Exploring the diversity and potential of host-associated microorganisms in Hypnum leptothallum to enhance plant resilience to drought stress for karst desertification restoration

Abstract

Background

Karst rocky desertification poses a serious threat to the ecosystems of southwest China, where Hypnum leptothallum plays a crucial role in forming microbial crusts essential for restoration. However, the use of native microbial applications in this area remains largely unexplored.

Results

In this study, the host-associated microbial communities of H. leptothallum from four severely desertified regions in Guizhou Province were characterized using high-throughput sequencing. The results revealed conserved α- and β-diversity, with dominant bacterial phyla being Pseudomonadota (34–47%) and Actinomycetota (23–35%), and fungal phyla being Ascomycota (57–83%) and Basidiomycota (14–32%). Subsequent carbon-source preference analysis guided the formulation of specialized media (e.g., α-D-lactose, N-acetyl-D-glucosamine) to isolate culturable strains, with cross-referencing identifying 14 bacterial and 36 fungal species consistently shared between sequencing and cultivation. Functional evaluation demonstrated bacterial dominance in inorganic phosphorus solubilization (68% of strains), protease synthesis (76%), ammonia production (56%), and indole-3-acetic acid biosynthesis (62%), while fungi excelled in organophosphorus solubilization. Further drought tolerance and gametophyte co-culture assays identified 10 drought-resistant bacterial strains and 16 strains significantly enhancing H. leptothallum growth within 7 days.

Conclusions

These functionally validated strains, particularly drought-adapted and growth-stimulating species closely related to Rhodococcus erythropolis, provide targeted microbial resources for developing synthetic inoculants to optimize artificial crust propagation in karst restoration.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40793-026-00849-z.

Background

Karst rocky desertification (KRD), recognized as one of China’s major ecological challenges, is a severe land degradation process prevalent in southwestern China (Yunnan, Guizhou, and Guangdong provinces) [1]. Driven by unsustainable human activities, KRD leads to critical ecological crises, including soil erosion, ecosystem collapse, and water scarcity [2–5]. While local governments have implemented vegetation restoration and agroforestry measures with some success, significant challenges remain—with 7.223 million hectares still requiring management at the end of 2021 [6–9]. The urgent need for developing more sustainable and effective remediation strategies is highlighted by the persistent severity of rocky desertification and its associated impacts.

Although most vascular plants struggle to colonize harsh rocky desertification terrains due to xeric conditions and limited substrates, some bryophytes demonstrate remarkable adaptability by successfully establishing populations on bare carbonate rock surfaces in karst ecosystems [4]. As the most primitive group of higher plants, bryophytes serve as pioneer communities in ecological succession within natural ecosystems [10]. Through their unique physiological architecture and reproductive strategies, these organisms effectively maintain water balance and optimize nutrient utilization under extreme environmental conditions [4, 11]. These bryophytes exhibit a distinctive growth pattern characterized by compact, intricately twisted, and helical structures that frequently coalesce into dense cushion-like aggregations or form extensive prostrate colonies [7]. Their remarkable physiological capacity for rapid hydration and exceptional moisture retention is complemented by evolutionary adaptations in both morphological architecture and stress-response mechanisms [12]. These synergistic biological features collectively enhance their ecological resilience, enabling survival under extreme abiotic stresses, including thermal extremes, periodic desiccation, and calcareous substrate conditions [4]. Numerous bryophyte species exhibit sophisticated reproductive modalities (e.g., spore propagation, fragmentation, bulbils, and gemmae) [13]. The diversified reproductive strategies effectively enable their population persistence and expansion in arid desert environments characterized by water scarcity and harsh conditions [14].

