Long-term conservation tillage with reduced nitrogen fertilization intensity can improve winter wheat health via positive plant–microorganism feedback in the rhizosphere

Abstract Microbiome-based solutions are regarded key for sustainable agroecosystems. However, it is unclear how agricultural practices affect the rhizosphere microbiome, plant–microorganism interactions and crop performance under field conditions. Therefore, we installed root observation windows in a winter wheat field cultivated either under long-term mouldboard plough (MP) or cultivator tillage (CT). Each tillage practice was also compared at two nitrogen (N) fertilization intensities, intensive (recommended N-supply with pesticides/growth regulators) or extensive (reduced N-supply, no fungicides/growth regulators). Shoot biomass, root exudates and rhizosphere metabolites, physiological stress indicators, and gene expression were analyzed together with the rhizosphere microbiome (bacterial/archaeal 16S rRNA gene, fungal ITS amplicon, and shotgun metagenome sequencing) shortly before flowering. Compared to MP, the rhizosphere of CT winter wheat contained more primary and secondary metabolites, especially benzoxazinoid derivatives. Potential copiotrophic and plant-beneficial taxa (e.g. Bacillus, Devosia, and Trichoderma) as well as functional genes (e.g. siderophore production, trehalose synthase, and ACC deaminase) were enriched in the CT rhizosphere, suggesting that tillage affected belowground plant–microorganism interactions. In addition, physiological stress markers were suppressed in CT winter wheat compared to MP. In summary, tillage practice was a major driver of crop performance, root deposits, and rhizosphere microbiome interactions, while the N-fertilization intensity was also relevant, but less important.


Introduction
The increasing demands for plant-based products together with the r educed av ailability of sites for a gricultur e has led to mor e intensiv e farming pr actice ov er the last decades (Hoang et al. 2023 ).This intensification, often ac hie v ed by high use of a gr oc hemicals, has negative consequences for soil quality and overall ecosystem functions (Kopittke et al. 2019 , Timmis andRamos 2021 ).To counter act the incr easing degr adation of soils, less intensive, so-called conservation farming practices including diverse crop rotations, r educed tilla ge combined with lo w er pesticide and fertilizer use have been promoted (Hobbs et al. 2008 ).
Soil micr oor ganisms play an essential r ole for soil structur e and fertility (Fierer 2017 , Banerjee and van der Heijden 2023 ).Furthermor e, soil micr oor ganisms r epr esent the r eservoir fr om whic h the plant assembles its rhizosphere (RH) microbiome (Philippot et al. 2013 ).T he RH, i.e .the carbon-enriched thin layer of soil surr ounding r oots and influenced by root activity (Berg and Smalla 2009 ), is a hot spot for microbial activity and is considered one of the most complex and diverse ecosystems (Hinsinger et al. 2009, Raaijmakers et al. 2009 ).T he RH microbiome pla ys a pivotal role in plant de v elopment and health (Ber endsen et al. 2012, Mendes et al. 2013 ).For instance, an impr ov ed plant toler ance to abiotic and biotic stressors can be the result of beneficial plantmicr oor ganism inter actions in the RH (P anke-Buisse et al. 2015, Mommer et al. 2016 ).Ther efor e, engineering the soil and RH microbiome is crucial for sustainable agroecosystems (Toju et al. 2018, Trivedi et al. 2021 ).
Plants hav e e volv ed v arious mec hanisms to modulate their RH micr obiome (Ber endsen et al. 2012, Lar een et al. 2016 ), including the release of root exudates and other organic rhizodeposits, which can account for up to 11% of photosynthetically fixed carbon (Jones et al. 2009 ).Low molecular weight (LMW) root exudates, released either by diffusion or via controlled secretion, comprise the full range of primary metabolites as well as a wide range of secondary plant compounds.Gener all y, LMW sugars in root exudates represent an easily available carbon source for micr oor ganisms.Exuded carboxylates such as malate and citrate are c hemo-attr actants and pr efer ential carbon sources for N 2 -fixing RH bacteria.Mor eov er, they can act as metal chelators and can thereby contribute to the mobilization of micronutrients and sparingly soluble phosphorus (P) in soils.Ad ditionally, carbo xylates can neutralize toxic aluminum as well as ada ptiv el y modify r oot arc hitectur e (Canarini et al. 2019 , Neumann andLudewig 2023 ).Amino acids are important nitrogen (N) sources for microorganisms .Furthermore , they can function as precursors for microbial production of phytohormones, signals in adaptations for N acquisition and as metal chelators (Canarini et al. 2019 , Neumann andLudewig 2023 ).Exuded secondary metabolites comprise also div erse bioactiv e compounds (Philippot et al. 2013 ), suc h as benzoxazinoids (BXs; Kudjordjie et al. 2019, Cadot et al. 2021 ), coumarins (Stassen et al. 2020 ), flavones (Yu et al. 2021 ), triterpenes (Huang et al. 2019 ), and camalexin (K oprivo v a et al. 2019 ), whic h wer e found to contribute to r oot-micr obiome inter actions and various crosskingdom relationships.
Pr e vious studies demonstr ated that the legacy of a gricultur al practices affect the RH microbiome (Sommermann et al. 2018, Babin et al. 2019, Cerecetto et al. 2021 ), and thereby the performance and health of plants (Chowdhury et al. 2019, Babin et al. 2021, Bourceret et al. 2022, Flemer et al. 2022 ).Using lettuce as a model in minirhizotrons with root observation windows, Neumann et al. ( 2014 ) and Windisch et al. ( 2021 ) sho w ed that the soil type and the fertilization strategy affected the root exudation profiles and accumulation of RH metabolites likely as a result of plant interactions with the soil-specific microbiomes.Howe v er, r ealistic insights into the dynamics of root exudate releases and turnover under field conditions and related interactions with pathogenic and beneficial soil micr oor ganisms r emain lar gel y unexplored (Kawasaki et al. 2018, Oburger et al. 2022 ).Consequently, factors influencing plant-micr oor ganism inter actions and processes in a gr oecosystems ar e not yet fully understood (Canarini et al. 2019 ).
In order to exemplarily address these knowledge gaps and provide a holistic understanding of soil-plant-micr oor ganism interactions in a gricultur al settings, we installed r oot windows, similar to large rhizotrons, in winter wheat plots of a long-term field experiment managed under contrasting tillage types [cultivator (CT) vs. mouldboard plough (MP)] and intensities of N-fertilization and pesticide/gr owth r egulator use [intensiv e (Int) vs. extensiv e (Ext)].An interdisciplinary a ppr oac h enabled us to assess r oot gr owth c har acteristics and RH metabolites in situ along with the soil and RH microbiome .Abo veground, shoot biomass , plant nutritional status, and physiological stress indicators were monitored to assess plant health.

