Soil microbiome analysis of Uruguayan grasslands and croplands reveals losses of microbial diversity and necromass recycling traits

Abstract

Background

Soil microbiomes are critical regulators of nutrient biogeochemical cycles, contributing significantly to ecosystem services that support plant productivity. In this study, we investigated the effects of agricultural rotations on soil microbial communities in Uruguayan grasslands, comparing cropland soils with native grasslands grazed by livestock. By employing advanced metagenomic techniques, we characterized the diversity and functional potential of the soil microbiome, with particular emphasis on its roles in carbon, nitrogen, and phosphorus cycling. Additionally, we assessed functional genes associated with microbial necromass recycling, a key process for maintaining soil health and fertility.

Results

Our analysis revealed a significant decrease in fungal diversity and a restructuring of the fungal community in agricultural soils, highlighting the profound impact of farming practices on soil biology. In contrast, while we did not observe a similar decline in bacterial diversity, there was a noticeable shift in its composition. Agricultural soils showed a reduced abundance of bacteriophages, which are associated with bacterial necromass formation, as well as a decline in enzymes involved in microbial necromass decomposition. This suggests potential long-term consequences for soil carbon dynamics and crop productivity. Additionally, croplands exhibited a marked decrease in genes and enzymes involved in nitrogen and phosphorus cycling, indicating diminished functionality and resilience for these essential nutrient processes compared to grassland soils.

Conclusions

Our findings emphasize the need for a sustainable approach to agriculture that preserves microbial diversity and functionality, ensuring the resilience of soil ecosystems. By comparing soil microbiomes across different land use types, this research provides novel insights into the mechanisms through which agriculture alters soil ecosystems and offers guidance for enhancing soil management practices to support environmental sustainability and agricultural productivity.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40793-025-00696-4.

Background

Temperate grasslands, crucial for ecological balance and agricultural productivity, formerly covered around 8% of the Earth surface. However, only around 4.6% of temperate grasslands are protected. Therefore, this type of ecosystem is thought to be one of the most altered terrestrial environments [11]. In Uruguay there has been a net loss of 0.9 Mha of grasslands, considering the period between 2001 and 2019 [3]. Economic factors and rising prices of agricultural commodities have led to the conversion of native grasslands into croplands [48]. This shift in land use introduces management practices such as tillage and fertilization, which may fundamentally alter the native soil microbiome. The soil microbiome is critical for nutrient cycling, organic matter decomposition, and overall ecosystem health [65], with microbial necromass playing a key role in soil carbon dynamics [13]. However, little is still known about the impact of replacing South American native grasslands with croplands on the soil microbiome and its nutrient cycling potential.

Soil microbiomes are fundamental to soil organic carbon (SOC) dynamics, acting as key drivers in both the protection and decomposition of SOC. These microbial communities influence SOC formation and stabilization through a variety of mechanisms. They decompose organic matter, breaking down complex organic compounds into simpler forms that can be immobilized by themselves, or assimilated by plants and other soil organisms, thus driving nutrient cycling and energy flow in ecosystems [28]. Distinct roles of bacterial and fungal communities have been described, where bacteria predominantly utilize plant materials rich in nitrogen [64], while the interaction between microorganisms and soil fauna enhances the incorporation of organic residues into stable SOC pools [41]. Microbial activity also contributes to the protection of SOC by promoting aggregate formation through the production of amphipatic proteins, precipitation of minerals and other physical interactions with organic matter, shielding carbon from rapid decomposition [29, 33, 54]. Moreover, microbial necromass—a significant proportion of SOC—plays a critical role in forming mineral-associated organic matter, a key component of long-term carbon storage in soils [36]. Despite these insights, the mechanisms through which agricultural practices influence microbial diversity, necromass formation and recycling, and microbial control on SOM formation remain incompletely understood, posing challenges for predicting the effects of land-use changes on soil health and crop productivity [14, 15]. Bridging these knowledge gaps is essential for advancing sustainable land management and maintaining ecosystem services.

Methodological advances in sequencing technologies and bioinformatics provide a suite of tools for in-depth analysis of these complex communities avoiding biases generated by culturing techniques [20]. Different approaches are employed to dissect the structure and functional potential of soil microbiomes. Using targeted amplicon sequencing, such as the 16S rRNA gene for bacteria and the ITS region for fungi, it is possible to elucidate the composition and diversity of these communities with high specificity [31, 47]. Furthermore, using shotgun metagenomic sequencing it is possible to capture the entire genetic repertoire of the soil microbiota, which allows for the identification and quantification of functional genes such as those encoding Carbohydrate-Active enZymes (CAZymes). Comparative microbiome analysis may reveal overlooked changes in microbial components contributing to microbial necromass formation and recycling upon land use change.

