while the grain-fed animals were between 18 and 22 months. This would allow for more time of fat-soluble vitamins to accumulate in tissue in the case of grass-fed animals; however, most prior studies find that diet is the most important factor in determining carotenoid and alpha-tocopherol levels in livestock with forages providing higher levels of these compounds^69–71.

In contrast, the abundance of B-vitamins (e.g., B1, B2, B5, and B6) in feed did not directly translate to beef (Fig. 6). For example, nicotinamide mononucleotide was higher in forage samples, but its abundance increased in grain-fed beef. However, beef cattle primarily rely on the synthesis of B-vitamins by rumen microbes rather than direct dietary intake. The higher levels of B-vitamins in grain-fed beef could also potentially explained by the higher concentration of non-structural carbohydrates in TMR (Supplemental Table 2), which have been reported to enhance the synthesis of several B-vitamin metabolites in the rumen⁷².

In conclusion, this study investigated connections along the soil-plant-animal-human nutrition continuum in Southern US grass-fed beef systems as compared to grain-fed beef control. Our findings suggest that grass-finishing enhances compounds with potential health benefits and improves metabolic health pathways of cattle. This was related to higher levels of phenolic antioxidants and very long-chain fatty acids in forages compared to TMR samples, though other factors such as the ability to engage in innate behavior and physical activity could also have contributed to the improved metabolic health status of cattle. Whether observed differences have an appreciable effect on human health is currently unknown; however, various randomized controlled trials with grain-finished beef, consumed as part of high-quality diets (such as the Mediterranean and DASH Diet), find improvements in cardiometabolic health compared to participant’s habitual and/or control diets^73,74, suggesting that grain-fed is compatible with good health when consumed as part of an overall “healthy” diet. Nonetheless, others have found that grass-fed meat consumption can increase circulating omega-3 fatty acids in the blood of consumers⁷⁵, which is associated with further improvements in metabolic health⁷⁶. Additionally, recent work found that consuming grass-fed beef or grain-fed beef results in differences in postprandial metabolomes, which indicates that the differences in metabolomes between grass-fed and grain-fed beef can directly lead to differences in circulating metabolites within consumers⁷⁷. How this would impact long-term health is currently not known. We also found improved metrics of soil health, including elevated SOM and mineral content in pasturelands when compared to paired cornfields, suggesting that grass-finishing of livestock on biodiverse pastures in rotational grazing systems can have soil health benefits compared to growing corn for feedlot-finishing of livestock. However, it is acknowledged that trade-offs may exist between pasture- and feedlot-finishing of cattle in terms of broader economic and environmental factors²⁰. Limitations of this study include a relatively small sample size with three grass-fed beef operation and a single grain-fed feedlot operation, and it is likely that heterogeneity exists both within and between production systems. Future work should include a broader number of both grass-fed and grain-fed operations to understand these variations. Nonetheless, our work provides initial evidence of cascading beneficial effects along the soil-plant-animal-human nutrition continuum. The study provides valuable information for consumers to make informed choices about the meat they purchase and for producers seeking to improve the quality of their beef products through improvements in soil health, forage and feed quality, and animal metabolic health.

Methods

Site selection

A total of three grass-fed beef operations in the Southern US, located within 500 ± 240 km of each other, were selected with operations based in Surry County, North Carolina (Southern Piedmont Highlands [SPH]; 36.86° N, 79.49° W); Buncombe County, North Carolina (Southern Blue Ridge [SBR]; 35.51° N, 82.38° W); and Perry County, Alabama (Blackland Prairie [BLP]; 32.56° N, 87.45° W). Furthermore, a cornfield within ≤13 km of each of the three farms was selected as control sites for the study. Soil and landscape positions from each pasture-based livestock operation were matched closely with neighboring crop fields that produced corn for commodity markets, including livestock feed. Geological and climatic conditions are described in further detail by Franzluebbers et al.²⁸, which is a companion paper that studies soil health and root-zone enrichment characteristic differences between the pastures and matched cropland fields.