Study site and experimental design
The long-term field trial in Bernbur g (German y; 51.82 • N, 11.70 • E), which was established at Anhalt University of Applied Sciences in 1992 on a loess c hernozem ov er limestone (Deubel et al. 2011 ), served as the study site .T he field trial consists of five 1.2 hasized plots, which are administered by a yearly crop rotation comprising grain maize ( Zea mays ), winter wheat 1 ( Triticum aestivum ), winter barley ( Hordeum vulgare ), winter r a peseed ( Brassica napus ssp.napus ), and winter wheat 2. Each plot is managed with conventional MP (20-30 cm ploughing depth, soil inversion) or conserv ation cultiv ator tilla ge (CT, 12-15 cm depth, flat soil loosening) under either intensive N-fertilization and pesticide/growth r egulator a pplication according to usual farming pr actice (Int) or r educed, extensiv e N-fertilization without addition of fungicides and gr owth r egulators (Ext).T hus , on eac h plot four a gricultur al practices (MP-Int, MP-Ext, CT-Int, and CT-Ext) eac h in four r eplicates were compared.In this study, we focused on winter wheat 2 (cultivar Lemmy) in the season 2018/2019, which received until sampling 130 kg N ha −1 or 60 kg N ha −1 in Int or Ext, r espectiv el y ( Text 1 , Supporting Information ).
Root windows were installed in spring 2019 at EC 29 in the marginal strips of each treatment ( n = 4; Figure S1 , Supporting Information ; Neumann et al. 2009 ).In brief, 50-cm-deep soil profiles were cut with steel plates next to the young wheat plants and cov er ed with plexiglass plates, which were fixed with square timbers .T he windo ws w er e thermall y isolated with two layers of Styrodur panels and cov er ed with field soil and plastic sheets.

Sampling
Appr oximatel y 6 weeks after root window installation and shortly before flo w ering (EC 59), sufficient root development was observed ( Figure S2A , Supporting Information ) and root windows were removed for microsampling of RH soil solutions (Windisch et al. 2021 ).Sampling at v egetativ e gr owth sta ge ensur ed high r oot exudation activity (Marschner 1995 ).Briefly, moist sorption filters (MN818, 5 Ø mm, Mac her ey & Na gel, Dür en, German y; Figur e S2B , Supporting Information ) were placed on apical root zones (two filters: 1-2 cm behind the root tip) and subapical root segments (two filters: 8-10 cm behind the root tip).For each root window, sampling was performed along five different roots growing at the observ ation plane.Furthermor e, sor ption filters wer e placed on soil without visible root contact [root-affected (RA) soil] in order to determine bac kgr ound noise .T he sor ption filters of eac h r eplicate and root zone were pooled after four hours of collection, immediatel y tr ansferr ed to 2 ml methanol 80% (v/v) for stabilization and k e pt at −80 • C until HPLC/HPLC-MS analyses for metabolite profiling (Windisch et al. 2021 ; see below).
After sorption filter collection, flag leaves of three different plants were sampled and pooled for plant gene expression, phytohormone and str ess-r elated metabolite anal ysis (see below).The leav es wer e cut into small pieces (5 cm) and tr ansferr ed in 15 ml tubes containing RNAlater solution (Ambion, Life technologies, Carlsbad, USA).The leaf tissue was stored overnight at 4 • C to allo w thorough RN Alater penetr ation and then tr ansferr ed to −20 • C for long-term stor a ge.
For nutrient anal ysis, fla g leav es of three plants were sampled and oven-dried at 60 • C for 3 days and subsequently stored in a desiccator for another 2 days (see below).
For standardized root sampling, six winter wheat plants were excav ated fr om eac h r oot window and divided into shoot and root.RA soil, defined as soil loosely adhering to roots, was obtained by shaking of roots and subsequently sieved to 2 mm and stored at −20 • C until total microbial community (TC)-DNA extraction.For one replicate of treatment CT-Int, no RA soil was mistakenly sampled resulting in this case in only three instead of four replicates.
For RH microbiome analysis we used complete root systems of winter wheat, fr om whic h, after brief washing with sterile tap water, microbial cells and strongly adhering soil particles were recov er ed fr om 5 g r oots by thr ee times 1 min Stomac her tr eatment and centrifugation according to Sc hr eiter et al. ( 2014 ).The resulting RH pellets were k e pt at −20

Determination of a bov e-and belowground plant gro wth par ameters
Shoot dry mass (SDM) and root dry mass (RDM) was determined after drying to mass constancy at 65 • C with 15% r elativ e humidity.The deepest root penetration visible along the observation window was recorded as rooting depth (RD).The WinRHIZO r oot anal ysis system (Regent Instruments , Quebec , Canada) was used to measure total root length (TRL) and fine root length (FRL; 0-0.2 mm diameter) of excav ated r oots by optical scanning.Root hair size (RHS) was determined after magnification of highresolution digital photographs taken from roots growing along the observation windows (Zeiss Axio vision software , Oberkochen, Germany).

RN A extr action and plant gene expression analysis
Leav es wer e r etrie v ed with sterile for ceps, excess RN Alater solution was r emov ed and samples were immediately submerged in liquid nitrogen and pulverized.The RNeasy Plant Mini Kit (Qiagen GmbH, Hilden, Germany) was used to extract RNA from 100 mg lea ves .After quantification of RN A b y NanoDr op spectr ophotometer (Thermo Fisher Scientific), cDN A w as synthesized from 2 μg of RNA with the High-Capacity cDNA Re v erse Tr anscription Kit with RNase Inhibitor (Applied Biosystems, Foster City, USA).
Based on an extensive literature search, 30 target genes from wheat ( T. aestivum ) were selected for expression analyses ( Table S1 , Supporting Information ).These genes are associated with biotic and abiotic stress responses [e.g.genes involved in signalingpathw ays mediated b y salic ylic acid (SA), jasmonic acid (JA), and ethylene] and in the first line of defense (MAP kinases , peroxidase , catalase, and superoxide dismutase).In addition, several genes in-volved in N-metabolism and Fe-transport in wheat were analyzed.The r efer ence genes coding for ubiquitin (Ubi) and elongation factor 1 α (EF1 α) were used for normalization.The comparative CT method (Livak and Schmittgen 2001 ) was a pplied.The tar get and endogenous control genes were validated and only primers with 100% ( ± 10%) efficiency were used.For this purpose, the target gene primers and the endogenous controls were used to amplify RN A (cDN A) from winter wheat with a concentration range of 100 ng to 100 pg in a series of 10-fold dilutions in triplicate .T he qPCR was performed with Power SYBR Green Supermix using a pe-qSTAR 96Q thermal cycler (PEQLAB Biotechnologie GmbH, Erlangen, Germany) as described previously (Chowdhury et al. 2019 ).A total of 28 primer pairs, that met the criteria, were selected for further analyses ( Table S1 , Supporting Information ).QPCR of four biological replicates was performed in three technical replicates (Chowdhury et al. 2019 ).Data were first normalized to the endogenous control and logarithmically transformed to fold change differences .T he standard error of the mean was calculated from the av er a ge of the biological replicates.

Analysis of plant hormones and stress-related metabolites
The plant hormones (JA and SA) and selected stress metabolites wer e anal yzed in homogenized and shoc k-fr ozen plant tissue.UHPLC-MS analysis of phytohormones in shoots was carried out as described by Moradtalab et al. ( 2020 ).Total phenolics were determined, after extraction with 80% v/v methanol, spectr ophotometricall y at 750 nm, using the Folin method (Swain and Hillis 1959 ).Pr oline anal ysis was conducted spectr ophotometrically at 520 nm after acetic acid and acid ninhydrin derivatization (Moradtalab et al. 2018 ).The 1,1-diphenyl-2-picrylhydrazyl radical method was used to evaluate the free radical scavenging activity of total antioxidants (T-AO; Moradtalab et al. 2020 ).Ascorbate peroxidase (APX,EC 1.11.1.11)activity was recorded by spectrophotometric method (Boominathan and Doran 2002 ).A Spectrophotometer U-3300 (Hitachi, Tok y o, J apan) w as used for all spectrophotometric determinations.

