Impact of Organic and Conventional Agricultural Management on Subsurface Soil Microbiota in Mediterranean Vineyards

Abstract

The impact of long-term organic (ECO) versus conventional (CON) agricultural management on subsurface soil microbiota diversity and soil physicochemical properties remains unclear in Mediterranean vineyards. This study evaluated long-term ECO and CON effects in the Alt Penedès terroir (Spain), focusing on subsurface soil microbial diversity and soil characteristics. ECO increased the fungal-to-bacterial ratio and ammonium-oxidizing bacteria but reduced total subsurface soil bacterial populations and soil organic carbon. While ECO did not enhance annual yield production in the vineyard, fungal abundance, and ammonium-oxidizing archaea, it slightly increased the overall alpha diversity (Shannon and Inverse Simpson indexes) and significantly altered taxa composition in subsurface soil with a more robust and modular community. Crop management, soil texture, training system, and rootstock, but not vine variety, significantly influenced beta diversity in subsurface soil. The Mantel test revealed subsurface soil texture, Ca²⁺/Mg²⁺ ratio, and salinity as the main key soil drivers shifting the microbial community (beta diversity), while C/N and topsoil organic matter significantly correlated with bacterial abundance; NH₄⁺ correlated with fungal abundance; and N-Kjeldahl, pH, and Mg²⁺/K⁺ correlated with alpha diversity. Integrating soil microbiota and physicochemical monitoring allowed us to confirm the positive effect of long-term agroecological practices on subsurface soil health and to identify the critical factors shaping their microbial communities in Mediterranean vineyards.

1. Introduction

Agriculture plays an ambiguous role in planetary sustainability since it is a major factor in global climate change and biodiversity [1]. To that end, the Food and Agriculture Organization (FAO) is currently promoting agroecology as an alternative practice for food and farming that can potentially tackle multiple crises in the agri-food system, contribute to Sustainable Development Goals, counteract climate change, and meet the world’s demand for food [2]. The promotion of agroecology is also crucial in the Mediterranean, where the negative environmental and social impacts of agriculture are being particularly strongly felt under climate change. Under Mediterranean conditions, in a general sense, soil erosion is the main cause of the decline in soil fertility since soil losses severely reduce organic carbon and water storage capacity [3]. Increasing the generally low carbon content of Mediterranean soils is an important greenhouse gas mitigation strategy and is a priority for preventing erosion and improving soil fertility. Significant carbon sequestration rates have been observed after the application of recommended practices and organic management in Mediterranean cropping systems [4].

The concept of terroir refers to a specific area where the interaction between physical and biological environments and vitivinicultural practices gives products their distinctive characteristics [5]. It encompasses soil, topography, climate, landscape, biodiversity, plant material, variety, and rootstock [6]. Over the past 30 years, the impact of soil and crop quality on terroir has been extensively studied, particularly in relation to grapevine varieties in Mediterranean wine production [7]. Other well-documented factors include rootstock, vine microbiome, water availability, climate, soil properties (including microbiome), and viticultural–oenological practices [8,9]. However, the role of agroecological practices in enhancing the efficiency and resilience of Mediterranean vineyards under climate change remains not fully understood [10].

Soil microbiomes play essential roles in agroecosystems by influencing soil fertility, crop productivity, and stress tolerance [11]. Growing interest in the role of soil microbial biodiversity in vineyard terroir is reflected in initiatives by the International Organization of Vine and Wine [12]. The potential role of soil microbiota in the terroir and its linkage to the physicochemical properties (i.e., soil organic carbon, soil carbon storage, and nutrient cycling) has only been reported recently, and mainly at the topsoil zone (first 20 cm of soil) [13,14,15,16,17]. In addition, other studies have been reported focusing on the effects of different crop management and fertilization strategies, including assessments of soil microbiota, the majority of which are focused on the topsoil [18,19,20]. Factors like organic matter, oxygen levels, and temperature fluctuations significantly affect microbial diversity in the topsoil. However, the subsurface soil layer (20–40 cm), close to the rhizosphere, is often overlooked despite its potential importance, as microbial diversity sources at this depth may play a significant role in plant health, root interactions, and also in wine quality and terroir.

The subsurface soil microbiota in agricultural soils can interact with the root system of crops and, in the case of vineyards, influence the final wine terroir and soil ecosystem services, including soil C sequestration. The subsurface zone (below 20 cm of depth) seems to encompass a more stable environment that seems to be more appropriate for assessing the real long-term effects of agricultural management on physicochemical soil parameters and soil microbiota, i.e., the key drivers that provide soil ecosystem services. Salomé et al. [21] reported that the driving forces affecting C dynamics are different in the topsoil and subsoil (subsurface soil), the organic carbon of subsoil being more unstable and mineralized under soil disruption than in the topsoil, as previously described in a global study [22], as well as in Mediterranean agroecosystems [23]. Therefore, special attention should be paid to how the subsurface soil reacts under different long-term techniques of soil management as a means of improving our knowledge of global C dynamics in agricultural soils. To fully harness the regenerative potential of vineyard soils and improve critical ecosystem services, such as carbon sequestration and nutrient availability, and biocontrol strategies, it is essential to investigate how long-term agroecological practices affect subsurface microbial diversity and its potential functions. Soil carbon storage capacity has recently been highly related to the diversity and function of microbial communities and even to the microbial necromass; thus, it holds similar importance to the mineralization of existing labile organic matter [24]. In vineyards, a perennial system with deep-rooting woody plants, soil microorganisms rely heavily on root-derived carbon (C) inputs such as root biomass and exudates, which not only sustain microbial growth but also shape the composition and function of soil microbiota, particularly in the rhizosphere [25,26]. These biological processes also enhance microbial biomass and activity in the subsurface soil zones, contributing to carbon accumulation and soil health [24]. Subsurface soil, characterized by high mineral content, lower disturbance, and more stable organic matter inputs, provides favorable conditions for storing long-term, stabilized carbon while supporting a relatively stable and diverse microbial community [27]. Importantly, both topsoil and subsurface soil serve as significant sources of nutrients and carbon sinks due to their ability to stabilize organic carbon inputs, including roots, decaying plant matter, and microbial necromass [28,29,30]. Despite this, the impact of agricultural practices on the subsurface soil’s organic carbon and microbial diversity in temperate agroecosystems is scarcely known.

The primary objective of this study was to gain insights into the effects of different vineyard management practices on subsurface soil bacterial diversity and soil organic carbon (SOC) within the terroir of the Alt Penedès (Northeastern Spain) in commercial vineyards. The specific aims were to (i) assess the long-term impact (>10 years) of organic (ECO) and conventional (CON) agricultural practices on bacterial diversity, taxonomic composition, the abundance of bacterial and fungal communities, and populations of ammonium-oxidizing microorganisms in vineyard subsurface soil; (ii) determine the main physicochemical soil parameters affected by long-term ECO and CON practices, in temperate commercial vineyards; and (iii) identify the key soil and subsurface soil physicochemical factors that drive microbial community structure and diversity in these vineyard systems.

2. Materials and Methods

2.1. Case Study

As a field case study to assess the impact of vineyard practices on soil quality and microbial diversity, six vineyard plots (three organic and three conventional) were selected within a 3 km radius of Sant Sadurní d’Anoia, in the Alt Penedès region, a Mediterranean terroir shared by DO-Penedès, DO-Cava, and DO-Catalunya in Catalonia (Spain). The whole area lies on marls and calcareous sandstones, which create mainly basic soils. The climate in the Alt Penedès is typically Mediterranean. The annual precipitation is around 550 mm, higher in fall and lower in summer. Mean daily temperatures are about 7 °C in winter and 24 °C in summer, without frost occurring between May and October. From 1950 to the present, precipitation has decreased by 13%, but when looking specifically at summer, it has decreased by 25% in the last 70 years. The mean annual temperatures have increased by 0.27 °C per decade, while the mean temperature in summer has increased by 0.4 °C per decade, resulting in an increase of already over 2.5 °C. In 2020, Alt Penedès had 17,310 ha of vineyards, with 70% under organic management, leading the way in terms of the surface area of ecological vineyards (Consell Català de la Producció Agrària Ecològica). The dominant grape varieties, Macabeu, Xarel·lo, and Parellada, make up 53.2% of the cultivated area.

