Herbicide-treated soil as a reservoir of beneficial bacteria: microbiome analysis and PGP bioinoculants in maize

Abstract

Background

Herbicides are integral to agricultural weed management but can adversely affect non-target organisms, soil health, and microbiome. We investigated the effects of herbicides on the total soil bacterial community composition using 16S rRNA gene amplicon community profiling. Further, we aimed to identify herbicide-tolerant bacteria with plant growth-promoting (PGP) capabilities as a mitigative strategy for these negative effects, thereby promoting sustainable agricultural practices.

Results

A bacterial community analysis explored the effects of long-term S-metolachlor application on soil bacterial diversity, revealing that the herbicide’s impact on microbial communities is less significant than the effects of temporal factors (summer vs. winter) or agricultural practices (continuous maize cultivation vs. maize-winter wheat rotation). Although S-metolachlor did not markedly alter the overall bacteriome structure in our environmental context, the application of enrichment techniques enabled the selection of genera such as Pseudomonas, Serratia, and Brucella, which were rare in metagenome analysis of soil samples. Strain isolation revealed a rich source of herbicide-tolerant PGP bacteria within the culturable microbiome fraction, termed the high herbicide concentration tolerant (HHCT) bacterial culture collection.

Within the HHCT collection, we isolated 120 strains that demonstrated significant in vitro PGP and biocontrol potential, and soil quality improvement abilities. The most promising HHCT isolates were combined into three consortia, each exhibiting a comprehensive range of plant-beneficial traits. We evaluated the efficacy and persistence of these multi-strain consortia during 4-week in pot experiments on maize using both agronomic parameters and 16S rRNA gene community analysis assessing early-stage plant development, root colonization, and rhizosphere persistence. Notably, 7 out of 10 inoculated consortia partners successfully established themselves and persisted in the maize root microbiome without significantly altering host root biodiversity. Our results further evidenced that all three consortia positively impacted both seed germination and early-stage plant development, increasing shoot biomass by up to 47%.

Conclusions

Herbicide-treated soil bacterial community analysis revealed that integrative agricultural practices can suppress the effects of continuous S-metolachlor application on soil microbial diversity and stabilize microbiome fluctuations. The HHCT bacterial collection holds promise as a source of beneficial bacteria that promote plant fitness while maintaining herbicide tolerance.

Introduction

The escalating growth of the world's population has increased the need for food production, underscoring the critical role of soil, a non-renewable resource essential for food security. Soils harbor diverse microorganisms vital for recycling organic matter, water filtration, and soil rejuvenation [1]. Agriculture currently relies on the extensive use of chemical fertilizers and pesticides with undeniable benefits. Herbicides, one of the most widely used pesticides, are classified based on their activity, timing and application method, mechanism of action, and chemical structure [2]. Comprising around 30% of worldwide pesticides, chiral herbicides present a risk to both human health and the environment [3]. In particular, S-metolachlor, a representative chiral chloroacetanilide pre-emergent herbicide, was extensively employed in the EU and Serbia to manage annual grasses and broadleaf weeds in the cultivation of over 70 crops, notably maize, soybean, peanuts, sorghum, and cotton [4]. S-metolachlor is moderately persistent in soil, with a half-life ranging from 2.5 to 289 days, depending on factors such as temperature, moisture, and microbial activity [5]. While biodegradation is the primary method for metolachlor breakup in soil, its persistent nature presents a challenge, especially in cooler, drier soils where degradation can be significantly slower [6]. Several studies have reported that S-metolachlor can persist in the soil for an extended period, potentially impacting subsequent planting seasons [7]. The excessive use of herbicides has resulted in environmental pollution, persistence, accumulation, decrease in soil fertility, and adverse effects on non-target organisms, including microorganisms [8]. The widespread and extensive use of herbicides raises global concerns about their environmental persistence and impacts [5, 9].

The ability of soil microbes to degrade S-metolachlor varies. Some microorganisms are able to utilize the herbicide as a carbon source, while others are more sensitive to its presence [10, 11]. Studies have also shown that S-metolachlor can alter enzymatic activities essential for nutrient cycling, which can affect the long-term fertility of soils exposed to repeated herbicide treatments [12].

The main concern revolves around herbicide influence on the soil's web of life, where they disrupt the delicate interactions among microbes, modify the soil’s diversity, affect interspecies dynamics, and interfere with essential enzymatic activities crucial for nutrient cycling [13]. Given these concerns, it is critical to explore how S-metolachlor affects soil microbial communities and their ability to maintain crucial ecosystem functions.

Meeting the growing global demand for food while avoiding the harmful effects of chemicals on the environment is one of today's most pressing challenges. This underlines the need to develop more efficient, cost-effective, and sustainable agricultural techniques that promote soil fertility, ensure high yields, and reduce dependence on agrochemicals. Bioinoculants, which consist of microorganisms, are proving to be promising tools not only for promoting plant growth but also for soil remediation [14]. This results in the need to identify beneficial bacteria that are resistant to the harmful effects of herbicides. Microorganisms have remarkable genetic flexibility and can rapidly adapt to various environmental stressors, including exposure to xenobiotics, by increasing their biotransformation capacity in a given environment to the point where they can utilize it as their sole carbon source [15, 16]. Therefore, soil extensively treated with herbicides can act as a valuable and abundant reservoir of herbicide-tolerant PGP bacteria [17, 18]. These bacteria have the potential to benefit plants in polluted environments, effectively enhancing soil fertility and aiding in the protection against residual pesticide contamination, a prevalent issue in agricultural soils. Beneficial plant growth-promoting bacteria (PGPB) directly improve plant health and growth by providing various compounds such as phytohormones, siderophores, and enzymes that aid nutrient uptake from the environment [19, 20]. They also benefit plants indirectly by competing with pathogens for available nutrients and inducing systemic plant resistance to disease [21]. To exhibit PGP properties, bacteria must be able to colonize the root rhizosphere [22], a nutrient-rich zone that attracts beneficial microbes. Due to the great microbial diversity found in the soil and plant microbiome, researchers are increasingly interested in PGPB which can be used as a beneficial bioinoculant. Bioinoculants should exhibit diverse beneficial properties for plants and distribute complicated tasks among different bacterial populations. In addition, multifunctional inoculants consisting of several strains have the potential to foster interactions among various species, offering opportunities for the shaping and managing of microbiomes within agricultural systems [23].

Our study aimed to understand the impact of the herbicide S-metolachlor on the structure of the soil microbiome community by 16S rRNA gene amplicon community profiling of treated and untreated field plots in different seasons and cropping conditions. Additionally, we aimed to identify herbicide-resistant bacteria that possess PGP potential and could serve as possible bioinoculants combining the most promising isolates into consortia that would exhibit all plant-promoting properties for more sustainable agricultural practices. Finally, the study evaluated the efficacy and persistence of the multi-strain consortia via both 16S rRNA gene community analysis and agronomic parameters in more relevant agricultural environments, i.e. in planta on maize. Further experiments in the field and optimization of fermentation will translate the promising results of the study into real solutions for sustainable agriculture.

