Five years of oxen grazing enhances soil carbon and structure in alpine vineyards

Abstract

Integrating livestock with crop farming can greatly enhance agricultural sustainability and accelerate the agroecological transition. This study investigated the five-year effects of oxen grazing in a vineyard in South-Tyrol (Italy). Grazing occurred from autumn to spring over five consecutive years at a density of 5–7 oxen ha^− 1. An adjacent site remained ungrazed. Soil samples were collected and analyzed for soil carbon pools (elemental analyzer), compaction (bulk density), soil structure (micro–water-stable aggregates, µWSA < 63 μm; macro-water-stable-aggregates, MWSA < 250 μm), plant available elements (ICP-MS), total microbial biomass (fumigation and extraction) and microbial abundance (qPCR). The results showed that, despite both fields being pedogenically similar, oxen grazing improved soil C. Oxen grazing increased total organic carbon (+ 14%), total nitrogen (+ 12%), carbon/nitrogen ratio (+ 2%), dissolved organic carbon (+ 11%) and dissolved carbon (+ 11%). Available elements and soil bulk density did not change, while soil structure even improved as evidenced by the increase of µWSA (+ 14%) in the oxen-grazed site. This observation is supported by the increase in bacterial abundance (+ 1%) as they are typically present in µWSA, while MWSA and fungal abundance together with microbial biomass remained stable across the two sites. Our findings highlight the potential of combining viticulture with pasture as a strategy to enhance soil health and C, with no evident negative effects. Strengthening the integration and cooperation between viticulture and livestock farming could play a key role in advancing sustainable agriculture for the agroecological transition.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-35761-6.

Introduction

Historically, agricultural systems that integrated both livestock and crop production were widespread globally until the 20th century. This combination allowed animals to utilise crop residues, maximising land use while simultaneously enriching the soil with manure. However, the rise of mechanisation led to the decline of animals as a source of labour, and the widespread use of chemical fertilisers increased agricultural productivity, decoupling crop production from the need for manure as an organic fertiliser. Additionally, the reduction in the number of cultivated species resulted in simplified and shorter crop rotations. Previously, long and diverse rotations were the primary method for pest management, but the advent of intensive pesticide use rendered them unnecessary¹. In many developed countries, farms have become increasingly specialized in either crops or livestock, leading to a decline in mixed farming systems². In Europe, the integration of livestock and crop production has persisted more commonly in less favourable regions, such as areas with poor market access or challenging soil and climatic conditions, including mountainous regions³.

However, the agricultural sector is undergoing an agroecological transition aimed at re-establishing diversified, circular systems that enhance ecosystem services, reduce dependence on external inputs, and improve resilience to climate change.

Recent work on agroforestry with integrated livestock highlights the potential of agriculture–pasture–animal systems to enhance both production and ecosystem functioning. Silvopastoral systems show that combining trees, forages and grazing livestock can improve forage quality, animal productivity, carbon (C) sequestration, soil and water regulation, biodiversity and animal welfare, positioning them as a more sustainable alternative to specialized forestry or livestock system^4,5. Other analyses of integrated crop–livestock–forestry systems further indicate that reintegrating grazing animals into tree–crop rotations can mimic natural trophic structures, restore disrupted nutrient cycles, and support long-term yield stability and climate mitigation⁶. At the same time, a growing body of evidence from sub-Saharan Africa demonstrates that agriculture–livestock–forestry systems can simultaneously improve farmer income, food and feed security, soil health and C storage, while underscoring that the magnitude of these benefits—and the balance of trade-offs such as competition or greenhouse-gas emissions⁷.

Within this context, integrating livestock grazing with perennial crops, such as grapevines, is re-emerging as a key strategy to close nutrient cycles, enhance soil health, control weeds naturally, and strengthen on-farm biodiversity while maintaining productivity⁸. Vineyards in Europe cover approximately 3.2 million hectares, with Spain, France, and Italy accounting for the majority of this area⁹. Although the potential importance of these practices is well established, the body of literature addressing their effects on soil health is still limited. A few existing studies have mainly focused on small livestock, such as sheep and geese, whereas the potential role of larger animals, such as oxen, remains unexplored^10–12. Large livestock (e.g. cattle, oxen) can have a stronger impact on agroecology and soil health than small livestock because their greater body mass, trampling and distinct grazing behavior more powerfully shape vegetation, soil structure and nutrient cycling. This higher “leverage” means they can potentially deliver larger benefits (weed control, nutrient return, habitat heterogeneity) but also greater risks (compaction, erosion) in systems like vineyards, so they require targeted, system-specific research rather than extrapolation from sheep-based studies^13,14.