Five sampling points were selected per plot (three organic and three conventional plots) using aerial imagery (Figure 1).

All the vineyards had been continuously managed in a specific way for at least 10 years (

Table 1. Characteristics and general practices for conventional (CON) and organic (ECO) managed vineyards. Data were obtained from the owner of each plot for the last four years (2017–2020).

Characteristics and Practices	Plots
	CON 1	CON 2	CON 3	ECO 3	ECO 8	ECO 12
Surface (ha)	2.5	0.99	2.12	1.8	1.14	0.67
Age (years)	25	25	11	14	26	33
Variety	Xarel·lo	Macabeu	Xarel·lo	Xarel·lo	Macabeu	Xarel·lo
Rootstock	110R	110R	SO4	110R	110R	161-49C
Vine spacing (m)	2.8 × 1.2	3 × 1.4	2.8 × 1.2	2.6 × 1.20	3 × 1.2	2.6 × 1.2
Training system	Doble cordon	Doble cordon	Doble cordon	Single cordon	Doble cordon	Gobelet
Continued management (years)	25	25	11	10	16	10
Yield (kg·ha−1)	6548	8768	8697	8716	8079	9194
Tillage	Always	Always	Always	Combined with mowing or grazing	Combined with mowing or grazing	Combined with mowing or grazing
Cover crop	No	No	No	Temporary (spontaneous) [31]	Temporary (spontaneous) [31]	Temporary (spontaneous) [31]
Fertilizer 1	Organo-mineral Fertilizer (4-6-10) 2 13 kg N ha−1 year−1 8.5 kg P ha−1 year−1 27 kg K ha−1 year−1	Organo-mineral Fertilizer (4-6-10) 2 13 kg N ha−1 year−1 8.5 kg P ha−1 year−1 27 kg K ha−1 year−1.	Organo-mineral Fertilizer (4-6-10) 2 13 kg N ha−1 year−1 8.5 kg P ha−1 year−1 27 kg K ha−1 year−1	Composted cow manure and biodynamic amendments 10 kg N ha−1 year−1 10 kg P ha−1 year−1 20 kg K ha−1 year−1	Composted cow manure and biodynamic amendments 10 kg N ha−1 year−1 10 kg P ha−1 year−1 20 kg K ha−1 year−1	Composted cow manure and biodynamic amendments 10 kg N ha−1 year−1 10 kg P ha−1 year−1 20 kg K ha−1 year−1

¹ The nitrogen load (kg N ha⁻¹ year⁻¹) is calculated based on the average annual value during the previous three years preceding the present study. ² Organo-mineral fertilizer (commercial formulation 4-6-10) consisted of 4% (w/w) nitrogen (N), 6% (w/w) phosphorus pentoxide (P₂O₅, equivalent to 2.61% P), 10% (w/w) potassium oxide (K₂O, equivalent to 8.3% K), and 30% (w/w) of composted organic carbon. Composted cow manure composition comprised 0.5%N, 0.5%P, and 1% K.

). Except for ECO 12, the oldest plot, all vineyards were trained in cordon. Yields were neither high nor significantly different between management types (ECO: 8663 ± 559 kg·ha⁻¹ vs. CON: 8004 ± 1261 kg·ha⁻¹; p = 0.445) (Table 1). However, it should be taken into account that winter (204 mm) and spring (281 mm) in 2020 were both very rainy compared to the mean rainfall (112 mm and 137 mm, respectively, for winter and spring averages 2011–2022), which led to great downy mildew pressure and resulted in a lower-than-expected yield across the terroir. The ECO vineyards use a combination of cover crops, mowing, and summer tillage, with occasional grazing or horse-powered tillage. Fertilizer is organo-mineral in the CON vineyards and composted cow manure in the ECO vineyards. Fertilization strategies in both management systems were implemented at equivalent annual nitrogen load (13 kg N/ha/year for CON vineyards and 10 kg N/ha/year for ECO vineyards). Pruning wood was left on the ground and incorporated into the soil with the first tillage of the year under both management types.

The soil texture of the plots ranged from silty clayey loam to sandy loam (a usual heterogeneity in agricultural soils in commercial plots), with a total of 59% of composite samples within the loam texture range (

Table 2. Subsurface soil texture from organic (ECO) and conventional (CON) vineyard plots. Values represent the mean and standard error from five composite samples per plot, across six plots.

	Plot		Soil Texture
		Soil Texture (USDA)	Clay (%)	Silt (%)	Sand (%)
Conventional	CON_1	Loam	17.82 ± 1.42	41.66 ± 4.63	40.52 ± 5.70
	CON_2	Loam	20.76 ± 1.40	38.94 ± 3.44	40.30 ± 4.57
	CON_3	Sandy Loam	13.98 ± 2.29	29.00 ± 4.90	57.02 ± 7.10
Organic	ECO_3	Loam	18.96 ± 2.45	36.88 ± 1.79	44.16 ± 4.22
	ECO_8	Silty Clay Loam	28.32 ± 1.64	49.44 ± 4.01	22.24 ± 5.21
	ECO 12	Clay Loam	25.10 ± 2.74	39.38 ± 3.87	35.52 ± 5.87

Linear mixed models revealed non-significant differences between CON vs. ECO soil texture plots as previously described [31] (see details in Table S5).

and Figure S1). The soil texture was loam in CON 1, CON 2, and ECO 3 plots; sandy loam in CON 3; silty clay loam in ECO 8; and clay loam in ECO 12 (Table 2).

2.2. Soil Sample Collection

At each selected point (five selected points per plot (Figure 1)), four soil subsamples were collected around a vine (40–80 cm from the trunk in the N–S and E–W directions) using a manual auger with a 5 cm diameter and a 14 cm long cylindrical drill (Eijkelkamp). The first core was taken from 0 to −20 cm depth, with the upper 5 cm discarded and the −5 to −20 cm portion retained. From the same hole, a second core was then extracted to collect soil from −20 to −40 cm depth. The four subsamples from each depth were combined to create separate composite samples for the topsoil (−5/−20 cm) and subsurface soil (−20/−40 cm). For the −20/−40 cm composite, a subsample was placed in a 50 mL tube, kept at 4–8 °C during sampling, and then frozen at −80 °C. The remaining soil from both depths was sealed in plastic bags, air-dried, and stored at room temperature. Both −5/−20 cm and −20/−40 cm samples were analyzed for soil organic carbon (SOC) and oxidizable organic matter (OOM).

2.3. Physicochemical Assessment of Soil Samples

The soil samples were analyzed following standard reference guidelines: soil texture (USDA discontinuous sedimentation of four fractions) was determined by gravimetry; SOC (% of dry matter) and oxidizable organic matter (% of dry matter) by potentiometric titration (ISO 14235:1998) [32]; total nitrogen Kjeldahl (% dry matter) by volumetry (ISO 11261:1995) [33]; and total N-NO₃ and N-NH₄, and phosphorus Olsen by UV-VIS spectrophotometry following ISO/TS 14256-1:2003 [34] and ISO 11263:1994 [35], respectively. Calcium, magnesium, potassium, and sodium were extracted with ammonium acetate and analyzed by inductively coupled plasma spectrometry (ICP-AES). pH (1:2.5) and electrical conductivity (1:5) were determined by potentiometry and conductimetry following ISO 10390:2021 [36] and ISO 11265:1994 [37], respectively. Total sulfur (S), sulfate (SO₄), iron (Fe), copper (Cu), manganese (Mn), zinc (Zn), and molybdenum (Mo) were extracted with EDTA and analyzed using inductively coupled plasma spectrometry (ICP-OES).

