Host genotype and edaphic factors shaped bacterial communities associated with native and endemic medicinal Artemisia species in arid environment Authors: Rezaul Karim, Jean Legeay, Mohamed Bammou, Mohamed Hijri, Bulbul Ahmed Published: 2026 ABSTRACT BackgroundPlant-associated microbiomes are key contributors to plant nutrition and stress tolerance, particularly in arid ecosystems where extreme abiotic conditions strongly shape plant–microbe interactions. Despite this, the abiotic drivers of microbiome assembly across different plant compartments in wild medicinal species from these environments remain poorly understood. In this study, we investigated bacterial community structure across multiple niches, including bulk soil, rhizosphere, root, and shoot, in three wild Artemisia species (Artemisia herba-alba Asso., Artemisia negrei L., and Artemisia mesatlantica Maire), the latter two being endemic to arid regions of Morocco.ResultsUsing amplicon sequencing, we observed diverse bacterial associations with each plant niche harboring distinct taxa. Host plant species significantly influenced bacteriome composition (p = 0.026), particularly in Artemisia mesatlantica, which hosted the most specific bacterial taxa compared to its other relatives. Plant compartment emerged as key drivers of belowground bacterial community assembly with additional structuring by host species and edaphic factors. Soil pH, calcium carbonate content, organic matter and electrical conductivity were strongly correlated with shifts in bacterial diversity and composition, emphasizing the role of soil physicochemical properties as an environmental filter under extreme alkaline and arid conditions. Despite these species- and environment-specific variations, a conserved core bacteriome was identified across all Artemisia species, and compartments except shoot, comprising of Bacillus, Microvirga and Rhizobium.ConclusionOur results demonstrate that the interplay between host identity and soil properties orchestrates distinct, yet functionally coherent bacterial communities in wild Artemisia species. Bacterial taxa identified as core are well-known for their roles in plant growth promotion, biocontrol and production of bioactive secondary metabolites. The persistence of this core bacterium comprising of Bacillus, Microvirga and Rhizobium suggests a stable association across hosts and compartments, potentially reflecting conserved ecological roles, although functional contributions were not directly assessed in this study. Overall, our findings reveal how the interplay between soil properties and host identity shapes the assembly of distinct, yet compositionally consistent bacterial communities in wild medicinal Artemisia species.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s40793-026-00920-9. Introduction The assembly of plant-associated microbiomes is increasingly recognized as a key determinant of plant health, productivity and environmental adaptation [20, 55]. Microbial communities inhabiting distinct plant compartments, particularly the rhizosphere and endosphere, play pivotal roles in nutrient cycling, abiotic stress tolerance, and pathogen suppression, often functioning as an extension of the plant’s phenotype [25, 57]. Microbial composition is shaped by plant compartment, host genotype, and soil physicochemical properties [17, 62]. Despite growing interest, the dynamics of microbiome assembly in wild medicinal plants, especially in arid and extreme environments, remain underexplored, although these associations are likely important for ecological adaptation [16]. Understanding these interactions in natural systems is essential to harness native microbiomes for conservation, ecological restoration, and sustainable cultivation of medicinal plants. The genus Artemisia (Asteraceae) comprises over 500 species [1]. In North Africa, particularly in Morocco, Artemisia species are ecologically dominant in arid and semi-arid habitats [6]. Some species, such as Artemisia mesatlantica Maire, are endemic to Morocco [7]. This species is adapted to alkaline, nutrient-poor, and arid soils, making it a compelling model for studying plant-microbe interactions under environmental stress. Similarly, Artemisia herba-alba Asso. and Artemisia negrei L. are common in Moroccan steppe ecosystems, where they contribute to soil stabilization. In arid ecosystems, soil physicochemical properties are primary determinants of plant-associated microbiome assembly. In particular, variables such as pH, salinity, and nutrient availability act as strong environmental filters shaping microbial recruitment and persistence in the rhizosphere and root compartments, with soil pH emerging as a dominant predictor of bacterial community composition across spatial scales [19, 30]. These edaphic controls are particularly critical in dryland systems, where microbial consortia can enhance plant resilience by facilitating nutrient cycling and producing bioactive compounds [67]. Identifying core microbiota consistently associated with a host across environments provides insights into conserved functional interactions that support plant health [47, 53]. However, the core and variable microbiomes of wild Artemisia species in North African ecosystems remain uncharacterized. In this study, we investigated the bacterial community structure across four niches: bulk soil, rhizosphere, root and shoot of three Artemisia species viz. Artemisia herba-alba (native to Morocco), Artemisia negrei and Artemisia mesatlantica (endemic to Morocco) in the semi-arid High Atlas region at two sites, Errachidia and Tinghir of Morocco. Using high-throughput 16 S rRNA gene amplicon sequencing in combination with comprehensive soil physicochemical analysis, we aimed to (i) define species- and niche-specific patterns of bacterial diversity and community structure, (ii) identify core bacterial taxa, and (iii) evaluate the influence of environmental factors on bacterial community assembly. We hypothesized that (1) bacterial community composition would differ among plant niches and between Artemisia species, with endemic species harboring more specialized associations; (2) a conserved core bacteriome would persist across species and niches, reflecting functional redundancy essential for host adaptation to arid conditions; and (3) soil physicochemical properties, particularly pH, salinity, and nutrient availability, would act as major drivers shaping belowground bacterial communities. By elucidating plant-microbe-environment interactions in wild Artemisia, our study provides new insights into the microbial ecological observations of arid-land plants and establishes a basis for microbiome-based strategies in conservation and dryland agriculture. Materials and methods Sample collection, processing, and soil physicochemical analysis Bulk soil, rhizosphere, root and shoot of three Artemisia species, Artemisia herba-alba Asso. (widely distributed in Morocco), Artemisia negrei L. (endemic to Morocco) and Artemisia mesatlantica Maire (endemic to Morocco), were collected in June 2024 from the semi-arid High Atlas Region (latitude 31.85677–31.98795°N; longitude 4.99051–5.46546°W), at Errachidia and Tinghir, Morocco (Fig. S1). Botanical identification was