Material and methods

Soil sampling

The bulk soil (slightly calcareous chernozem with 30% silt, 17% clay, and 53% sand [24]) was sampled from the experimental agricultural plots (44.87013°N, 20.33186°E) of the Maize Research Institute “Zemun Polje”, Serbia. The plots, subjected to S-metolachlor treatment for 13 consecutive years, underwent either maize continuous cropping (MCC) or maize-winter wheat (MWW) rotations. The plots were sampled in the year in which maize was planted in the crop rotation. Control plots (MCC-C and MWW-C) were devoid of herbicides and manually weeded. Each treatment comprised 4 sampled plots, with 4 samples collected from each plot. For each sample, 6 spots were sampled and pooled. A sterile scoop was used to collect approximately 10 g of soil from a 15 cm depth at each spot, and the samples were pooled into a sterile container. After sterile air drying for 3 h, the samples were homogenized in a sterile mortar. Metagenomic DNA isolation was performed using 0.5 g of the homogenized soil, while the rest was used for S-metolachlor content analysis. The analysis of S-metolachlor content was performed by the Municipal Health Institute Belgrade, Serbia, using an accredited method based on Masiá et al. [25]. The analysis was performed using an Agilent 1290 Infinity II UPLC system equipped with a ZORBAX SB-C18 column (4.6 × 50 mm, 1.8 μm). Mass spectrometry was performed using an Agilent 6495C MS/MS with an AJS ESI ion source. Winter samples were collected in mid-December 2021, 8 months after herbicide application and one month after deep plowing, while summer samples were collected 4 weeks after herbicide application in mid-May 2022 to investigate the herbicide’s effect on microbial diversity at different time points and to assess its impact under both integrative cropping approach and continuous maize cropping.

Metagenomic DNA (mDNA) isolation and 16S rRNA gene amplicon library preparation and sequencing from soil and maize root samples

Metagenomic DNA (mDNA) was extracted from dried, homogenized bulk soil samples from agricultural plots for soil metagenomic analysis and maize root samples from bioinoculation experiments to assess root colonization by herbicide-tolerant PGP consortia. For each treatment and time point, roots were carefully cleaned from adhering soil and rinsed twice with sterile water. Two grams of clean (rinsed) roots were excised, macerated, and flooded with liquid nitrogen to aid cell wall disruption and ensure efficient DNA extraction. DNA extraction from the soil and root samples was performed using the DNeasy PowerSoil Pro Kit (QIAGEN) according to the manufacturer’s protocol. DNA quality was assessed using the Nanodrop ONE Spectrophotometer (Thermo Scientific), and the mDNA was stored at − 80 °C until sequencing.

For 16S rRNA gene amplification, the V3 and V4 hypervariable regions were targeted using Illumina-compatible barcoded primers and PCR conditions following the manufacturer’s protocol (Illumina Inc., San Diego, CA, USA). Individual barcoded libraries were amplified by PCR (Kapa HiFi HotStart ReadyMix from La-Roche, Basel, Switzerland) using the manufacturer’s recommended protocol with a modified annealing temperature of 55 °C) and long primers (16S_Amplicon_PCR_Fw: TCGTCGGCAGCGTCAGATG TGTATAAGAGACAGCCTACGGGNGGCWGCAG; 16S_Amplicon_PCR_Rv: GTCTCGTGG GCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC) [26] incorporating the Illumina adapter sequences, which allowed pooling of multiple samples in one sequencing run. All steps, including PCR amplification, clean-up step with AMPure XP Bead Clean-up (A63880l; Beckman Coulter Inc., Brea, CA, USA), second PCR reaction that incorporated dual indexes and Illumina sequencing adapters using the Illumina Nextera XT Index Kit v2, followed by a final AMPure XP Bead Clean-up, were performed in-house. Amplicon size, integrity, and purity were verified using Bioanalyzer (Agilent Inc., Santa Clara, CA, USA), and the library concentration was determined by fluorimetric quantification using Qubit 2 (Invitrogen Inc., Carlsbad, CA, USA), and finally, library sequencing was performed using 2 × 250 bp MiSeq in-house. The raw sequencing data have been deposited in NCBI's Sequence Read Archive (SRA) and are accessible through BioProject ID PRJNA1091599. The metadata files are given in Supplementary Material (SM1.1, SM1.2).

Sequence data processing and bioinformatic analyses

The.fastq files were processed in qiime2 [27], using the DADA2 [28] plugin for clustering reads into Amplicon Sequence Variants (ASVs). For quality control of the sequencing reads, we used the tools q2-dada2 or q2-deblur within the QIIME 2 framework and the package ShortRead in the programming software R 4.4.1 [29]. The taxonomic assignments were done based on the SILVA database [30] (release 138) for the microbiome of soil samples under herbicide or non-herbicide treatments and a customized dataset of 16S rRNA gene sequences from the 10 isolates used in this study for evaluation of the consortia ability to persist in the rhizosphere and root endosphere. The cd-hit program, with a 98% sequence similarity parameter, was employed for clustering and comparing the ASVs based on sequence similarity. All the ASVs from root samples not match any of the 16S rRNA gene sequences of the customized dataset were removed from the analysis. Obtained datasets from both soil and maize root samples were imported in R using the package qiime2R [27], and the subsequent analyses and plots were drawn using either phyloseq, microbiome, vegan, or reshape2 R-packages [31]. DESeq2 [32] and MaAsLin2 [33] were used for differential abundance analysis when required. The prevalence of each taxon was determined at the Phylum and Genus levels. We excluded the rare taxa to determine the top taxa and only included taxa present in at least 10% of samples. Community richness and diversity were determined using the QIIME diversity core-metrics-phylogenetic command for alpha and beta diversity analysis in the QIIME2 package, and the vegan package in R. Alpha diversity was measured using the Shannon index. To check if the data were normally distributed the Shapiro–Wilk test was run, while for the variances between groups, the Levene test was adopted. Significant differences in alpha diversity were calculated using Welch’s one-way ANOVA, followed by Tukey’s multiple comparisons tests. Beta diversity was analyzed using the Bray–Curtis dissimilarity metric to quantify differences in community composition among samples. The dissimilarity matrix was visualized using non-metric multidimensional scaling (NMDS), a rank-based ordination method that preserves the relative ordering of dissimilarities in a low-dimensional space. NMDS was performed with k = 2 dimensions and a maximum of 100 random starts (trymax = 100) to avoid local minima in the stress optimization process. To obtain size effects and significances for compositional differences between sample categories the permutational multivariate analysis (adonis2, vegan) was used (PERMANOVA tests). We performed a linear discriminant analysis (LDA) using the DESeq package in R to identify differential abundances of bacterial taxa between the two conditions tested. In this analysis differences in the relative abundance of taxa between herbicide and no-herbicide conditions were calculated using the Kruskal–Wallis test. The trend identified by the Kruskal–Wallis test was then tested using the Wilcoxon test. For the comparison, the alpha value of the Wilcoxon test was set at 0.01. For clarity, the results of the LDA were presented as columns rather than individual points, with different ASVs of the same genus that are significantly more abundant in one of the two investigated conditions represented by vertical lines within the column.