All grass-fed beef farmers used adaptive rotational grazing practices, moving animals every 1–7 days into new paddocks on pasture with year-round grazing on all grass-fed beef farms. During the winter months, the animals on the SBR and BLP farms were provided ad libitum supplemental hay, while the animals on the SPH farm were not supplemented with feed. Animals on all three grass-fed farms had ad libitum access to a mineral block. The grain-finished beef samples and TMR samples were obtained from a producer in the Midwest USA (South Dakota). During the cow-calf and stocker phase, the feedlot-finished cattle grazed on native pastures owned or leased by the feeder/finishing operations in South Dakota, USA. During the finishing phase, the grain-finished cattle were kept in a feedlot located at the same feeding/finishing operation for ~130 days.

Soil collection

Soil samples were collected from each grass-fed beef operation and paired croplands in July-August 2021, followed by a second collection 60 days later. On each grass-fed farm, pastures were sampled several days prior to cattle entering the pasture to ensure ungrazed forages were collected and soil samples matched the forage collection sites. Four distinct sites were sampled on each pasture and paired corn fields, and per collection site five subsamples were composited, separated 10 m in cardinal directions from the central point. Soil was sampled at 0–10 cm depth with a 4 cm inside diameter push probe. Further details on the soil collection procedures can be found in Franzluebbers et al.²⁸. For each farm or cropland, we collected ~1500 g of soil in a single bag, which was mixed to ensure a homogenous sample. Soil samples were immediately put on ice and frozen at −20 °C for further analysis. Approximately, 400 g of soil from each collection site was subsequently shipped to Logan Labs (Lakeview, OH) utilizing their Standard Soil Analysis package.

Forage collection

Pasture forages were sampled three times at 60-day intervals throughout the grazing season of 2021, resulting in a total of 13 samples that were collected and tested (n = 13), with n = 5 samples collected from the SPH farm, n = 3 from the BLP farm, and n = 5 from the SBR farm. The final forage and TMR collections fell within ≤15 days of harvesting the animals. The research team collaborated with farmers to ensure the collection of forages represented the forages the cattle selected and were collected via a pasture walk with the farmers, collecting plant samples every 15–20 meters while covering the entire pasture. The fresh plants were cut at the animals’ reachable height using stainless steel blades and initially collected in a thirty-gallon bag, which was filled up completely during the pasture walk. After collecting the pasture, the forage samples were laid out on a tarp, further cut up, and mixed to ensure a homogenous sample. Three one-gallon bags were subsequently filled up with forage samples and immediately put on ice and frozen at −20 °C for further analysis. Total mixed ration samples (TMR; n = 4) were procured from the Midwest feedlot and were collected only once in November 2021 and coincided with the final collection date of the forages. A one-gallon bag of forage sample and TMR sample was subsequently shipped to Cumberland Valley Analytical for protein, fiber and mineral analysis utilizing their Standard Wet Chemistry package. Prior to metabolomics analysis, a 25-gram portion of each forage sample and TMR was weighed, lyophilized, and processed into a powder under liquid nitrogen, and stored at −80 °C for future metabolomics analysis.

Pastures were composed of a diverse mixture of forages on all farms, including tall fescue (Lolium arundinaceum [Schreb.] S.J. Darbyshire), red clover (Trifolium pratense L.), orchardgrass (Dactylis glomerata L.), and crabgrass (Digitaria sanguinalis [L.] Scop.) at the SPH farm; tall fescue, orchard grass, plantain (Plantago lanceolata L.), and white clover (Trifolium repens L.) at the SBR farm; and tall fescue, ryegrass, plantain, and chicory (Cichorium intybus L.) at the BLP farm²⁸. Studied cropland fields were all planted with corn (Zea mays L.) with conservation tillage at the Piedmont and Blue Ridge locations and with conventional tillage at the BLP location with routine fertilizer and pesticide application. Pastures and croplands were in respective land uses for at least 10 years. The TMR samples consisted of ground corn (~55–60%), distiller’s grain (~10–15%), grass hay/corn silage (~30%), vitamin/minerals pre-mix (~2.6%), and rumensin (~0.4%).