The integration of large livestock could however potentially enhance both production diversity and ecosystem services, including improved crop yields, farm profitability, pest and weed management, and soil health. Notably, livestock utilize only a small fraction of the nutrients they consume, with 60–99% returned to the soil through excreta (urine and dung)¹⁵. Thus, the combination of viticulture with oxen grazing minimizes nutrient losses and reduces the dependence on external inputs¹⁶. Additionally, incorporating grazing into vineyard management can help lower the overall environmental footprint of wine production^8,16.

This study investigates the effects of oxen grazing in vineyards in South Tyrol, Italy, focusing on how large livestock influence soil chemical and biological properties. Specifically, we examine three key aspects: Soil Compaction, C and Nitrogen (N) Dynamics and microbial abundances. Excessive hoof trampling can lead to soil compaction, reducing pore space, water infiltration, and thus overall soil health. Our first hypothesis is that oxen grazing will increase soil compaction (H1). Grazing can enhance soil organic matter and carbon content through excreta deposition, influencing soil structure and biological activity. Thus, secondly, we hypothesised that oxen grazing would increase soil total organic carbon (TOC, H2). Livestock can alter soil microbial abundance by modifying soil properties and nutrient cycling due to excreta and trampling. Therefore, our last hypothesis is that oxen grazing will significantly increase the fungal and bacterial abundance (H3). By addressing these hypotheses, this study provides insights into the role of large livestock in sustainable vineyard management and their impact on soil health.

Materials and methods

Sample collection and storage

Soil samples were collected in the late spring of 2024 in the vineyards of the ”Alois Lageder” winery in Margreid, South Tyrol, Italy (46.275002°N, 11.212469°E, Fig. 1). The vineyard, planted with Vitis vinifera L., cv. Chardonnay in 2010, has divided into two fields: one where oxen grazing has been practiced annually during late autumn (approximately 8000 m²), winter and early spring for approximately five years (starting in autumn 2019) (OG site = Oxen-grazed site), and another without oxen grazing (NG site = Non-grazed site, approximately 7500 m²). The oxen density was 5–7 oxen per hectare. Both fields are managed without irrigation and underwent the same agronomical practice (e.g. fertilisation). The grass in the NG and OG site was cut only after the oxen were removed from the fields in late spring. The climate is continental, with an average annual temperature of 12.7° C and annual rainfall of 895.6 mm (nearest weather station: Salorno, Italy). The two study sites, are situated adjacent to each other (within a few meters), separated by a narrow buffer zone. Both sites are subjected to the same physical, chemical and biological weathering, as well as the same climatic conditions and are described as mainly quaternary alluvial gravels originating from carbonate (dolomite/limestone) and porphyry sources (https://natura-territorio.provincia.bz.it/it/geobrowser-maps). The soil in both experimental sides is classified as loamy according to the United States Department of Agriculture (USDA) texture classification, with sand, loam, and clay percentages varying by no more than 3% between sites (Table 1).

Fig. 1.

Sampling location (Margreid, South Tyrol, Italy) and grazing oxen in the experimental vineyards. The map was created using QGIS (version 3.34.0, https://www.qgis.org).

Table 1.

Soil parameters in non-grazed and oxen-grazed sites (n = 15; mean ± SE).

Parameter	Not-grazed (NG)	Oxen-grazed (OG)
Clay (%)#	20 ± 1n.s.	19 ± 1n.s.
Silt (%)#	36 ± 2n.s.	35 ± 2n.s.
Sand (%)#	44 ± 3n.s.	46 ± 3n.s.
Texture	Loam	Loam
Bulk Density (g/cm3)	0.88 ± 0.02n.s.	0.88 ± 0.03n.s.
pHH2O#	7.53 ± 0.03n.s.	7.46 ± 0.02n.s.
pHCaCl2	7.10 ± 0.01n.s.	7.08 ± 0.02n.s.
TOC (%)	4.06 ± 0.16**	4.64 ± 0.11**
TN (%)	0.40 ± 0.02*	0.45 ± 0.01*
C/N	10.05 ± 0.07*	10.26 ± 0.06*

Asterisks indicate statistically significant differences between treatments assessed by a Student’s t-test (*p < 0.05; **p < 0.01; ***p < 0.001). A ^# symbol indicates that data was not distributed normally and a Mann–Whitney U test was performed. If no significant difference was detected, n.s. is displayed.