Bacterial culture collection isolation

To obtain the culturable part of the soil bacteriome, we used an enrichment method with S-metolachlor as a selective agent. The soil samples for forming bacterial collection were taken from the same MCC and MWW experimental plots as the samples for mDNA isolation. For MWW, the samples were taken from the plots in the year in which maize was grown in the rotation. The herbicides were applied at full label rate (FLR) on the same day as the seed and granules of pyrethroid insecticide tefluthrin (FORCE 1.5G, 15 g/kg of teflutrin, Syngenta). No other pesticides were applied to the plots. The samples were taken 7 days after sowing/spraying. Eight samples were taken from each plot type (approximately 10 g of bulk soil) and pooled. Samples were collected with an ethanol-rinsed spade from a depth of 5–10 cm avoiding any plant material and stored in a sterile container. The samples were then homogenized using a sterile pestle and mortar and 0.5 g aliquots were measured. 0.5 g of fresh soil was added to 50 mL of Minimal Salt Medium (MSM) [34] supplemented with 20 mM glucose, 12 μg/mL of S-metolachlor (Dual Gold 960EC, Syngenta, Switzerland) and 75 μg/mL of cycloheximide (Difco, Detroit, MI, USA). The concentrations of herbicide added to the media were calculated to represent FLR application. The isolates were obtained by enrichment with herbicide after short (3 days) and extended (7 days) cultivation at 30 °C on a rotary shaker (180 rpm). After 3 or 7 days 10% inoculums were transferred to fresh media for the second cultivation batch. From the second batch, serial dilutions were plated on SCA (Starch Casein Agar, HiMedia Laboratories, Mumbai, India) supplemented with 150 µg/mL nalidixic acid to support the growth of Gram-positive bacteria or on MacConkey agar (Difco, Detroit, MI, USA) to support the growth of Gram-negative bacteria. Plates were incubated at 30 °C for 2 days and inspected for growth daily. Colonies were then transferred to fresh SCA or MacConkey plates and minimal medium plates with herbicide as the sole carbon source (MSM agar supplemented with 12 μg/mL of Dual Gold 960EC). Colonies were inspected for phenotype on both rich and minimal media, and phenotypically distinct colonies that grew on both media were picked and cultivated in Luria–Bertani (LB) medium supplemented with herbicides for making glycerol stocks.

HHCT collection characterization

Resistance to the herbicide S-metolachlor was confirmed by assessing growth on MSM medium agarose plates supplemented with 2.28 g/L S-metolachlor (Dual Gold 960EC), replacing the usual carbon addition with glucose. Furthermore, the ability to grow on herbicide was additionally confirmed by cultivation of all selected clones in liquid MSM media with S-metolachlor as the sole carbon source in concentrations that replaced the carbon content of 20 mM glucose for 5 days at 30 °C/180 rpm, re-inoculated 3 consecutive times with both 0.5 and 0.1% inoculum transfer to fresh media. In the end, the liquid culture was spot-inoculated on the LB/herbicide plate to asses for growth. Strains that did not exhibit growth were excluded from the HHCT collection and subsequent experiments. The collection was subsequently thoroughly characterized to exclude possible siblings from further examination. Biochemical phenotyping involved assessing different carbon source utilization on MSM supplemented with 10 g/L of 12 different sugars (glucose, galactose, lactose, xylose, maltose, raffinose, fructose, mannose, sucrose, agarose, glycerol, and mannitol), extracellular enzyme production in assays using urea [35], citrate [36], starch [37], and hydrogen peroxide [38], and the utilization of 10 different monoaromatic compounds (benzene, toluene, phenol, catechol, ethylbenzene, m-xylene, p-xylene, chlorobenzene, 3-methylbenzene, nitrobenzene) as sole carbon and energy source on MSM agarose plates as presented in Supplementary Material (SM2).

In vitro evaluation of plant growth-promotion (PGP) traits in herbicide-tolerant bacteria

The entire HHCT collection was tested for both direct and indirect PGP activities. Each of the assays was performed in triplicate for each bacterial strain. The growth temperature was 30 °C for all strains.

Direct PGP mechanisms

The ability to grow under nitrogen limitation (diazotroph), phosphate solubilization, indoleacetic acid (IAA) production, and siderophore production were assessed. The growth conditions for the isolates varied depending on the strain and included cultivation in LB or Tryptic Soy (TS) [39] medium spiked with 0.23 g/L S-metolachlor to ensure the maintenance of herbicide tolerance.

Growth in nitrogen-free media was assessed using Jensen’s medium [40]. Overnight cultures were washed in phosphate-buffered saline (PBS) [41], and 5 μL of each isolate suspension was spread, a maximum of three strains per plate. The plates were incubated for 8 days. Phosphate solubilization activity was determined on the National Botanical Research Institute’s phosphate growth medium (NBRIP) [42]. The halo around the bacterial colonies was measured after 5–7 days. The colorimetric Salkowski assay [43] was used for the quantitative determination of IAA. Isolates were grown overnight in LB medium supplemented with 5 mM tryptophan, centrifuged, and 1 mL of the supernatant was collected and mixed with 2 mL Salkowski reagent, 1 μL orthophosphoric acid and incubated in the dark for 15–30 min. The optical density was measured after 30 min at 530 nm. Siderophore production was determined using the chromium azurol blue agar (CAS) assay [44]. The plates were incubated for 5 days. Siderophore production was observed as a yellow-colored halo around the colonies.

Indirect PGP mechanisms

Exopolysaccharide (EPS) production, motility (swimming and swarming), and extracellular enzyme production (cellulolytic, pectinolytic, lipolytic, proteolytic) were assessed.

The production of EPS was evaluated by growing the bacteria on yeast extract mannitol (YEM) medium [45] for 48 h. The ability of each isolate to produce EPS was determined visually by observing the presence or absence of mucoid features. To assess swarming activity, bacteria were point-inoculated to the center of the M8 media plate supplemented with 0.5% agar [46], The plates were incubated for 20 h without turning them over. The growth radius was measured to quantify the swarming motility. The same experimental protocol was used to study swimming on M8 media supplemented with 0.3% agar [47]. Cellulase production was determined as described in Slama et al. [48]. The plates were incubated for 2 days. To evaluate the pectinolytic activity, the bacteria were cultured on a pectin agar (PA) medium [49]. After incubation for 24 h, the plates were flooded with a 1% solution of cetyltrimethylammonium bromide (CTAB). The proteolytic activity was evaluated on LB agar plates supplemented with skim milk (12 g/L). The diameter of the clearing zones was measured after 24 h of incubation. Lipolytic activity was evaluated on LB agar plates supplemented with 1% tributyrin. The diameter of the clearing zones was measured after 24 h of incubation.