performed through macroscopic comparison using reference specimens and taxonomic keys, and subsequently confirmed by Professor Hamid Khamar (Scientific Institute, Rabat, Morocco). The voucher specimens are deposited in the National Herbarium of the Scientific Institute of Rabat under the voucher numbers RAB114680 (Artemisia herba-alba), RAB114681 (Artemisia negrei) and RAB114682 (Artemisia mesatlantica). A total of 40 samples were collected: four replicates each of rhizosphere, root, and shoot per species, and four bulk soil samples. Sampling was conducted across multiple sites, with A. herba-alba collected from two distinct sites, A. negrei from one site, and A. negrei and A. mesatlantica co-occurring and sampled within the same site. Thus, part of the sampling design includes spatial overlap between species, allowing partial disentanglement of host and site effects, while other samples remain site-specific. Samples were placed in plastic bags or sterile Falcon tubes and then were transported in icepack-cooled containers. Roots were washed with tap water to remove soil particles, rinsed with sterile distilled water, and dried on sterile paper towels. Soil samples were sieved (2 mm, 10 mesh) to remove root fragments, residues, and stones, then were sent to the Agricultural Innovation and Technology Transfer Center (AITTC) (Benguerir, Morocco) for physicochemical analysis. Parameters measured were pH, electrical conductivity (EC), soil texture (clay, silt, sand), total nitrogen (TotalN), P2O5, total potassium (TotalK), organic matter (OM), total calcium carbonate (CaCO₃), nitrate (NO₃⁻) ammonium (NH₄⁺), chloride (Cl⁻), exchangeable oxides (K₂O, Na₂O, CaO, MgO), and trace elements (Cu, Mn, Fe, Zn, B). Analytical methods used (Aliyat et al., 2026) are mentioned in Table S1. DNA extraction and PCR amplification Root and shoot tissues were flash-frozen in liquid nitrogen and ground to a fine powder using a sterile mortar and pestle. Approximately 100 mg of homogenized tissue was transferred into 2 mL microtubes and further disrupted using a TissueLyzer II (QIAGEN, Germany). DNA was extracted using the DNeasy® Plant Pro Kit (QIAGEN, Germany), following the manufacturer’s instructions. For bulk and rhizospheric soils, 250 mg of soil was processed using DNA extraction with the DNeasy® PowerSoil® Pro Kit (QIAGEN, Germany). DNA concentrations were assessed with a BioSpectrometer (Eppendorf, Germany), and integrity was confirmed by electrophoresis on a 1% (w/v) agarose gel and visualized under UV light. Extracted DNA was stored at -20 °C. To minimize amplification of chloroplast DNA, bacterial 16 S rRNA genes were amplified from the V5-V6 regions [24] using the primers with Custom Sequence (CS) adapters: CS1-799 F (ACACTGACGACATGGTTCTACA-AACMGGATTAGATACCCKG) and CS2-1115R (TACGGTAGCAGAGACTTGGTCT-AGGGTTGCGCTCGTTG). PCR reactions were performed in a final volume of 25 µL containing 12.5 µL Platinum™ Direct PCR Universal Master Mix (ThermoFisher, Temara, Morocco). Thermal cycling conditions were: 95 °C for 3 min, 34 cycles of 95 °C for 30 s, 55 °C for 30 s and 72 °C for 1 min; and a final extension at 72 °C for 5 min. Amplifications were performed on a Mastercycler X50s (Eppendorf, Genrmany). Amplicons were verified on 1% agarose gels visualized with iBright FL1500 Imaging System (ThermoFisher, Temara, Morocco). A non-template controls (NTC) were included in each run. Library preparation and high-throughput sequencing PCR products were purified using AMPure XP beads (Beckman Coulter, USA) washed twice with ethanol, air-dried and resuspended in 10 mM Tris (pH 8.5). A secondary PCR reaction with unique indexing primers (Fluidigm, Markham, ON, Canada) was performed to add sample-specific barcodes and sequencing adapters. The indexed amplicons were purified again using AMPure XP beads, quantified with the Qubit DNA HS kit (ThermoFisher, Temara, Morocco), and pooled in equimolar concentrations. Sequencing was performed on al Illumina Miseq platform (MegaFlex, Casablanca, Morocco) using MiSeq Reagent Kit V3 (600 cycles). Data processing and statistical analysis Raw sequencing reads were processed in R version 4.5.0 [51] Base call (BCL) files were converted to fastq format using MiSeq Reporter with bcl2fastq pipeline. Reads were quality-filtered and denoised using DADA2 pipeline [14]. Sequences with expected errors > 2 were discarded; forward and reverse reads were truncated at 50 bp and 100 bp, respectively. Chimeras were removed using the consensus method. Taxonomic assignment of amplicon sequence variants (ASVs) was performed with the SILVA database (version 138.2) [43] using the default parameters of “assignTaxonomy”. Data were transformed to compositional format using phyloseq package [35, 36]. Alpha diversity (Shannon, Simpson, Chao1 and Pielou) was calculated with phyloseq. Analysis of variance (ANOVA) was used to test overall differences among species and niches, while pairwise comparisons were assessed with the Kruskal-Wallis test. Beta diversity was calculated using Bray-Curtis dissimilarities and visualized via Principal Coordinate Analysis (PCoA) using the vegan package [39]. Significance of community dissimilarities was tested with PERMANOVA (ADONIS, 999 permutations, Hellinger-transformed data. Relative taxonomic abundances were determined 4 using dplyr package [63]. The PICRUSt2 tool was used to predict abundance of KO (Kegg Orthology) functional pathways, using the ASVs generated by dada2. The potential correlation of KO pathways with environmental parameters was done using the vegan package of R and were plotted using the ggpicrust2 package of R. Venn diagram was constructed to determine the shared and unique taxa between the species and sample types using ggvenn package (v.0.1.10). Environmental correlations with taxa were tested using Mantel tests (Bray-Curtis dissimilarity) and visualized with linkET (version 0.0.7.4). Spearman correlation heatmaps were also generated. Core taxa were defined as ASVs with ≥ 70–90% prevalence and ≥ 0.1% relative abundance. Co-occurrence network computation and representation was performed with the package NetCoMi of R (https://doi.org/10.1093/bib/bbaa290) NetCoMi [41]. Analyses were conducted on the complete dataset as well as on subsets representing individual Artemisia species or plant compartments. To facilitate network visualization, analyses were restricted to the 100 most abundant genera in each subset. Prior to network inference, count data were normalized using centered log-ratio (CLR) transformation following zero replacement with a pseudo-count. Microbial association networks were inferred at the genus level using the SPIEC-EASI method, which estimates sparse conditional dependence relationships in compositional microbiome data. Keystone taxa were identified based on betweenness and eigenvector centrality. Results Soil physicochemical properties Soils exhibited pronounced heterogeneity in texture, chemistry, and nutrient status across sampling sites (Table S1). Texture ranged from sandy (82% sand, BS (bulk soil)1_ Artemisia herba-alba) to more balanced compositions (30% clay, BS2_ Artemisia herba-alba). All samples were strongly alkaline (pH 8.6–8.95), consistent with high total CaCO₃ (38.9–60.5%) and