In total, 15 samples per treatment (NG, OG) of topsoil (0–30 cm) were collected with an Edelman auger from the inter-rows of vineyards without oxen grazing (NG) and of vineyards in the presence of oxen grazing (OG). After each replicate, the auger was sterilized using alcohol and samples were transported in sterile vials. Samples were stored at -20 °C for DNA extraction, at -5 °C for the microbial biomass, while the samples for the physico-chemical analyses were dried at 60 °C until constant weight and sieved to 2 mm.

The two study sites are situated adjacent to each other (within a few meters), separated by a narrow buffer zone. Both sites are subjected to the same physical, chemical and biological weathering, as well as the same climatic conditions.

and are described as mainly quaternary alluvial gravels originating from carbonate (dolomite/limestone) and porphyry sources (https://natura-territorio.provincia.bz.it/it/geobrowser-maps). The soil in both experimental sides is classified as loamy according to the United States Department of Agriculture (USDA) texture classification, with sand, loam, and clay percentages varying by no more than 3% between sites (Table 1). Thus, variations in the pedological background of the two sites are minimal or non-existent.

Physico-chemical analyses

Soil texture was determined using a combined sieve–hydrometer method after dispersion with sodium pyrophosphate (Na₄P₂O₇). Briefly, 100 g of sieved soil were placed in a 1-L graduated cylinder with 100 mL of distilled water and 25 mL of 0.1 M Na₄P₂O₇, shaken thoroughly, and brought to volume. After a second shake, hydrometer readings were taken at 30 s (sand and silt fractions) and 2 h (clay fraction), together with suspension temperature. The proportions of sand, silt and clay were then calculated using standard particle-size equations. The soil texture was then determined plotting the results of the three fractions in the USDA soil texture triangle¹⁷. Bulk density (BD) was measured after removing the grass layer (approximately 10–12 cm) by pressing a 5 × 8 (height x diameter) cm soil cylinder into the soil. Bulk density was then determined as the ratio of the oven-dry mass of the soil core to its volume, calculated by dividing the dry weight of the soil (105 °C, constant mass) by the known core volume (g cm⁻³).

Soil pH

Soil pH was determined using a portable pH meter (pH7 Vio, Metrocal, Italy) after extraction of 10 g of air-dried, sieved soil with 25 mL of either 0.01 M CaCl₂ solution or deionized water. The resulting suspensions were centrifuged at 2000 g for 3 min, and the supernatants were subsequently filtered through 0.45 μm membranes prior to pH measurement.

Soil available elements

CaCl₂ extracts for pH analysis were filtered at 0.45 μm (cellulose Whatman syringe filters) and acidified with 2% ultrapure HNO₃. Available element concentrations were measured by ICP-MS (Agilent 7800, USA) using 1 ppm lanthanum (La) as the internal standard, and results were normalized to extraction sample weight¹⁸.

Aggregate stability

Water-stable aggregates (WSA) were measured using a wet-sieving apparatus (Eijkelkamp, The Netherlands). For each sample, 4 g of air-dried, 2-mm sieved soil were pre-moistened in cups fitted with 63-µm or 250-µm sieves and oscillated under deionized water for 3 min. The water-dispersible fraction was collected, dried and weighed (Wd_w). The remaining material on the sieve was then dispersed in 100 mL of 0.5 mol L⁻¹ NaOH for 5–8 min; the dispersed fraction was dried, blank-corrected (–0.20 g), and weighed (Wd_s). WSA (%) was calculated as the proportion of aggregates resistant to water dispersion relative to the total dispersible fraction. Results are reported as µWSA (63 μm) and MWSA (250 μm), with all measurements performed in triplicate according to the following formula:

1

Dissolved organic carbon and nitrogen

Dissolved organic carbon (DOC) and dissolved nitrogen (DN) were extracted by the method of Jones and Willett (2006)¹⁹, in which 2 g of air-dried soil sample were extracted with 40 mL ultra-pure water by shaking for 30 min and subsequent centrifuging at 9000 rpm for 20 min at 4 °C. The supernatant was filtered through a 0.45 μm glass fibre filter. The clear solution was then analysed with a total organic carbon analyser (TOC–V CPN, Shimadzu, Japan²⁰.