2.4. Microbial Community Assessment: Microbial Population Abundance

DNA extraction of soil material (0.5 g) from each composite sample was performed using a DNeasy PowerSoil Pro kit (Qiagen, Hilden, Germany). Gene copy numbers of 16S rRNA (total bacteria), ITS1 rRNA (total fungi), ammonium-oxidizing bacteria (amoA gene of AOB), and archaea (amoA gene of AOA) were quantified by real-time qPCR of each DNA extract (15 for CON and 15 for ECO managed vineyards). The analyses were carried out using Brilliant II SYBR ^®Green qPCR Master Mix (Agilent Technologies, Madrid, Spain) in a real-time PCR system MX3000-P, as described elsewhere [38,39]. For the standard curve of each target gene, four gBlocks^® Gene Fragments were designed. Ten-fold serial dilutions from synthetic genes were subjected to qPCR assays in duplicate with a linear range between 10¹ and 10⁸ gene copy numbers per reaction to generate standard curves. qPCR reactions fitted quality standards, efficiencies were between 90% and 110%, and R² was above 0.985. All results were processed by MxPro™ QPCR software (version 3.0.0, Strategene, La Jolla, CA, USA) and treated statistically.

2.5. Microbial Community Assessment: 16S rRNA-Metabarcoding and Bioinformatics

Briefly, 16S rRNA paired-end amplicon sequencing targeting the V3–V4 region was performed in 30 soil samples using Illumina MiSeq (2 × 300 bp) at Molecular Research DNA facilities. Amplification used primers 341F and 805R. Raw paired-end FASTQ files were processed with Cutadapt in QIIME2 to remove primers and then exported to RStudio (version 4.2.2) and analyzed with the DADA2 package [40]. Sequences were quality-filtered (truncLen: 260 for R1, 240 for R2), denoised, merged (minimum 12 bp overlap), and screened to discard reads with ambiguities or expected errors >2 (maxEE). Chimeras were removed using the removeBimeraDenovo function, and amplicon sequence variants (ASVs) were inferred based on DADA2’s error model. Taxonomic assignment of ASVs was performed using the naïve Bayesian method against the SILVA 138 database, with an 80% bootstrap confidence threshold.

Alpha diversity indices (Shannon, Inverse Simpson, Richness, and Chao1) were calculated from rarefied samples (6393 reads/sample) using Mothur. Community composition and relative abundance of main bacterial groups were visualized using MicrobiomeAnalyst and the Phyloseq (version 3.16) on RStudio (version 4.2.2).

Beta diversity was assessed by calculating Bray–Curtis dissimilarity from rarefied and normalized ASV tables (total sum scaling), and ordinated using principal coordinates analysis (PCoA). The effects of crop management practices on community structure were tested using PERMANOVA (adonis2) and ANOSIM (anosim), both with 999 permutations in the Vegan (version 2.6) on RStudio (version 4.2.2).

Differential abundance analysis was conducted on rarefied ASVs using the Wilcoxon rank-sum test (Mann–Whitney), with p-values adjusted by false discovery rate (FDR). ASVs with >5 reads in at least 20% of samples were retained. Analyses were carried out using MicrobiomeAnalyst and MicroEco (version 1.7.0) R packages. To identify the biomarkers associated with ecological (ECO) and conventional (CON) vineyard management, linear discriminant analysis effect size (LEfSe) [41] was implemented through Microeco [42]. Taxa showing significant differences in relative abundance were detected using the Kruskal–Wallis test, followed by Wilcoxon rank-sum tests. Taxa with logarithmic LDA scores ≥2.0 were selected as biomarkers.

Functional prediction of prokaryotic traits was performed using FAPROTAX [36] implemented through Microeco [43], by matching ASVs with metabolic functions based on curated databases of biogeochemical roles.

Main environmental drivers (Euclidean distances) that better correlate with bacterial community composition (Bray–Curtis matrix at ASV level) were determined by means of Mantel tests. Matrix values were vectorized and tested using Spearman’s correlation and 9999 permutations in the Vegan (version 2.6) R Package.

Microbial network interactions and the structural robustness of subsurface soil bacterial communities were assessed by means of co-occurrence network analysis [44], comparing both crop management systems. Correlation matrices were generated using Hmisc (version 5.2-3), and microbial networks were built and analyzed with igraph (version 1.3.5) and tidygraph (version 1.3.1), calculating properties such as node/edge count, modularity (Louvain algorithm), clustering coefficient, and global path length. Network robustness was tested via simulated random and targeted node removal, and visualizations were created using ggraph (version 2.2.1) and ggplot2 (version 3.5.2).

2.6. Statistical Analysis

All statistical analyses were performed using R version 4.2.2. To assess the effect of crop management (organic: ECO vs. conventional: CON) on soil physicochemical parameters, one-way ANOVAs were conducted (stats in RStudio 4,2.2), with management type as the independent variable. Residuals were visually inspected, and heteroscedasticity was addressed using the ncvTest function from the car version 3.1.3) R package.

Microbial abundance (log₁₀-transformed qPCR data), gene population ratios (ITS:16S rRNA, amoA-AOB:amoA-AOA), and alpha diversity indices (Sobs, Chao1, Shannon, Inverse Simpson, Reads, and Coverage) were tested for normality and homogeneity of variance using the Shapiro–Wilk and Levene’s tests (p > 0.05). When assumptions were met, parametric tests (one-way ANOVA, post hoc Tukey’s test, and t-tests for pairwise comparisons) were applied at a 5% significance level. Otherwise, non-parametric tests (Kruskal–Wallis and Mann–Whitney U) were used. Associations between physicochemical and microbial variables (abundance and diversity) were assessed via Spearman’s rank correlation using the rcorr function from the Hmisc R package (version 5.2-3). Corresponding p-values were calculated using the corr.test function (from the psych package (version 2.4.3), with FDR correction applied. A correlation matrix plot was generated using the ggcorrplot R package (version 0.1.4.1).

3. Results and Discussion

3.1. Impact of Crop Management on the Subsurface Soil Physicochemical Soil Parameters

The results are summarized in

Table 3. Physicochemical and microbial population abundance parameters of subsurface soils (−20/−40 cm) comparing organic (ECO) and conventional (CON) management systems. p-values are from linear models, The number of degrees of freedom is 28. Mantel test results are also included for each variable.