Biocontrol potential

The biocontrol potential of the bacterial strains was investigated against two common plant pathogens, Fusarium spp. AG1, a fungal rice pathogen from ICGEB fungus culture collection and the bacteria Pseudomonas fuscovaginae UPB0736 [50]. Two different assays were used to evaluate biocontrol activity: a dual culture plate assay [51, 52] for antifungal activity and an agar well-diffusion assay [53] for antibacterial activity. These assays were performed in at least triplicate. For the antibacterial evaluation against P. fuscovaginae, agar plates were prepared in combination with soft and solid agar. First, 10 mL of solid agar was poured into the Petri dish. As soon as it had solidified, 15 mL of soft agar (0.7% agar) containing a 0.1% pathogen overnight culture was layered on top. After the soft agar had solidified, wells were made in the soft agar layer using a wide-mouth sterile pipette tip. 10 μL of the LB overnight culture of each HHCT strain was added to the wells. The plates were incubated at 30 °C and the inhibition zones were measured after 24 h. To evaluate biocontrol activity against fungal pathogens, the assay involved co-cultivation of each bacterial strain with the fungal pathogen on potato dextrose agar (PDA) [51] plates. Mycelial disks (6 mm in diameter) of freshly grown phytopathogenic fungi were placed in the center of the plate, while each bacterial strain was streaked to the edge of the plate. The efficacy of the biocontrol was determined by measuring the growth inhibition zone or growth patterns after 4–7 days of incubation at 30 °C.

Molecular identification of the HHCT culture collection by 16S rRNA gene sequencing

To taxonomically characterize the entire HHCT collection, we amplified the 16S rRNA genes by colony PCR using universal primers fD1Funi 16S (5’-AGAGTTTGATCCTGGCTCAG-3’) and rP2Runi 16S (5’-ACGGCTACCTTGTTAGGACTT-3’), flanking the approximately 1505 bp region of the 16S rRNA gene [54]. GoTaq G2 Flexi DNA polymerase was employed following the manufacturer’s instructions (Promega). For colony PCR, a colony suspension in 50 μl of sterile H₂O was heat-incubated (10 min at 98 °C), and the following PCR conditions were applied: 94 °C for 5 min, followed by 30 cycles of 94 °C for 40 s, 54 °C for 40 s, 72 °C for 1 min, and a final extension at 72 °C for 5 min before cooling to 4 °C. PCR products were purified using the Gel Extraction and PCR Clean-Up System kit (Euroclone S.p.A) and sequenced by GATC (Eurofins Genomics Company, Germany) using the primers 907R (5’-CCGTCAATTCMTTTRAGTTT-3’) or 518F (5’-CCAGCAGCCGCGGTAATACG-3’). All nucleotide sequences were deposited in GeneBank under the accession numbers PP552877-PP552978, OM475755-OM475763, and ON306427-ON306433. Taxonomic identification of HHCT strains was accomplished through alignment with sequences from the Ribosomal Database Project via BLAST analysis at NCBI (http://www.ncbi.nlm.nih.gov).

The design of HHCT-consortia with multiple PGP features

The selection of strains for the formation of the consortia was guided by criteria based on the strong presence of direct PGP traits, their diversity and complementarity, and the principle that each consortium should primarily consist of strains of known rhizosphere-colonizing genera recognized for their PGP traits, such as Pseudomonas, Enterobacter, and Serratia. After an extensive in vitro screening of the herbicide-resistant bacterial collection (Supplementary Material SM3), 10 PGP strains were selected and grouped into three consortia (C1, C2, and C3), each containing 2 to 4 strains based on the complementarity of PGP activities that would ensure optimal growth promotion and stress resilience in maize. The consortia contained the already mentioned genera and three less exploited genera – Sphingobacterium, Sphingobium, and Stenotrophomonas. The small number of strains in each consortium was deliberately chosen to develop a viable, scalable, and potentially field-applicable microbial bioinoculum. The composition of the consortia is shown in Supplementary Material SM4.

Whole genome sequencing and genome mining of selected strains

The genomes of 10 bacterial strains used in consortia studies were sequenced to characterize them further. DNA was extracted using the Norgen Biotek Corp. Bacterial Genomic DNA Isolation Kit (Cat. No. 17900, Norgen Biotek Corp., Thorold, ON, Canada). Quality assessment was conducted using the Nanodrop ONE Spectrophotometer (Thermo Scientific, Waltham, MA, USA) and electrophoresis in 0.8% agarose gel. The complete genomes were sequenced at SeqCenter (Pittsburgh, PA, USA) with the Illumina NovaSeq 6000 platform using 151 bp paired-end reads and following the tagmentation Illumina Nextera XT protocol (Illumina Inc., San Diego, CA, USA). The assembly was performed with Unicycler v. 0.5.0 and the assembly statistics were recorded with QUAST v.5.2.0 [55]. The assembled genomes were uploaded in the JGI Integrated Microbial Genomes and Metagenomes (IMG/M) database and automatically annotated, using the annotation pipeline IMG Annotation Pipeline v.4.16.6 [56] and deposited under the IMG Submission IDs 329,542, 329,543, 329,544, 323,564, 323,566, 323,567, 323,568, 323,569, 323,572, 323,573. Functional annotation and phylogenetic characterization were performed by the IMG/M and DFAST platform [57]. Genome mining was performed within the IMG/M platform to identify genetic determinants associated with plant-beneficial traits and herbicide tolerance based on literature reports and Pfam ID.

PGP-HHCT consortia effects on maize seeds germination

It was of interest to test the 3 designed consortia on maize plants for their ability to promote plant growth and health. The in planta experiments were conducted with the seeds of commercial hybrid maize seeds of ZP Maize 606 [58], a high-yielding variety widely used in Europe and Serbia, which were obtained from the Maize Research Institute “Zemun Polje”. For evaluation of the seed germination promotion, 100 maize seeds per treatment were bioprimed with the microbial consortia C1 (consisting of 2 strains), C2 (consisting of 4 strains), and C3 (consisting of 4 strains). To prepare the suspension of bacterial cells for biopriming all strains were cultured overnight in LB supplemented with 0.23 mg/L S-metolachlor. Given the incapacity of Sphingobium strains to grow in LB, they were cultivated in 1/5 TS medium supplemented with 240 µl/L S-metolachlor. A 20 ml liquid culture of each strain was harvested by centrifugation at 3000 rpm for 10 min. After removing the supernatant the cell pellet was resuspended in PBS and the procedure was repeated. The optical density at 600 nm (OD₆₀₀) was measured, and the concentration was adjusted to 0.4 OD₆₀₀ for each strain. To ensure optimal biopriming conditions, the maize seeds were thoroughly washed with sterile H₂O. 10 ml of each strain suspension (OD₆₀₀ 0.4) combined with other consortia members was mixed with 25 maize seeds in a 50 ml sterile tube. The mixture was incubated for 1 h at room temperature. 100 maize seeds per consortium were germinated in a plate on moist absorbent cotton in three replicates and incubated at 30 °C, while the untreated control seeds were treated with a sterile PBS solution. The seedlings were kept moist by watering with sterile water every two to three days. The average percentage of germinated seeds was measured.