Total carbon and nitrogen analysis

Dried soil was ball-milled for 6 min at 30 Hz (Mixer Mill MM 400, Retsch, Italy), and ~ 0.2 g of homogenized material was weighed into silver capsules. Samples were acidified with 1 mol L⁻¹ HCl and heated to remove carbonates before combustion in an elemental analyser (EA Flash 1112, Thermo Scientific, Germany) for TOC and TN determination. The oxidation and reduction furnaces operated at 1020 °C and 900 °C, respectively, and H₂O was removed with a Mg(ClO₄)₂ trap²¹.

Soil microbial biomass

Soil microbial biomass was determined by fumigation with chloroform. Approx. 5 g of fresh soil (stored at 4 °C) were weighed in aluminium cups and placed in a desiccator. Samples were then fumigated with ethanol-free chloroform for 24 h and then extracted with 20 mL of K₂SO₄ 0.5 mol L^− 1. A non-fumigated control was immediately extracted with 20 mL of K₂SO₄ 0.5 mol L^− 1. Both fumigated and non–fumigated extracts were then analysed using a total organic carbon analyser (TOC–V CPN, Shimadzu, Japan). Microbial carbon biomass was then calculated as follows:

2

where EC is (organic C extracted from fumigated soils) – (organic C extracted from non-fumigated soils), k_EC is a constant representing the extractable part of microbial biomass C after fumigation. The k_EC has been reported for different soils and ranges from values below 0.2, for soils collected from depths greater than 40 cm²², to 0.45, for agricultural soils²³. In our case, a k_EC of 0.45 was used.

DNA extraction and microbial quantification

Total DNA was extracted from 0.25 g of sample using the Dneasy^® PowerSoil^® extraction kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions and then stored at -20°C. Extracted DNA was quantified using a Qubit 4.0 fluorometer (Life Technologies Corporation) in combination with the Qubit™ 1X dsDNA Broad Range Assay Kits (Invitrogen, Milan, Italy), and normalised to 10 ng µL^{− 1} for all samples so that they could be processed for microbial abundance quantification by qPCR. Bacterial and fungal abundances were quantified through the amplification of the 16S rRNA gene and the fungal ITS2 region, respectively. Amplifications were performed using the primer pairs 341F_16S / 805R_16S (5’-CCTACGGGNGGCWGCAG-3’ / 5’-GACTACHVGGGTATCTAATCC-3’²⁴ for bacteria and ITS3_KYO2 / ITS4 (5’-GATGAAGAACGYAGYRAA-3’ / 5’-TCCTCCGCTTATTGATATGC-3’²⁵ for fungi in a total volume of 20 µl containing 1 µl of total DNA, 10 µl of KAPA SYBR Fast qPCR Kit Master Mix 2X Universal, 0.6 µl of each primer, 7.8 µl of sterile ultrapure water. Each reaction was conducted using a Biorad CFX96 thermocycler (Biorad, Hercules, California, United States) with the following cycling conditions: For bacteria 3 min at 95 °C followed by 39 cycles of 10 s at 95 °C, 20 s at 60 °C, and finally a melting curve starting from 6 5 °C to 95 °C with temperature increase of 0.5 °C and a transition rate of 5s. For fungi, 3 min at 95 °C followed by 39 cycles of 10 s at 95 °C, 20 s at 53 °C and 30 s at 72 °C, and finally a melting curve starting from 65 °C to 95 °C with temperature increase of 0.5 °C and a transition rate of 5 s. Standards were obtained by amplifying ad hoc synthetic sequences of known concentration, molecular weight and copy number (IDT, Coralville, Iowa, USA) using previously described protocols. All samples and ten-fold serial standards dilutions were run in triplicate along with qPCR negative controls.