In this study, we compared soil microbial communities from four productive farms in Uruguay, which have undergone at least a decade of agricultural rotations, with adjacent native grasslands grazed by livestock. Uruguay is covered by native grasslands, which remain nearly 50% of the territory, but have been rapidly replaced in the last decades by other land uses such as croplands. Utilizing high-throughput metagenomic sequencing, we characterized the diversity and functional attributes of bacterial and fungal communities, with a particular focus on their roles in carbon, nitrogen, and phosphorus cycling. In order to understand microbial contribution to SOC formation, microbial necromass recycling traits were specifically assessed through the detection and quantification of genes encoding enzymes that exclusively target peptidoglycan and/or chitin. Our work contributes to bridge the knowledge gap in soil microbial ecology concerning the long-term impacts of agricultural rotations on soil microbial communities, offering insights that can inform sustainable land management practices to preserve soil health and ensure continued agricultural productivity.

Methods

Uruguay is a country located in the temperate zone of South America, characterized by its rolling plains and fertile soils, with a climate that is conducive to extensive grassland ecosystems. These grasslands, known locally as'pampas', are a predominant feature of Uruguay’s landscape, covering a significant portion of its territory. The soils of these pampas are primarily Mollisols, known for their thick, organic-rich surface horizon, which contributes to their high fertility and makes them some of the most productive agricultural soils in the world [17]. Uruguay’s climate is classified as humid subtropical, with warm summers, cool winters, and relatively uniform rainfall throughout the year, conditions that have historically supported both native grassland ecosystems and agricultural productivity [24].

The grasslands of Uruguay have been subject to extensive agricultural activities, with long-term livestock grazing and agricultural rotations (afforestations and croplands) significantly altering their soil properties and biotic communities. Native grasslands are dominated by species such as the herbaceous Paspalum notatum and Stipa species, these native grasslands host a diverse flora that is adapted to the local climate and soil conditions [59]. The native grassland ecosystem supports a range of wildlife and serves as a natural resource for grazing livestock, which is a traditional component of Uruguayan culture and economy [27]. Agricultural rotations in Uruguay typically involve a sequence of summer crops like soybean (Glycine max) and maize (Zea mays), followed by winter crops such as wheat (Triticum aestivum) and barley (Hordeum vulgare). Crops are often fertilized using Nitrogen, in the form of urea, and Phosphorus. These rotations often include periods where the land is left to fallow or is used for livestock grazing, integrating agricultural and livestock production systems [49]. This blend of crop and livestock farming, known as mixed farming, has shaped the use of land and the management of soil resources in Uruguay [23], where croplands are usually found nearby grazed native grasslands, under similar soils, and only separated by fences.

Soil sampling and DNA extraction

We selected four fields of different farms located in the agricultural region of Uruguay with a history of at least 10 years of agricultural rotations (AR) with adjacent native grasslands (NG) grazed by livestock. Sites were sampled either in autumn or spring (southern hemisphere) of 2020 or 2021. Sampled sites were located in the west side of Uruguay (Additional file 2: Fig. S1), which is considered the main agricultural region in the country, encompassing four departments: Florida (L), Soriano (P), San Jose (M) and Salto (SA). At the time of sampling, the agricultural rotation of site L had an oilseed rape crop at the flowering stage, while sites P and SA were under soybean fallow, and site M was under maize fallow. At each site, an area of 20 by 20 m was delimited inside the two land cover patches (grasslands or croplands), located adjacent one to the other, separated by a fence. Soil pits were established to check for similarity between soils (horizons depth, colors, etc.) and animal trails or machinery compaction zones were avoided. Soil texture and nutrients analysis were also performed to check that soils were comparable (Additional file 1: Table S1). At each land use patch, 10 soil cores of 15 cm depth were randomly taken inside the delimited area in order to make a composite sample. Three composite samples were obtained for each site and land use sampled, making a total of 24 composite samples. Soil was preserved at ~ 4 °C for transportation to the laboratory and kept at –20 °C until further processing. Soil composite samples were air dried at room temperature for 4 days. After that, samples were grounded with a manual soil mortar and thoroughly mixed. Grounded samples were sieved in a 2 mm mesh in order to leave out vegetable debris and other major size components that were not soil particles. Grounded and sieved soil samples were thoroughly mixed again and further processed for DNA extraction, which was done using PowerSoil Lyzer kit (Qiagen). We used 300 mg of each soil sample and followed manufacturer instructions using a combined mechanical and chemical approach to lyse bacterial and fungal cells. In order to achieve the mechanical disruption of the cells the FastPrep- 24 5G (MP Biomedicals) was used at 6 m/s for 40 s.