Analysis of root exudates and RH metabolites
HPLC-profiling of organic acids , sugars , and amino acids in the RH soil solutions in the 80% methanol extracts of the sorption filters was conducted as described by Windisch et al. ( 2021 ).For phenolic compounds and BXs, identification was performed with positi ve/negati ve switching LC-MS on a QExactive Plus Electr ospr ay Mass Spectr ometer (Thermo Fisher Scientific) coupled to an Agilent 1290 Ultra Performance Liquid Chromatography System ( Table S2 , Supporting Information ).Quantitativ e anal ysis of identified compounds was conducted by comparison with kno wn standar ds using a Shimadzu LC10 HPLC system ( Table S2 , Supporting Information ).

Microbial community DNA extraction and amplicon sequencing
TC-DN A w as extr acted fr om complete RH pellets or 0.5 g RA soils (w et w eight) using FastPrep FP24 bead-beating system (twice, each 30 s, 5.5 m s −1 ) and FastDNA Spin Kit for soil (MP Biomedicals, Santa Ana, USA) according to manufacturer's recommendations.Extracted TC-DNAs were purified using GeneClean Spin Kit (MP Biomedicals).TC-DNA yield and quality was c hec ked by a gar ose gel electr ophor esis and stor ed at −20 • C.
For bacterial community profiling based on the V3-V4 region of the 16S rRNA gene (16S), PCR was performed as described in Babin et al. ( 2021 ) with primers 341F/806R.Adding of Illumina sequencing adapters and sample-specific dual indexes, and pr epar ation of libr aries wer e done as described in Fernandez-Gnecco et al. ( 2022 ).Sequencing was carried out on an Illumina MiSeq platform using Reagent Kit v2 (2 × 250 bp; Illumina, San Diego, USA) following manufactur er's instructions.16S r eads wer e pr ocessed into amplicon sequence variants (ASVs) using D AD A2 version 1.10.0(Callahan et al. 2016 ), which were taxonomically annotated with SILVA SSU rel.132 database (Quast et al. 2013 ) as described in more detail in Fernandez-Gnecco et al. ( 2022 ).ASVs with less than five r eads acr oss the full data set wer e excluded fr om anal ysis.Furthermore, ASVs classified as chloroplasts, mitochondria (both < 0.3% of total reads) or present in the negative control were discarded.This resulted in a final number of 7811 ASVs and on average 28 711 quality reads per sample, which was sufficient to cover the diversity in all samples ( Figure S3A , Supporting Information ).
Profiling of fungal communities was carried out based on high-throughput amplicon sequencing of the Internal Transcribed Spacer (ITS2) region using a sample-specific barcoded primer pair (ITS86F/ITS4; White et al. 1990, Op de Beeck et al. 2014 ; Table S3 , Supporting Information ) as pr e viousl y described (Sommermann et al. 2018, Babin et al. 2021 ).In brief, the amplification was performed with three PCRs per sample (10 ng TC-DNA each) at different annealing temperatures (56 • C ± 2 • C) using the Q5 High-Fidelity 2x Master Mix (New England Biolabs , Ips wich, USA) at 25 cycles.Subsequentl y, tec hnical r eplicates wer e mixed and purified by MinElute PCR Purification Kit (Qiagen) with a final elution volume of 12 μl 10 mM Tris-HCl (pH 8.5).The concentration of each sample was checked by a Qubit ® 3.0 Fluor ometer (Invitr ogen, Carlsbad, USA) and mixed equimolarly.ITS2 sequencing was conducted as pr e viousl y described (Babin et al. 2021 ) on the Illumina MiSeq platform in paired-end mode (2 × 300 bp).Different trimming steps (barcodes , primers , and adapters) including the FASTX toolkit5 and raw sequence merging using FLASH (Mago č and Salzberg 2011 ) were conducted.With a local GALAXY Bioinformatics Platform in combination with the fungal UNITE database v8.0 (UNITE Community 2019 ), the database-dependent strategy (Antweiler et al. 2017 ) was performed.For this, sequences of all samples were aligned with the database (e-value ≤ 0.001) k ee ping onl y r esults fulfilling the following conditions: min.alignment length ≥ 200 bp and min.similarity ≥ 97%.To generate the ASV abundance table, the SH-numbers of the UNITE database were used as identifiers counting the sequences per assignment.Finally, a total of 1361 ASVs could be r etrie v ed with an av er a ge of 86 958 reads per sample ( Figure S3B , Supporting Information ).

Shotgun metagenome sequencing
In order to profile the functional potential of the RH microbiome, shotgun metagenomic libraries of TC-DN A w ere prepared using NEBNext Ultra II FS DNA Library Prep Kit (New England Biolabs) according to the manufactur er's pr otocol with the following modifications.Enzymatic shearing was performed using 100 ng DNA of each sample and an optimized incubation time of 15 min.The NEBNext Adaptor for Illumina was diluted (1:10) and no size selection was performed on the adaptor ligated DNA.The purified samples were amplified for ten cycles using the i7 and i5 primers of the NEBNext Multiplex Oligos for Illumina (Dual Index Primers Set 1) (New England Biolabs).The final products were purified by Ma gSi-NGSPr ep Plus kit (Steinbr enner, Wiesenbac h, German y) twice using two different library-to-bead ratios (0.6 and 0.8, respectiv el y).The size distribution and concentration of the libraries wer e e v aluated on a Fr a gment Anal yzer Automated CE System (Adv anced Anal ytical Tec hnologies, Or angebur g, USA) using the DNF-474 High Sensitivity NGS Fr a gment Anal ysis Kit (1-6000 bp).The libraries were diluted to 1 nM and sequenced on a NextSeq550 sequencer (Illumina) using the NextSeq 500/550 High Output Kit v2.5 (300 cycles) after equimolar pooling.
The shotgun metagenome sequencing produced a high-quality output with a sequencing depth of 16 million reads per sample.Raw reads obtained from NextSeq sequencing were processed on a GALAXY Bioinformatics platform.They were trimmed and Illumina adaptor sequences were removed using TrimGalore (Galaxy Version 0.6.3;Krueger et al. 2021 ) with min.read length 50 bp and min.Phred quality = 20.For the r ead-based anal ysis, forw ar d and r e v erse r eads wer e mer ged using Illumina pair ed-end r ead mer ger (PEAR Galaxy Version 0.9.6.1;Zhang et al. 2014 ).PhiX contamination was r emov ed using bbduk (v ersion 38.93; Bushnell et al. 2017 ) with kmer length = 31.Trimming and merging of reads resulted in 178 750 308 reads covering 91.7% of all sequences .Co verage of the metagenomic data set was estimated using Nonpareil (Galaxy Version 3.1.1.0;Rodriguez-R and Konstantinidis 2014 ) in alignment mode .T he co v er a ge was 7%-10%, whic h indicates a r ead-based anal ysis ( Figur e S7 , Supporting Information ).
Pr epr ocessed r eads wer e taxonomicall y classified using the MGX pipeline [Kraken with customized Prokaryotic protein database plus DIAMOND with the National Center for Biotechnology Information's nonredundant (NCBI-nr) protein database] implemented on the MGX platform (Jaenicke et al. 2018 ) with default parameter (e = 10 −5 , identity = 80%).46.8% of r eads wer e classified as bacteria and considered for further anal ysis.Onl y a small fraction of archaea (0.52%) and fungi (0.047%) could be classified, whic h wer e not enough to be r epr esentativ e and were therefore not further analyzed.
To analyze potential plant-beneficial functions, a customized database was established.Functions were selected based on a pr escr eening using the online tool for metagenome analysis MG-RAST (Keegan et al. 2016 ).Protein sequences of functions, which could be r ele v ant for plant-microbe interactions with potential dir ect (e.g.nitr ogen fixation) or indir ect (e.g.antimicr obial compounds) beneficial effects for plant growth and health ( Table S4 , Supporting Information ), wer e extr acted based on KO numbers from the latest version of the KEGG database (November 2022) using a customized R-Script and stored in the customized database .T he database was extended for additional potential plant-beneficial functions described in liter atur e (Glic k 2012 , Kuzmanovi ć et al. 2018, Cania et al. 2019 ) and downloaded from KEGG database (November 2022) ( Table S4 , Supporting Information ).Pr epr ocessed r eads wer e run a gainst the customized database using DIAMOND (Version 2.0.15;Buchfink et al. 2015 ) with alignment mode = blastx, mor e-sensitiv e mode, e = 10 −5 , identity = 80% and query cov er a ge = 50%.