Statistical analysis

The statistical analysis to compare the two treatments (NG vs. OG) was carried out in R²⁶. Data normality was assessed using the Shapiro–Wilk test. Normally distributed data were analyzed with Student´s t-test, while non-normally distributed data were evaluated using the Mann–Whitney U test. Statistical significance was considered at p < 0.05. Data were visualized using the ggplot2 package²⁷. In every Figure, it is given which test was used for each parameter. All results will be presented as means ± standard error (SE) of 15 replicates. Additionally, effect sizes were calculated as Hedges’ g for each parameter by comparing NG vs. OG. All statistical parameters with p-values and Hedges’ g values are reported in Supplementary Table 1.

A Mantel test was performed using the vegan²⁸ and linkET²⁹ packages to evaluate the influence of soil physicochemical properties on both WSA. Bray–Curtis and Euclidean distance matrices were used to compute pairwise dissimilarities and assess correlations. Additionally, Pearson’s correlation coefficient was used to explore the relationships among individual soil chemical and physical variables and the relative abundance of µWSA 63 μm and MWSA 250 μm.

Results

Physico-chemical analyses

Physico-chemical data are presented in Tables 1 and 2.

Table 2.

Concentrations of Mg, P, K, Fe, Mn, Cu, and mo in CaCl₂ soil extracts (n = 15, mean ± SE) from non-grazed and oxen-grazed sites.

Parameter	Not-grazed (NG)	Oxen-grazed (OG)
Mg (mg g− 1 DW)#	117.48 ± 4.52n.s.	136.67 ± 11.72n.s.
P (mg g− 1 DW)#	0.77 ± 0.1n.s.	1.01 ± 0.24n.s.
K (mg g− 1 DW)#	41.66 ± 2.04n.s.	48.83 ± 5.28n.s.
Fe (mg g− 1 DW)#	1.19 ± 0.12n.s.	1.53 ± 0.21n.s.
Mn (µg g− 1 DW)#	0.29 ± 0.01n.s.	0.38 ± 0.04n.s.
Cu (µg g− 1 DW)#	0.01 ± 0.01n.s.	0.01 ± 0.01n.s.
Mo (µg g− 1 DW)#	0.02 ± 0.01n.s.	0.03 ± 0.02n.s.

Asterisks indicate statistically significant differences between treatments based on Student’s t-test (*p < 0.05; **p < 0.01; ***p < 0.001). A ^# symbol indicates that data was not distributed normally and a Mann–Whitney U test was performed. If no significant difference was detected, n.s. is displayed.

Total and dissolved organic carbon and nitrogen

The soil TOC was found to be significantly higher in the OG site as compared to the NG (4.64 ± 0.11% for OG vs. 4.06 ± 0.16% for C; p = 0.005). The same was found for the TN with the OG site having 0.45 ± 0.01% of TN while the NG site showed 0.40 ± 0.02% (p = 0.015) (Table 1). The soil content of dissolved carbon (DC) (43.75 ± 3.27 mg L^− 1 for OG vs. 39.32 ± 5.41 mg L^− 1 for NG site; p = 0.012) and dissolved organic carbon (DOC) (42.91 ± 3.17 mg L^− 1 for OG vs. 38.53 ± 5.36 mg L^− 1 for C; p = 0.012) increased in the OG site too (Fig. 2). On the contrary, dissolved nitrogen (DN) (4.19 ± 0.42 mg L^− 1 for OG vs. 3.80 ± 0.64 mg L^− 1 for NG site; p = 0.057) and dissolved inorganic carbon (DIC) (0.84 ± 0.18 mg L^− 1 for OG vs. 0.79 ± 0.12 mg L^− 1 for C; p = 0.385) content did not differ between the sites (Fig. 2).

Fig. 2.

Dissolved Organic Carbon (DOC), Dissolved Carbon (DC), Dissolved Inorganic Carbon (DIC) and Dissolved Nitrogen (DN) content in mg L^-1 in non-grazed and oxen-grazed sites (n = 15). Asterisks over to the boxes indicate statistically significant differences between treatments assessed by Student´s t-test (* = p < 0.05; ** = p < 0.01; *** = p < 0.001). If no significant difference was detected n.s. is displayed.