In karstic rocky desertification areas, bryophytes usually coexist with algae, bacteria, fungi, and inter-root fine particles on exposed rocks to form moss crusts [4, 7]. Many studies have shown that moss crusts can improve soil nutrient status, influence water infiltration and evaporation, promote the reproduction and survival of soil organisms and vegetation succession, and increase biodiversity [7]. These crusts have an irreplaceable role in the restoration and reconstruction of karstic rocky desertification ecosystems. Notably, among bryophyte crusts prevalent in karst rocky desertification regions, Hypnum leptothallum formations demonstrated superior performance in critical ecological metrics compared to counterparts (Brachythecium perminusculum, Bryum dichotomum, and Cyrto-hypnum pygmaeum) [15]. Specifically, they exhibited 12–53% greater biomass accumulation, 2.8-fold higher hydraulic conductivity, and 34% enhanced erosion resistance and transpiration rates (0.93 g/ (kg·h)). These physiological advantages correlated with significantly higher soil enzyme activities (urease: 31.8-fold; catalase: 13.9-fold) [15], establishing H. leptothallum as a keystone species for enhanced soil stabilization and hydrologic regulation in degraded karst ecosystems. In addition, H. leptothallum crust is widely distributed and highly adaptable to different environments, making it a key species for ecological restoration of karst desertification [16–18]. However, the H. leptothallum crust grows slowly in its natural environment. Consequently, the artificial cultivation of H. leptothallum crust for ecological restoration in rocky desertification areas might offer a novel approach to rocky desertification management [19]. Current artificial moss crust cultivation methods primarily involve: (1) controlling environmental factors such as light, temperature, and moisture [19]; (2) optimizing nutrient concentrations, moss species, substrates, and inoculation techniques [20–22]; and (3) exogenous application of plant-growth-promoting bacteria (PGPB) to enhance crust region recovery [23–25]. Nevertheless, these methods are still time-consuming, require a lot of manual work, and do not produce satisfactory outcomes, underscoring the pressing requirement for quicker and more scalable techniques to enhance the restoration of large moss crusts.

The intricate interplay between plants and microorganisms plays a pivotal role in plant growth and stress mitigation through diverse biochemical mechanisms. Many studies have shown that microbial communities enhance plant resilience to environmental challenges through biofilm formation, extracellular polymer (EPS) production, nitrogen fixation, phytohormone synthesis, and modulation of ACC deaminase activity [26]. The symbiotic relationship has been utilized in biotechnological applications, where exogenous phytoproducing bacteria (e.g., Actinomyces bovis, Bacillus megaterium, Chlorella vulgaris, and Microcoleus vaginatus) have been employed to cultivate artificial moss crusts [23–25]. Previous studies have shown that the addition of demonstrating efficacy in enhancing artificial moss crust development through improved height, density, and coverage metrics in both laboratory and field settings [23–25]. However, the ecological adaptability and functional sustainability of exogenous microbial inoculants under extreme environmental conditions remain poorly understood.

The native moss-associated microbiota exhibits remarkable environmental stress tolerance and plays critical roles in ecosystem functioning. Particularly in moss crust systems, microbial extracellular enzymes of moss regulate organic carbon degradation and mineralization in moss crusts, and drive nutrient cycling processes fundamental to these fragile ecosystems [27]. While microbial community structures associated with bryophytes have been characterized across diverse habitats (polar regions, deserts, forests, grasslands, alpine meadows, and karstic rocky deserts) [4, 27, 28], significant knowledge gaps persist regarding the functional application of indigenous moss microbiota in artificial crust cultivation.

Recent findings reveal that microorganisms associated with karstic rocky desertification mosses (H. leptothallum and curly-leaved wetland mosses) exhibit host-specificity rather than geographic determinism [4]. This suggests an evolutionarily conserved symbiotic relationship between moss hosts and their microbiota, presenting novel opportunities for targeted microbial isolation and application. Despite this potential, the development of host-specific microbial consortia for ecological restoration remains largely unexplored, particularly in karst desertification remediation. This study aims to decipher host-specific moss-microbe symbiosis in karst ecosystems, establishing microbiome-driven restoration frameworks. These functional host-associated microorganisms will enhance crust resilience, demonstrating practical bioaugmentation strategies for desertification control while advancing understanding of microbial functional redundancy and niche partitioning in extreme environments.

Experimental procedures

Sampling data collection

To characterize the microbial communities via high-throughput sequencing, Hypnum leptothallum specimens and associated rhizospheric particles were collected from four severe karst rocky desertification regions in Guizhou Province, China: Wangmo County (Sanglang Township, Rongli Village; RL), Zhenfeng County (Geshang Township, Dajiang; ZF), Guanling County (Huajiang Township, Rengjiawan; HJ), and Zhenning County (Huangguoshu Township, Wengzhai Village; WZ). (Supplementary Tab. S1).