active lime (1.6–14.8%). Electrical conductivity was low to moderate (0.06–0.17 mS/cm), suggesting limited salinity stress. Nitrate (NO₃⁻) concentrations ranged from < 1 mg/kg to 39.2 mg/kg, while ammonium (NH₄⁺) was generally low, except in BS3_ Artemisia negrei (10.4 mg/kg). Organic matter (OM) (0.96–2.72%) and total nitrogen (TotalN) (0.08–0.18%) were moderate with higher values in BS4_Artemisia negrei and Artemisia mesatlantica, indicating localized nutrient enrichment. Phosphorus (15–41 mg/kg), potassium (212–378 mg/kg), and micronutrients, such as Fe (2.3–10.2 mg/kg), Mn (1.9–10.2 mg/kg), and Zn (0.15–0.75 mg/kg) showed variable distributions. Overall, gradients in pH, carbonate content, and nutrient availability are expected to impose strong selective pressures on bacterial community composition and function. Bacterial alpha diversity patterns across compartments and species Bacterial alpha diversity across niches (bulk soil, rhizosphere, root, and shoot) of the three Artemisia species revealed both conserved and species-specific patterns (Fig. 1, S2-2). For Shannon diversity, Artemisia herba-alba exhibited significant differences among compartments (p = 0.0092), with shoots harboring significantly lower diversity than roots (p < 0.01) indicating strong ecological filtering and limited microbial inputs aboveground, while other pairwise comparisons were nonsignificant (Fig. 1A). In Artemisia negrei, overall variation was also significant (p = 0.0239), driven mainly by a decline in diversity from root to shoot (p < 0.05) (Fig. 1B). In contrast, Artemisia mesatlantica showed no significant differences (p = 0.077), although diversity values were numerically higher in roots and rhizospheres compared to shoots (Fig. 1C). These trends were further assessed using the Observed Richness index (Observed ASVs), which revealed a significant compartmental effect in Artemisia herba-alba (p = 0.0062) (Fig. 1D) between roots and shoots, with shoots harboring higher numbers of observed ASVs than roots (adjusted p = 0.0065). In contrast, only marginally no significant effects of compartment were detected in Artemisia negrei (p = 0.06) and Artemisia mesatlantica (p = 0.053) (Fig. 1E, F), and no pairwise comparisons remained significant after multiple-testing correction. Overall, observed richness varied among plant compartments, with the strongest compartmental differentiation occurring in A. herba-alba. Collectively, bacterial alpha diversity showed clear compartment-specific patterns, particularly in A. herba-alba, where both Shannon diversity and observed richness varied significantly among niches. In contrast, community evenness remained stable across compartments and species, indicating that compartment effects primarily influenced diversity and richness rather than the distribution of taxa. Fig. 1Principal coordinates analysis (PCoA) of Hellinger-transformed bacterial communities across Artemisia species and sample types. PCoA plot showing beta diversity of microbial communities from bulk soil, rhizosphere, root, and shoot compartments associated with three Artemisia species (Artemisia herba-alba, Artemisia negrei, and Artemisia mesatlantica). The ordination is based on Aitchison distances calculated from CLR-transformed ASV data. Points are colored by sample type: red (bulk soil), green (rhizosphere), blue (root), purple (shoot), and shaped by plant species: circle (Artemisia herba-alba), square (Artemisia negrei), triangle (Artemisia mesatlantica). Ellipses represent 95% confidence intervals for sample type clusters. PERMANOVA results indicate that sample type explains a greater proportion of the variance (R² = 0.13, p = 0.001) than species identity (R² = 0.06, p = 0.026) Principal coordinates analysis (PCoA) of Hellinger-transformed bacterial communities across Artemisia species and sample types. PCoA plot showing beta diversity of microbial communities from bulk soil, rhizosphere, root, and shoot compartments associated with three Artemisia species (Artemisia herba-alba, Artemisia negrei, and Artemisia mesatlantica). The ordination is based on Aitchison distances calculated from CLR-transformed ASV data. Points are colored by sample type: red (bulk soil), green (rhizosphere), blue (root), purple (shoot), and shaped by plant species: circle (Artemisia herba-alba), square (Artemisia negrei), triangle (Artemisia mesatlantica). Ellipses represent 95% confidence intervals for sample type clusters. PERMANOVA results indicate that sample type explains a greater proportion of the variance (R² = 0.13, p = 0.001) than species identity (R² = 0.06, p = 0.026) Beta diversity shaped by both niche and host species Principal Coordinates Analysis (PCoA) of bacterial communities revealed clear clustering by compartment across all three Artemisia species (Fig. 2). Root and shoot samples of A. herba-alba and A. negrei appeared to group more closely, whereas A. mesatlantica showed greater dispersion, particularly in rhizosphere and shoot compartments. However, these patterns are based on visual inspection of the ordination. PERMANOVA analysis indicated that compartment was the primary driver of variation (R² = 0.13, p = 0.001), with a smaller but significant contribution of host species (R² = 0.06, p = 0.026). Interactive effect between species and niches was nonsignificant (R²=0.149, p = 091) (Table S2). Fig. 2Genus-level relative abundance of bacterial communities across different compartments and Artemisia species. Stacked bar plots represent the relative abundance of dominant bacterial genera in bulk soil, rhizosphere, root, and shoot compartments of three Artemisia species: Artemisia herba-alba, Artemisia negrei, and Artemisia mesatlantica. Each bar corresponds to one compartment–species combination. Colors represent different bacterial genera, with notable variation in community composition observed across both plant compartments and species Genus-level relative abundance of bacterial communities across different compartments and Artemisia species. Stacked bar plots represent the relative abundance of dominant bacterial genera in bulk soil, rhizosphere, root, and shoot compartments of three Artemisia species: Artemisia herba-alba, Artemisia negrei, and Artemisia mesatlantica. Each bar corresponds to one compartment–species combination. Colors represent different bacterial genera, with notable variation in community composition observed across both plant compartments and species Bacterial community composition across compartments At the phylum level (Fig. S4A), Pseudomonadota, Actinobacteriota, Chloroflexota, Bacteroidota, and Bacillota dominated across all samples, but their relative abundances varied. In Artemisia herba-alba, Actinobacteriota, Pseudomonadota, and Patescibacteriota were prevalent in roots and shoots, while Chloroflexota dominated bulk and rhizosphere soils. In Artemisia negrei, Actinobacteriota dominated in bulk soil, whereas Pseudomonadota and Actinobacteriota were enriched in the rhizosphere, roots, and shoots. In Artemisia mesatlantica, Pseudomonadota, Actinobacteriota, and Patescibacteriota dominated in root, shoot, and rhizosphere niches, while bulk soil was enriched by Chloroflexota. At the family level (Fig. S4B), communities were highly