DNA extractions were quality checked by electrophoresis and DNA quantification was performed with a Qubit 2.0 Fluorometer (Invitrogen). Eight equimolar pools of DNA extractions from the same sites and land uses were generated for shotgun sequencing, while total DNA dilutions from the 24 composite samples were used for bacterial and fungal molecular markers sequencing in BGI Americas facilities (San Jose, USA). Sequencing was done with short-read DNBseq technology using 100-bp pair-end read lengths for shotgun metagenomic pooled samples, obtaining 12.04 ± 0.03 Gb of raw data per sample. 300-bp length reads were generated for V3-V4 regions of 16S rRNA genes obtaining 78,673 ± 4230 paired reads per sample. ITS1 region was amplified with ITS1f-ITS2 primers and sequenced obtaining 67,765 ± 933 paired reads per sample. Two of the DNA extractions obtained from L-AR failed ITS1 region sequencing, thus fungal communities comparison has 22 samples 10 from agricultural rotations and 12 from native grasslands.

Bacterial communities’ analysis

Raw sequences were quality-filtered, dereplicated and ASVs were resolved using DADA2 R package custom pipeline [8]. SILVA- 138 database was used to compare sequences and assign taxonomy to ASVs [57]. Then, the phyloseq R package was used to continue the downstream microbial communities’ analysis from the obtained counts table [44]. ASVs count tables were filtered to obtain ASVs present in at least 10% of the samples. Alpha diversity was computed as the richness of bacterial genus present in each sample, to detect statistical differences one-way ANOVA was computed with post-hoc T-test and visualized with ggplot2 R package [72]. For beta diversity analysis, Bray–Curtis dissimilarity index was computed from normalized community profiles. Permutational multivariate analysis of variance (PERMANOVA), implemented in adonis function from vegan package, was used to test for differences in the distribution of bacterial taxa for the different samples [16]. After that, a non-metric multidimensional scaling (NMDS) analysis enabled the visualization of the clustering of samples in a bidimensional space, and environmental metadata was fitted using vegan R package. Core microbiome analysis was done using Microbiome R package custom pipeline [32].

In order to detect differential abundance of ASVs for each land use treatment, generalized linear models were used, implemented in Maaslin2 package [42]. Genus-level microbiome data was transformed to consider compositional bias of sequencing data using centered log-ratio (CLR) transformation. Sampling site, pH and sampling season were used as random effect variables in this analysis.

Fungal communities’ analysis

ITS analysis was similar to 16S, but with slight variations for fungi-specific molecular marker pipelines. Sequences were filtered, dereplicated and ASVs were resolved using DADA2 R package pipeline [8]. The UNITE database was used to compare sequences and assign taxonomy to ASVs [1]. After the creation of the ASV count table phyloseq R package was used to continue the downstream communities’ analysis [44]. Alpha diversity was computed as the richness of bacterial genus present in each sample, with a minimum prevalence threshold of 10% and transformed to counts per million accounting for sequencing depth variations. For beta diversity analysis Bray–Curtis dissimilarity index was computed from filtered count tables. A non-metric multidimensional scaling analysis was used to visualize the ordination of samples in a bidimensional space, environmental metadata was fitted using envfit function of vegan (v 2.6–8) R package [16], the constructed plot was visualized with ggplot2 R package. In order to detect significant differences between the distribution of species obtained from fungal communities of different land uses, a permutational multivariate analysis of variances was implemented using adonis function with the Bray–Curtis distance matrix.

In order to detect differential abundance of taxa for each land use, we first collapsed ASV tables into count tables at the genus level, then we applied generalized linear models analysis, implemented in Maaslin2 package [42]. Microbiome data was transformed to consider compositional bias of sequencing data, this transformation was done using centered log-ratio (CLR) method. Sampling site, pH and sampling season were indicated as random effect variables. Additionally, detected fungal genus were classified into guilds using FUNGuild software. This software uses literature references of different genus and classifies them into fungal guilds depending on its trophic mode and substrate utilization [50]. Trophic modes abundance tables were collapsed and Generalized Linear Models implemented in Maaslin2 package were used to detect differences in abundance between different land uses, as previously explained.