Meat collection

The pasture-finished animals were between 25 and 28 months of age at time of slaughter, while the grain-finished animals ranged from 18 to 22 months. Slaughter took place in a United States Department of Agriculture (USDA)-regulated facility in the presence of an accredited inspector. Animals were stunned before being slaughtered by rapid exsanguination. A total of 3 beef ribeye steak (longissimus dorsi) samples from separate carcasses were collected from the SBR and BLP pasture-finished farms, while 2 samples were collected from the SPH farm in November of 2021, representing the grass-fed beef samples (Bf_Grass). Additionally, 8 beef samples were collected from a Midwest US feedlot (MWF) during November 2021, representing the grain-fed beef samples (Bf_Grain). Furthermore, all the farmers raised the same breed of cattle (Black Angus). Upon arrival of meat samples to the laboratory at Utah State University, meat samples were ground in a meat grinder, pulverized under liquid nitrogen, and stored at −80 °C. Since this work evaluated cattle from commercial ranches and did not involve intervention by the research team, Institutional Animal Care and Use Committee (IAUAC) approval was not necessary. The sample size was set to 8, allowing for a true discovery rate ranging from 96–99%, assuming a fold difference of 1.1 or 1.3 in grass-fed and grain-fed beef, respectively, with a p ≤ 0.05 and a q ≤ 0.3.

Metabolomics profiling

A schematic representation of the study flow is provided on Fig. 8. Members of the research team conducting the analysis were blinded to the experimental groups. Sample preparation for metabolomics was carried out as described previously⁶¹. Briefly, 100 mg of each sample (plant and meat) was weighed, and recovery standards were added for quality control. Proteins were precipitated with methanol, and supernatants were harvested by centrifugation at 15,000 × g. Extracts were divided into five fractions: two for analysis by separate reverse phase (RP)/UPLC-MS/MS methods with positive/negative ion electrospray ionization (ESI) mode, one for HILIC/UPLC-MS/MS with negative ion ESI mode, and one for backup. Sample extracts were dried under a nitrogen stream using TurboVap (Zymark) and reconstituted in the mobile phases described below.

Fig. 8. Workflow of metabolomica analysis.

a Cattle were finished either on pasture or with a total mixed ration (TMR) in confinement. b Meat samples (Longissimus dorsi) were collected, pulverized under liquid nitrogen, and processed for extraction using a MicroLab STAR system. c Metabolomic profiling was conducted using liquid chromatography-mass spectrometry (LC-MS/MS) coupled with Orbitrap MS/MS. d Metabolite peaks were identified, and MS/MS spectra were acquired. e Acquired MS/MS spectra were annotated by matching with reference databases. f Data were analyzed using Metabolon Pathway Explorer, ChemRICH, and MetaboAnalyst 6.0 for pathway and enrichment analysis.

The UPLC-MS/MS platform utilized a Waters Acquity UPLC coupled with a Thermo Scientific Q-Exactive mass spectrometer, interfaced with a HESI-II probe and Orbitrap analyzer. Four aliquots were analyzed under different conditions. The first aliquot, optimized for hydrophilic compounds, was eluted from a C18 column (UPLC BEH Amide 2.1 × 150 mm, 1.7 µm, Waters) using water and methanol with 0.05% PFPA and 0.1% FA. The second aliquot, optimized for hydrophobic compounds, was eluted from the same C18 column using methanol, acetonitrile, water, 0.05% PFPA, and 0.01% FA. The third aliquot was analyzed under basic negative ESI conditions, eluted from a separate C18 column with methanol and water containing 6.5 mmol/L Ammonium Bicarbonate at pH 8. The fourth aliquot was analyzed under negative ESI conditions, eluted from a HILIC column using water and acetonitrile with 10 mmol/L Ammonium Formate at pH 10.8. MS analysis alternated between MS and data-dependent MSn scans, covering m/z 70–1000 at a resolving power of R = 35,000.