compartmentalized and species dependent, with a large fraction of unclassified families. Bulk soils showed minimal interspecific variation, dominated by Bacillaceae, unknown and AKIW781. Rhizospheres were richer with Beijerinckiaceae, Oxalobacteraceae, Comamonadaceae, Micrococcaceae and Nocardioidaceae particularly abundant in Artemisia mesatlantica and Artemisia negrei. At the genus level (Fig. 3), bulk soils were were dominated by Bacillus, with Chloronema, the most abundant in Artemisia herba-alba. Rhizospheres exhibited higher diversity, enriched by Bacillus, Rubrobacter, Sphingomonas, Streptomyces, Massilia, and Microvirga with Sodalis distinctively enriched in Artemisia herba-alba. Root bacterial communities showed strong host filtering: TM7a, Nocardioides, Massilia, Rhizobium, Sphingomonas, Devosia, Streptomyces and Bacillus were common, with species-specific enrichments such as Bacillus in Artemisia mesatlantica roots and Deinococcus in Artemisia negrei roots. Shoots hosted the lowest diversity but consistently included Bacillus, TM7a, Thermus, Streptomyces, Blastococcus, Devosia, and Arthrobacter. Species-specific enrichments were observed, including Curtobacterium in A. mesatlantica shoots, Sodalis in Artemisia negrei roots, and Rhizobium in Artemisia herba-alba roots, indicating that plant species identity shapes bacterial communities by selectively enriching distinct microbial taxa. Collectively, these results highlight niche-specific structuring alongside both conserved and species-specific bacterial associations. Fig. 3Spearman correlation between dominant bacterial genera and environmental variables.A Partial Mantel test and Pearson correlation analysis among environmental variables and bacterial community structure. The lower triangle shows Pearson correlation coefficients among environmental factors, with positive correlations in blue and negative in red. The size of the squares reflects the magnitude of the correlation, and outlined squares indicate statistically significant correlations (p < 0.05). The upper triangle includes partial Mantel test results linking each environmental variable to bacterial community composition (center node “Bacteria”), controlling for spatial effects. Line thickness reflects Mantel’s r values, and color indicates. Mantel’s p-values: orange (< 0.01), green (0.01–0.05), and gray (≥ 0.05). Strong correlations (Mentel’s r > = 0.4) highlight the most influential environmental drivers. B The heatmap shows Spearman correlation coefficients between the 20 most abundant bacterial genera and measured soil physicochemical parameters. Positive correlations are shown in blue, and negative correlations in red. The intensity of the color indicates the strength of the correlation. Asterisks denote statistically significant correlations (p < 0.05). The color gradient ranges from − 0.5 to + 0.5. Environmental variables include soil texture (clay, silt, sand), pH, electrical conductivity (EC), macro- and micronutrients (e.g., NNO₃⁻, NNH₄⁺, P₂O₅, TotalN K₂O, CaO, MgO, Fe, Zn, Cu, Mn, B, Total K), organic matter (OM), active lime, and total CaCO₃ Spearman correlation between dominant bacterial genera and environmental variables.A Partial Mantel test and Pearson correlation analysis among environmental variables and bacterial community structure. The lower triangle shows Pearson correlation coefficients among environmental factors, with positive correlations in blue and negative in red. The size of the squares reflects the magnitude of the correlation, and outlined squares indicate statistically significant correlations (p < 0.05). The upper triangle includes partial Mantel test results linking each environmental variable to bacterial community composition (center node “Bacteria”), controlling for spatial effects. Line thickness reflects Mantel’s r values, and color indicates. Mantel’s p-values: orange (< 0.01), green (0.01–0.05), and gray (≥ 0.05). Strong correlations (Mentel’s r > = 0.4) highlight the most influential environmental drivers. B The heatmap shows Spearman correlation coefficients between the 20 most abundant bacterial genera and measured soil physicochemical parameters. Positive correlations are shown in blue, and negative correlations in red. The intensity of the color indicates the strength of the correlation. Asterisks denote statistically significant correlations (p < 0.05). The color gradient ranges from − 0.5 to + 0.5. Environmental variables include soil texture (clay, silt, sand), pH, electrical conductivity (EC), macro- and micronutrients (e.g., NNO₃⁻, NNH₄⁺, P₂O₅, TotalN K₂O, CaO, MgO, Fe, Zn, Cu, Mn, B, Total K), organic matter (OM), active lime, and total CaCO₃ Shared and unique taxa among species and compartments Across the three Artemisia species, considering all plant niches combined to capture species-level microbial signatures, 117 taxa (31.7%) were shared, while each species also harbored unique members: 42 in A. herba-alba (11.4%), 68 in A. negrei (18.4%), and 51 in A. mesatlantica (13.8%) (Fig. S5A). Pairwise comparisons revealed 28 taxa (7.6%) shared between Artemisia herba-alba and Artemisia negrei, 10 (2.7%) between Artemisia herba-alba and Artemisia mesatlantica, and 53 (14.4%) between Artemisia negrei and Artemisia mesatlantica. Across compartments (Fig. S5B), 32 taxa (8.7%) were common to all niches, while roots hosted the largest number of unique taxa (102; 27.6%), followed by rhizosphere (38; 10.3%), shoots (37; 10.0%), and bulk soil (7; 1.9%). Roots and rhizospheres also showed notable overlap (41 taxa; 11.1%). These findings indicate strong host- and niche-specific assembly, with roots and rhizospheres functioning as key zones of microbial enrichment and diversification, whereas bulk soil represents the primary reservoir from which plant-associated communities are recruited. Core bacteriome across species and compartments Core taxa (≥ 80% prevalence, ≥ 0.1% abundance) showed species- and compartment-specific patterns (Fig. 4). In Artemisia herba-alba, five core bacteria were identified: Bacillus, Microvirga, and Steroidobacter enriched in bulk soil, rhizosphere, and roots, while Shinella and Rhizobium were restricted to roots (Fig. 4A). Artemisia negrei and Artemisia mesatlantica harbored more diverse core bacteriome across compartments (Fig. 4B-C). To evaluate the robustness of the core microbiome definition, a sensitivity analysis was performed using alternative prevalence thresholds (70% and 90%). Key genera, including Bacillus, Microvirga, and Rhizobium, were consistently retained across thresholds, indicating a stable core bacteriome, whereas less prevalent taxa were sensitive to threshold variation (Fig. S6A-E). In Artemisia herba-alba, bulk and rhizosphere bacteriome were enriched in Blastococcus, Arthrobacter, Mycobacterium, Rubrobacter, Nocardioides, and unknown taxa, while roots were dominated by Rubrobacter, Truepera, and Wangia. In Artemisia mesatlantica, rhizospheres were enriched in Marmoricola, Arthrobacter, Rhizobium, and Microvirga, while roots were dominated by extremophiles such as Rubrobacter, Truepera, and Thermus. Bulk soil bacteriomes remained distinct, with lower prevalence of plant-associated genera. Comparative analysis revealed Blastococcus and