H. leptothallum specimens and associated rhizospheric particles were collected from three levels of karst rocky desertification regions in Guizhou Province, China: Zhenfeng County (Geshang Township, Dajiang; XY), Yanhe County (Miaoping Shang, Qinjian Village, Zhongzhai Town; YH), and Huaxi District (Shanshulin, Shibianjing Village; HX) for characterizing carbon source utilization using Biolog-ECO technology. All sampling was conducted in October. (Supplementary Tab. S1).

Sampling followed a standardized five-point protocol. At each site, five 1 m × 1 m quadrats were established (non-planar rock surfaces were sampled using horizontal projections). Three subsamples (~ 4 g each) per quadrat were combined to form one composite sample, yielding five composite samples per site (20 total) [4]. Samples were stored in sterile centrifuge tubes, transported to the laboratory on ice, and maintained at 4 °C before microbial community analysis [26]. Amplification controls (no-template controls) were incorporated in all sample batches to confirm the absence of contamination, ensuring that the sequenced amplicons derived from the sample DNA rather than laboratory reagents or environmental impurities.

Biolog-ECO metabolic profiling

Fresh H. leptothallum plants, collected in October, were cleaned of surface impurities and soil, then 2 g of the material was weighed and transferred into a sterile mortar with 5 mL of sterilized water for thorough grinding. After a 5-min incubation, the supernatant was collected to create the mixed microbial suspension of H. leptothallum-associated microorganisms. This suspension was serially diluted up to 10⁻³, and subsequently, 150 μL of the microbial suspension was pipetted into each detection well of the Biolog Ecoplates™ (ECO 1506, Biolog Inc., Hayward, CA, USA) [29]. Plates were incubated at 25 °C in the dark for 7 d. Absorbance (590 nm) was measured daily at room temperature using a full-wavelength microplate reader (model 1900991S, Thermo Scientific, Waltham, MA, USA) [30]. Average absorbance rate of change (AWCD), reflecting total microbial metabolic activity, was calculated as: AWCD = [∑(C_i − R)]/31, where C_i is the optical density (OD) of substrate-containing wells, R is the OD of control wells, and 31 is the total carbon source. The efficiency of the moss microorganisms in utilizing the different carbon sources is characterized by the AWCD [31].

Isolation of culturable microorganisms

To isolate culturable microorganisms, the plant material source matched that used for high-throughput sequencing. However, the H. leptothallum specimens from the four severe karst rocky desertification regions were combined and mixed to create a composite sample before microbial isolation. The bacterial and fungal isolates obtained were labeled with codes starting with ‘MHXX’ for bacteria and ‘MHXZ’ for fungi.

Epiphytic microbes. Sampling was conducted from October to April of the following year. The stem segments of H. leptothallum (2 g) were weighed and then added to a sterile conical flask with 98 mL of 1% Tween 80 solution [32]. The flask was set on an orbital shaker (model SPH-2102C, Shanghai, China) and shaken at 150 rpm/min for 60 min. Subsequently, a 15-min static incubation period was observed for inoculum preparation. The suspension was subsequently diluted 1000-fold and inoculated into six bacterial media (1/3 × Luria–Bertani (LB) medium, 1/2 × LB, LB, 1/3 × Tryptic Broth (TB) medium, 1/2 × TB, TB) and nine fungal media (1/3 × potato dextrose agar (PDA) medium, 1/2 × PDA, PDA, 1/3 × Sabouraud Dextrose Agar (SDA) medium, 1/2 × SDA, SDA, 1/3 × Czapek–Dox Agar (CDA) medium, 1/2 × CDA, CDA) for cultivation (Supplementary Figs. S4-5).