Da ta anal yses and sta tistics
ITS/16S: the effect of tillage practice and N-fertilization intensity on micr obial comm unity composition was tested by PER-MANOVA (10 000 permutations) based on Bray-Curtis dissimilarity matrix (ITS: count data; 16S: log r elativ e abundance) and nonmetric multidimensional scaling (NMDS) was used for ordination.Alpha-div ersity (species ric hness, Pielou, Shannon) was calculated by 100 times random subsampling to the lowest number of reads (ITS: 51 670; 16S: 18 544).The effects of tillage practice and N-fertilization intensity on alpha-div ersity wer e tested b y tw o w ay-ANOVA follo w ed b y post hoc Tuk e y's HSD test ( P < .05).If ANOVA assumptions failed, data were transformed by Tuk e y's Lad der of Po w er.Heatma ps r epr esent the 30 most abundant microbial genera based on relative abundances (Euclidean distances).Differences in relative abundances in the RH on ASV le v el wer e tested by likelihood r atio tests under negative binomial distribution and generalized linear models (edgeR) for tilla ge pr actice (MP vs. CT) and N-fertilization intensity (Int vs. Ext) following de v elopers' r ecommendation (filter criterion: ASV present in min.three samples; each comparison group n = 8).
Meta genome: low abundance featur es wer e r emov ed by filtering the data for at least two hits in two samples.Data were analyzed based on relative abundance of reads by dividing the classified reads by the total number of reads per sample.Results were converted into per centages b y multiplying with 100.PERMANOVA anal ysis (Br ay-Curtis dissimilarity, 10 000 permutations) was performed to detect effects of tillage practice, Nfertilization intensity and their possible impact on microbial community and functional composition.Visualization of dissimilarities between samples was carried out with NMDS using the Bray-Curtis distance matrix.Differential analysis was performed with edgeR (Chen et al. 2016 ) filtered for at least one hit in three samples.
Plant: the effects of tilla ge pr actice and N-fertilization intensity on the metabolic profiles of shoots, RH soil solutions and RA soils, as well as on the nutrient status of the plants were analyzed by means of a tw o-w a y ANOVA.T he assumptions of homogeneity of variance and normality of the residuals were visually c hec ked with the performance pac ka ge (Lüdec ke et al. 2021 ).If the conditions for ANOVA failed, the data were Tuk e y-transformed and c hec ked a gain.If the r equir ements wer e still not met, an ART-ANOVA was performed.A pairwise mean comparison was performed on the basis of P ≤ .05using Tuk e y-HSD test.PERMANOVA (10 000 permutations) analyses were performed based on Bray-Curtis dissimilarities calculated from Ct values of 28 genes and 15 LMW organic compounds and secondary metabolites of the RH soil solution and principal coordinates analysis (PCoA).
Corr elation anal ysis: K endall r ank corr elations of metabolites of the apical RH soil solution against ASVs were carried out using the kendalltau function (Knight 1966 ) of the python pac ka ge scipy (Virtanen et al. 2020 ).To reduce the number of calculations only metabolites with significant differences between the treatments and tilla ge r esponders (edgeR; r elativ e abundance > 0.5%) were consider ed.Corr elations with K endall's tau ≥ 0.5 or ≤ −0.5 with P < .01 were considered as significantly correlated and visualized in a correlation matrix.

Plant biomass, root growth characteristics, nutritional status, and grain yield
Winter wheat grown under Ext exhibited a significantly higher RDM than under Int (Table 1 ).TRL and SDM were lowest under Int, regardless of the tillage practice.Overall, Ext significantly increased RHS and RD over Int, while CT significantly increased FRL over MP (Table 1 ).Final grain yields determined for the whole field experiment ranged between 6.3 and 6.6 t ha −1 without significant differences among agricultural practices (data not shown).
Elemental concentrations in flag leaves were affected by Nfertilization intensity.Plants grown in CT-Ext soil exhibited a significantly lo w er N, Mg, Ca, and S concentration in the leaf tissue compared to the other treatments.CT-Ext plants had the lo w est C and K concentrations in the leaf tissue.Only marginal differences wer e observ ed for leaf micronutrient contents, with all above the deficiency threshold (Table 2 ; Table S5 , Supporting Information ).Onl y K le v els wer e belo w the deficienc y threshold (Campbell 2000 ) across all treatments (Table 2 ).
No symptoms of leaf rust, but symptoms of powdery mildew caused by Blumeria graminis were visible on the lo w er leaves and stem.Although se v erity of the infestation was not exactl y quantified, a stronger visible infestation in MP and no or only weak infestation in CT treatments was observed.

Stress-related metabolites and gene expression in wheat leaves
The tilla ge pr actice was identified as a dominant factor determining metabolic stress adaptations (Table 3 ; Table S6 , Supporting Information ).Lo w er concentrations of stress hormones (JA and SA) and metabolic stress indicators (T-AO, proline, and APX) were detected in leaves of plants under CT (Table 3 ).N-fertilization intensity influenced str ess-r elated metabolites, especiall y in the CT tr eatments.Significantl y lo w er concentr ations of str ess hormones and pr oline wer e detected in CT-Int compar ed to CT-Ext plants (Table 3 ; Table S6 , Supporting Information ).
The influence of the different agricultural practices on the expr ession le v els of 28 selected genes involved in str ess r esponses was analyzed by qPCR in lea ves .In line with PERMANOVA analysis (Table 4 ), PCoA sho w ed that samples clustered accor ding to tillage practice with minor differences between N-fertilization intensities (Fig. 1 A).Pairwise comparison of gene expr ession r e v ealed that 21 out of 28 genes sho w ed significantly enhanced expression in plants from MP compared to CT soils in both Int and Ext fertilization ( Figure S4 , Supporting Information ).Nitrate reductase 1 ( TaNR1 ) and nitrite reductase ( TaNIR ) were identified as significantl y incr eased in Int compared to Ext treatment in both CT and MP tilla ge pr actice, according to Tuk e y's HSD analyses (data not shown).