Arthrobacter as core genera in rhizospheres across all species, with highest relative prevalence in Artemisia mesatlantica (Fig. S7A). In roots, Rhizobium was the only core genus, again most abundant in Artemisia mesatlantica (Fig. S7B). These patterns highlight species-specific enrichment of core taxa, with Artemisia mesatlantica supporting higher overall abundance. Fig. 4Core bacterial composition across plant compartments in three Artemisia species. Stacked bar plots showing the mean relative abundance (normalized) of bacterial genera in bulk soil, rhizosphere, and root compartments of Artemisia herba-alba (A), Artemisia negrei (B), and Artemisia mesatlantica (C). Each bar represents the average microbial composition across replicates per compartment. Only the most abundant genera are shown; low-abundance and unclassified taxa are grouped as "Unknown." Core bacterial composition across plant compartments in three Artemisia species. Stacked bar plots showing the mean relative abundance (normalized) of bacterial genera in bulk soil, rhizosphere, and root compartments of Artemisia herba-alba (A), Artemisia negrei (B), and Artemisia mesatlantica (C). Each bar represents the average microbial composition across replicates per compartment. Only the most abundant genera are shown; low-abundance and unclassified taxa are grouped as "Unknown." Predicted functional profiling To further explore the potential functional attributes of bacterial communities, PICRUSt2 analysis was performed to infer metabolic pathways across compartments (Fig. S8A-D). The results revealed distinct functional profiles between bulk soil, rhizosphere, root, and shoot compartments. Notably, pathways related to metabolism (e.g., amino acid metabolism, lipid metabolism), secondary metabolite biosynthesis, and signal transduction were enriched in rhizosphere and root-associated communities compared to bulk soil. In contrast, shoot-associated communities displayed reduced representation of several metabolic pathways. Additionally, pathways associated with stress response and environmental adaptation showed higher predicted abundance in root-associated microbiomes, whereas bulk soil communities exhibited a broader but less specialized functional profile. Environmental drivers of bacterial community structure Mantel tests (Fig. 5A) were conducted to evaluate the influence of environmental variables on bacterial community structure. Several factors displayed strong and statistically significant correlations with community composition. For instance, EC, active lime, organic matter (OM), NNO₃⁻, total CaCO₃, TotalN, zinc and P₂O₅ were the most influential variables (Mantel’s r > 0.2, p < 0.01), shown by thick connecting lines to “Bacteria.” These variables were also strongly intercorrelated (e.g., pH with active lime and CaCO₃; EC with Na₂O and MgO), suggesting overlapping soil chemical gradients shaping microbial assemblages. The positive association of OM and TotalN with bacterial community structure further highlights the influence of nutrient availability on microbial community composition. Taken together, these analyses identify soil pH, salinity (EC), nitrogen species (NNO₃⁻, NNH₄⁺), carbonates (CaCO₃, active lime), and organic nutrients (OM, TotalN) as key environmental factors associated with variation in overall bacterial community structure, as well as with the distribution of dominant bacterial genera. To complement these findings, we analyzed correlations between dominant bacterial genera and soil physicochemical properties. The spearman correlation heatmap (Fig. 5B) revealed distinct associations between environmental parameters and the most abundant bacterial taxa. Notably, Chloronema showed strong negative correlations with site, latitude, electrical conductivity (EC), sand content, active lime, total CaCO3, organic matter, P2O5, manganese, zinc and total nitrogen, suggesting a potential suppression in nutrient-enriched or alkaline soils. In contrast, Nocardioides, Arthrobacter, Truepera and Massilia and Deinococcus were strongly positively correlated with boron, zinc, manganese, potassium oxide, total nitrogen, and total calcium carbonate. Additionally, Sphingomonas, Microvirga, and Arthrobacter were positively associated with salinity-related variables such as EC, pH, and Na₂O, indicating their tolerance of or preference for saline-alkaline conditions (Fig. 5A). Fig. 5Spearman correlation between dominant bacterial genera and environmental variables. A Partial Mantel test and Pearson correlation analysis among environmental variables and bacterial community structure. The lower triangle shows Pearson correlation coefficients among environmental factors, with positive correlations in blue and negative in red. The size of the squares reflects the magnitude of the correlation, and outlined squares indicate statistically significant correlations (p < 0.05). The upper triangle includes partial Mantel test results linking each environmental variable to bacterial community composition (center node “Bacteria”), controlling for spatial effects. Line thickness reflects Mantel's r values, and color indicates. Mantel's p-values: orange (<0.01), green (0.01–0.05), and gray (≥0.05). Strong correlations (Mentel’s r >= 0.4) highlight the most influential environmental drivers. B The heatmap shows Spearman correlation coefficients between the 20 most abundant bacterial genera and measured soil physicochemical parameters. Positive correlations are shown in blue, and negative correlations in red. The intensity of the color indicates the strength of the correlation. Asterisks denote statistically significant correlations (p < 0.05). The color gradient ranges from –0.5 to +0.5. Environmental variables include soil texture (clay, silt, sand), pH, electrical conductivity (EC), macro- and micronutrients (e.g., NNO₃⁻, NNH₄⁺, P₂O₅, TotalN K₂O, CaO, MgO, Fe, Zn, Cu, Mn, B, Total K), organic matter (OM), active lime, and total CaCO3. Spearman correlation between dominant bacterial genera and environmental variables. A Partial Mantel test and Pearson correlation analysis among environmental variables and bacterial community structure. The lower triangle shows Pearson correlation coefficients among environmental factors, with positive correlations in blue and negative in red. The size of the squares reflects the magnitude of the correlation, and outlined squares indicate statistically significant correlations (p < 0.05). The upper triangle includes partial Mantel test results linking each environmental variable to bacterial community composition (center node “Bacteria”), controlling for spatial effects. Line thickness reflects Mantel's r values, and color indicates. Mantel's p-values: orange (<0.01), green (0.01–0.05), and gray (≥0.05). Strong correlations (Mentel’s r >= 0.4) highlight the most influential environmental drivers. B The heatmap shows Spearman correlation coefficients between the 20 most abundant bacterial genera and measured soil physicochemical parameters. Positive correlations are shown in blue, and negative correlations in red. The intensity of the color indicates the strength of the correlation. Asterisks denote statistically significant correlations (p < 0.05). The color gradient