Endophytic microbes. The stem segments of H. leptothallum (0.5 g) were rinsed under a slow-flowing water stream for 12 h to remove soil. Following that, they were disinfected with 75% ethanol for 5 min and 5% sodium hypochlorite solution for another 5 min. After each disinfection, the segments were rinsed with sterile water 5–7 times. The final rinse water (100 µL) was plated onto 1/3 × LB medium and 1/2 × PDA to confirm the absence of microbial growth. The sterilized stem segments were ground and diluted with 10 mL of sterile water, then inoculated into 1/3 × LB medium and 1/2 × PDA medium for further culturing. Preparation of carbon source enrichment medium. The screened carbon sources were added to the base medium with varying concentration gradients (sugars, amines, amino acids, esters, alcohols: 1–5 g/L, acids: 0.1–0.5 g/L) based on the preference of H. leptothallum microorganisms for special carbon sources on Eiolog-ECO ecological plate, and then screened for the suitable concentration to be added. The isolation of H. leptothallum microorganisms was continued at the appropriate addition concentration, while the 1/3 × LB medium without the addition of special carbon sources was used as a control. After incubating at 28 °C for 3 d for bacteria and 25 °C for 7 d for fungi, colonies with varied morphology were chosen. These colonies were then purified by streaking, preserved in 25% glycerol (− 80 °C), and documented photographically.

Molecular identification and morphological characterization

Molecular identification. The bacterial strains were incubated in LB liquid basal medium (1/3 × Luria–Bertani (LB) medium) at 28 °C for 3 days, while the fungi were incubated in PDA solid basal medium (1/2 × potato dextrose agar (PDA) medium) at 25 °C for 7 days. Bacterial genomic DNA was extracted using the Bacterial Genomic DNA Extraction Kit (Solarbio, China), and fungal genomic DNA was extracted with the Fungal Genomic DNA Extraction Kit (Solarbio, China) according to the manufacturer’s guidelines. Bacterial primers used were 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1492R (5′-TACGGCTACCTTGTTACGACTT-3′) [33], while fungal ITS Primers included ITS1 (5′-TCCGTAGGTGGAACCTGCGG-3′) and ITS4 (5′-TCCTCCGCTTATTGATATGC-3′) [34, 35]. PCRs were performed in 25 or 50 μL volumes using Phata Taq Polymerase (Vazyme Biotech, China), with genomic DNA as the template. The reactions included an initial denaturation step at 95 °C for 3 min, followed by 35 cycles of denaturation at 95 °C for 15 s, annealing at Tm °C for 15 s, and extension at 72 °C for X seconds (set to 30 s per kb). A final elongation step was carried out at 72 °C for 5 min. PCR products were subsequently purified using the EasyPure Quick Gel Extraction Kit (Vazyme Biotech, China), following the manufacturer’s instructions. The amplification products were sent to the DynaPro (Beijing, China) company for sequencing. The sequences were sent to NCBI (https://www.ncbi.nlm.nih.gov) for Blast comparison against taxonomically classified sequences. To enhance our understanding of the phylogenetic diversity of epiphytic and endophytic microbes associated with H. leptothallum, a total of 88 bacterial and 135 fungal isolates were selected for identification using phylogenetic trees. Maximum likelihood phylogenetic analyses were performed in MEGA11 using the suggested partition parameters to identify the optimal tree topology. The bootstrap support values were calculated based on 500 search replicates and then compiled in TreeGraph2. The relative abundance of 16S rRNA or ITS variants in all sequences of the collection and in the corresponding sequences from metabarcoding analysis for each isolate was determined based on a 97% similarity in the 16S rRNA or ITS sequence.

Morphological identification. Bacterial colony morphology was assessed by streaking on solid LB medium and then incubating at 30 °C for 48 h. Characteristics such as size, shape, color, surface texture (smooth or rough), opacity, margin (entire or undulate), and elevation (flat or convex) were documented. Gram staining was conducted following standard procedures: heat-fixed smears were stained with crystal violet, treated with Gram’s iodine, decolorized with 95% ethanol, and counterstained with safranin. Cellular morphology, including cell shape, size, and arrangement (cocci, rods, singles, chains, or clusters), was observed under a light microscope using an oil-immersion objective (1000 × magnification).