RH metabolites
PCoA analysis sho w ed that the metabolite pr ofiles differ ed str ongl y among r oot zones (Fig. 1 B).The le v el of all measured metabolites was highest in the apical root zone ( Tables S6 and S7 , Supporting Information ) and differentiation of samples according to tilla ge pr actice was visible (Fig. 1 B).Ther efor e, the a pical r oot zone was investigated in more detail to elabor ate, whic h metabolites contributed to the differentiation.
CT treatments had higher levels of specific metabolites like succinic acid, aspar a gine, tryptophan, and tr ehalose, compar ed to MP (T able 5 ; T able S6 , Supporting Information ).Various secondary Table 1.Effect of long-term a gricultur al pr actice on shoot dry mass (SDM), root dry mass (RDM), total root length (TRL), fine root length (FRL), root hair size (RHS), and rooting depth (RD) of winter wheat (cv.Lemmy; EC 59).MP -mouldboard plough, CT -cultivator tillage, Ext -extensive N-fertilization intensity without fungicides and growth regulators, and Int -intensive N-fertilization intensity with pesticides and growth regulators.SDM, RDM, and TRL data are adjusted to 10 tillers, which represents the average number of tillers in one plant.
Values ar e pr esented as means ± standard deviation of four replicates.Means not sharing an y letters ar e significantl y differ ent by the Tuk e y-test ( P ≤ .05).metabolites including benzoic acid, flav onoids (quer cetin and naringenin), phenolic acids (caffeic acid, cinnamic acid, and paracoumaric acid), and a range of BXs were detected.For BXs solely metabolites of the unstable BX DIMBOA were detectable in the RH soil solution, such as MBOA and 6-Methyl-2H-1,4-benzoxazin-3(4H)-one (MeBOA).Tillage practice significantly affected the composition and quantity of the BXs profiles.A higher MBOA concentration was detected in the apical zones of CT treatments, while MeBOA and benzoic acid wer e significantl y lo w er compared to MP treatments (Table 5 ; Table S6 , Supporting Information ).Apical root zones of plants grown in MP-Int exhibited the highest concentrations of MeBOA and benzoic acid.In the RH soil solution collected from older, basal root zones, secondary metabolites tended to be higher in CT treatments (Table 5 ).

Phylogenetic composition of the microbiome
The microhabitat (RA soil or RH) was identified as the driving factor for the bacterial/archaeal and fungal beta-diversity ( R 2 = 27% and R 2 = 40.5%,r espectiv el y; both P < .001;PER-MANOVA).This was also a ppar ent at the le v el of the major Table 4. Effect of long-term a gricultur al pr actices on plant gene expr ession in leav es of wheat plants (cv.Lemmy, EC 59) as well as on taxonomic and functional composition of microbial communities expressed as explained variance (EV).Gene expression is based on 28 selected genes expressed in lea ves .Taxonomic composition of microbial communities is based on 16S rRNA gene or ITS amplicon sequencing in the root-affected (RA) soil and rhizosphere (RH).Functional composition is based on metagenome sequencing of RH and analyzed using a customized database for plant-beneficial bacterial functions ( n = 321; filtered tw o hits, tw o samples, no log transformation).Analysis is based on PERMANOVA using Bray-Curtis distances (10 000 permutations).Significant effects ( P -value = P ) are marked with * P < .05,* * P < .01,and * * * P < .001.ns -not significant.abundant gener a ( Figur e S5A , Supporting Information ).Typical bacterial/arc haeal gener a detected in RA soil had the closest affiliation to Nitrososphaeraceae , Acidobacteria subgroup 6 and RB41.In contrast, Bacillus , Devosia , Pedobacter , and Massilia were more pr e v alent in the RH ( Figure S5A , Supporting Information ).For fungi, the genera Mortierella , Chaetomium , Fusarium , Gibellulopsis , and Solicoccozyma dominated the RA soil while in the RH Zymoseptoria , Pseudogymnoascus , Mycosphaerella , and Filobasidium prev ailed ( Figur e S5B , Supporting Information ).In addition, effects of tilla ge pr actice wer e a ppar ent, especiall y in RH on the le v el of the most abundant fungal genera such as Rhizopu s and Trichoderma or Mortierella and Chrysosporium enriched in CT or MP treatments, r espectiv el y ( Figur e S5 , Supporting Information ).Bacterial and archaeal alpha-diversity was significantly affected by N-fertilization intensity (Shannon, Richness) or tillage and tillage x N-fertilization (Pielou) in RA soil.While Pielou indices were higher in RA soil under CT than under MP, richness and Shannon indices were highest in RA soil under Ext fertilization independent of the tillage practice ( Table S8 , Supporting Information ).No significant effects of the a gricultur al pr actice on bacterial and archaeal alpha-diversity were observed in the RH ( Table S8 , Supporting Information ).Similarly, for fungi only a sig-nificantl y higher ric hness in Ext compar ed to Int tr eatments was observed in RH.Ho w ever, in RA, CT-Ext soils exhibited a significantly higher fungal richness or Shannon diversity compared to MP-Int or CT-Int, r espectiv el y ( Table S9 , Supporting Information ).

Gene expression
In both RA soil and RH, tilla ge pr actice exhibited a stronger effect on the micr obial comm unity composition than N-fertilization intensity (PERMANOV A, T able 4 ).The effect of N-fertilization intensity was stronger for fungi than for bacteria.A significant interaction effect between both agricultural practices was found on the fungal RH communities (Table 4 ).
The NMDS ordination of microbial communities substantiated the results of the PERMANOVA showing a distinct separation between CT vs. MP.For archaea/bacteria, CT samples displayed a higher variability among replicates than MP.In contrast to archaea/bacteria, an additional clustering based on N-fertilization intensity was observed for fungal communities in both microhabitats (Fig. 2 A, B, D, and E).
Since the RH microbiome has important impacts on plant de v elopment and health, differ entiall y abundant taxa (ASVs) among tilla ge pr actices or N-fertilization intensity were identified.Many of the bacterial and archaeal taxa that were significantl y incr eased in r elativ e abundance in the RH of MP in contr ast to CT tr eatment belonged to Actinobacteria and Acidobacteria e.g.sequences with closest affiliation to Micrococcaceae , Agrom yces , RB41, Acidobacteria Subgr oup 6, but also to Terrimonas , Nitrospir a , and Nitrososphaer aceae (T able 6 ; T able S10 , Supporting Information ).Taxa with higher r elativ e abundance in CT compared to MP belonged to Bacillus , Sphingomonas , Devosia , Pseudoxanthomonas , and Sacc harimonadaceae (P atescibacteria) (Table 6 ; Table S10 , Supporting Information ).No bacterial/archaeal ASVs were found that responded to either Int or Ext in the RH (FDR < 0.05).

Functional profiling of the RH microbiome
Tillage was the main driver for bacterial plant-beneficial functions, follo w ed b y the combination of tillage and N-fertilization intensity (Fig. 2 C; Table 4 ).A total of 350 genes were detected based on the customized database of plant-beneficial functions.From these, 45 were differentially abundant with 24 genes enriched in CT and 21 enriched in MP treatments.Under CT, genes coding for ACC deaminase, sider ophor e pr oduction, spermidine, and trehalose synthase were enriched in RH (Fig. 3 ).In contrast, genes for biofilm production and quorum sensing exhibited significantly higher r elativ e abundances in the RH of MP plants (Fig. 3 ).In RH of both tillage practices, different genes encoding for N-metabolism, aromatic compound degradation and secondary metabolite production were differentially abundant.
For linking functions with taxonomy, sequences of the differentially abundant functions were extracted and taxonomically classified.In order to explore whether they contributed to the positiv e plant-micr oor ganism inter actions observ ed under CT (Tables 1 and 3 , Fig. 4 ), the focus was put on 12 genes associated with functions exclusiv el y incr eased in CT treatments ( Figure S8 , Supporting Information ).These were genes encoding ACC deaminase and mbtA , mbtB , mbtC , mbtD , mbtE , and mbtF, encoding the sider ophor e mycobactin, potC involv ed in spermidine pr oduction, as well as otsA , otsB , treT , and treZ involved in trehalose production.Between 50% and 70% of the extracted sequences were classified on genus le v el.Among the ten most detected genera associated with these genes, the majority belonged to Actinobacteria/Actinomycetota, such as Mycolicibacterium , Nocardia , Nocardioides , Rubrobacter , and Streptomyces ( Figure S8 , Supporting Information ).Furthermor e, these genes wer e associated with the genera Devosia , Mesorhizobium , and Microbacterium , which were also detected as positive responders in CT based on 16S amplicon sequencing.