In planta experiments: the influence of PGP-HHCT consortia on the promotion of early-stage development of maize and their ability to colonize the root and persist in the rhizosphere

The experimental setup lasted 27 days, with observations carried out in the growth chamber at 24 °C, controlled humidity, and a 15 h photoperiod. The seeds were planted in unsterilized substrate to simulate more realistic natural environmental conditions. The substrate consisted of a soil-sand mixture (3:4). Each treatment included 20 replicas of plants. Seeds and bacterial inoculum were prepared according to the previously described protocols for the seed germination experiment. After biopriming, the individual seeds were placed 2.5 to 4 cm deep in the pots using sterile tweezers and watered thoroughly to ensure optimal moisture for seed germination. A subsequent inoculation with bacterial consortia was performed seven days after seed planting, when the plants were about 1 cm above the soil, along the stem. The bacterial inoculum was prepared following the same protocol for biopriming, applying 1 ml of PBS consortium suspension per inoculation. Eight days after the second inoculation, the initial harvest of 4 plants was conducted. From each plant, 2 g of roots were collected for metagenomic DNA isolation and preserved at −80 °C for subsequent 16S rDNA sequence analysis evaluating consortia bacterial members’ capacity to colonize roots and survive and persist in the native plant microbiome in pot throughout the experiment, as well as for effects on root microbiome diversity and composition. The raw sequencing data have been deposited in NCBI’s Sequence Read Archive (SRA) and are accessible through BioProject ID PRJNA1091599. Colonization ability was assessed at two time points: 8 days after the second bacterial inoculation and at the end of the experiment, 20 days after the second bacterial inoculation (4 weeks after the initial seed biopriming). The remaining parts of the initially harvested plants (roots and shoots) were thoroughly cleaned and oven-dried to allow measurement of the relevant agronomic parameters—dry biomass and length of roots and shoots. The same assessment was performed for all remaining plants at the end of the 27 day period, with root and shoot parameters measured for final comparisons.

Statistical analysis

Statistical analysis was performed using PRISM software version 10.1.1. (GraphPad Software, Inc., San Diego, CA, USA). For the comparison of data obtained from different groups, treatments, and controls, the one-way analysis of variance (ANOVA) was employed, followed by Tukey’s multiple comparisons tests. A family-wise alpha threshold of 0.05 and a confidence level of 95% were applied to assess the statistical significance of observed differences. The p-values representation was as follows: ‘ns’ for ≥ 0.05, ‘*’ for < 0.05, ‘**’ for < 0.01, ‘***’ for < 0.001, and ‘****’ for < 0.0001.

Results

Bacterial diversity, community structure, and taxonomic composition of S-metolachlor treated and non-treated soils

To understand the impact of the long-term application of herbicide S-metolachlor on the bacteriome biodiversity in the soil, we conducted 16S rRNA gene amplicon community profiling. Following a quality check and removal of reads classified as chloroplasts and mitochondria, the number of reads ranged from 1194–31,145, with an average of 14,624.59 per sample (rarefaction curve, Supplementary Material SM5). The alpha diversity, using the Shannon index, ranged from 5.19–6.53, with an average of 5.57 ± 0.64. Noteworthy variations among different conditions and time points were observed and were statistically validated by one-way ANOVA, as described in the Material and Methods section (Fig. 1A and supplementary figure SM6 for Chao1, Simpson, and Observed indices). In the summer samples, significantly higher biodiversity was observed between soil samples cultivated solely with maize (MCC-C) compared to the ones cultivated in rotation maize and wheat (MWW) (Supplementary Material SM7). However, the difference between MCC-C and both MCC and MWW became significant during the winter. Our observations were that the time-point and the cropping system have a more pronounced influence on bacterial community diversity than herbicide treatment. This suggests that S-metolachlor does not significantly impact bacterial community diversity on a larger scale. However, a significant difference was noted in MCC samples under herbicide application during the winter season, pointing to a less biodiverse soil bacterial community when S-metolachlor diminished in soil. S-metolachlor content analysis of soil samples indicated that in winter, there was significantly less herbicide present in the soil, less than 0.009 mg/kg, compared to samples collected in summer. Specifically, summer samples contained 0.02 mg/kg S-metolachlor (MCC_Control), 0.25 mg/kg (MCC), 0.03 mg/kg (MWW_Control), and 0.11 mg/kg (MWW).

Fig. 1

Bacterial diversity, community structure, and taxonomic composition of S-metolachlor treated and non-treated soils. Alpha diversity (A); Beta diversity (B); Taxonomic composition at the phylum level—ten the most abundant phyla (C); Twenty the most abundant genera (D); Differentially represented bacterial taxa in herbicide-treated versus non-herbicide-treated samples (p-value < 0.05) indicating potential biomarkers associated with herbicide treatment (E)

Variation in bacterial taxa composition between different treatment/time points (beta diversity) indicated that summer soil samples sustained a distinct bacterial community compared to those collected in winter, regardless of herbicide application or type of cultivation. This difference was evident as they formed a distinct cluster on the first axes of the NMDS plot (Fig. 1B). Temporal factors played a significant role in community structure (SM7), as samples collected from December and May cluster separately on the principal axis (Fig. 1B). A slight shift in structure was also observed between soil samples cultivated with maize (MCC) under herbicide and no-herbicide treatments during winter (SM7, p-value < 0.05). In contrast, such a pattern was not observed for soil samples cultivated with wheat, where the community remained stable across all conditions tested.

Taxonomically, Actinobacteriota dominated the soil microbiome across all conditions tested, accounting for 53% abundance, followed by Proteobacteria (23%), Acidobacteriota (15%), Firmicutes (8%), and other phyla contributing 1% in total (Fig. 1C). Despite fluctuations in relative abundance, certain taxa, including Bacillus, Microlunatus, and Blastococcus, consistently dominated the soil microbiome across conditions. Top taxa accounted for 60% of the total relative abundance and accounted for ≥ 1% of abundance within each sample (Fig. 1D).

Several bacterial taxa significantly correlate with the presence or absence of the herbicide. A linear discriminant analysis (LDA) between the bacterial community composition of samples treated with and without herbicide was performed to identify potential biomarker taxa between the two conditions. A graphical representation of the differentially represented taxa (p-value < 0.05) from the two conditions is shown in Fig. 1E. The vertical lines within columns represent different ASVs from the same genus significantly more abundant in one of the two conditions. Among the significantly differentially represented taxa, 31 were more abundant in samples under herbicide treatment, while only 7 were in the not-treated ones. The genus Rhodoplanes was found to be significantly correlated with the absence of herbicide, the genus Gaiella with the presence of herbicide, while genera Chryseolinea, Sphingomonas, Microvirga, Bacillus, Bryobacter, Pirellula, Nocardioides, and Dongia were observed in both conditions.

To conclude, despite fluctuations in community composition, temporal factors and the type of cropping system (monoculture vs. rotation) exerted a more pronounced influence on bacterial biodiversity than the presence of the herbicide S-metolachlor. While the herbicide did not significantly affect the overall community structure, specific differences were observed, especially during winter, suggesting a temporal and herbicide-concentration dependence.

Phenotypic and genotypic characterization of the HHCT collection

From agricultural soil treated with the herbicide S-metolachlor for 13 years, the HHCT collection of 120 bacterial strains was isolated. To characterize phenotypically and biochemically these strains and reduce the presence of siblings, the entire collection was tested for various metabolic traits, such as the ability to utilize various carbon sources and the production of extracellular enzymes (Supplementary Material SM2).