Aggregates

Micro-Water Stable Aggregates (< 63 μm) were found to be 11% more abundant in fields with oxen grazing (74% in NG site vs. 85% in OG site; p-value < 0.001). Instead, there was no significant difference in MWSA (< 250 μm) between the sites (69% in NG site vs. 67% OG site; p-value 0.27) (Fig. 3).

Fig. 3.

Water Stable Aggregates < 63 μm (WSA_63µm) and < 250 μm (WSA_250µm) expressed as % in non-grazed and oxen-grazed sites; n = 15. Asterisks over to the boxes indicate statistically significant differences between treatments assessed by Student´s t-test (* = p < 0.05; ** = p < 0.01; *** = p < 0.001). A # symbol indicates that data was not distributed normally and a Mann–Whitney rank-sum test was performed. If no significant difference was detected n.s. is displayed.

Microbial biomass

The microbial biomass was found to be 94.83 ± 5.40 mg C kg^− 1 in the non-grazed and 100.35 ± 5.90 mg C kg^− 1 in the oxen-grazed site and not statistically different from each other (p = 0.510) (Fig. 4).

Fig. 4.

Copy number (log₁₀(CP)) of the ITS region for fungal species and 16 S rRNA gene for bacteria, and the Microbial Biomass (mg C kg^-1) in non-grazed and oxen-grazed sites; n = 15. Asterisks over to the boxes indicate statistically significant differences between treatments assessed by Student´s t-test (* = p < 0.05; ** = p < 0.01; *** = p < 0.001). A # symbol indicates that data was not distributed normally and a Mann–Whitney U test was performed. If no significant difference was detected n.s. is displayed.

DNA analysis

The fungal abundance (ITS) is not different between the two sites (p = 0.627), with an average value of 5.38 ± 0.06 in the NG site, and 5.39 ± 0.09 in the OG site (Fig. 4). Conversely, bacterial abundance (16 S) is slightly higher (p = 0.040) in OG than in NG site, with values of 6.95 ± 0.07 and 6.90 ± 0.07, respectively.

Discussion

Within the framework of the agroecological transition, practices such as livestock grazing in vineyards have been proposed and, in some contexts, implemented. However, their potential benefits remain insufficiently documented due to the limited availability of scientific literature. Here, we present a unique case in which we investigated the effects of five years of oxen grazing on a vineyard in the wine region of South Tyrol (Italy). Both sites are subjected to the same weathering and climatic conditions, as well as parent rock material. The soil in both experimental sides is classified as loamy according to the United States Department of Agriculture (USDA) texture classification, with sand, loam, and clay percentages varying by no more than 3% between sites (Table 1). Thus, variations in the pedological background of the two sites are minimal or non-existent.

Our first hypothesis was that oxen grazing would increase soil compaction (H1). Previous studies suggest that heavy grazing can cause localised soil compaction^30,31. However, no evidence of soil compaction was observed, as soil bulk density did not show any statistically significant differences between the two sites (Table 1). Given the well-documented impact of large-animal trampling on soil compaction³², several factors might provide an explanation. First, the moderate grazing pressure at our experimental site (only 5–7 oxen per hectare) was lower than that reported in other studies³², potentially keeping compaction below detectable levels. Moreover, the oxen did not stay in the vineyard the whole year but were transferred to the pasture in late spring until grape harvest in autumn. Additionally, this study was conducted in a vineyard rather than a traditional pasture, which is dominated by grass species and scattered trees. Vines are known for their exceptionally well-developed root systems³³, which significantly alter soil structure and contribute to soil decompaction^34,35. Another factor is the increased input of organic matter from animal manure. Notably, TOC, DOC and DC content were significantly higher at the OG site compared to the NG site (Table 1; Fig. 2). Since TOC is known to mitigate soil compaction and enhance soil resilience to compaction³⁶, TOC accumulation at site OG likely played a substantial role in counteracting compaction effects. Therefore, the combination of the vines’ extensive root systems and the increased TOC content at the OG site likely offset the compaction that might have resulted from oxen grazing (Table 1). The organic matter input provided by oxen could also help mitigate soil compaction caused by heavy machinery, thereby supporting sustainable agriculture and contributing to the agroecological transition.