Parameters	CON (n = 15) Estimate ± SE	ECO (n = 15) Estimate ± SE	F-Statistic	p-Value	Mantel Test Statistic	Mantel Test (p-Values)
Clay (%)	17.50 ± 1.42 a	24.10 ± 2.01 b	10.770	0.003	0.405	0.000
Ca2+/Mg2+	46.44 ± 3.59 a	26.42 ± 5.07 b	15.570	0.000	0.371	0.000
EC (dS/m)	0.15 ± 0.00 a	0.16 ± 0.00 b	5.271	0.029	0.368	0.001
Sand (%)	45.90 ± 3.72 a	34.00 ± 5.26 b	5.190	0.031	0.353	0.000
K+ (mg/kg)	153.53 ± 21.48 a	381.00 ± 30.38 b	5.890	0.022	0.264	0.005
Mg2+ (mg/kg)	149.53 ± 35.53 a	342.00 ± 50.25 b	14.680	0.001	0.239	0.020
SOC Ratio (5–20)/(20–40)	0.99 ± 0.09 a	1.61 ± 0.13 b	22.200	0.000	0.216	0.001
Ca2+/K+	57.1 ± 6.82 a	33.00 ± 9.64 b	6.246	0.019	0.209	0.002
OOM Ratio (5–20)/(20–40)	1.07 ± 0.08 a	1.64 ± 0.12 b	23.760	0.000	0.201	0.007
C/N	8.33 ± 0.42 a	5.59 ± 0.59 b	21.850	0.000	0.181	0.008
OOM 20–40 cm (%)	1.04 ± 0.11 a	0.73 ± 0.11 b	4.574	0.041	0.149	0.045
Diff SOC (%) (top-sub)	−0.04 ± 0.04 a	0.24 ± 0.05 b	31.090	0.000	0.141	0.051
Diff OOM (%) (top-sub)	−0.02 ± 0.06 a	0.44 ± 0.07 b	31.100	0.000	0.141	0.039
Na (mg/kg)	16.30 ± 1.66 a	23.10 ± 2.35 b	8.378	0.007	0.140	0.111
Fe (mg/kg)	74.50 ± 5.24 a	35.10 ± 7.41 b	28.150	0.000	0.121	0.114
SOC 20–40 cm (%)	0.61 ± 0.09 a	0.42 ± 0.06 b	4.625	0.040	0.118	0.099
NO3− (mg/kg)	3.67 ± 0.46 a	2.29 ± 0.57 b	5.474	0.027	0.096	0.194
Cu (mg/kg)	19.40 ± 2.07 a	12.50 ± 2.93 b	5.612	0.025	0.092	0.134
Zn (mg/kg)	3.47 ± 0.30 a	2.47 ± 0.39 b	6.275	0.018	0.045	0.270
NH4+ (mg/kg)	4.24 ± 0.30 a	5.90 ± 0.42 b	15.710	0.000	−0.019	0.565
S (mg/kg)	352.45 ± 58.33 a	186.00 ± 82.49 a	4.054	0.054	0.140	0.117
SO4 (mg/kg)	22.13 ± 5.43 a	37.20 ± 7.69 a	3.843	0.060	0.204	0.038
Silt (%)	36.50 ± 2.54 a	41.90 ± 3.60 a	2.225	0.147	0.405	0.000
Mo (mg/kg)	0.10 ± 0.00 a	0.11 ± 0.01 a	1.312	0.262	0.157	0.051
Mg2+/K+	1.26 ± 0.21 a	1.55 ± 0.29 a	1.021	0.321	0.297	0.002
Ca2+ (mg/kg)	6591.00 ± 105.92 a	6451.00 ± 149.80 a	0.868	0.360	0.232	0.002
P Olsen (mg/kg)	13.95 ± 1.62 a	14.97 ± 2.29 a	0.198	0.660	0.186	0.009
Mn (mg/kg)	83.70 ± 10.10 a	85.40 ± 14.28 a	0.015	0.904	0.128	0.070
pH	8.43 ± 0.02 a	8.43 ± 0.03 a	0.000	1.000	0.264	0.118
N-Kjeldahl (%)	0.08 ± 0.01 a	0.08 ± 0.01 a	0.256	0.617	0.067	0.2291
EUB16S rRNA (log10 gene copies/g)	9.48 ± 0.07 a	9.24 ± 0.08 b	5.6863	0.024	0.136	0.080
FUNG ITS (log10 gene copies/g)	8.33 ± 0.08 a	8.54 ± 0.07 a	3.9513	0.057	0.198	0.007
ITS/16S rRNA	0.09 ± 0.02 a	0.21 ± 0.02 b	11.582	<0.001	0.085	0.214
AOB (log amoA gene copies/g)	5.47 ± 0.14 a	5.84 ± 0.09 b	5.0963	0.041	0.182	0.033
AOA (log amoA gene copies/g)	5.06 ± 0.10 a	5.27 ± 0.08 a	3.0582	0.080	0.149	0.046
AOB/AOA	3.03 ± 0.49 a	4.72 ± 0.95 a	1.3642	0.243	0.113	0.145

Different letters represent significant pairwise differences between treatments within each parameter (Tukey’s HSD p < 0.05). Mann–Whitney rank-sum test (U); Kruskal–Wallis one-way analysis of variance (H value); one-way ANOVA. Mantel test values indicating the correlation of soil variables with changes in soil microbial beta diversity (ASV distribution) are included.

. Regarding the effects of crop management on soil organic carbon (SOC) and oxidized soil organic matter (OOM), topsoil and subsurface soil were also assessed. The SOC and OOM contents of the subsurface soil were both lower (p < 0.05) in ECO (0.73 and 0.42%, respectively) than in CON (1.04 and 0.61%, respectively), with values that are frequently found in this terroir [23]. Interestingly, no increase in SOC or OOM was observed in the topsoil compared to the subsurface soil in CON (SOC 5–20 cm: 0.59% and OOM 5–20 cm: 1.02%), whereas in ECO, higher values of SOC and OOM were observed in the topsoil (SOC 5–20 cm: 0.64% and OOM 5–20 cm: 1.10%). However, a high heterogeneity in the topsoil was observed (OOM 5–20 cm 1.02% in CON vs. 1.10% in ECO, p = 0.44 and SOC 5–20 cm: 0.59% in CON vs. 0.64% in ECO, p = 0.36), when compared with subsurface soil depth, where significantly lower contents of SOC and OOM (p = 0.040 and 0.041, respectively) were observed in ECO management (SOC 0.42% and OOM 0.73%) when compared with CON management (SOC: 0.61% and OOM: 1.04%) (Table 3). These differences could indicate that the greater organic matter amendment in ECO remains detectable in the topsoil (where the organic matter is mostly immobilized in the case of loam soils) but is less detectable in the subsurface soil, which suggests that the edible amended organic matter that reaches the lower soil surface is completely degraded due to the existence of a more metabolically active microbiota, as also described in [18]. The proximity of root systems and metabolites from root exudates and a significantly lower C/N ratio in ECO, with higher N-NH4, could ensure that the microbial activity remains higher in the subsurface soil under ECO management, leading to a decrease in total SOC. The stabilized organic carbon fraction, as the mineral-associated organic matter (MAOM) in total SOC, should be further studied to ascertain the long-term effect of ECO management on the accumulation of stable carbon fractions in both the top- and subsurface soils [45]. Thus, ECO significantly increases the organic matter in the topsoil, which suggests that this type of management practice contributes to the recovery of organic matter in the topsoil, improving carbon sequestration in the topsoil but not in the subsurface soil, whereas under CON, these practices contribute to a reduction in the SOC and OOM in the topsoil.

The total nitrogen content (N-Kjeldahl) was not high and did not show any differences between management types (p > 0.05). Nevertheless, when examining the sources of nitrogen, CON had a higher N content of nitrate origin and ECO of ammonium origin, a finding related to the fertilizers applied under these respective management types (Table 3). The C/N ratio was significantly higher (p < 0.05) in CON (8.63%) than in ECO (5.59%), but there were no significant differences (p > 0.05) in phosphorus, sulfur, sulfates, calcium (high due to the calcareous lithology), manganese, molybdenum, or pH (which was moderately basic around 8.4). Potassium, magnesium, sodium, and electric conductivity were higher (p < 0.05) in ECO than in CON, whereas copper, iron, and zinc were higher (p < 0.05) in CON.

Even though there was variability between plots (Table 3), the differences between management types were clear and reflected some of the expected results in terms of the accumulation of organic matter at the surface and the source of nitrogen. Interestingly, SOC and OOM in the subsurface soil were lower in ECO (p < 0.05), which could be related to the greater capacity of soil microbiota to transform labile organic matter, such as root exudates, in subsurface depth soil under ECO management, as described by [46]. Root exudates have been reported to be key elements in certain ecosystem functions such as nutrient availability, drying–wetting events [47], and carbon cycling. They act as an organic substrate source for subsequent carbon sequestration and the generation of soil aggregate processes in the subsurface soil [48,49]. Agroecological farming practices often involve the use of compost and other organic amendments, which can also help enhance the microbial activity in the soil, resulting in higher losses of edible SOM in the case of an unbalanced low C:N ratio. Although the quick degradation of soil organic matter may seem a disadvantage, it can, in fact, be beneficial for soil health. When organic matter is broken down quickly, it releases the nutrients that are available to plants, which can improve crop yields and soil microbiota metabolic activity. Additionally, the byproducts of organic matter microbial transformation, such as humic and fulvic acids, can help improve the soil structure, stabilize organic matter as mineral-associated organic matter (MAOM) in the mineral fraction, and increase the water-holding capacity [45]. In addition, in a recent study, a significant increase in soil organic carbon was observed mainly in the surface soil (not in the subsurface soil) in temperate vineyards using cover crops [50]. The effects of soil management could begin to have an impact on the subsurface soil organic carbon content within a decade of their implementation [22]. It is noteworthy that, in a recent long-term time-course assessment (35 years), only a significant increase in mineral-associated oxidable organic matter and organic carbon (MAOM and MAOC) was observed when continuous organic fertilization was implemented in tempered agricultural soils, when compared with synthetic fertilization, where a loss of SOC was revealed [51].