Functional trait characterization of culturable strains

To assess the plant growth-promoting traits of culturable strains, we conducted functional characterization through four biochemical assays: (1) phosphate solubilization and protease production. The isolated strains were spot-inoculated on tricalcium phosphate (inorganic phosphorus) and skim milk agar (protease detection) media. Phosphate-solubilizing activity and proteolytic activity were determined by the formation of clear halo zones around colonies after incubation. The solubilization/proteolysis was quantified using the halo diameter to colony diameter ratio (D/d) [36] (Supplementary Tab. S2). (2) NH₃ production. Strains were cultured in peptone water medium at 28 °C for 72 h. Ammonium ion concentrations were semi-quantified through colorimetric analysis: yellow ( −), pale orange ( +), orange (+ +), and orange-red (+ + +) corresponded to no, weak, moderate, and strong ammonia production, respectively [36] (Supplementary Tab. S2). (3) Siderophore production. Siderophore synthesis was evaluated using the Chrome Azurol S (CAS) (Coolaber Co, China) assay. Cell-free supernatant was mixed 1:1 (v/v) with CAS detection solution. Absorbance at 630 nm was measured for test samples (AS) and blank controls (Ar). Siderophore production was calculated as: Siderophore unit (%) = [(Ar − As)/Ar] × 100 [36] (Supplementary Tab. S2). (4) auxin-like compounds production. auxin-like compounds synthesis was quantified using the Salkowski (H₂SO₄) colorimetric method. Strains were cultured in King’s B medium supplemented with 0.02% L-tryptophan. The fermentation broth was combined with Salkowski’s colorimetric solution in equal amounts and incubated in the dark for 30 min. The absorbance value of each strain’s fermentation broth was then measured at a wavelength of 530 nm, and the content of auxin-like compounds produced by the strains was calculated based on the standard curve of auxin-like compounds. A standard curve for auxin-like compounds was constructed to measure the auxin-like compounds content of the strains, which was then employed to assess their auxin-like compounds production capacity [37] (Supplementary Tab. S2). (5) Evaluation of drought tolerance of isolated strains. A 100 μL aliquot of a standardized bacterial suspension (OD600 = 1.0) was inoculated into a liquid culture medium supplemented with 10% polyethylene glycol 6000 (PEG6000), with uninoculated medium serving as the control. Following a 48-h incubation period at 28 °C, bacterial proliferation was assessed through OD600 measurement. Fungal plugs (4 mm diameter) were transferred to 1/2 PDA solid medium containing 10% PEG6000, using PEG-free medium as a control. After 5-day incubation at 25 °C, the colony diameters were measured to assess drought resistance. Both protocols employed PEG6000 to simulate osmotic stress equivalent to − 0.5 MPa water potential [38].

Promotional effect of isolated strains on the gametophyte growth of H. leptothallum

Six clumps of H. leptothallum group seedlings with uniform growth were selected and inoculated onto Knop medium plates, cultured for 7–8 days [39] (Supplementary Tab. S2). Contamination-free group seedlings exhibiting consistent growth (average plant height—2 mm) were re-selected as inoculation materials. Thirty-six bacterial strains and four fungal strains were separately inoculated into 1/3-strength LB and 1/2-strength PDB liquid media. Bacteria were incubated at 30 °C for 1–2 d, while fungi were cultured at 28 °C for 5–7 d. Bacterial suspensions were centrifuged and adjusted to 10⁹ CFU mL⁻¹ (OD600 = 0.2) [26]. Fungal colonies were fragmented into minimal uniform spheres. For inoculation, 2 μL of bacterial suspension or three fungal spheres were applied to the roots of each gametophyte tissue-cultured seedling cluster of H. leptothallum. Control groups received equivalent volumes of sterile water. All inoculated seedlings were maintained at 25 °C in a light incubator for 5–7 d. Plant height and fresh weight were evaluated to ascertain the strain-specific effects on growth promotion.

High-throughput sequencing

Microbial genomic DNA was extracted from H. leptothallum-associated communities using the Total DNA Extraction Kit (MP Biomedicals). The V5-V7 hypervariable regions of bacterial 16S rRNA genes were amplified with primers 799F (5′-AACMGGATTAGATACCCKG-3′) and 1193R (5′-ACGTCATCCCCACCTTCC -3′) [40], while fungal ITS regions were amplified using primers ITS1F (5′-CTTGTCATTTAGAGAAGTAA-3′) and ITS2R (5′-GCTGCGTTTCTTCATCGATGC-3′) [34, 35]. PCR reactions (20 μL total volume) contained: 2 μL DNA template, 2 μL each forward/reverse primer (5 μM), 4 μL 10 × TransStart FastPfu Buffer, 2 μL dNTPs (2.5 mM), 0.4 μL TransStart FastPfu Polymerase (TransGen Biotech), 8.6 μL ddH₂O. The reaction parameters were: 95 °C for 3 min; 25 × (95 °C for 30 s; 45 °C for 30 s; 72 °C for 45 s); and 72 °C for 10 min. 3 replicates were performed for each sample. Triplicate PCR products per sample were pooled, verified by 2% agarose gel electrophoresis, and quantified using the QuantiFluor™-ST system (Promega). Equimolar amplicon pools were prepared for Illumina paired-end sequencing (Novogene).