ranges from –0.5 to +0.5. Environmental variables include soil texture (clay, silt, sand), pH, electrical conductivity (EC), macro- and micronutrients (e.g., NNO₃⁻, NNH₄⁺, P₂O₅, TotalN K₂O, CaO, MgO, Fe, Zn, Cu, Mn, B, Total K), organic matter (OM), active lime, and total CaCO3. Bacterial co-occurrence networks across species Co-occurrence network analysis revealed distinct yet modular bacteriomes in Artemisia herba-alba, Artemisia negrei, and Artemisia mesatlantica (Fig. S9 A-C). In Artemisia herba-alba (Fig. S9A), the network had a modular architecture, with one major cluster formed comprising Belnapia, Skermanella, and Friedmaniella, with a notable very strong negative correlation between two Belnapia ASVs. In Artemisia negrei (Fig. S9B), the network was denser and more highly connected, revealing a central role for Pseudarthrobacter, Blastococcus, Rubellimicrobium and Skermanella, along with highly connected unclassified taxa. This structure suggests a cooperative consortium supporting plant adaptation under extreme aridity. The Artemisia mesatlantica network (Fig. S9C) also displayed a modular architecturem with the most abundant ASVs gathered in a single cluster comprising Bacillus, Rhizobium, Shinella and Pseudomonas as hub taxas. Comparative analysis across the three species (Fig. S9D) showed a higher connectivity than host-specific networks, pointing to a conserved microbial core with generalist tendencies between the three species. Together, these findings suggest that Artemisia species maintain distinct yet functionally convergent microbial networks, in which keystone taxa both known and uncharacterized may enhance host resilience to harsh desert environments. Together, these findings indicate that Artemisia species harbor distinct yet structured microbial networks, with both shared and species-specific keystone taxa, which may reflect functional convergence across hosts. Discussion Our study reveals that bacterial community assembly in Artemisia species is strongly shaped by plant compartments with major contributions from both host genotype and local edaphic factors, highlighting the combined influence of environmental filtering and host-associated processes. Importantly, our results suggest a hierarchical assembly process in which environmental filtering imposed by alkaline and arid soil conditions acts as a primary driver, while host filtering further refines microbial community composition across species. From all three species, Artemisia herba-alba, Artemisia negrei, and Artemisia mesatlantica, we observed both conserved and species-specific patterns in bacterial alpha and beta diversity, taxonomic composition, and environmental filtering highlighting the dynamic nature of plant–microbe interactions at the soil-plant interface. However, because sampling was partly site-specific and only partially overlapping among species, the observed host effects should be interpreted with caution, as local environmental variation may also contribute to the observed differences in microbial community structure. Filtering of bacterial communities in diverse niches Consistent with prior findings in both wild and cultivated plant systems [12, 17], we observed strong spatial structuring of bacterial communities alongside plant compartments. The rhizosphere and root compartments harbored higher taxonomic richness and compositional particularity compared to shoot and bulk soil. Although Shannon diversity did not consistently differ among bulk soil, rhizosphere, and root compartments this likely reflects the relatively stable distribution of taxa (evenness) across these niches, as supported by the absence of significant differences in Simpson and Pielou indices. In contrast, richness-based metrics (Chao1) revealed clear compartmental effects, indicating that filtering processes primarily influence taxonomic composition and presence–absence patterns rather than the relative abundance structure of communities. This pattern reflects the combined influence of soil-driven environmental filtering and host-mediated recruitment processes. Root communities exhibited reduced diversity but greater specificity, a pattern commonly attributed to selective colonization through root exudates and immune regulation [33, 34, 65]. Our findings extend current knowledge by showing that bacterial compartmentalization is not uniform across Artemisia species. Specially, Artemisia herba-alba harbored relatively low-diversity, whereas both Artemisia negrei and Artemisia mesatlantica supported richer and more complex communities enriched in ecologically functional taxa such as Rubrobacter, Streptomyces, Rhizobium, and Trupera (Fig. 3). These genera are known for roles in nutrient cycling, antibiotic production, and stress tolerance [9, 56], although their functional roles were not directly assessed in this study. Taken together, these results directly support our first hypothesis, confirming that bacterial community composition differs not only across plant niches but also between Artemisia species, with Artemisia negrei and Artemisia mesatlantica exhibiting more distinct community composition patterns. Parallel studies on medicinal plants have similarly documented compartment driven microbial structuring, for example in Panax ginseng [22, 66] and in desert leguminous shrub species, namely, Caragana microphylla, Hedysarum mongolicum, and Hedysarum scoparium [68], where root associated microbiomes were enriched in taxa linked to secondary metabolite biosynthesis and stress resilience. In Artemisia annua, the rhizosphere microbiome has been implicated in modulating artemisinin biosynthesis [2, 54], suggesting that the microbiome might influence key phytochemical pathways in closely related species. These findings indicate that microbiome assembly in Artemisia is governed by the interplay between environmental filtering and host filtering, with soil conditions constraining the available microbial pool and host species selectively recruiting from this pool. Bacteriome differentiation in Artemisia. Although multiple niches associated with Artemisia employed the strongest influence on microbiome structure, host species identity also contributed to microbiome differentiation, although this effect was secondary to strong environmental filtering driven by soil physicochemical conditions, although this effect occurred alongside strong environmental filtering driven by soil physicochemical conditions [40, 61]. Differences in the abundance of genera, such as Bacillus, Devosia, and Rubrobacter between Artemisia herba-alba, Artemisia negrei and Artemisia mesatlantica suggested divergent recruitment approaches that may reflect differences in ecological or environmental adaptation. For example, the consistent presence of thermotolerant taxa, such as Trupera and Thermus in Artemisia mesatlantica root may indicate the niche adaptation to harsh and arid environments, which is a known feature of Artemisia mesatlantica [3]. Likewise, the enrichment of Massilia and Nocardioides in Artemisia negrei could reflect unique metabolic preferences or symbiotic associations adapted to local edaphic conditions. These host-genotype consequences are likely mediated through both structural (e.g., root