3.2. Impact of Crop Management on the Subsurface Soil Microbiota

The abundance rates of the main microbial populations (total bacteria, fungi, and ammonia-oxidizing bacteria (AOB) and archaea (AOA)) under ECO and CON management are shown in Figure 2 and Table 3. ECO decreased significantly the total bacterial populations (p = 0.024) but led to an increase in the ITS/16S ratio (p = 0.001) and the total AOB (p = 0.033).

A greater AOB abundance has been also described for topsoils managed with agroecological practices due to more N-NH₄⁺ availability after organic fertigation and higher nitrification rates and N_NH₄⁺ affinity (higher Km and Vmax of AOB compared to AOA); nevertheless, the higher ITS/16S ratios in ECO could be related to the greater inputs of lignocellulosic materials, plant litter and root exudates from cover crops, and a better soil structure due to fewer tillage events [52,53]. No significant effects of management were detected for total fungi (p = 0.057) or for ammonium-oxidizing archaea (AOA; p = 0.080). Interestingly, SOC (0.61–0.42%) and OOM (1.04–0.73%) were higher in CON (p = 0.040 and p = 0.041) than in ECO. This coincides with the greater total bacterial population in CON soils, which was not related to clay content (higher in ECO than in CON (p = 0.003)). The lower C content in ECO subsurface soils could be due to the higher metabolic rate in ECO soils, better adapted to greater inputs of organic matter with high C:N, which could promote high mineralization rates of organic matter. In addition, conventional practices lead to drier soils that may promote greater exudate production in root systems and have an impact at subsurface depth, as described elsewhere [52], which could be related to the larger bacterial populations detected in ECO subsurface soils. C accumulation in soils is dependent on a balance between fresh C inputs, microbial biomass turnover, and SOC respiration. Soil microbiota increases SOC respiration when organic carbon is obtained from organic fertigation, cover crops, and root exudates, which can boost the synthesis of chemically diverse and more stable SOM [54]. The observation that ECO reduces organic carbon in the subsurface soil but increases it in the upper layer is consistent with the hypothesis of accelerated mineralization due to microbial activity; however, a direct measurement of the stable carbon fraction (MAOM) should be conducted in further research to better understand carbon fractionation in SOC in the different soil depths during long-term time-course studies.

Our results confirmed the great abundance of bacteria and fungi in the subsurface zone, as well as the low prevalence of AOP (AOA + AOB). Interestingly, the AOA and AOB abundances in subsurface soil samples were low (0.004–0.1%), which contrasts with the high AOA abundance (1–10%) reported by [55] in topsoils. However, the low specificity of AOP primers used [56] for the Nitrososphaera genus could also explain the low relative abundance of AOA in the vineyard subsurface soil, as well as the effect of comparing different genes (16S rRNA vs. amoA) that could be represented in a different number of copies per cell. This suggests that, although the contents of SOC and OOM were relatively low (in the range of arid–semiarid soils), the use of organic materials could be due to surface crop management as well as to root exudates in the subsurface. The decrease in AOP and nitrification rates with soil depth has also been reported in agricultural soils due to the lower N-NH₄⁺ availability in the subsurface [57,58,59].

Regarding the effects of management on alpha bacterial diversity and the taxonomic composition of vineyard soils, a total of 2972 ASVs were identified in the samples, with 15,059 reads/sample for CON (n = 15) and 15,693 reads/sample for ECO (n = 14) (Table 3). To perform alpha and beta diversity assessments, rarefied read samples were used for downstream analysis. ECO significantly increased both the Inverse Simpson index (P_MW = 0.012) and the Shannon Index (P_MW = 0.029) (

Table 4. Reads, coverage, and alpha diversity indexes on rarefied samples (6393 reads) of soil bacterial populations comparing organic (ECO) and conventional (CON) management systems.

Indexes	Mean ECO (n = 14)	Mean CON (n = 15)	Mann–Whitney (ECO vs. CON p-Value)	Statistic (U)
Reads (contigs)	15,693 ± 3051	15,059 ± 6125	0.918	102
Coverage	0.899 ± 0.008	0.898 ± 0.003	0.913	102
Sobs (Richness)	1810 ± 132	1789 ± 65	0.285	80
Chao1 (Richness)	1845 ± 119	1790 ± 103	0.1225	94
Shannon (H) (Diversity)	6.80 ± 0.27 a	6.76 ± 0.06 b	* 0.029	62
Inverse Simpson (Diversity)	489.2 ± 88.1 a	390.3 ± 167.9 b	* 0.012	50

Rarefied indexes at 6393 reads. Mean values ± standard deviation (SD). * denotes significant differences between management systems (p < 0.05). Different letters represent significant pairwise differences between treatments within each parameter.

and Figure S2). However, the richness indexes (Chao1 and Sobs) were not significantly affected by management (P_MW = 0.1225 and P_MW = 0.285, respectively).

The significant low increase observed in the alpha diversity in ECO management (Shannon of 6.80 in ECO vs. 6.76 in CON; Inverse Simpson of 489 in ECO vs. 390 in CON) in the present study seems to confirm the great stability of the subsurface soil bacterial diversity in the vineyard soil under the same edaphoclimatic conditions even with different rootstocks, vine varieties, and spatial distances between commercial plots (Figure 1 and Table 1). The high stability of the bacterial diversity in soil vineyards under conventional tillage versus 2–8-year cover crops has also been reported by [8,55] in the 0–30 cm depth range. The low carbon sequestration observed by these authors, and also in our study, could be related to low C/N due to the leguminous plant species used in cover crops or edible organic matter from manure rich in N, which could decrease the optimum C/N ratio by boosting subsurface soil SOC mineralization [60].

A total of 13 phyla and 34 classes of bacterial taxa above 0.1% of relative abundance (RA) were identified (see Figure 3 for phyla and classes; Table S1 for classes). The dominant phyla above 5% RA present under both CON and ECO management systems were Proteobacteria (31–34%), Actinobacteria (34–30%), Acidobacteria (14–15%), Gemmatimonadetes (6–4%), and Bacteroidetes (5–6%). Only four phyla were differentially abundant: Proteobacteria (p = 0.010), Actinobacteria (p = 0.034), Gemmatimonadetes (p = 0.031), and Firmicutes (p = 0.031). ECO increased the number of Proteobacteria, whereas CON increased the Actinobacteria, Gemmatimonadetes, and Firmicutes (p < 0.05). The phylum composition was in a similar range and proportion to that recently described in other studies on soil vineyards in Spain [55].

Regarding beta diversity, a PCoA ordination of dissimilarity (Bray–Curtis distance at the ASV level) revealed that management helps differentiate the soil microbial communities at the ASV level (Figure 4). The ANOSIM and PERMANOVA assessments (

Table 5. ANOSIM assessment of the changing effect of different factors on soil bacterial beta diversity.

Parameter	R	p-Value
Management (CON vs. ECO)	0.3603	0.0001
Plot (all)	0.5483	0.0001
Plot in CON	0.448	0.0001
Plot in ECO	0.4651	0.0006
Variety (all)	−0.0748	0.8216
Variety in ECO	−0.04541	0.5687
Variety in CON	0.1047	0.2114
Soil Texture	0.3419	0.0022
Rootstock	0.3022	0.0061
Training system	0.2316	0.0356

) confirmed the significance of the effect of management on the microbial community structure (ANOSIM: p = 0.0001; PERMANOVA: p < 0.001). Grapevine variety did not influence the microbial community structure of the whole assessment (ANOSIM: p = 0.8216) (Table 5).