After excluding phenotypically identical isolates the entire HHCT collection was molecularly characterized by amplifying and sequencing the 16S rRNA gene to exclude potential plant and human pathogens from further experiments. The resulting sequences revealed that only three phyla made up the collection with Proteobacteria being predominant (95%) and Actinomycetota and Bacteroidota contributing with 2.5% each. Pseudomonas, Serratia, and Brucella were the most abundant genera in the collection (Fig. 2A). When compared to metagenomics data of MCC and MWW plots from summer, from which the collection was made, all HHCT genera except Pseudomonas were rare, below the threshold of detection. Moreover, Pseudomonas was undetectable in all MWW samples and detected only in MCC samples with relative abundance ranging from 0.040 to 0.311% in different MCC metagenomes.

Fig. 2

Genera composition (A) and screening of direct and indirect PGP mechanisms and biocontrol potential (B) of the entire HHCT collection counting 120 strains

The HHCT collection was further assessed through in vitro assays for various plant-beneficial associated phenotypes (Fig. 2B, Supplementary Material SM3). Direct PGP mechanisms tested were the ability to grow in nitrogen-limited conditions, phosphate solubilization, and production of IAA and siderophores (Fig. 2B). The results show that 32% of the collection exhibited 3 or 4 direct PGP mechanisms tested. The diazotrophic phenotype was observed in 55% of the strains grown on nitrogen-free Jensen's medium, while 32% of the isolates showed phosphate solubilization. The production of IAA was detected in 43% of the isolates. The Blue Agar CAS assay revealed that 77% of the HHCT collection had the potential to synthesize siderophores.

In addition to direct PGP activities, we also evaluated indirect activities, such as the ability to synthesize extracellular enzymes, including lipolytic, proteolytic, cellulolytic, and pectinolytic activities, as well as exopolysaccharide production and motility ability (Fig. 2B). Lipolytic activity was observed in 42.5% of the HHCT collection grown on tributyrin-supplemented medium, while 29% showed proteolytic ability on skim milk agar plates. 57.5% of the collection exhibited cellulolytic activity with carboxymethyl cellulose, while 50% showed pectinolytic properties on the PA medium. 40% of the bacterial collection showed production of exopolysaccharides. Swimming and swarming were observed in 52.5 and 21% of the isolates, respectively, which could indicate that the HHCT collection has the potential to colonize the rhizosphere and enhance nutrient uptake by plants.

The biocontrol potential of the isolates was also evaluated against two common plant pathogens, the fungal Fusarium spp. and the bacterial P. fuscovaginae. Out of the total isolates screened, 22 exhibited antifungal activity, while 13 showed antibacterial activity. Two strains had both activities—Serratia sp. KS-15, and Pseudomonas donghuensis KPS-14.

Summarizing our assessment of the HHCT collection, Pseudomonas, Serratia, and Brucella were the predominant genera, with 32% of the collection having at least 3 direct PGP mechanisms. In addition, the collection showed a diversity of indirect PGP mechanisms, including extracellular enzyme production and biocontrol activities against fungal and bacterial plant pathogens.

After in vitro screening and molecular characterization of the herbicide-resistant bacterial collection, the selection of strains for consortia formation was based on both their herbicide tolerance and their complementary PGP properties. We strategically grouped 10 strains into three consortia consisting of 2 to 4 strains from genera recognized for PGP and rhizosphere colonization, all showing high and complementary in vitro PGP and biocontrol activity, to maximize the range of beneficial PGP functions and compare their consistency and performance in planta. The composition of the consortia is shown in Supplementary Material SM4 and a summary of the relevant phenotypes of the consortia members can be found in Fig. 3. The first consortium (C1) was a “minimal” consortium consisting of only two strains. One strain showed high activity for all direct PGP mechanisms tested (Enterobacter mori KPS-16), while another (Pseudomonas donghuensis KPS-14) had both biocontrol activities. The second consortium (C2) was based on Serratia strains (KS-9 and KS + 7) and additionally contained Pseudomonas putida KPS-3 with 3 direct and 7 overall PGP traits, and Sphingobacterium prati KS-7. Both Serratia strains also showed potential biocontrol activity. The third consortium (C3) was based on two Enterobacter mori strains (KM + 1 and KM + 3) and additionally contained Sphingobium rhizovicinum (KPM + 5) and a Stenotrophomonas rhizophila (KPS-13). C3 combined strains that exhibit all direct and indirect PGP characteristics, thus ensuring a comprehensive spectrum of potential benefits.

Fig. 3

Overview of beneficial PGP phenotypes of consortia members chosen for further in planta evaluation

Each consortium was tested for its efficacy under controlled conditions in in pot experiments in maize seed germination, early-stage plant development, and rhizosphere and root endosphere colonization ability.

Consortia members’ genome mining for PGP traits and herbicide-tolerance

Genomes of the consortia members were mined for genes encoding plant beneficial properties and genes involved in herbicide tolerance mechanisms, general detoxification, and metabolism of xenobiotics including metolachlor and other, similar herbicides (e.g. atrazine, 2,4-D) (Table 1). We detected the gene tfdA associated with chloroacetamide herbicide degradation, and the genes atzA, atzC, and atzE linked to atrazine breakdown, where atzC was present in all strains. Furthermore, we found genes like yfcG, gstA, kat, ald, and p450, which are instrumental in the detoxification of xenobiotics, including herbicides with kat and ald being also omnipresent. Sphingobacterium sp. KS-7 from C2 was the only strain lacking a set of genes for detoxification via glutathione S-transferase. In summary, each consortium possessed in total various features that could be beneficial for sustainable agricultural practices.

Table 1 Genomic screening of PGP and herbicide-tolerance related genes

In planta experiments in growth-chamber on maize inoculated with PGP-HHTC consortia

The seed germination assay revealed the benefits of all three consortia on maize seed germination. Notably, consortia C1 and C2 exhibited a 15% enhancement in germination rate, while consortium C3 demonstrated a 9% increase compared to untreated control seeds, which exhibited a germination rate of 65%.

In terms of the promotion of early-stage plant development, C1 and C2 were found to be the most effective (Fig. 4A) in growth-chamber conditions. C1 and C2 promoted plant height by 17.6% and 10.4%, respectively, while C3 did not significantly affect plant height compared to the non-treated control. Regarding root length, all three consortia positively influenced it (Fig. 4B), with C1 having the greatest effect (27.7%), followed by C2 (25.8%) and C3 (3.4%).

Fig. 4

Evaluation of the promotion of early-stage plant development by herbicide-tolerant bacterial consortia, effects on plant height(A), root length (B), shoot dry biomass (C), and root dry biomass (D). In planta experiments on maize were performed in a growth chamber with 20 plant replicas per treatment. * Indicates a significant difference between treatments (p < 0.05) obtained using the ANOVA test. The representation of p-values is as follows: ‘*’ for p < 0.05, ‘**’ for p < 0.01, ‘***’ for p < 0.001, and ‘****’ for p < 0.0001

In terms of shoot biomass (Fig. 4C), all three consortia influenced biomass positively, with C1 having the greatest effect (47%), followed by C2 (37.8%) and C3 (8.2%).

Root biomass was however decreased by all three consortia at the end of the experiment (Fig. 4D). Interestingly, at the beginning of growth (8 days) at the first measurement, all three consortia positively influenced root biomass.