architecture) and biochemical (e.g., exudate composition) mechanisms [10, 46, 60]. Our study showed that Artemisia mesatlantica and Artemisia negrei shared more common genera across rhizosphere, root, and shoot compared to Artemisia herba-alba, suggesting partially overlapping microbial assemblages, which may reflect shared host traits or similar environmental conditions. Notably, these two species were sampled from a shared site, which may partly explain their higher microbial overlap. These differences emphasized the influence of plant genotype on shaping distinct yet partially overlapping bacterial community along the soil plant interface. All together, these results demonstrated that bacterial community assembly is strongly structured by plant compartments and species identity. While the bulk soil bacterial taxa remain stable, distinct differentiation appeared in rhizosphere, root and shoot with Artemisia mesatlantica and Artemisia negrei hosting more diverse and abundant bacterial communities than Artemisia herba-alba. These patterns highlighted a potential role of plant genotype in shaping the assembly of microbial consortia across compartments, although environmental filtering likely acts concurrently. Moreover, this genotype specific microbial recruitment corresponded with the findings in other medicinal plants, such as Salvia miltiorrhiza [15], Glycyrrhiza uralensis [32] and Ocimum sanctum [52], where host genetic variation correlated with differential microbiome composition and functional traits. These patterns underscore possible co-association patterns between medicinal plants and their microbiomes, which may influence phytochemical profiles and ecological fitness. Core taxa and functional implications Despite interspecific variation, certain genera were consistently detected across compartments and hosts, including Bacillus and Microvirga (Fig. 4). Blastococcus emerged as core genera in rhizospheres (Fig. S7A), while Rhizobium was the only core genus consistently detected in roots across all species (Fig. S7B). These patterns support our second hypothesis, confirming that a conserved core bacteriome persists across species and niches, suggesting the presence of conserved taxa across species and compartments, although their functional roles were not directly assessed. The stability of key taxa (e.g., Bacillus, Microvirga, and Rhizobium) across multiple prevalence thresholds further supports the robustness of the identified core microbiome and indicates that these patterns are not driven by arbitrary threshold selection (Fig. S6A-E, S7A-E). The persistence of this core microbiome indicates consistent association across hosts and compartments, potentially relevant to plant-associated processes, in line with previous reports (Rachid et al., 2022; [53, 64]). To further explore potential functional attributes, PICRUSt2-based predictions revealed compartment-specific enrichment of metabolic pathways, with rhizosphere and root-associated communities showing higher representation of pathways related to metabolism, stress response, and secondary metabolite biosynthesis (Fig. S8A-D). These predicted functional profiles are consistent with the taxonomic composition of dominant genera identified in this study. However, these results should be interpreted cautiously, as they represent inferred functional potential rather than direct measurements. Notably, Artemisia mesatlantica exhibited a more distinct and enriched microbiome compared to the other species, which may indicate stronger microbial recruitment specificity. This pattern could reflect adaptation to more extreme or localized environmental conditions, where tighter host–microbe associations are favored. Functionally, Bacillus is well known for promoting drought tolerance, producing antimicrobial compounds, and suppressing plant diseases [18, 58], while Microvirga [37, 45] and Rhizobium [21, 23, 26] have been reported to contribute to nitrogen fixation and other plant growth-promoting activities. The consistent presence of these taxa, particularly in the rhizosphere and root, indicates the presence of taxa previously reported to possess plant-beneficial traits. In accordance with these results, some studies on medicinal plants, such as Echinacea purpurea and Curcuma longa highlighted that core bacterial taxa including Streptomyces and Bacillus are integral to pathogen resistance and secondary metabolite enhancement [8, 27, 50, 59]. These observations suggest that the core microbiome of Artemisia may be associated with plant health and ecological performance; however, these functional roles should be considered as hypotheses requiring validation through metagenomic, transcriptomic, or culture-based approaches. This suggested that the core microbiome of Artemisia may be associated with plant health and ecological performance, although functional validation is required. Interestingly, Artemisia herba-alba exhibited a more constrained and compartmentalized bacteriome, consistent with its lower diversity and reduced taxonomic complexity compared to the other species. This pattern may reflect adaptation to arid and nutrient-poor steppe environments, where stronger environmental constraints and host filtering could lead to a narrower microbial niche breadth. In contrast, Artemisia mesatlantica and Artemisia negrei shared more overlapping taxa across compartments, highlighting a pronounced host effect on microbiome assembly. This pattern may reflect host-specific traits, such as root exudate composition or physiological characteristics, that promote broader microbial recruitment and reduced compartmental filtering compared to Artemisia herba-alba. Such differences might have ecological implications for nutrient acquisition, stress tolerance and secondary metabolite production which are known to differ among Artemisia species [28]. Soil physicochemical properties shape Artemisia bacterial microbiome Soil physicochemical properties emerged as strong determinants of microbiome structure. Our results showed that edaphic parameters such as pH, salinity (EC), active lime, and nutrient levels (Total N, OM, P₂O₅) were significantly correlated with both dominant bacterial genera and overall community composition. These findings are consistent with consistent with and reinforce previous studies demonstrating that soil physicochemical properties are primary drivers of microbiome structure [19, 29] and strongly support our third hypothesis, stipulating that soil physicochemical properties, particularly pH, salinity, and nutrient availability, act as environmental filters that shape bacterial communities prior to and alongside host-mediated selection. For instance, the alkaline pH and high carbonate content in our samples likely favored the proliferation of alkaliphilic genera such as Rubrobacter and Blastococcus. Similarly, correlations between nitrogen species and diazotrophic taxa, including Rhizobium and Microvirga, underscored the central role of nitrogen dynamics in structuring rhizosphere communities [11, 48]. These environmental effects were validated through both Spearman correlations and Mantel tests (Fig. 5), highlignting the complex interplay between host identity, niche differentiation, and soil