Samples were collected during winter (August 2018) from two regions with different climatic conditions, according to measurements with weather station imetos 3.3 (Werksweg 107, 8160 Weiz, Austria) (Table S1). On one side, Gualtallary (G), located in Uco Valley, at the foothills of the Andes Mountains, 1245 m.a.s.l., and classified at zone III for Winkler index. The second selected region, Santa Rosa (SR), is located in the traditionally productive region in the east of Mendoza city, 607 m.a.s.l., classified as Winkler V region (Figure 1A) (Abatzoglou et al., 2018). Five soil samples were collected from the native shrubland surrounding a vineyard in Gualtallary (G1–G5, Figure 1B, –3340777 to 3340868; –69.22336 to –60.22537), three soil samples from native shrubland in Santa Rosa (SRU1-SRU3, Figure 1C, –3327912 to –3327985; –6815158 to –6815624) and six samples of soils cultivated with Vitis vinifera cv. Malbec (SRC1-SRC6, Figure 1C).

Table

(A) Mendoza vineyard regions classified according to the Winkler index. (B) Zoom in on the Gualtallary map (Winkler Index III, orange marks) and (C) zoom in Santa Rosa map. SR samples were collected from grapevine-cultivated soils (Winkler Index V, blue marks) and uncultivated soil surrounding the vineyard (Winkler Index V, green marks).

DNA extraction was performed using a Macherey Nagel Nucleospin soil DNA kit following the manufacturer’s instructions. DNA integrity was evaluated in 1 % agarose gel electrophoresis, and minimal quality of 260/280 absorbance ratio > 1.8 was quantified in Thermo Scientific^TM NanoDrop^TM 2000. Library preparation and sequencing were performed at the Integrated Microbiome Resource sequencing facility at Dalhousie University (Halifax, Canada). Briefly, for prokaryotic microorganisms, the hypervariable regions V4-V5 were amplified using primers 515FB (5’-GTGYCAGCMGCCGCGGTAA-3’) and 926R (5’-CCGYCAATTYMTTTRAGTTT-3’). In the case of Fungi, the Internal Transcribed Spacer (ITS) regions, located between the 5.8S and 28S genes of ribosomal ARN (ITS2), were amplified using primers ITS86F (5’-GTGAATCATCGAATCTTTGAA-3’) and ITS4R (5’-TCCTCCGCTTATTGATATGC-3’). Sequencing was carried out in Illumina MiSeq^TM paired-end 2 × 300bp format. The resulting raw reads were deposited at BioProject in NIH-NCBI (PRJNA833442).

Organic matter (OM) in soil samples was determined according to the Walkley–Black method. Total nitrogen (N) was measured using the Kjeldahl method. Available phosphorus (P) was quantified through Bray's method. Exchangeable potassium (K) was determined based on the ammonium acetate method. All methods were performed as previously described by Page et al. (1982).

In shrublands surrounding vineyards, NDVI was calculated as a quantitative estimation of vegetation growth through the QGIS software (QGIS Development Team, 2021) from a Sentinal-2 image. Only the red band (665 ηm ± 15 ηm) and the Near-Infrared (NIR) band (842 ηm ± 15 ηm) were used to compute the classical NDVI index as (NIR-red)/(NIR+red) (Rouse Jr et al., 1974).

Quality filtration, merging of paired-end reads, chimaeras remotion, clustering in Operational Taxonomic Units (OTUs) and the taxonomic assignment was performed with mothur v1.45.0 (Schloss et al., 2009) hosted in High-Performance Computer at CICESE (Ensenada, Mexico) and based on the protocol designed by Biocore-Center for Genomic Regulation, Barcelona (

https://github.com/biocorecrg/microbiome_procedures/

).

The main criteria for quality filtering of 16S reads was to preserve only reads with no ambiguous bases in the merging region of paired reads and with a length in the range of 409-416 bp. Uchime was used for chimaera identification and remotion (Edgar et al., 2011). In the ‘screen.steps' function in mothur, the homopolymers higher than 8 bp were removed; subsequently, remaining reads were grouped on OTUs based on 97 % similarity, and once assigned the taxonomy to OTUs, those classified as “Chloroplast”, “Mitochondrion” or “unknown” were removed. The reference used for the taxonomic assignment was SILVA SSU132 (Quast et al., 2013).

The quality filtering criteria for ITS2 reads was to preserve only those with nonambiguous bases in merging regions of paired reads and having a maximal length of 350 bp. Vsearch was used for chimaera identification and remotion. The OTUs were generated with abundance-based greedy clustering with sequences sharing 95 % identity (Rognes et al., 2016). The dynamic version of UNITE ‘Species Hypothesis' database (UNITE Community, 2019) was used for taxonomic assignment to OTUs. All the OTUs assigned to the “unknown” and/or “Protista” taxon were removed. Statistics test and graphs were obtained with phyloseq v1.36.0 (McMurdie et al., 2013) and phylosmith v1.0.6 (Smith, 2019).