Under the same crop management, the use of different vine varieties did not result in any differentiation in the structure of the soil microbial populations (bacterial beta diversity) (Table 5). However, in each individual variety (Xarel·lo or Macabeu), the management type (ECO versus CON) did have a significant impact on the microbiota. Interestingly, a rootstock influence on bacterial soil beta diversity was observed, which confirms a recent study published by Darriaut et al. [61], in which these authors describe the influence of rootstocks on the alpha and beta diversity of the bacterial and fungal microbiota of the root compartment, although they failed to note differences in the bulk soil. The assessment of bacterial diversity in the subsurface soil (subsoil) could help identify the effect of root interaction with the bulk soil microbiota in deeper zones. It is worth noting that we were probably able to detect the influence of plots on beta diversity due to the physicochemical heterogeneity within each plot, despite being in the same edaphoclimatic region.

The main classes of taxa detected in the CON-ECO management were Actinobacteria (26–23%), α-Proteobacteria (19–22%), Acidobacteria (12%), Gemmatimonadetes (6–4%), and β-Proteobacteria (5%), followed by γ-Proteobacteria, Thermophilia, δ-Proteobacteria, Acidimicrobiia, Cytophagia, and Chitinophagia (above 2% RA). Of these, only Gemmatimonadetes (p-value = 0.007), γ-Proteobacteria (p-value = 0.019), and Thermoleophilia (p-value = 0.031) differed between management systems (Figure 3). Among the most predominant classes (five classes > 5% RA), only the Gemmatimonadetes differed significantly between management systems, being higher in CON (p-value = 0.0074). Of the 16 classes above 1% RA in CON-ECO, only 4 had significantly different abundances between management systems: Gemmatimonadetes (6.4–4.1%; p-value = 0.0074), Thermoleophilia (Actinobacteria) (4–3%; p-value = 0.0315), γ-Proteobacteria (3.7–4.5%; p-value = 0.0186), and Rubrobacteria (Actinobacteria) (1.4–0.8%; p-value = 0.0250). Interestingly, four out of the nine differentially abundant classes belong to the least predominant classes (<1% RA): Bacilli (Firmicutes) was the most significant differentially abundant class (0.4–0.2%, with the lowest p-value = 0.00002), followed by Cyanobacteria (0.4–1.9%; p-value = 0.0054), Thermomicrobia (Chloroflexi) (0.4–0.2%; p-value = 0.0385), Flavobacteria (Bacteroidetes) (0.2–0.3%; p-value = 0.0385) and Phycisphaerae (Planctomycetes) (0.6–0.9%; p-value = 0.0315). In summary, the most significantly different classes were Bacilli > Cyanobacteria > Gemmatimonadetes > γ-Proteobacteria > Rubrobacteria > Thermomicrobia > Phycis-phaerae (Planctomycetes) > Thermoleophilia. Under ECO management, more significant predominance was observed in classes γ-Proteobacteria, Cyanobacteria, Phycisphaerae, and Flavobacteria, and less predominance was detected in Bacilli, Gemmatimonadetes, Rubrobacteria, Thermomicrobia, and Thermoleophilia. Organo-mineral fertilizer use in CON (Table 1) and the annual incorporation of pruning wood into the soil in both ECO and CON could help maintain the high abundance and microbial diversity in CON.

In terms of the core microbiome in soil vineyards, although 2973 different ASVs were detected during the assessment, only 409 ASVs were identified in all samples (680 ASVs in at least 95% of samples), and only 216 ASVs above 0.01% RA were detected in all samples (342 ASVs in 95% of samples) (Table S2). Only five ASVs >0.1% RA were shared in all samples (13 ASVs in 95% of samples): Arthrobacter/Pseudoarthrobacter (PGPR); Dongia (PSB); Acidobacteria GP4 (soil bacteria linked to the use of complex carbon substrates) [62]; Skermanella (inorganic fertilization with high nitrate and ammonium soil concentrations) [63,64]; and phyla >1% RA in all soil samples belonging to Proteobacteria, Actinobacteria, Acidobacteria, Gemmatimonadetes, Bacteroidetes, Chloroflexi, Planctomycetes, and Verrucomicrobia. In the core microbiome at the genus level, the most predominant genera and shared individual ASVs >0.3% RA in rarefied samples were Arthrobacter/Pseudoarthrobacter (Actinobacteria), Dongia (α-Proteobacteria), Bradyrhizobium (α-proteobacteria), Microcoleus (Cyanobacteria), Skermanella (α-Proteobacteria), Sphingomonas (α-Proteobacteria), Acidobacterium (Acidobacteria), and Gemmatimonas (Gemmatimonadetes).

The linear discriminant analysis effect size (LEfSe) of soil bacterial abundance at the genus level (Figure 5) showed an increase in soil-promoting plant bacterial genera in ECO (p < 0.05): Dongia (gluconic acid phosphorous-solubilizing bacteria (PSB) in the soil rhizosphere) [65], Ohtaekwangia (disease-suppressive soil rhizosphere bacteria against Fusarium oxysporum) [66,67], Bradyrhizobium (symbiotic N-fixing bacteria) [68], Iamia (smut resistance and denitrification, N₂O-producing bacteria during manure composting) [69], Hyphomicrobium, Frankia (N-fixing bacteria), Dactylosporangium, Pseudomonas (plant growth-promoting rhizobacteria (PGPR) with pathogen-suppressive capability and phytohormone-producing bacteria) [70], Algoriphagus, Mesorhizobium (symbiotic N-fixing bacteria in rhizobia-legumes) [71], Xylanibacterium (disease suppression in organic managed soils) [72], Marmoricola, Roseiflexus, Reyranella, Pirelulla, Prosthecobacter, Thiobacillus, and Porphyrobacter. In a recent study, an increase in Dongia and Bradyrhizobium was also observed in cover crop management in Mediterranean orchards [73]. Conversely, in CON fewer increased PGPR genera were detected than in ECO (p < 0.05): Arthrobacter (well-known soil PGPR, with genera associated with PSB, N fixation, and tolerance to abiotic stress such as drought and salinity) [71], Gemmatimonas, Blastococcus (soil nitrogen cycle in poor soils after wildfires) [74], Conexibacter, Bacillus (a PGPR with capacity to work as a biocontrol of grapevine fungal diseases, and phytohormone producer, PSB and siderophores) [75,76], Acidomicrobium, Adhaeribacter, Bosea, Chelatococcus, and Thermomicrobium.

Functional annotation of prokariotic taxonomy (FAPROTAX) revealed significant differences in favor of the organic (ECO) treatment compared to the conventional (CON) management, particularly with respect to biological nitrogen fixation and ureolytic activity (Figure 6).

Faprotax and LEfSe data revealed an enhanced nitrogen bioavailability under ECO conditions. Additionally, chitinolytic activity was also higher in ECO, which is associated with an improved potential suppression of fungal pathogens. Specifically, the ECO treatment stimulated ureolytic microorganisms such as Mesorhizobium, Afipia, and Methylophilus, with Humibacter and Azotobacter consistently present across both treatments. Regarding nitrogen-fixing organisms, ECO favored the presence of Mesorhizobium and Bradyrhizobium, while Azotobacter and Methylocystis remained constant and relatively minor in abundance. Chitinolytic activity under ECO was primarily attributed to Lysobacter, a genus frequently reported in the literature for its role in pathogen suppression. Moreover, hydrogen-oxidizing bacteria (HOBs), known to contribute to the degradation of recalcitrant soil organic matter [77], were more abundant under ECO conditions. Key HOB representatives included Hydrogenophaga, with Ancylobacter showing a weaker increase. Potential phototrophic capacity, particularly involving Acidihalobacter, was also elevated in ECO. Interestingly, this genus possesses genes associated with the Calvin cycle with carbon fixation potential and has been described to participate in dark carbon fixation processes in intertidal sediments [78]. In contrast, specific functions were more prominent under conventional management. Xylanolytic activity, linked to the presence of Xylanobacter, was enhanced under CON conditions—possibly due to the burial of pruning residues in all vineyard fields. Additionally, manganese oxidation was more prevalent in CON, with Geodermatophilus identified as a key genus. This organism is known for its oxidative stress resistance and adaptation to extreme environments, such as desert-like conditions [79].

The co-occurrence network analysis revealed distinct structural differences between conventional (CON) and ecological (ECO) management regimes, with important implications for ecological resilience and adaptability (Figure 7 and

Table 6. Co-occurrence network metrics of bacterial community in subsurface soil under CON and ECO management systems.