Correla tion betw een microbial responders and RH metabolites
The correlation analysis between microbial responder taxa and metabolites in the RH soil solution sho w ed tw o main clusters (Fig. 4 ).Cluster 1 contained fungal, archaeal, and bacterial taxa that wer e positiv el y corr elated with succinic acid, catec hin, MBOA, tryptophan, trehalose, and asparagine.Notably, fungal ASVs of Table 6.Bacterial and archaeal species (ASV) with differential relative abundance in the rhizosphere (RH) of winter wheat (cv.Lemmy, EC 59) grown in mouldboard plough (MP) or cultivator tillage (CT) soils (FDR < 0.05).Only discriminative species with mean > 0.5% are displayed.Significantl y enric hed taxa ar e marked in bold.Mean ± standard de viation is shown ( n = 8).

Phylum
Genus Species ASV MP CT cluster 1 mainly belonged to Ascomycota and were affiliated among others to Trichoderma spp.and Penicillium spp.(Fig. 4 ; Table S14 , Supporting Information ).Bacterial ASVs of cluster 1 were affiliated to Pr oteobacteria, P atescibacteria, and Firmicutes (e.g.Bacillus ).In contrast, cluster 2 contained bacterial and fungal taxa, whic h wer e positiv el y corr elated with MeBOA and benzoic acid and negativ el y with the other compounds .T he bacterial ASVs were affiliated to e .g. Actinobacteria, Nitrospirae , and Acidobacteria (Fig. 4 ; Table S13 , Supporting Information ).For fungal ASVs, this was true for r epr esentativ es of the Ascomycota and Mortierellomycota.

Reduced N-fertilization intensity supported wheat root development
The root development is strongly influenced by nutrient availability and fertilization (Pierret et al. 2007 ).Bilalis et al. ( 2015 ) sho w ed that reduced nutrient availability stimulates root properties required for nutrient acquisition.In the present study, winter wheat under Ext exhibited a larger root system compared to Int, regardless of the tilla ge pr actice, likel y supporting the uptake of water and nutrients e v en under reduced availability.The r eductiv e assimilation of nitr ate as the pr edominant N source in mineral fertilized soils requires the successive activities of nitrate reductase 1 ( TaNR1 ) and nitrite reductase ( TaNIR ) for conv ersion of nitr ate to nitrite and finally to ammonium (Costa-Broseta et al. 2021 ).The higher amount of available N consistently induced higher expression of these genes, including the nitr ate tr ansporter gene ( TaNPF7.1 , NRT1-PTR-Famil y), in the Int compared to Ext treatments ( Figure S4 , Supporting Information ).Aside fr om K, whic h w as belo w the deficienc y le v el acr oss all tr eatments, ther e wer e no specific nutrient deficiencies.Consequently, winter wheat plants acquired sufficient nutrient amounts, leading to comparable shoot biomass across treatments regardless of N-fertilization intensity (Table 1 ).This highlights that wheat compensated the reduced N availability (Ext) by extending the root arc hitectur e and physiological properties.As wheat under Ext and r educed tilla ge (CT) pr oduced the same yield like plants under Int and MP, current N-fertilization practices for wheat still appear too high, with potential negative consequences for the environment, due to leakage.

Agricultur al pr actice altered RH metabolites and plant-microorganism interactions
Root exudates and related RH metabolites are k e y dri vers of interactions at the plant-microorganism-soil interface (Hu et al. 2018, Kudjordjie et al. 2019, Oburger et al. 2022 ) influenced by various factors (Neumann and Ludewig 2023 ).The root window setup combined with microsampling of RH soil solutions (Neumann 2006 ) allo w ed to tr ac k spatial variations along different root zones and relate them to the RH microbiome and plant performance.In our study, tillage practice, but not N-fertilization intensity, pr edominantl y influenced RH metabolite patterns, especially in the young apical root zones (Fig. 1 B).
Tryptophan was significantly increased in the RH of plants grown under CT (Table 5 ).Many RH microorganisms use tryptophan as a precursor for indole acetic acid (IAA) production (Spaepen andVanderleyden 2011 , Vurukonda et al. 2016 ) and ther eby stim ulate the gr owth of fine later al r oots, as observ ed also in this study (Table 1 ).Ele v ated tryptophan concentrations were also positively correlated with an enrichment of Bacillus in the RH (Fig. 4 , Table 6 ), containing strains described to produce IAA (Özdal et al. 2016 ).Inter estingl y, micr obial genes involv ed in IAA pr oduction wer e not among the differ entiall y abundant genes (Fig. 3 ).Ho w e v er, it has been reported that microorganisms can also impr ov e later al r oot formation independent of IAA production (Yu et al. 2021 ) by modulating the hormonal balances of the host plant via changes in plant hormonal metabolism (Moradtalab et al. 2020 ).In this context, also the increased abundance of ACC deaminase genes in RH of CT plants could be r ele v ant for the stimulation of fine root development (Table 1 ) b y lo w ering the plant ethylene le v el acting as an antagonist (Glick et al. 2007 ).
Other RH metabolites, such as succinic acid, trehalose, and aspar a gine, wer e higher in the CT treatments and correlated with the incr eased r elativ e abundance of potentiall y beneficial micr oor ganisms ( Bacillus and Trichoderma ; Fig. 4 ).Succinic acid acts as a c hemoattr actant and carbon source for various beneficial micr oor ganisms and is r ele v ant for functions in biocontrol and plant defense responses (Sampedro et al. 2015 ).Trehalose represents a signaling molecule regulating bacterial and fungal growth, de v elopment, and virulence (Sharma et al. 2020 ), but it is also produced by micr oor ganisms in high quantities, promoting adaptive responses to abiotic stress in plants (Vurukonda et al. 2016, Kosar et al. 2019 ).Mor eov er, tr ehalose can induce r esistance to powdery mildew in wheat (Reignault et al. 2001 ), which is in line with the her e observ ed r educed symptoms of powdery milde w in the CT treatments .T he increased abundance of trehalose synthase genes in the RH of CT plants (Fig. 3 ) suggests a microbial origin for the ele v ated tr ehalose le v els.
Apart from some phenolic acids and fla vonoids , the predominant secondary metabolites detected in the RH soil solution were differ ent BXs, deriv ed fr om the young a pical r oot zones particularly of CT plants .T hese compounds ar e pr oduced by most cer eal crops at an early growth stage and in response to stress events as defense substances with allelopathic, insecticidal (antifeeding), and antimicrobial effects (Hu et al. 2018, Kudjordjie et al. 2019 , Sc handry and Bec k er 2020 ).Ad ditionally, the y can stimulate growth of beneficial soil microorganisms, decrease plant growth of competitors, increase JA signaling and plant defenses, and suppr ess herbivor e performance in the next plant gener ation (Hu et al. 2018, Oburger et al. 2022 ).In soils, these highl y bioactiv e substances are rapidly converted into more stable derivates, such as APO, MBO A, MeBO A, and AMPO, through both microbial and nonmicr obial pr ocesses (Sc hütz et al. 2019, Obur ger et al. 2022 ).In the CT treatments, MBOA was predominant, while in the MP treatments, other BXs (MeBOA) dominated, which could indicate differences in microbial degradation between CT and MP tillage practice .T he accumulation of MBOA was positively correlated with potentiall y anta gonistic micr oor ganisms ( Bacillus and Tric hoderma ) (Tables 6 and 7 ; Fig. 4 ).T his , together with the fact that MBOA has toxic effects on various fungal pathogens (Cotton et al. 2019, Kudjordjie et al. 2019 ), such as Fusarium species (Glenn et al. 2001 ), may explain the reduced relative abundance of putative plant pathogenic fungal genera in the CT-Ext treatment ( Zymoseptoria and Gibellulopsis ; Figure S5B , Supporting Information ).