The increase in TOC content (Table 1) confirms our second hypothesis (H2). Most parameters measured, which are related to soil carbon, including TOC, C/N ratio, DC, and DOC, were significantly higher in the OG site as compared to the NG site (Table 1; Fig. 2, high effect size, Supplementary Table 1). Also, TN increased significantly. The increase in TOC is very likely linked to the higher input of organic matter from animal manure³⁷. This result further supports sustainable agricultural practices and contributes to the ongoing agroecological transition. However, findings from previous studies are inconsistent and relatively scarce. One long-term study reported that sheep–vineyard systems increased both labile/active carbon pools and mineral-associated organic C³⁷. In contrast, a short-term study on sheep grazing in Mediterranean vineyards found no significant effect on soil carbon accumulation¹¹. To date, no study has investigated the impact of large livestock, as in our case.

Interestingly, beyond directly mitigating soil compaction, the rise in TOC (approx. +15% in OG vs. NG site) also improved the stability of soil µWSA (< 63 μm) (Fig. 3, high effect size, Supplementary Table 1). The aggregate stability increased significantly in the grazed site, rising from 74% to 85% compared to the NG site. However, the increase in TOC appeared insufficient to enhance MWSA (> 250 μm) stability, which remained constant between sites (67–69%) (Fig. 3). Indeed, while TOC controls directly aggregate stability through chemical and biological binding, it has a stronger effect on smaller aggregates (< 63 μm) than on the larger aggregates (> 250 μm). These microaggregates rely on persistent TOC–mineral interactions (e.g. clay–humus complexes, microbial necromass bound to Fe/Al oxides) compared to the larger aggregates (≈ 250 μm) which are held together more by temporary physical agents like roots, hyphae and fresh organic matter. New text.

Interestingly, OG (+ 15% in OG vs. C site) also improved the stability of soil µWSA (< 63 μm) (Fig. 3, high effect size, Supplementary Table 1) increasing from 74% to 85% compared to the C site, while MSWA (> 250 μm) stability, remained constant between sites (67–69%) (Fig. 3). We expected the increased TOC in the OG site to be the cause for the increased µWSA stability as these microaggregates rely on persistent TOC–mineral interactions (e.g. clay–humus complexes, microbial necromass bound to Fe/Al oxides) compared to the larger aggregates (> 250 μm) which are held together more by temporary physical agents like roots, hyphae and fresh organic matter^38–40. However, the no correlation was found between TOC and µWSA (Supplementary Figs. 1 and 2).

The available elemental content of the soil CaCl₂ extract in the OG site did not increase significantly (Table 2). While changes in soil metal concentrations are not expected, P and K may increase as they are commonly returned to the soil through animal dung and urine^15,41,42. The mobility of P in soil is generally very low, while K, although somewhat more mobile than P, remains relatively limited. This is because both elements are retained in the soil: P through strong sorption to Fe- and Al-(hydr)oxides as well as Ca compounds, and K primarily through weaker electrostatic interactions on cation exchange sites. In addition, both elements can also be bound to soil organic matter⁴³. In this study, we applied a CaCl₂ extraction (pH extract), which primarily reflects the immediately soluble and mobile fractions of these elements. In contrast, organically bound elements are not directly bioavailable. The observed 15% increase in TOC (Table 1) likely enhanced the retention of these elements within organic matter, thereby contributing to their stabilization and increasing their potential for long-term storage.