To identify the genus indicator for each group of samples (G, SR, SRC or SRU), a linear discriminant analysis was carried out with lefser v1.2.0 R package (Khleborodova, 2021). Functional predictions were obtained with picrust2 v2.4.2 (Douglas et al., 2020). Specifically, 16S copy numbers of the more phylogenetically related bacteria having their genome sequenced were considered for functional predictions. The information on genes in the related bacteria and the OTU counts were employed to predict the abundance profile of metabolic pathways based on the MetaCyc database (

https://metacyc.org/

). Aldex2 v1.24.0 (Fernandes et al., 2014) R package was used to identify the statistically different pathways when comparing groups. FunGuild v1.0 was used to parse fungal OTUs into guilds based on their taxonomic assignments using the FunGuild database that groups taxa based on the bibliographic information of the experimentally validated functions for fungi (Nguyen et al., 2016).

Microorganisms are the main drivers of biogeochemical cycles in soil which is critical for plant nutrition (Rousk and Bengtson, 2014). With the aim of linking microorganisms’ abundance with the availability of nutrients for plants, the content of macronutrients, OM and C:N ratio were determined on soil samples. All variables, except for phosphorous, were in higher abundance in SRU soils than in SRC and G soils (Table 1). OM was different in all the collection sites (ANOVA p-value < 0.01), reaching the lowest values in soils of Gualtallary and intermediate values in cultivated soils of Santa Rosa, whereas the richest contents were observed in uncultivated soils from Santa Rosa.

Table

titre du tableau
Collection sites	Nitrogen (mg/kg)	Phosphorous (mg/kg)	Potassium (mk/kg)	Organic matter (% g/g)	C/N ratio
Gualtallary uncultivated soils (G)	458 ± 34b	9.1 ± 1,5a	207 ± 31b	0.2 ± 0.02c	2.6 ± 0.3b
Santa Rosa cultivated soils (SRC)	580 ± 57b	12.5 ± 1.9a	303 ± 35b	0.82 ± 0.13 b	8.1 ± 1.0a
Santa Rosa uncultivated soils (SRU)	1255 ± 314a	18 ± 7.7a	1554 ± 289a	1.93 ± 0.49 a	9.1 ± 0.8a

Different letters indicate statistical differences (ANOVA p-value < 0.01)

The output from sequencing was 899,391 pair end (PE) reads for prokaryotes and 792,396 PE reads for fungi (Table S2). After quality filtering, 468,974 and 603,649 reads for prokaryotes and fungi, respectively, were used for clustering in operational taxonomic units (OTUs). Of 1065 defined bacterial OTUs (at 97 % similarity threshold), 28.3 % were unable to be classified at any taxonomic level (Table S2). The OTUs defined for fungi were 2820 (at 95 % similarity threshold), and 78 % could not be classified at any taxonomic level (Table S2).

Regarding bacterial alpha-diversity, SR samples showed a higher number of species, Chao1 and Shannon values than G samples. Moreover, samples from SRC soils showed the highest Shannon alpha diversity (Figure 2A). In contrast, fungal alpha diversity was not different between groups (Figure 2B).

Table

Bacterial (A) and fungal (B) diversity in each group of samples from Gualtallary (G) and grapevine cultivated (SRC) and uncultivated (SRU) soils from Santa Rosa.

The relationship between microbial and biogeochemical cycles is underexplored in arid soils. In the present study, we evaluated the microbial composition of different Winkler-classified sites as well as grapevine-cultivated and uncultivated soils. Based on our results, a model of the microorganisms’ roles in ecological and functional contexts was proposed. The identification of nitrifier denitrification microbial activity in SR grapevine cultivated soils suggest that ammonium input promotes the growth of such taxa, and this activity could favour nitrous oxide release, impacting the environment and reducing the efficiency of fertilisation. The identification of chlorophyll synthesis activity in G soils highlights the importance of phototrophs (possibly forming or promoting the formation of BSC) in those unperturbed soils and heavily exposed to UV radiation. All the hypotheses proposed here regarding the functional microbial’s roles naturally need to be validated with specific experiments to document the presence and importance of the microorganisms in the soil. The associations of agronomic handling in the vineyard with the changes in microbiome structure and functional profiles would contribute to select crop practices that preserve and promotes microbial diversity and ecological equilibrium. Microorganisms with a putative beneficial impact on ecological and agricultural interest were also described in this study. This information will be useful to develop future bioinoculants that exploit the biologically well-adapted microbial resources from the particularities of the viticultural soils of Mendoza.

The authors acknowledge Clara Elizabeth Galindo-Sanchez for kindly sharing the CICESE HPC resources and Sylvia Camacho Lara for all the informatics administration support required for this work. Authors acknowledge Luciana García for sharing Nanodrop2000 at Instituto Nacional de Vitivinicultura (INV). The authors thank Gabriela Ruiz for editing the image of the schematic model. MP acknowledges the postdoctoral fellowship from Agencia Nacional de Promoción de la Investigación, el Desarrollo Tecnológico y la Innovación (Argentina). The authors acknowledge the helpful comments of anonymous reviewers.