parameters in shaping microbial community structure. Similar studies in the past in medicinal plants growing in alkaline and arid soils reported analogous environmental filtering effects, for instance in Glycyrrhiza uralensis [42] and Astragalus membranaceus [31]. Such edaphic influences are critical to consider for microbiome manipulation and restoration efforts in degraded or marginal lands. Together, these results indicate that microbiome assembly in Artemisia is governed by a hierarchical process in which broad-scale environmental filtering by soil properties precedes and constrains host-specific microbial selection. Assembly of cooperative bacterial networks in arid environments Our findings show that Artemisia roots, shoots, and rhizospheres host highly integrated, positively co-occurring bacterial networks. The absence of hub taxa in the root and rhizosphere compartments suggest that microbial network organization differs among species, potentially reflecting differences in environmental context or community structure in these compartments. In congruence with the results suggesting lower host specificity of Artemisia mesatlantica and Artemisia negrei, the co-occurence network analysis showed a higher clustering coefficient and natural connectivity in Artemisia herba-alba compared to the other species. The hub taxa were different between the three species, suggesting that different mechanisms of selection are at play in the different Artemisia species. Functionally, hub taxa are often considered keystone species that disproportionately influence community structure by maintaining network connectivity, facilitating microbial interactions, and stabilizing community dynamics. In plant-associated microbiomes, such taxa may contribute to nutrient acquisition, production of bioactive compounds, and enhancement of host stress tolerance. However, the fact that only one hub taxa was found for each species indicates that the synergistic interactions of bacterial taxa within the sampled Artemisia plants may indicate a balance between cooperative and competitive interactions within these communities. The dense, modular, and cross-linked architectures observed particularly in Artemisia herba-alba and Artemisia negrei, suggested cooperative or niche-complementary interactions that may enhance microbiome stability and functional resilience under arid, nutrient-poor conditions [5]. Keystone taxa from multiple phyla likely mediate cross-compartment connectivity, supporting coordinated plant–microbe functions [53], while the more compartmentalized but still cooperative network of Artemisia mesatlantica points to species-specific assembly patterns [38]. Bulk soil networks, by comparison, were dominated by isolated, niche-specialized clusters, underscoring the plant’s role in fostering integrated microbial consortia through selective recruitment and facilitation [17, 49]. These structural signatures suggest that Artemisia exerts strong ecological filtering, yielding microbiomes adapted to persistence under extreme environmental conditions in extreme environments [44]. Ecological and applied perspectives pattern Understanding the bacteriome assembly patterns in wild Artemisia species holds both ecological and applied relevance. The genus Artemisia includes numerous medicinal and aromatic plants with its roles in traditional medicine, phytoremediation and arid land agriculture. The association with stress resilient microbes, such as Streptomyces, Sphingomonas and Trupera may contribute to their adaptation to harsh environments and potentially associated with bioactive compound production. Harnessing these microbial associations could inform sustainable agriculture, especially in arid and alkaline soils where Artemisia naturally thrives. Moreover, the observed genotype specific bacterial signatures may be leveraged for microbiome informed breeding strategies, considering by selecting host genotypes with beneficial microbiome recruitment traits and it might also be possible to potentially improve plant performance and resilience under environmental stress [13]. However, comparable efforts in other medicinal plants, such as Panax ginseng and Artemisia annua have shown promise in manipulating root associated microbiomes to enhance secondary metabolite yields and stress tolerance [23, 66]. Integrating native microbiomes into agronomic practices could thus enhance the productivity and resilience of medicinal plants under climate change scenarios. These results move beyond geographic specificity and provide a broader ecological framework for understanding how plant-associated microbiomes are assembled under extreme environmental constraints, highlighting the relative roles of environmental and host filtering in shaping microbial diversity and function. Conclusion and future directions Our findings demonstrated that bacterial community composition in Artemisia species was shaped by a combination of host genotype, microbial niches and edaphic filtering. While some core genera, such as Streptomyces and Bacillus persist across compartments and hosts, substantial differences in microbial richness, composition, and recruitment strategies were evident among Artemisia herba-alba, Artemisia negrei and Artemisia mesatlantica. The enrichment of thermophilic and nitrogen cycling taxa in roots and rhizospheres of Artemisia negrei and A. mesatlantica suggested that these species may harbor bacterial communities potentially better adapted to arid and nutrient variable environments. These host-associated differences, coupled with strong categorization underscore the potential of leveraging native plant microbiomes for ecological adaptation and sustainable agriculture in drylands. However, several limitations should be acknowledged. Our study focused exclusively on bacterial communities and did not assess fungal and archaeal components of the microbiome, which may also influence plant health and fitness. Functional interpretations remain hypothetical in the absence of metagenomic, or transcriptomic validation and the geographic and temporal scope was limited to a single season. To elaborate on these findings, future work should incorporate multiomics approaches, expanding taxonomic coverage and conduct manipulative controlled and field experiments to unravel causal relationships in plant microbe soil interactions. Additionally, isolating key taxa enriched in specific hosts or compartments, such as Trupera, Microvirga, and Rubrobacter could yield valuable microbial resources for microbiome engineering and phytobiome-informed plant breeding. Overall, this work provides a foundational understanding of the core and variable microbiota in wild and Artemisia species and offers a platform for future efforts to harness native microbiomes for ecosystem resilience and agricultural innovation. Future studies integrating co-located sampling designs and functional approaches (e.g., metagenomics or metabolomics) are needed to disentangle host-specific effects from environmental influences and to validate the functional roles of the identified taxa. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3 Supplementary Material 4 Supplementary Material 4 Supplementary Material 5 Supplementary Material 5