Shotgun metagenomics analysis

A shotgun metagenomic analysis was performed to evaluate the presence of enzymes related to carbon, nitrogen, and phosphorus cycles. We also used this data to detect and quantify the presence of bacteriophages as important drivers of bacterial necromass formation. Shotgun metagenomes were sequenced from a pool of the three composite samples obtained for each site and treatment in equimolar amounts. Raw reads were filtered using Trimmomatic software with the options SLIDINGWINDOW:10:20 LEADING:25 TRAILING:25 MINLEN:75 [5]. Remaining paired-end reads were used as input for Megahit software, with the preset options –meta_large, in order to assemble the reads into contigs [35]. The quality of the assembly was checked using metaQUAST software [46].

Functional characterization of soil microbiomes of the different samples was implemented in a customized pipeline. First, metagenomic ORFs were predicted using Prodigal software [26]. To detect Carbohydrate-Active enzymes (CAZymes) that are responsible for the assembly, breakdown, and modification of carbohydrates and glycoconjugates, we used dbCAN-seq software, which uses an updated database of CAZyme hidden Markov models to compare against annotated ORFs [78]. After that, we used MMseqs2 software for clustering identical ORFs and obtaining a dereplicated CAZymes gene catalog [66]. To quantify the abundance of ORFs in the gene catalog, we used MMseqs2 software to map reads of each sample against the CAZYmes genes. These gene-specific abundances were normalized with total library sizes and expressed in FPKMs (Fragments Per Kilobase per Millon mapped reads) for each gene and sample. CAZyme abundance tables were used to compute shannon index with Vegan R package [16], as a measure of alpha diversity. Additionally, degradative CAZyme families with detected differential abundances between land uses were used for substrate mapping using the information available in Zheng, et al. [78].

A similar gene catalog approach was used for quantifying Nitrogen cycle genes. NCycdb database was used with diamond software to detect Nitrogen cycle relevant genes with 10⁻⁵ E-value and 70% of identity and query cover inclusion thresholds [68]. Then MMseqs2 was used to dereplicate and map filtered reads against the gene catalog. R software was used to obtain the normalized abundance of each gene, which was summarized in a count table expressed in FPKM. Maaslin2 software was used to detect differential abundances for each gene between samples of different land uses [42], using a False Discovery Rate threshold of 0.25 and adjusting p-values with Benjamini–Hochberg method.

Additionally, the phosphorus cycling genes database was used to detect and quantify genes involved in phosphorus cycling. Prodigal ORFs were used as input and searched against PCycdb files using diamond software [76]. Filtering thresholds used were 10⁻⁵ e-value and 70% of identity and query cover. After that, the same custom pipelines were used to map reads and quantify FPKM variation through the different samples. In this case, differential abundances were detected using Maaslin2 software [42]. Differentially abundant genes were grouped in the different metabolic processes described in databases and visualized using ggplot2 R package [72]. For the detection of phage contigs we used PhaBOX wrapper software [61], which is designed for the detection and classification of phage-derived sequences. Contigs abundances were obtained by mapping filtered metagenomic reads with MMseqs2 software against phage contigs. The number of mapped reads was normalized by contig length and total reads generated for each metagenomic library.

Results

Fungal communities´ structure and diversity

Our study showed a notable decline in fungal diversity within agricultural rotations compared to native grasslands (Fig. 1, Panel B). The ITS1 region sequencing unveiled a total of 1,715 fungal ASVs, with a core mycobiome dominated by Fusarium spp., reflective of the overall dominance of the Ascomycota phylum in these ecosystems, followed by the Basidiomycota and Mortierellomycota Phylum (Fig. 1, Panel A and Additional file 3; fig S2). Beta diversity patterns, elucidated through NMDS analysis, further distinguished between the fungal communities of different land uses, with a discernible separation along the NMDS1 axis correlating with variations in soil organic carbon (SOC), soil apparent density (AD) and phosphorus (P) content (Fig. 1, Panel C). This distinction is indicative of the environmental impact on fungal community structure, with agricultural practices leading to a distinct microbial composition. Furthermore, our analysis identified 22 fungal genera significantly associated with native grassland communities, including genera like Cephaliophora and Collarina, which exhibited the most pronounced fold changes (Fig. 1, Panel D). In contrast, genera such as Alternaria, Stemphilium and Pyrenophora were significantly associated with agricultural soils, suggesting an altered landscape that favors pathotrophs (FDR < 0.15) (see Additional file 5: Fig. S4). In contrast, native grasslands showed a non-significant trend with higher relative abundance of symbiotrophs genera (FDR < 0.15). This trend was not followed by Tetracladium genus, which has been recently described as a beneficial fungal root endophyte and is significantly associated to agricultural soils in this dataset (Fig. 1, Panel D). The observed shift in fungal community composition and diversity underscores the profound influence of agricultural management practices on soil fungal communities.