Metabolites were identified by comparing ion features in the samples to a reference library of chemical standards, considering retention time, molecular weight (m/z), preferred adducts, in-source fragments, and associated MS spectra⁷⁸. The data were curated for quality control using Metabolon’s proprietary software, with library matches checked and corrected as needed. Peaks were quantified using the area-under-the-curve method. Data normalization was performed to correct for instrument inter-day tuning differences by setting the medians to one (1.00) and normalizing each data point proportionately, a process known as “block correction.” This preserved variation between samples while allowing metabolites with different raw peak areas to be compared on a similar scale.

Data analysis

Soil physicochemical properties data of pastures and corn fields were compared using paired student t-tests with a p ≤ 0.05. Metabolomics data for feed and beef samples were normalized by median, log10-transformed, and mean-centered. Missing values were imputed using the minimum observed value for each compound. Differences in compounds between total-mixed ration and pasture samples, as well as grass-finished and grain-finished beef samples, were computed using the Wilcoxon rank-sum test with p ≤ 0.05. To account for multiple comparisons, False Discovery Rate (FDR) adjusted statistics (q-values) were applied, with a q ≤ 0.05 indicating high confidence in the t-test results⁷⁹. PLS-DA and Volcano plots were developed for feed and beef metabolomics data (FDR = 0.05). Heatmaps of the top 50 compounds were created using Pearson distance measure, Ward clustering, and PLS-DA VIP settings with MetaboAnalyst 6.0. Cluster analysis was conducted using ChemRICH software by ontology mapping and structural similarity using InChiKeys and SMILES (https://chemrich.fiehnlab.ucdavis.edu/class.html). Sub-pathway clusters with FDR ≤ 0.05 were retained for analysis and reported. Annotated metabolites were investigated for potential health effects and bioactivities using the Chemical Abstracts Service numbers in PubChem and FooDB databases, along with Google Scholar and PubMed searches. Figures were created using Biorender (https://biorender.com/) and/or Graphpad (PRISM Graphpad, Boston, MA, USA).

Author contributions

M.A.: Investigation, Data Analysis, Writing—Original Draft. M.H.P.: Conceptualization, Methodology Investigation, Writing—Review & Editing. J.R.: Conceptualization, Methodology, Investigation, Writing—Review & Editing. A.F. Conceptualization, Methodology, Investigation, Writing—Review & Editing. S.N.Y.: Conceptualization, Methodology, Investigation, Writing—Review & Editing. S.L.K.: Conceptualization, Methodology, Writing—Review & Editing. F.D.P.: Conceptualization, Methodology, Writing—Review & Editing. J.R.B.: Methodology, Investigation, Data Analysis, Writing—Review & Editing. S.v.V.: Conceptualization, Methodology, Investigation, Formal Analysis, Writing—Original Draft. All authors read and approved the final manuscript.

Data availability

The datasets supporting the conclusions of this article are included within the article. In particular, the raw metabolomics data used for statistical analysis and interpretation is with study ID# ST003546. Chromatographic data files can be provided by the corresponding author upon reasonable request.

Competing interests

S.v.V. acknowledges current grant support from USDA-NIFA-SARE (2020-38640-31521; 2021-38640-34714), USDA-ARS (USDA-2022-58-3064-2-007), the Greenacres Foundation, Applegate LLC, Perdue Foods LLC, and the Bionutrient Institute for (co-funded) projects that link agricultural production systems—animal and cropping systems—to the nutritional/metabolite composition of animal and plant foods. S.v.V. also reports travel honoraria and speaker fees related to presentation related to their research. S.v.V. is a non-paid member of the Scientific Advisory Committee of the Food and Agriculture Organization of the United Nations. F.D.P. reports receiving honoraria for his talks about behavior-based management of livestock. All other authors report no competing interests.

Ethics approval and consent to participate

This work evaluated general practices of commercial operations with no intervention by the research team, while animals were harvested in a USDA-inspected facility. Institutional Animal Care and Use Committee (IACUC) approval was therefore not necessary.

Supplementary information

The online version contains supplementary material available at 10.1038/s41538-025-00471-2.