	CON (n = 15)	ECO (n = 14)
Average path length	1.93	2.01
Network diameter	5	5
Global clustering coefficient	0.712	0.535
Average local clustering coefficient	0.725	0.553
Modularity (Louvain)	0.172	0.331

). Networks under CON management exhibited higher global (0.712) and local (0.725) clustering coefficients, indicating tightly connected species groups. While such clustering may enhance stability under familiar environmental conditions, it can limit flexibility in response to novel stressors. In contrast, the ECO network showed substantially higher modularity (0.331 vs. 0.172 in CON), suggesting a more compartmentalized structure. Modularity is a key feature linked to increased resilience, as it enables the containment of disturbances and supports independent adaptive responses within modules. Although ECO networks had slightly longer average path lengths (2.01 vs. 1.93), this may facilitate broader ecological interactions and enhance the system’s capacity to reorganize under shifting conditions. Together, these patterns suggest that ecological management fosters a network architecture more conducive to resilience and ecological plasticity, potentially enabling species communities to better withstand environmental stress and exploit new ecological niches.

3.3. Main Drivers That Impact Subsurface Soil Microbiota

The association of physicochemical parameters with microbial population abundance and their alpha diversity was assessed by means of Spearman’s rank correlation [r_s] (

Table 7. Spearman’s rank correlation [rs] matrix of physicochemical soil parameters compared with microbial abundance (bacteria and fungi) and alpha diversity indexes (significant variables with p-adjusted values (<0.05) and [rs] > 0.4 or [rs] < −0.4 are shown in bold).

Parameter	Statistic	log10 16S rRNA gene Copies (n = 30)	log10 ITS Copies (n = 30)	Ratio ITS/16S rRNA (n = 30)	Sobs Richness (n = 29)	Chao1 Richness (n = 29)	Shannon (n = 29)	Inv. Simpson (n = 29)
Clay (%)	p-adjusted (FDR)	0.338	0.662	0.156	0.866	0.831	0.527	0.409
	[rs]	−0.257	0.132	0.349	0.058	−0.070	0.178	0.224
Ca2+/Mg2+	p-adjusted (FDR)	0.108	0.647	0.071	0.859	0.893	0.701	0.376
	[rs]	0.389	−0.138	−0.426	−0.061	0.047	−0.116	−0.240
EC (dS/m)	p-adjusted (FDR)	0.670	0.845	0.732	0.739	0.351	0.968	0.856
	[rs]	−0.129	−0.066	0.104	−0.100	−0.251	−0.016	0.062
Sand (%)	p-adjusted (FDR)	0.341	0.986	0.422	0.963	0.733	0.808	0.701
	[rs]	0.255	−0.008	−0.217	−0.018	0.102	−0.078	−0.116
K+ (mg/kg)	p-adjusted (FDR)	0.739	0.358	0.408	0.378	0.806	0.110	0.129
	[rs]	−0.100	0.248	0.225	0.239	0.079	0.387	0.371
Mg2+ (mg/kg)	p-adjusted (FDR)	0.128	0.682	0.090	0.902	0.824	0.732	0.389
	[rs]	−0.372	0.125	0.405	0.041	−0.073	0.105	0.235
Ca2+/K+	p-adjusted (FDR)	0.603	0.410	0.339	0.365	0.814	0.099	0.114
	[rs]	0.150	−0.223	−0.256	−0.245	−0.076	−0.397	−0.384
C/N	p-adjusted (FDR)	0.041	0.667	0.013	0.701	0.491	1.000	0.318
	[rs]	0.466	−0.130	−0.541	0.116	0.190	−0.001	−0.267
OOM 20–40 cm (%)	p-adjusted (FDR)	0.129	0.893	0.220	0.144	0.121	0.162	0.794
	[rs]	0.370	0.047	−0.312	0.357	0.378	0.344	0.083
Na (mg/kg)	p-adjusted (FDR)	0.230	0.435	0.050	0.597	0.351	0.900	0.875
	[rs]	−0.307	0.210	0.453	−0.153	−0.252	−0.042	0.055
Fe (mg/kg)	p-adjusted (FDR)	0.161	0.314	0.019	0.585	0.582	0.394	0.133
	[rs]	0.346	−0.269	−0.516	−0.157	−0.158	−0.232	−0.367
SOC 20–40 cm (%)	p-adjusted (FDR)	0.107	0.893	0.200	0.180	0.150	0.191	0.844
	[rs]	0.389	0.046	−0.323	0.334	0.353	0.328	0.066
NO3− (mg/kg)	p-adjusted (FDR)	0.252	0.325	0.051	0.816	0.835	0.890	0.183
	[rs]	0.298	−0.264	−0.450	0.076	0.068	−0.050	−0.332
Cu (mg/kg)	p-adjusted (FDR)	0.097	0.893	0.137	0.207	0.094	0.410	0.996
	[rs]	0.398	0.049	−0.364	0.319	0.402	0.223	−0.003
Zn (mg/kg)	p-adjusted (FDR)	0.351	0.960	0.323	0.465	0.338	0.701	0.916
	[rs]	0.251	0.019	−0.264	0.199	0.257	0.116	−0.035
NH4+ (mg/kg)	p-adjusted (FDR)	0.733	0.021	0.056	0.422	0.362	0.190	0.136
	[rs]	−0.103	0.508	0.444	0.216	0.246	0.328	0.365
S (mg/kg)	p-adjusted (FDR)	0.418	0.409	0.159	0.603	0.900	0.547	0.977
	[rs]	0.219	−0.224	−0.347	0.150	0.042	0.171	0.013
SO4 (mg/kg)	p-adjusted (FDR)	0.036	0.794	0.241	0.935	1.000	0.893	0.742
	[rs]	−0.476	−0.083	0.303	0.028	0.000	−0.047	0.099
Silt (%)	p-adjusted (FDR)	0.269	0.900	0.462	0.875	0.805	0.693	0.404
	[rs]	−0.288	−0.041	0.200	0.055	−0.080	0.121	0.227
Mo (mg/kg)	p-adjusted (FDR)	0.816	0.407	0.435	0.598	0.659	0.659	0.667
	[rs]	−0.075	0.226	0.210	0.152	0.133	0.133	0.130
Mg2+/K+	p-adjusted (FDR)	0.180	0.875	0.285	0.045	0.065	0.039	0.544
	[rs]	−0.334	−0.055	0.281	−0.461	−0.432	−0.471	−0.172
Ca2+ (mg/kg)	p-adjusted (FDR)	0.15	0.598	0.521	0.378	0.362	0.370	0.893
	[rs]	0.353	0.153	−0.180	0.239	0.246	0.243	0.047
P Olsen (mg/kg)	p-adjusted (FDR)	0.778	0.955	0.893	0.406	0.581	0.255	0.475
	[rs]	−0.087	0.021	0.046	0.226	0.158	0.296	0.196
Mn (mg/kg)	p-adjusted (FDR)	0.567	0.982	0.732	0.732	0.491	0.732	0.733
	[rs]	−0.163	−0.011	0.104	−0.105	−0.190	−0.105	−0.103
pH	p-adjusted (FDR)	0.763	0.567	0.856	0.017	0.009	0.049	0.410
	[rs]	−0.092	−0.163	0.063	−0.525	−0.560	−0.454	−0.223
N-Kjeldahl (%)	p-adjusted (FDR)	0.542	0.358	0.926	0.140	0.230	0.034	0.138
	[rs]	0.173	0.248	0.031	0.361	0.307	0.480	0.363
OOM 5–20 cm (%)	p-adjusted (FDR)	0.021	0.701	0.129	0.250	0.263	0.139	0.269
	[rs]	0.509	0.116	−0.370	0.299	0.291	0.362	0.288
SOC 5–20 cm (%)	p-adjusted (FDR)	0.019	0.673	0.133	0.263	0.275	0.148	0.282
	rs	0.518	0.127	−0.367	0.291	0.286	0.355	0.282

p-values were adjusted to compensate for the false discovery rate (FDR).