Evaluating herbicide-resistant bacterial consortia: root colonization, persistence, and microbiome dynamics in maize pot experiments

To investigate how the microbiome was affected by the application of the three consortia and their persistence within the root compartment, the 16S rRNA gene amplicon rhizosphere microbiome profiling was performed. In the plant growth-promoting experiments described above, the rhizosphere and root endosphere colonization ability were examined at two-time points at 8- and 20-days after the second inoculation with the bacterial consortia (4 weeks in total). The C1 and C2 showed an uneven distribution of species in 8 days post-inoculation, however after the next 20 days they reached an even distribution with the Shannon index between 6.5 and 6.75 (Fig. 5A). Beta diversity analysis revealed that the total root microbiome composition and structure were not significantly affected following the inoculation with all three consortia used in this study (Fig. 5B). Seven out of ten inoculated strains forming the three consortia were detected at both time points tested with comparable abundance, proving that most of the bacterial isolates tested can colonize and persist in the maize root system after 4 weeks (Fig. 5C). While Enterobacter and Pseudomonas were detected in similar relative abundance at the two-time points in the consortia C1 and C2, Sphingobacterium was more abundant in roots after 20 days. Interestingly, in C3, Pseudomonas was detected at low abundance on the 8th day but completely disappeared after 20 days. Stenotrophomonas sp. KPS-13 showed no ability to colonize, as we could not detect its presence in the root compartment at any of the time points tested.

Fig. 5

Maize root colonization after 8- and 20-days post-consortia inoculation, based on metabarcoding analysis; Alpha diversity (A); Effects of different consortia on beta-diversity of the root microbiome (B); Relative abundance of inoculated strains after 8- and 20-days post-consortia inoculation (C)

Discussion

The widespread use of herbicides in conventional agriculture has led to their persistence in the environment, significant environmental degradation [86], and health risks [87], necessitating the exploration of alternative approaches for more sustainable crop management that, among others, include the application of plant-growth-promoting (PGP) bacteria as biofertilizers [88]. The main question concerning soil health revolves around herbicide influence on the network of interactions among soil microbiota, modification of their diversity, abundance, interspecies dynamics, and enzymatic activities essential for nutrient cycling [13].

To investigate the possible effect of the herbicide S-metolachlor on the soil microbial community diversity and structure, 16S rRNA gene amplicon sequencing was employed. Results indicate an increased robustness of the bacterial community to herbicide treatment, suggesting potential adaptation of soil microbes to herbicide exposure over time, consistent with existing literature [15, 89]. Temporal and soil cultivation factors, including seasonal variations and crop rotation practices, contribute to shaping the composition and dynamics of the soil microbiome, exerting more influences on microbial diversity than herbicide treatment alone [90, 91]. However, the herbicide effect becomes more evident throughout the year, as control plots without herbicide treatment showed higher biodiversity, indicating that the presence of S-metolachlor decreases bacterial community richness. Another factor contributing to the lessening of herbicide exposure in the soil is the propensity of the herbicide to disperse in the environment [5] and the tendency to leach into groundwater, as observed in our field conditions where control plots after four weeks contained a small amount of herbicide. Consequently, the winter samples contained only residual concentrations of S-metolachlor. Our findings suggest the potential benefits of crop rotation compared to continuous maize cultivation and emphasize the potential benefits of crop rotation in mitigating the temporal effects of herbicide use for sustainable agricultural practices [91, 92].

The dominance of Actinobacteriota, Proteobacteria, Acidobacteriota, and Firmicutes is consistent with wider studies of the soil microbiome, with genera such as Bacillus, Microlunatus, and Blastococcus, previously reported to be able to tolerate herbicides, consistently predominant under the conditions observed [93, 94]. Although we do not observe major differences between herbicide-treated and untreated samples in the most abundant bacterial genera, we have detected an enrichment of 31 genera that may be better adapted to life in the presence of herbicides (Fig. 1E). This adaptation could be transient due to metabolic and enzymatic plasticity [95] of the general microbial response to chemical contaminants such as antibiotics, pesticides, and herbicides [9, 95,96,97], including an increase in oxidative stress enzymes [15], or the presence and activity of various pumps and porins in the cell membrane [98]. The adaptation can also be the result of possessing various genes carrying dioxygenase for the degradation of aromatics that can influence herbicide degradation [99] located on plasmids [100] that can be acquired by HGT.

By integrating both culture-independent and culture-dependent techniques, we have obtained data on the soil microbiome and its response to herbicide exposure. The culture-dependent method, in which herbicide-tolerant strains were isolated from herbicide-treated soil in vitro, revealed that the dominant genera in the HHCT collection were Pseudomonas, Serratia, and Brucella. The abundance of Serratia and Brucella in our soil samples was below the limit of detection, while only Pseudomonas had a detectable relative abundance (below 0.311% in the MCC and MWW microbiome). This low abundance can be attributed to the intrinsic property of microbial communities, which are often dominated by a few species [101, 102]. However, the application of high-throughput sequencing methods should have detected numerous species with low abundance [103]. Apart from the general problem of the culturability of soil microorganisms, where only a small proportion of bacteria can be successfully cultivated in vitro [104], the dominance of these genera in the HHCT collection suggests that our selection method has filtered rare genera into a herbicide-tolerant portion of the culturable microbiome. These genera are extensively studied for their PGP traits and are known for their significant roles in plant–microbe interactions and beneficial effects on plants [105,106,107], as well as for their ability to tolerate herbicides [95, 105, 108,109,110]. The bacterial collection we isolated mainly consisted of γ-proteobacteria, which are known predominant root-associated plant bacteria across various plant species [111]. Interestingly, other HHCT isolates also belong to the fraction of the soil bacteriome with a very low abundance. The enrichment of certain genera/species in our bacterial collection is consistent with the findings of Massot et al. [112], who reported that glyphosate exposure can influence bacterial isolation by favoring certain genera/species over others. This suggests that herbicide enrichment may lead to a bias in the selection of bacterial types during isolation. Unsurprisingly, 57.5% of the collection comprises strains belonging to the Pseudomonas and Serratia genera. Pseudomonas is known to be an indicator of negative anthropogenic influence on the environment [113] and both genera are extremely resilient to the influence of xenobiotics owing to the presence of efficient efflux pumps and various enzymatic defense mechanisms [15, 98, 114, 115]. HHCT collection exhibited a diverse range of PGP traits such as nitrogen fixation, phosphate solubilization, IAA production, and siderophore synthesis, essential for enhancing soil fertility and plant growth. Notably, a significant portion of the collection demonstrates diazotroph and phosphate-solubilizing capabilities that are important PGP phenotypes [116]. The production of IAA suggests potential benefits for root and overall plant growth [117], especially in crops with shallow roots or stressful environmental conditions [118]. The bacterial collection also exhibits a high capacity for synthesizing siderophores, crucial for iron acquisition by plants [119]. Extracellular enzymatic activities indicate a potential role in organic matter turnover and nutrient cycling, while exopolysaccharide production may contribute to soil aggregation and nutrient retention [120]. Motility abilities suggest colonization of the rhizosphere [121], enhancing nutrient uptake [122]. Recently, plant growth-promoting bacteria have gained high interest as potential biofertilizers and a sustainable alternative to chemical fertilizers since the increase in the use of chemical fertilizers in agriculture reached levels of concern in terms of sustainability and environmental impact [123]. Traditionally, bioinoculants, designed to enhance plant growth, have relied on identifying effective soil microbes with traits conducive to plant development [124]. By setting up this selection process so that tolerance to herbicides was a prerequisite and then determining PGP traits [19, 88, 125,126,127], we obtained a set of isolates with the dual trait of herbicide tolerance and plant growth promotion which can potentially be used in agroecosystems where conventional chemical inputs, including herbicides and fertilizers, are applied. This adaptability could enable the development of microbial solutions that thrive in different soil types, climates, and agricultural conditions offering resilience in the face of changing environmental conditions [15]. Considering all these challenges, our study further focused on generating herbicide-resistant bacterial consortia with diverse PGP traits, aiming to test them as beneficial plant probiotics [112, 128] for plant growth-promotion in maize. Some single-strain microbial bioinoculants have proven effective in agriculture (e.g., based on Trichoderma, mycorrhiza, or rhizobia [129]). However, challenges arise in the agricultural use of microbial inoculants, particularly in competing with native microbes and adapting to changing environmental conditions [130]. The utilization of microbial consortia comprising multiple strains with complementary beneficial traits can be a way to more efficient root colonization. The study revealed that the culturable part of the soil bacteriome, subjected to intensive treatment with the herbicide S-metolachlor, is a valuable source of PGP bacteria.