Our last hypothesis (H3) was that oxen grazing would significantly increase both fungal and bacterial abundance. However, microbial biomass did not show any significant differences between treatments (Fig. 4). Differently, qPCR analyses revealed an increase in bacterial abundance in grazed plots compared to the control (medium effect size, Supplementary Table 1), while fungal abundance remained stable across the different sampling sites (Fig. 4). The contrasting results between microbial biomass and qPCR are not surprising. qPCR is more sensitive to specific variations in gene targets (i.e., 16 S rRNA), whereas microbial biomass measurements are less sensitive and require significant changes to detect differences. Also, microbial biomass is not taxon-specific because it measures the whole microbial community in soil, including bacteria, fungi, unicellular algae and protozoa⁴⁴. The bacterial and fungal abundance may reflect differences in their ecological niches. The increase in µWSA underpins this observation as smaller aggregates (< 63 μm) tend to be dominated by bacterial communities, while larger aggregates (> 250 μm) are often stabilized by filamentous fungi⁴⁵. Within these µWSA, bacterial necro mass and extracellular polymers play a central role in forming persistent organo–mineral complexes that ensure long-term stability⁴⁰. Further confirmation of this is that the available elements have not changed (Table 2). The absence of changes in fungal abundance could also be due to the timing of sampling, or to the resilience of fungal communities. In vineyard soils, long-established root systems and consistent OM inputs from vine residues may maintain a relatively stable fungal population^46,47. Additionally, fungi, especially saprotrophic and mycorrhizal species, tend to respond more slowly to organic matter inputs as they rely on more recalcitrant C sources^47,48. On the contrary, bacteria respond more rapidly to changes in soluble substrates (increase in DOC, DC, Fig. 2) and thus exhibit increased abundance^49,50. Other studies on this subject are scarce. Only two studies conducted similar experiments. In vineyards, a long-term study with sheep grazing reported increases of 39% in bacterial microbial biomass and 65% in fungal biomass³⁷. In contrast, a short-term vineyard trial, again with sheep, found that grazing primarily affected microbial functioning rather than overall microbial abundance¹⁰.

A potential drawback of integrating cattle into vineyards is the risk of damage to vines and their root systems through browsing, trampling and associated soil⁵¹. In our case study, the farmer reported no visible damage to trunks, canopies or root zones nor negative effects about yield or grapevine quality. This is likely because oxen were kept at relatively low density and excluded from the vineyard during grape ripening (May-September). Nevertheless, under these conservative conditions, we cannot draw firm conclusions about whether grazing by oxen at higher stocking densities or during sensitive phenological stages would negatively affect grape yield or quality.

Our experiment was conducted on a single small-scale farm and was explicitly designed to assess soil health responses rather than the logistical or economic feasibility of upscaling this management model. Nevertheless, analogous integrated crop–livestock–forestry systems have already been implemented at commercial scales in regions such as the Brazilian Cerrado and Amazon, where they can match or outperform conventional large-scale monocultures in terms of economic performance while providing soil and environmental co-benefits⁵². Given the structural and economic specificities of viticulture, our results should therefore be interpreted as a proof of concept at a local farm scale, and future multi-farm, regionally diverse studies will be required to determine whether similar livestock-based management can be effectively and safely implemented in large-scale vineyard systems.

Conclusions

Using agricultural land for multiple purposes in parallel (e.g. combining agricultural production with livestock) could be considered a promising strategy for promoting efficient agroecology, especially given limited soil availability. In this context, our study aimed to assess the impact of oxen grazing in a vineyard, exploring its potential benefits for soil health while simultaneously enhancing land-use efficiency by integrating grape production with pasturing. Our findings indicate that oxen grazing increased soil organic matter, in particular TOC, TN, C/N, DOC and DC. Surprisingly, soil bulk density did not change, while soil structure even improved as evidenced by the increase of µWSA in the oxen-grazed site. This observation is supported by the increase in bacterial abundance as they are typically present in µWSA (Supplementary Fig. 3), while MWSA and fungal abundance, together with microbial biomass, remained stable across the two sites. While our assessment focused on fungal and bacterial abundance as broad indicators of microbial biomass, future studies integrating microbial community structure, necromass, and enzymatic activities will be essential to further elucidate the mechanisms linking grazing management to soil carbon stabilization in alpine vineyards.

Our results emphasise the potential benefits of integrating viticulture with pasture to improve soil health and land-use efficiency, when properly managed by selecting suitable crops (such as vines) and maintaining appropriate grazing densities, without any detectable drawback in our selected parameters. This approach is particularly relevant in the region of South Tyrol (Italy), where both vast amounts of viticulture and livestock farming are already present. The integration and cooperation of both types of agricultural business could significantly improve sustainable agriculture for an agroecological transition.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

I.F. conducted the experiments and contributed to manuscript writing. E.T. carried out the experiments and field sampling. R.T. and O.B. contributed to manuscript preparation. G.L. critically reviewed the manuscript. L.B. and T.M. supervised the project and provided manuscript revisions.

Funding

No external funding was used for this study.

Data availability

The raw data supporting this study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.