). Only subsurface soil C/N and topsoil OOM and organic carbon (OOM5-20 and SOC5-20) exhibited a strong positive correlation (r_s > 0.45 and p-adjusted < 0.05) with the abundance of subsurface soil bacteria (16S rRNA gene abundance), whereas sulfate content (SO₄) in subsurface soil was negatively correlated (r_s < 0.45, p-adjusted: 0.036). No other significant associations were depicted between environmental edaphic variables and the total subsurface soil bacterial abundance. Interestingly, only the NH₄⁺ subsurface soil content was positively correlated (r_s: 0.508, p-adjusted: 0.021) with fungal abundance (ITS1 subsurface soil abundance), as described in previous studies [80], but no association was observed with bacterial abundance, which could be explained by the low C/N of the vineyard soil (C:N < 10), as described in agricultural soils in EU [81]) and the bioavailability of lignocellulosic material, which is incorporated yearly in the vineyard soil. In addition, fungi/bacteria ratio (16S rRNA/ITS1) showed a slight positive correlation with Na and NH₄⁺ (r_s > 0.44, p-adjusted: 0.05) but exhibited a stronger negative correlation with C/N (r_s −0.54 p-adjusted: 0.013), confirming the positive effect of N on the fungal growth in the subsurface soil, Fe (r_s −0.52 p-adjusted: 0.019), and NO₃ (r_s: −0.45 p-adjusted: 0.05) (Table 7, Figure 6, and Table S3). Although pH and Mg²⁺/K⁺ and N-Kjeldahl were not different in both management systems (ECO vs. CON), they exhibited certain Spearman’s correlation with bacterial richness and diversity in subsurface soil. pH (r_s: −0.52/−0.57, p-adjusted: 0.017/0.009) and Mg²⁺/K⁺ (r_s: −0.46, p-adjusted 0.045) were negatively correlated with bacterial richness (Sobs / Chao1), whereas bacterial Shannon diversity was positively correlated with N-Kjeldahl (rs: 0.48, p-adjusted: 0.034) but negatively correlated with Mg²⁺/K⁺ (rs:−0.47, p-adjusted: 0.039) and pH (rs: −045, p-adjusted: 0.049) (Table 7).

In addition, specific associations between different important physicochemical soil variables were identified (Figure 8, Table S3). The positive correlations depicted between OOM 5–20cm and subsurface soil environmental variables are noteworthy (i.e., the positive correlations with SOC 20–40 (r_s > 0.7), C/N (rs: 0.465), N-Kjeldahl (rs: 0.691), P Olsen (r_s: 0.452), Ca²⁺, S, Cu, and Zn (r_s: 0.50–0.58), as well as an inverted correlation with subsurface soil pH (r_s: −0.532) (Figure 6; Table S3)). In addition, the significantly positive correlation observed in C/N (both in OOM 20–40 and SOC 20–40 with r_s > 0.85), N-NO₃⁻ (r_s: 0.768, p-adjusted: 0.0005), and N-Kjeldahl (r_s: 0.40, p-adjusted: 0.03) but a lack of correlation with NH₄⁺ (rs: −0.28, p-adjusted: 0.146) revealed high nitrification processes in the subsurface soil. Interestingly, Cu exhibited a strong positive correlation with accumulated OOM 20–40 and SOC 20–40 (r_s: 0.67/0.69, p-adjusted: 0.0005), which could explain the higher OOM and SOC in the subsurface soil of CON, with higher Cu content compared with ECO. However, the low Cu concentrations in the soil and its high pH (above 7,5) could mitigate the inhibitory effect of Cu on microbial metabolism, as previously described [82].

Finally, Mantel tests revealed that the effect of soil texture on changes in the beta diversity of microbial community (ASV distribution) was significant (ANOSIM, R: 0.3419, p = 0.0022; Mantel test clay/silt/sand, R: 0.378, p = 0.0001). The heterogeneity of soil texture in vineyards could greatly affect the interpretation of the effects of management on the diversity of microbial communities (Table 3), as described elsewhere in commercial vineyards [83]. CON showed higher values (p < 0.05) of sand, OOM at −20/−40cm, SOC −20/40cm, NO₃−, Cu, Fe, and Zn. Interestingly, ECO was related to a significant increase (p < 0.05) in electrical conductivity, which could be explained by a higher concentration of Na⁺, K⁺, and Mg²⁺, and N-NH₄⁺, which could be derived from cow manure-based compost on soils or leguminous plant species in the cover crop. ECO showed a higher level (p < 0.05) of SOC ratio (5–20 cm/20–40 cm), and OOM ratio (5–20 cm/20–40 cm), as well as differences between OOM and SOC in the topsoil versus subsurface soil. The results indicate that OOM could become highly degraded in the subsurface soil in ECO, probably due to the presence of a more active soil microbiota. The Mantel test revealed (in order of relevance) clay (texture), EC, silt (texture), K⁺, Mg²⁺, Ca/Mg, Ca/K, SOC, and OOM ratios (topsoil/subsurface), OOM Diff (topsoil–subsurface soil), and OOM to be the main drivers affecting the microbial community diversity. Organic carbon is a key component of soil organic matter, which is important for soil fertility and microbial activity. Microbial turnover is another key process in soil organic matter formation, and microbial diversity can have an impact on the chemical complexity and biogeochemical nature of soil organic matter [84,85]. Soil texture has been reported to be one of the main drivers of soil bacterial diversity in arable lands, grasslands, and forests in temperate climates [86]. In addition, soil erosion, which has an impact on both SOC and soil texture, has been shown to be another important driver diminishing microbial soil diversity [87]. Interestingly, Burns et al. [18] reported the importance of the relationship between TC and TN in the <53 μm and 53–250 μm soil fractions in terms of the bacterial community structure rather than in individual taxa for fine soil organic matter content. Some other relevant soil parameters in ECO versus CON showed no significant correlations (p > 0.05 in Mantel test) with microbial community changes in the beta diversity at ASV level (pH, NO₃⁻, NH₄⁺, Cu, Fe, Na, and Zn), likely affected by the heterogeneity among the soil sampled replicates in each plot.

4. Conclusions

Long-term agroecological management in Mediterranean vineyards slightly increased soil alpha diversity but influenced the microbial community structure and function in subsurface soil to a greater extent, particularly under organic management. Organic practices increased the fungal-to-bacteria ratio, AOB, and potential beneficial microbial functions (e.g., nitrogen fixation, chitinolysis, and carbon fixation) while reducing total bacterial abundance and altering key phyla, such as increasing Proteobacteria and decreasing Gemmatimonadetes and Firmicutes. Beta diversity and microbial composition were affected by both management and plot. Organic systems also supported a more diverse and modular microbial community, enriching more genera (21 vs. 7 under conventional). Soil texture and changes in the proportion of organic carbon at different soil depths were identified as the main environmental parameters modulating the microbial community in subsurface soils in the vineyards. Interestingly, N-NH₄⁺ and NTK, which were greater in organic managed soils, did not correlate significantly with changes in the soil microbiome population.

Although subsurface soil conditions are generally more stable than those of the topsoil, long-term experiments incorporating multiple geographical locations, with different soils, including a greater number of replicates to account for soil heterogeneity, are necessary to confirm causal relationships between soil drivers and microbial diversity and functions in the subsurface soil of commercial vineyards. The inclusion of functional biochemical indicators, such as microbial respiration and mineralization potential, will also further support the elucidation of these soil–microbe interactions. This study highlights the need for long-term time-course integrated assessments, combining soil physicochemical traits, microbial data, and crop management, to better understand subsurface soil microbiome dynamics during agroecological transitions, especially under climate change conditions.

The results obtained in this study provide a foundation for further research into the application of organic fertilization strategies in vineyards. This is particularly relevant for identifying the most effective types of organic fertilizers for dryland viticulture in Mediterranean climates. Moreover, the findings emphasize the importance of retaining pruning residues and leaf litter in the soil, as well as promoting the establishment of spontaneous cover crops. They also highlight the association between long-term agronomic practices and their beneficial effects on subsurface soil microbiota.