Agricultur al pr actices altered the RH microbiome of winter wheat
The RH microbiome strongly depends on the soil microbiome, which is influenced by different factors, such as agricultural practice (Bulgarelli et al. 2013 ).In our study, tillage practice was the main factor affecting the microbiome in the RA soil and RH (Table 4 ), indicating that the complete plant-micr oor ganism-soil system was affected.These observations are in line with results fr om pr e vious studies inv estigating wheat soils (Sommermann et al. 2018, Babin et al. 2019, Romano et al. 2023 ), which highlights the stable legacy of tillage practice likely caused by differences in physical soil properties (Schlüter et al. 2018 ) as well as differences in chemical composition (e.g.C and N stocks; Table S15 , Supporting Information ).N-fertilization intensity shaped the comm unities onl y to a minor extent with a stronger influence on fungi.This confirmed pr e vious studies (Sommermann et al. 2018, Bziuk et al. 2021 ) and could be related to the availability of N that influences the activity and growth efficiency, especially of saprotrophic fungi (Di Lonardo et al. 2020 ).Clearly, not only plants can influence the microbial community via root exudation, but micr oor ganisms can also alter the exudate composition (Korenblum et al. 2020 ).Our results further support such interdependence of root exudates and the microbiome (Fig. 4 ).
The effects of tilla ge pr actice on the composition of root exudates are likely related to specific patterns of substrate utilization by the microbiome.Bacterial and archaeal species in the RH of MP plants belonged mainly to Actinobacteria and Acidobacteria (Table 6 ; Table S10 , Supporting Information ), which ar e typicall y r egarded oligotr ophs (Fier er 2017 ).This might indicate that the selective effect of the plant is lo w er in MP than in CT.The acidobacterial genus RB41 has been fr equentl y isolated from soil and was reported as a tilla ge r esponder.In accordance with our results ( Table S10 , Supporting Information ), Kudjordjie et al. ( 2019 ) found a negative correlation between many acidobacterial species and BX compounds in maize roots.BXs have a selective impact on root and RH microbiota across different field locations (Niemey er 2009, Kudjor djie et al. 2019, Cadot et al. 2021 ).Taxa with higher r elativ e abundance in the RH of CT plants compared to MP plants belonged to typical RH genera such as Bacillus , Sphingomonas , Devosia , and Pseudoxanthomonas (Table 6 ; Table S10 , Supporting Information ), which are known to harbor members with plant-beneficial properties (Chowdhury et al. 2019 ).Additionall y, man y sequences with affiliation to Saccharimonadaceae (P atescibacteria) wer e enric hed in the RH of CT plants.A symbiotic lifestyle and cometabolism interdependencies were proposed for Patescibacteria (Lemos et al. 2019 ), which might favor these taxa in microbial hotspots such as the RH.In summary, this suggests that the driving effect of the plant on the soil bacterial/archaeal community was stronger in soils under CT than MP tillage.
In the RH of CT more fungal genera with known plantbeneficial members, such as Trichoderma , were present, whereas potential plant pathogens, like Olpidium , wer e mor e abundant in the RH of MP plants .T he genus Olpidium harbors typical pathogens for Brassicaceae (Lay et al. 2018 ) like r a peseed, which was the previous crop at the sampling site .Furthermore , Chrysosporium species wer e enric hed in MP and are known in the context of gibberellin production, which can enhance plant growth (Hamayun et al. 2009 ), but is detrimental for plants in high doses (Cen et al. 2020 ).Among the fungal responders in CT were many potential fungal antagonists such as Chaetomium , Penicillium , Trichoderma , and nonpathogenic Fusarium , which were often found in disease-suppr essiv e soils (r e vie w ed b y van Bruggen andSemenov 2000 , Mazzola 2002 ).Trichoderma species are well known as plant-growth promoters, mycoparasite and inducers of plant systemic resistance against pathogens (Harman et al. 2004, López-Bucio et al. 2015, Hafiz et al. 2023 ) and were previously identified in a growth chamber experiment with soils from the identical sampling site as responders to CT (Babin et al. 2021 ).The class Sordariomycetes was increased in the RH of CT plants with high BXs concentr ation (Fig. 4 ), whic h confirms pr e vious studies (Kudjordjie et al. 2019 ).
The effect of tillage on micr obial comm unities was not only observed on a taxonomic but also functional le v el.ACC deaminase activity, sider ophor e, tr ehalose, and spermidine production wer e enric hed in the RH of CT plants (Fig. 3 ).ACC deaminase lo w ers the le v el of the plant stress hormone ethylene and is, ther efor e, consider ed as beneficial (Dubois et al. 2018 ).Genes encoding ACC deaminase were present in several genera including Microbacterium , whic h was enric hed in CT samples and is wellknown to harbor plant-beneficial members (Vílchez et al. 2018, Freitas et al. 2019 ).Moreover, spermidine is a compound with several important functions , e .g. biofilm production, o verall bacterial fitness, plant growth promotion, and stress protection (Alavi et al. 2013, Xie et al. 2014, Chen et al. 2018 ).Among the classified genera, that encode spermidine production, was the responder genus Mesorhizobium from the RH of CT.In rhizobia, spermidine plays a role in symbiotic interactions and root nodule formation (Becerr a-Riv er a and Dunn 2019 ).Sider ophor es ar e ir onchelating compounds that are involved in pathogen suppression by iron competition (Gu et al. 2020 ) or by inducing systemic resistance (Zamioudis et al. 2015, Verbon et al. 2019 ).Se v er al gener a wer e associated with sider ophor e pr oduction, but no r esponder genus was included ( Figure S8 , Supporting Information ).T hus , the increased siderophore production in CT plants seems to be a concerted effect of se v er al beneficial taxa.Comparing amplicon and metagenomics data, responders encoding trehalose synthase genes were found ( Microbacterium and Devosia ; Figure S8 and Table S10 , Supporting Information ).Trehalose produced by Microbacterium sp.3J1 was pr e viousl y described to be involved in the protection of pepper plants against drought stress (Vílchez et al. 2018 ).Taken together, these responders could have contributed to the observed improved plant health in CT plants.