Fig. 1.

Diversity and Structure of Fungal Communities in Grassland and Agricultural Soils. Panel A Relative abundances of the ten most abundant fungal phylum in agricultural rotations (AR) versus native grasslands (NG) for each site. Panel B Alpha diversity boxplots of fungal communities, showing the observed ASV richness. The bar graph illustrates a comparative loss in fungal diversity in AR versus NG. Panel C: Beta diversity of fungal communities represented by an NMDS plot, highlighting the distinct separation of microbial communities according to land use. The plot demonstrates significant differences in ASV composition and abundance between AR and NG samples. Environmental variables fitted correspond to soil organic carbon (SOC), clay content (clay), Bray1 phosphorus content (P), soil apparent density (AD), earthworms biomass in the sampling area (EW_biomass), Magnesium content (Mg) and relative humidity in soil samples (%RH). Panel D: Differential Abundances of fungal genera between land uses, the stacked bar graphs show genera significantly associated with either NG (red) or AR (blue) soil communities, changes between land uses are expressed as the logarithm of Fold change between land uses (log2 FC). Uncertain genus level taxonomic classifications are noted as gen_Inc_sed

Bacterial communities’ structure and diversity

In contrast to fungal communities, we could not detect a difference in bacterial ASVs richness or alpha diversity indexes between land uses, although we found strong differences in the composition and identity of bacterial communities in native grasslands compared to croplands (Fig. 2, Panels A–C). There was a greater overall bacterial richness as compared to fungi ASVs since bacterial communities across both native and agricultural soils uncovered a total of 2,941 bacterial ASVs present in at least 10% of the samples sequenced (Fig. 2, Panel B). Remarkably, the core microbiome, consisting of 38 taxa present in all samples, was rich in genera such as Bradyrhizobium, Streptomyces, and Bacillus, which are recognized for their beneficial roles in soil ecosystem functioning (Additional file 4; Fig. 3). However, beta diversity assessments indicated pronounced dissimilarities in bacterial community composition, with NMDS ordination separating the bacterial profiles of agricultural rotations from native grassland soils. Environmental factors, including soil phosphorus and SOC contents, were identified as influential in structuring these communities, given its correlation with NMDS2 axis. These results suggest that land management practices are reshaping soil bacterial biodiversity (Fig. 2, Panel C). When focusing on bacterial genera, we discovered that 58 of them were significantly associated with either native grasslands or agricultural rotations (FDR < 0.15). Among these, genera belonging to the order Azospirillales, Bacillus, and Candidatus Udaeobacter, known for their roles in nutrient and necromass cycling, were more abundant in native grasslands. In contrast, genera like Microlunatus, Methylocystis and Rubrobacter were more prevalent in agricultural soils, indicating a possible shift in soil functions due to agricultural managements, such as fertilization (Fig. 2, Panel D).

Fig. 2.

Composition and Beta Diversity of Bacterial Communities in Grassland and Agricultural Soils. Panel A Relative abundances of the ten most abundant bacterial phylum in agricultural soils (AR) versus native grasslands (NG). Panel B Alpha diversity boxplots of bacterial communities, showing the observed ASV richness in NG versus AR. Panel C: NMDS ordination of bacterial communities’ beta diversity analysis. Environmental variables fitted correspond to soil organic carbon (SOC), Clay content (clay), Bray1 phosphorus content (P), soil apparent density (AD), Earthworms biomass in the sampling area (EW_biomass), Magnesium content (Mg) and relative humidity in soil samples (%RH). Panel D: Differential abundance of bacterial genera between land uses, the stacked bar graph shows genera significantly associated with either NG (red) or AR (blue) soil communities, changes between land uses are expressed as the logarithm of Fold change between land uses (log2 FC)