Bioinformatic and statistical analyses

All experiments included triplicate biological replicates, with data expressed as mean ± standard deviation (SD). Statistical significance (p ≤ 0.05) was determined using one-way ANOVA with Duncan’s post-hoc test in SPSS 28.0 (IBM).

Bioinformatic analysis of the microbiota of H. leptothallum was carried out using the Majorbio Cloud platform (https://cloud.majorbio.com). The sequencing data were processed for quality control, splicing. Then the optimized sequences were clustered into operational taxonomic units (OTUs) using UPARSE 7.1 [41, 42] with 97% sequence similarity level. The most abundant sequence for each OTU was selected as a representative sequence. To minimize the effects of sequencing depth on alpha and beta diversity measures, the number of 16S rRNA gene sequences from each sample was rarefied to 20,000, which still yielded an average Good’s coverage of 99.09%, respectively.

The taxonomy of each OTU representative sequence was analyzed by RDP Classifier version 2.2 [43] against the 16S rRNA gene database (Silva v138) using a confidence threshold of 0.7. The metagenomic function was predicted by PICRUSt2 (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States) [44] based on OTU representative sequences. PICRUSt2 is a software containing a series of tools as follows: HMMER was used to align OTU representative sequences with reference sequences. EPA-NG and Gappa were employed for the placement of OTU representative sequences within a reference tree. The castor was used to normalize the 16S gene copies. MinPath was utilized to forecast gene family profiles and pinpoint their locations within gene pathways, following the protocols of PICRUSt2.

The OTUs were annotated based on the distribution of species’ abundance to obtain pertinent species information. Venn diagrams were used to determine shared and distinct OTU data among various groups. Microbial community dissimilarity and similarity were analyzed using α and β diversity indices. Community α diversity was assessed using Chao and Shannon indices, while community β diversity was examined through non-metric multidimensional scaling analysis (NMDS) based on Bray–Curtis similarity distances. Multilevel discriminant analyses of species differences were performed with LEfSe to investigate the community structure variances [45]. The co-association network was analyzed using Network X, and microbial function was assessed using PICRUSt2 and FUNGuild [46, 47]. The microbial community co-occurrence in H. leptothallum crusts from different severe rocky desertification areas was analyzed using SparCC’s correlation coefficient [48]. Carbon source utilization preference, relative abundance, evaluation of growth-promoting traits, and high-throughput comparative analysis of data and visualization of H. leptothallum culturable microorganisms were conducted using the Origin Pro 2024 software package. Significance analysis of differences was carried out using SPSS 28.0 software. The results of the evaluation of the pro-life trait in culturable microorganisms were visualized using the Photoshop 2024 software package.

Results

Taxonomical composition of the host-associated microbiome in H. leptothallum crusts

To characterize the microbial community structure of H. leptothallum across varying severity levels of rocky desertification in Guizhou province, we analyzed high-throughput sequencing data from four geographic locations sampled in September 2022: Huajiang (Guanling County), Rongli village (Wangmo County), Dajiang village (Zhenfeng County), and Wengzai village (Zhenning County). After quality filtering and removal of chloroplast or mitochondrial sequences, operational taxonomic units (OTUs) were identified at 97% sequence similarity with a classification confidence threshold of 0.7. A total of 877,419 bacterial sequences (330,271,412 bases; average length: 376 bp) and 1,206,506 fungal sequences (296,836,982 bases; average length: 246 bp) were retained. Bacterial OTUs were taxonomically classified into 24 phyla, 54 classes,138 orders, 229 families, and 415 genera, while fungal OTUs were assigned to 12 phyla, 42 classes, 114 orders, 311 families, and 818 genera.