To better characterize consortia partners we searched for genes associated with both PGP/plant colonization [131, 132] and mechanisms known to be involved in herbicide degradation and detoxification, as well as those implicated in herbicide resistance mechanisms. Our findings revealed the presence of several key genes within the genomes of our bacterial isolates, indicating their potential for herbicide tolerance and detoxification. For instance, we observed the presence of genes atzA, atzC, and atzE, involved in the degradation of atrazine, another commonly used herbicide. While atrazine and S-metolachlor differ structurally, the presence of these genes suggests the potential for cross-reactivity or shared metabolic pathways that could confer tolerance to S-metolachlor [75,76,77, 133]. Furthermore, we revealed the presence of tfdA, which encodes an enzyme involved in the degradation of chloroacetamide herbicides [78]. Additionally, we identified genes such as yfcG, gstA, kat, ald, and p450, which play crucial roles in detoxifying xenobiotics, including herbicides, by facilitating their conjugation, oxidation, and metabolism, underscoring their potential for herbicide detoxification and tolerance [5, 79, 80, 82,83,84]. Overall, these data highlight the potential of tested bacteria for application in herbicide-contaminated environments and conventional agriculture for increased sustainability of agricultural practices.

Ten strains with a high potential for promoting plant growth assembled into three consortia designed to potentially provide a comprehensive range of beneficial traits [22] that effectively promoted seed germination and early-stage plant development in planta. The design of the consortium is one of the most challenging aspects of the development of bioinoculants. Different approaches have been tested mainly predicting the compatibility of two strains [134, 135]. Our approach was based on the complementarity of PGP traits and experience working with PGP bacteria. A small number of strains were deliberately selected in each consortium to develop a viable, scalable, and potentially field-applicable microbial bioinoculum. C1 was a “minimal” consortium consisting of only two strains that had between them a full range of direct PGP traits, most indirect PGP traits (all except cellulase), and both antibacterial and antifungal biocontrol activity. We felt that more than four strains would limit the scalability of industrial fermentation, especially if the strains need to be fermented separately. Therefore, C2 and C3 consisted of four strains. In planta results suggest that we successfully assembled consortia with the potential to improve germination rates, a crucial factor [136] and a critical stage in plant growth and productivity. The use of seeds with a lower germination rate underlined and confirmed this potential. The impact of the consortia on early-stage maize development varied, with some promoting shoot biomass (C1) and others influencing root length (C1 and C2). Strains from C1 (Pseudomonas sp. KPS-14 and Enterobacter sp. KPS-16) share between them the presence of all PGP genes listed in Table 1, except acdS for 1-aminocyclopropane-1-carboxylate deaminase. This gene is absent from C3 members as well as phnC critical for phosphate import and stress alleviating proA. The streamlined composition of the C1 of only two strains makes it highly feasible for application. This simplicity would not only facilitate scalability but enhance the ease of monitoring and managing its effects in diverse agricultural settings. Thus, C1 represents a compelling candidate for future field trials and commercialization efforts aimed at optimizing plant growth and soil health.

The rhizosphere microbiota plays a crucial role in plant adaptation to their environment and promotes a mutually beneficial relationship [125]. Bacteria equipped with successful colonization traits contribute to complex partners’ communication and effective competition within the environment [137]. The 16S rRNA gene community profiling of maize root microbiome from in pot cultivation revealed a conserved abundance of Enterobacter and Pseudomonas strains throughout the experiment highlighting their root colonization ability [138, 139]. Interestingly, other consortia members were detected at different time-points tested. Sphingobium and Serratia were most abundant in the earlier phase, 8 days post-inoculation. Despite its ability to colonize the plant [140] and high growth rate, Serratia were not detected in the later time point. Sphingobacterium was detected only 20 days after consortia inoculation suggesting that for its establishment in rhizosphere the positive interaction with other members of the root community may be needed. This indicates likely synergistic interactions among the PGP-active strains within the consortia, showcasing their collective impact on plant growth and underscoring the nuanced outcomes that arise from their combined presence [141]. The successful establishment and persistence of these consortia in the maize root system without adverse effects on maize endophyte biodiversity is crucial to their beneficial PGP potential [22, 125].

To conclude, this study sheds light on the intricate interplay between soil microbiome dynamics, herbicide exposure, and the potential of herbicide-tolerant bacterial consortia in promoting plant growth. Despite fluctuations in bacterial community composition, temporal factors, and agricultural practices exerted a more pronounced influence on soil microbiome biodiversity than the presence of the herbicide S-metolachlor. These findings not only contribute to our understanding of microbial ecology in agroecosystems but also underscore the practical implications of harnessing beneficial microbial consortia to optimize agricultural practices and mitigate the impacts of herbicides on soil health. The identified HHCT collection showcased diverse PGP traits, emphasizing their potential in sustainable agriculture. Through in planta experiments, the formulated multi-PGP consortia exhibited promising results in enhancing maize seed germination, early-stage plant development, and root colonization without significantly altering the resident root microbiome. Our findings contribute to the limited literature on herbicide-tolerant plant growth-promoting bacteria, specifically on the widely used herbicide S-metolachlor. Future research should focus on the optimization of bioinoculants and field trials to validate the effectiveness of HHCT consortia under real agricultural conditions. On the one hand, different types of crops and environmental conditions should be studied, and on the other hand, the interactions between consortia members to evaluate the interspecies interactions at the molecular level. In addition, incorporating these consortia into integrated soil management practices, including crop rotation and reduced herbicide use, could bring further advances in sustainable practices in conventional agriculture. The HHCT collection holds promise for the bioprospecting of herbicide degradation enzymes and the further exploitation of bacterial strains with bioremediation potential that could contribute to the reduction of residual herbicide accumulation in soils and thus support long-term agricultural sustainability.