Functional analysis of metagenomes

Our functional analysis of soil metagenomes, utilizing shotgun sequencing, showed significant differences in soil microbial mediated processes in soils under agricultural rotations compared to native grasslands (Fig. 3). In the case of carbon related enzymes, we could observe that there are no changes in the alpha diversity measures of CAZymes in soils under different land uses. However, we could reveal that land use change generates a significantly different pattern of abundance of CAZyme families (FDR < 0.25) (Fig. 3A). Specifically, 64 carbohydrate-active enzymes families were predominantly represented in grassland communities, underscoring a robust enzymatic toolkit for the biotransformation of complex carbohydrates. In fact, a subset of these enzymes, including 38 glycoside hydrolases (GH), 3 carboxyl esterases (CE), and 3 polysaccharide lyases (PL), play an important role in the breakdown of complex polysaccharides, cell wall components, and other critical organic matter components. In the case of agricultural rotations, 55 CAZyme families were significantly more abundant compared to grassland soils. This includes 36 GHs, 1 CE and 1 PL, which are involved in the degradation of complex carbon polymers. Enzyme substrate analysis enabled to elucidate that although there is an important change in predominant enzyme families, carbon polymers targeted are generally conserved under both land uses. This is observed in Fig. 3B, where each bar represents the count of family enzymes, significantly associated to each land use, that target different carbon compounds. When focusing on microbiome traits related to microbial necromass carbon dynamics we detected that the cumulative relative abundance of all gene families encoding enzymes targeting peptidoglycan and chitin, as main carbon polymers in fungal and bacterial cell walls, was significantly higher in NG microbial communities (Fig. 3C). Additionally, NG soils had a significantly higher abundance of contigs derived from bacteriophages, which play an important role in bacterial community dynamics as drivers of bacterial necromass formation (Additional file 6; Fig. S5). These results give evidence on the potential changes on carbon dynamics regarding microbial necromass turnover upon land use change.

Fig. 3.

Functional Gene Analysis in Soil Metagenomes. Panel A Volcano plot showing the differential abundances of CAZyme families across microbial communities, points in red represent CAZyme families with differential abundance in AR, while points in blue represent CAZyme families significantly more abundant in NG (FDR < 0.25). Panel B Barplot showing the number of enzymes, with differential abundance in either of the two treatments, targeting different carbohydrate substrates. Panel C Boxplots representing the cumulative relative abundance of all enzymes targeting chitin and peptidoglycan, as main microbial-derived carbon polymers, in AR and NG. Panel D Effect size barplot of nitrogen cycle genes with differential abundance in microbial communities in agricultural rotations (AR) and native grasslands (NG) (FDR < 0.25). Panel E Effect size barplot of phosphorus metabolism genes with differential abundance (FDR > 0.25) across different land uses. Positive effect sizes are shown for AR while negative values are for NG

In addition, our analyses highlighted a marked decrease in the abundance of some genes encoding enzymes involved in the nitrogen cycle in agricultural soils (Fig. 3D). Eighteen genes involving different nitrogen cycling processes were in a higher abundance in NG soils, while only two were favored in agricultural rotations. Genes involved in nitrate reduction, nitrogen fixation, and denitrification, such as nirB, nifH, and nirS, were more abundant in grassland soils, indicating an active nitrogen cycling that could contribute to the high nitrogen availability in these soils. This finding suggests that native grasslands possess a microbial infrastructure capable of supporting nitrogenous nutrient cycling, which is essential for sustaining plant growth and soil fertility. However, successive planting and managing of annual crops has diminished these capabilities, potentially affecting crops performances in absence of external N input.

In parallel, the study of phosphorus metabolism unearthed a differential abundance of genes involved in phosphorus cycling between native grasslands and agricultural soils. Twenty genes were significantly more abundant on NG, while only seven were more abundant on AR (Fig. 3E). Genes related to phosphonate and phosphinate metabolism including the C-P lyase phnK and phnT transporter, were significantly more prevalent in native grasslands. Apart from that, we could detect a shift in organic phosphorus metabolism comprising purine, pyrimidine and pyruvate metabolism genes. There was also a difference in transporters, detecting five genes differentially abundant in NG, while only pstC was more abundant in AR. This abundance patterns reflect a diversified phosphorus uptake capability, potentially promoting a more resilient phosphorus utilization and recycling in grassland soils. Such a comprehensive understanding of phosphorus cycling mechanisms underscores the critical role of microbial communities in maintaining the balance of essential nutrients, further highlighting the impact of land use on the functional potential of soil microbiomes.