Reduced tillage practice and N-fertilization intensity enhanced stress tolerance in winter wheat
Analysis of stress-related metabolites and genes in wheat leaves highlighted that tillage practice had a pronounced effect on metabolic str ess ada ptation.In contr ast to MP plants, lo w er concentr ations of str ess hormones (JA and SA) and stress indicators (T-AO, proline, and APX; Table 3 ) were revealed for CT plants.Moreov er, incr eased abundance of the ACC deaminase gene derived from bacteria in the RH of CT plants (Fig. 3 ) could have also contributed to the reduced stress responses in CT plants (Glick et al. 2007 ).
The ele v ated str ess le v el of MP plants w as underlined b y the differ entiall y enric hed expr ession of genes involv ed in str ess r esponses (Fig. 1 A) like the well-c har acterized pathogenesis-r elated gene PR1 ( β-1,3-glucanase) and Chitinase ( TaCHI ).These genes encode for proteins involved in hydrolysis of glucan and chitin, whic h ar e pr esent in fungal cell walls (v an Loon et al. 2006 ) and have been shown to be upregulated in Puccinia triticina infected wheat leaves (Casassola et al. 2015 ) and in Pyricularia oryzae infected rice leaves (Cruz et al. 2015 ).In our study, an enhanced expression of other defense-related genes lik e lipo xygenase ( TaLOX ), defensin ( Tad1 ), and allene oxide synthase ( TaAOS ), which are highly inducible by biotic stress factors (Manners et al. 1998 ), was observ ed.The cr oss-talk among SA, JA, and ET in the regulation of plant stress responses has been extensiv el y studied in model plants like Arabidopsis (Pieterse et al. 2009, Verhage et al. 2010 ).
The incr eased expr ession of T aSOD , T aCAT , and T aPER genes associated with leaf accumulation of T-AO and increased APX activity in the MP plants indicates an activation of defense mechanisms to detoxify free radicals and to alleviate the o xidati ve damage associated with the overproduction of reacti ve o xygen species (Miller et al. 2010 ).Our findings from wheat leaf gene expression suggest that compared to CT, plants grown in MP soil sho w ed enhanced activity of defense signaling pathwa ys , potentially triggered by the presence of a leaf pathogen (i.e .B .graminis ).T his observation was supported by an ele v ated concentr ation of the stress hormones JA and SA in the shoots (Table 3 ).The described scenario indicates a reduced influence of stress factors on wheat plants grown in CT soil, which was most likely supported by modifications of the interactions with the RH microbiome e.g.via root exudates.

Conclusion
In our study, we used a long-term field study site and in-field root windows to obtain a holistic view on the impact of agricultural pr actice (tilla ge, N-fertilization intensity) on the complex interaction between plant roots and the surrounding microbial comm unity in soil.Tilla ge pr actice str ongl y impacted on the m ultipartite interaction network with consequences for plant health and stress resilience.CT combined with reduced N-fertilization intensity resulted in a positive plant-microorganism feedback in the RH contributing to an impr ov ed plant performance compared to conv entional mana gement.Ho w e v er, car e should be taken as these observations cannot be generalized as indicated by longterm (2012-2016) higher yields in the conventional (9.6 t ha −1 MP-Int) compared to conservation practice (8.4 t ha −1 CT-Ext).We, ther efor e, suggest that the observed compensatory effects of conserv ation pr actices ar e mor e likel y to play a r ole under unfavorable conditions such as pathogen pr essur e or dr ought, whic h ar e predicted to occur more frequently in a changing climate.Altogether, this study contributes to a holistic understanding of soilplant-micr oor ganism inter actions in a gricultur al settings.

Figure 1 .
Figure 1.PCoA based on (A) the expression of 28 selected genes ( Ct values) and (B) of 15 low molecular weight (LMW) organic compounds and secondary metabolites of the apical and basal rhizosphere (RH) soil solution as well as the root-affected (RA) soil solution of winter wheat (cv.Lemmy, EC 59) grown in different long-term agricultural practices (MP -mouldboard plough, CT -cultivator tillage, Ext -extensive N-fertilization intensity without fungicides and growth regulators, and Int -intensive N-fertilization intensity with pesticides and growth regulators) of the long-term experiment in Bernburg.

Figure 2 .
Figure 2. NMDS analysis showing the effects of a gricultur al pr actice (MP -mouldboard plough; CT -cultiv ator tilla ge; Int -intensiv e N-fertilization intensity with pesticides and growth regulators; and Ext -extensive N-fertilization intensity without fungicides and growth regulators) in the long-term field experiment Bernburg on (A) root-affected (RA) soil, (B) rhizosphere (RH) archaeal/bacterial community structure, (C) RH potential plant-beneficial functional composition of the bacterial community (shotgun metagenomic sequences annotated with a customized database; Table S4 , Supporting Information ), and (D) RA soil and (E) RH fungal community structure of winter wheat (cv.Lemmy, EC 59).Stress indicates the ordination stress value of each individual NMDS analysis.

Figure 4 .
Figure 4. Corr elation (K endall) matrix of micr obial r esponders [r elativ e abundance (RelA) > 0.5%] and metabolites in rhizospher e (RH) soil solution of winter wheat (cv.Lemmy, EC 59) apical roots.Only the metabolites affected by the different long-term agricultural practices are shown (Table5).Organisms and metabolites are sorted according to correlation-based hierarchal cluster analysis.High positive correlations are represented by dark r ed, negativ e ones by dark blue circles, whereas the circle diameter is indicative for the strength of the correlation.X, not significantly correlated (tau < 0.5 or > −0.5 with a P > .01).The effect of tillage practice on the mean RelA of microbial responders is shown in a bar chart (cultivator tillage [CT] = light gray and mouldboard plough [MP] = dark gray).
• C until total microbial community (TC)-DNA extraction (see below).Roots after Stomacher treatment wer e stor ed at 4 • C until determination of r oot par ameters (see below).

Table 2 .
Impact of long-term a gricultur al pr actice on the n utrient status of winter wheat (cv.Lemm y; EC 59).MP -mouldboard plough tilla ge, CT -cultiv ator tilla ge, Ext -extensiv e N-fertilization intensity without fungicides and gr owth r egulators, and Int -intensiv e Nfertilization intensity with pesticides and growth regulators, DM-dry mass.Values are presented as means ± standard deviation of four replicates.Means not sharing any letters are significantly different by the Tuk e y-test ( P ≤ .05).

Table 3 .
Phytohormones (SA = salicylic acid and JA = jasmonic acid) and stress-associated metabolites (T-AO = total antioxidants and APX = ascorbate peroxidase) in leaves of winter wheat (cv.Lemmy, EC 59) grown under different long-term agricultural practices.MP -mouldboard plough tillage, CT -cultivator tillage, Ext -extensive N-fertilization intensity without fungicides and growth regulators, and Int -intensive N-fertilization intensity with pesticides and gr owth r egulators.Values ar e pr esented as means ± standard deviation of four replicates.Means not sharing any letters are significantly different by the Tuk e y-test ( P ≤ .05).FW -fresh weight.

Table 5 .
Effect of long-term a gricultur al pr actices on primary and secondary metabolites in RH soil solution of apical and basal roots of winter wheat (cv.Lemmy; EC 59).MP -mouldboard plough tillage, CT -cultivator tillage, Ext -extensive N-fertilization intensity without fungicides and growth regulators, and Int -intensive N-fertilization intensity with pesticides and gr owth r egulators.Values ar e pr esented as means ± standard deviation of four replicates.Means not sharing any letters are significantly different by the Tuk e y-test ( P ≤ .05).