Discussion

The substitution of native grasslands for croplands exerts significant influence on multiple ecological processes (Fig. 4). These effects can be attributed to various management practices, such as the replacement of diverse native plant species with exotic monocultures, or alterations induced by changes in management practices, such as the usage of herbicides, fungicides, fertilizers and plowing. We found a strong decline in fungal diversity in agricultural rotations, which may be explained by the intensive use of fungicides that are applied to crops and seeds to control fungal pathogens in agricultural systems of Uruguay [19, 52]. Soil fungal diversity has been found to be important for maintaining ecosystem stability and plant productivity at a global scale [38, 39]. There is evidence on the strong impact of different fungicides on soil microbial diversity, structure and functioning [25, 41]. In a recent article it was shown that combined and single addition of pesticides negatively affected litter decomposition rates in soil microcosms [45]. Apart from that, when grouping fungal genera by trophic mode, we detected a rise in the relative abundance of pathotrophs and a decline in symbiotrophs in agricultural soils. This effect has also been observed in other studies and has been strongly linked to phosphorus and nitrogen fertilization [34], which are part of the external inputs in Uruguayan agricultural rotations. In fact, reduced nitrogen fertilization is being proposed as a mechanism for plant disease prevention based on the effects of this type of fertilization on plant physiology and soil fungal communities [4, 62]. In contrast, we did not observe a decrease in bacterial diversity in AR soils, but bacterial communities were very dissimilar under both land uses studied. We could detect a decline in the abundance of iconic genus such as Bacillus and Streptomyces in agricultural soils, whose metabolic and genetic diversity has long been studied. Bacillus has been linked to nitrogen fixation in crop rhizosphere, as well as P solubilization, phytohormone production and even the control of fungal and insect pathogens, providing an important amount of ecosystem services that could enhance plant productivity [60].

Fig. 4.

Schematic representation of main differences obtained comparing microbiome data from native grasslands and agricultural rotations. Microbial traits involved in different soil processes, which are significantly more abundant under each land use, are described below the soil surface for each rotation. Microbial mediated processes and guilds are shown in bold, while microbial genus and functional genes involved are shown in italics. Carbon cycle processes are shown in blue, Nitrogen processes are shown in green and phosphorus processes are shown in red

Another important genus whose relative abundance has been decreased in agricultural rotations is Ca. Udaeobacter, which has been described as a soil heterotroph of small genome size with multiple auxotrophies [6]. This ubiquitous soil bacteria has been proposed to acquire essential aminoacids and vitamins from scavenging bioavailable microbial metabolites [73], a process known in soils as microbial necromass recycling, which is highly important for C stabilization [7]. Microbial necromass is an important fraction of soil carbon contributing to the mineral-associated organic matter fraction accounting up to half of its carbon content [12]. We could also detect that the abundance of bacteriophages in NG metagenomes is higher than in their agricultural counterparts. Bacterial viruses have been described as major contributors to the release of C and readily available microbial metabolites in a process named ‘viral shunt’ which can have different effects on SOC dynamics [30, 71, 75]. Microbial death pathways have been recently described as important drivers of microbial necromass composition [9]. There is also evidence that phage-mediated bacterial lysis increases C and N contents in microbial necromass [74], which may have an important effect in terms of soil fertility. Here we provide evidence that land use change may lead to an alteration in microbial death pathways, reflected by a variation in bacteriophages relative abundance. Further research is needed to understand soil phage dynamics and the effects of land use on the activity of these microbiome components.

Soils harbor an incredible diversity of microorganisms, which is mainly driven by aggregates structures; this poses a challenge for its study and characterization [20]. Although the regulation of microbial activity in environmental communities can be driven by complex and dynamic interactions of environmental and biological factors, the quantification of carbon-cycle functional genes in microbial communities can be used as a proxy for enzymatic potential activity [67]. When analyzing changes in CAZyme families’ abundances, we detected that there is an important functional redundancy in degradative CAZYmes in the surveyed soils, this has already been reported for bulk soil and plant rhizosphere microbiomes [36]. Additionally, by substrate analysis we detected that four families with higher abundance in NG soils are exclusively related to the hydrolysis of peptidoglycan and chitin. Families GH102, GH103, GH171 and GH23 exclusively use bacterial and fungal cell wall polysaccharides as substrates. Their reduced abundance in agricultural soils indicates a loss of microbial necromass recycling potential, which could be linked to the lower SOC content observed in agricultural soils [70]. A possible reduction in microbial necromass carbon stocks could lead to a decrease in the microbiome’s capacity to recycle necromass-derived polymers [10]. There is evidence that SOC fractions with different lability are differentially affected by land use change and soil carbon inputs [37, 77]. Microbial necromass carbon can constitute an important fraction of soil organic matter and it has been traditionally linked to mineral associated organic matter [13, 69]. Our metagenomic analyses indicate an alteration in the microbiome components that regulate microbial necromass carbon fluxes following land use change.

Nitrogen is another important macronutrient for soil function that is affected by agricultural practices. Metagenomic analysis revealed that some genes related to nitrogen cycle processes were in a higher abundance in native grassland soils, such as the NifH gene, which is responsible for nitrogen fixation. Through community profiling analysis, we also detected a higher abundance of a genus belonging to Azospirillales order along with the genus Bradyrhizobium, both could be contributing to the higher abundance of NifH gene. A high rate of biological nitrogen fixation by free-living organisms has previously been reported in association with C₄ grasses in South American native grasslands, particularly with rhizosphere-colonizing bacterial genus [43]. The presence of vegetation that can host these BNF organisms, like Desmodium incanum and Paspallum spp. [53, 56], as well as the fast cycling of plant materials triggered by grazers, who also excrete N into soils through dung and urine, might explain high nitrogen concentrations in NG soils without the need of external inputs in addition to nutrient extraction by crop harvesting [63]. Apart from that, genes responsible for processes of nitrogen loss in soils were also in a higher relative abundance in native grasslands. This is the case of nirK and nirS denitrification genes and dissimilatory nitrate reduction genes. If we consider that nitrogen fertilizers'production and usage is estimated to account for 5% of global greenhouse gas (GHG) emissions, the substitution of this external inputs by microbial based products coupled with suitable agricultural managements could be a sustainable strategy to be considered for agricultural production [21].

Finally, it is striking that although agriculture soils had higher phosphorus (P) contents, its respective microbial communities had a lower relative abundance of enzymes related to P metabolic processes, like purine (purBC and purHKLS) metabolism related genes, this might be driven by the fact that most of the P in soil is inorganic, while most of the genes present in the database are related to organic P cycling. Regarding P acquisition strategies in soil microbiomes, we detected a higher abundance of a C-P lyase multienzyme complex subunit (phnK), which are the enzymes that make phosphonates P available for bacteria [58]. Previous studies have proposed that phosphonate and phosphinate could be an important source of P for microbial communities in soils with low P availability [51]. The application of fertilizers likely contributes to elevated phosphorus levels in agricultural soils, concurrently leading to a decrease in the abundance of numerous P transporters within microbial communities. In the case of AR we could detect a higher abundance of the Microlunatus genus which contains species reported to be polyphosphate accumulating organisms (PAO), under aerobic conditions, used for phosphorus bioremediation in wastewaters and associated with high P availability in croplands [2, 38, 39]. The high abundance of PAO in croplands may lead to lower phosphate use efficiency (PUE) by the adaptation of the microbiome to P exceeding in soils. Additionally, the only phosphatase found in this dataset was phoD, with no differential abundance between land uses, this has been described as a prevailing gene in a recent study of global grasslands soil communities [22]. The positive effect on the abundance of microbial phosphorus cycling genes that we could observe in native grasslands has already been reported in previous grasslands surveys, and it has been specifically linked to livestock grazing [40].

Conclusions

Our research highlights the profound impact of agricultural rotations on soil microbial communities, signaling a critical need for sustainable agricultural practices. The evident decline in fungal diversity and shift towards pathotrophic dominance, together with the reduction in beneficial bacterial genera and the decline in soil microbially mediated processes, may compromise soil health and ecosystem services, such as nutrient cycling and disease suppression. These findings emphasize the importance of developing management strategies that support microbial functionality, potentially enhancing soil carbon sequestration and contributing to climate change mitigation, such as enhancing cash crops diversity, the use of a diverse set of service crop species [55] and the substitution of pesticides for biological inputs [18]. To ensure long-term agricultural productivity and environmental health, policy frameworks must integrate these insights, advocating for practices that maintain microbial functionality. Future directions should focus on understanding the mechanistic pathways of microbial turnover influenced by agricultural inputs, informing practices that sustain the delicate balance of soil ecosystems.

Author contributions

MG, GI and GP conceived the work, GI and GP contributed with funding acquisition and supervision of the data generation and analysis. AL contributed to the data generation and sampling sites location. PB contributed to the formal analysis and figure design. All authors read and approved the final manuscript. During the preparation of this work the authors used Chat-GPT 3.0 in order to correct and clarify English language. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

Funding

This research has been funded by Agencia Nacional de Investigación e Innovación [fondo POS_FSA_2019_1_1008860, Innovagro FSA_PP_2018_1_148651, FSA_PI_2018_1_148819 and FSA_1_2022_1_175475]; FOCEM—Fondo para la Convergencia Estructural del Mercosur [COF 03/11]; and Comisión Académica de Posgrado.

Availability of data and materials

Sequences generated of 16S rRNA genes, ITS region and shotgun metagenomic data are available at NCBI public repositories under the BioProject accession PRJNA1037557. R scripts for data analysis are available at (https://github.com/mgimenez720/Gimenez_etal_soil_microbiome) with its corresponding data files.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable

Competing interests

The authors declare that they have competing interests.