The predominant bacterial phyla in all samples were Pseudomonadota, Actinomycetota, Chloroflexota, Bacillota, Acidobacteriota, Bacteroidota, and Myxococcota, each with relative abundances exceeding 1%. Notably, Pseudomonadota (32.31%-47.19%) and Actinomycetota (33.32%-47.47%) were the most dominant, collectively constituting 65% of the total bacterial population. Other major bacterial phyla (> 1% abundance) included Bacteroidota (Fig. 1A, B). Fungal communities were predominantly composed of Ascomycota (72.91–86.69%) and Basidiomycota (7.28–18.70%), with Mortierellomycota and unclassified fungal taxa (unclassified_k__Fungi) also present at abundances > 1% (Fig. 1A, B).

Fig. 1.

Microbial taxonomic composition and OTU distribution in H. leptothallum crusts across the severe rocky desertification area. A, B: Phylum-level relative abundance of bacterial and fungal communities; C, D: Genus-level relative abundance of bacterial and fungal taxa; E, F: Venn diagram depicting shared and unique bacterial and fungal OTUs among four sampling sites: HJ (Huajiang), RL(Rongli), ZF (Zhenfeng), and WZ (Wengzhai)

At the genus level, 35 bacterial genera exhibited > 1% relative abundance. The top five taxa- Pseudonocardia (12.69–15.64%), Bacillus (3.14–6.38%), Sphingomonas, Microvirga, and norank_f_67-14 (12.69–15.64%) were collectively represented 25–30% of bacterial communities (Fig. 1C). Fungal communities comprised 53 genera (> 1% abundance), with Beauveria (0.25–23.55%), unclassified_p_Ascomycota (1.48–11.55%), Nigrospora, Trichoderma, and Mortierella being most prevalent. Notably, Punctulariopsis (4.88%) was exclusively detected at the Huajiang (HJ) site (Fig. 1D). Comparative analysis revealed greater homogeneity in bacterial community structure across sites compared to fungi, which displayed higher taxonomic diversity.

To further investigate the microbial composition of H. leptothallum in varying severity levels of rocky desertification areas, Venn diagrams at the OTU level were used to analyze microbial endemism. The results showed that 56.88% of bacterial OTUs (1,058 total) were shared among all four sites, with site-specific OTUs ranging from 1.40% (WZ) to 3.60% (RL) (Fig. 1E). In contrast, fungal communities exhibited stronger site specificity, with only 18.18% of OTUs (770 total) shared across all sites. Site-endemic fungal OTUs accounted for 9.54 –13.03% per location (Fig. 1F). The findings indicate a persistent presence of core bacterial taxa in severe rocky desertification areas, contrasting with the more variable nature of fungal communities in response to localized environmental conditions.

The diversity of bacterial and fungal taxa

We evaluated variability in the microbial community structure of H. leptothallum across different severe rocky desertification areas by analyzing α- and β-diversity among samples. At the OTU level, α-diversity indices (Chao and Shannon) showed no significant difference (P > 0.05) in bacterial and fungal communities among HJ, RL, WZ, and ZF samples (Fig. 2A–D). NMDS ordination revealed significant clustering of bacterial and fungal communities within sampling sites Fig. 2E, F). However, bacterial communities exhibited overlapping clusters at various sites, showing a lack of clear separation (ANOSIM: R = 0.155, P = 0.039; STRESS = 0.123). This indicates a structural similarity among the communities (Fig. 2E). Pairwise ANOSIM comparisons of bacterial communities identified significant differentiation only between WZ and ZF along NMDS1, while other inter-group differences were nonsignificant. In contrast, fungal communities displayed stronger site-specific clustering. Fungal groups, apart from HJ, RL, and WZ, which exhibited complete separation, did not display significant differentiation (ANOSIM: R = 0.3292, P = 0.001; STRESS = 0.185). Post-hoc analysis confirmed significant divergence among HJ, RL, and WZ along NMDS2, with no other inter-group difference detected (Fig. 2F). These results demonstrate that microbial communities share core compositional similarities across sites, with bacterial assemblages exhibiting greater structural homogeneity compared to fungi, which showed higher sensitivity to localized environments.

Fig. 2.

Alpha- and beta-diversity patterns of bacterial and fungal communities in H. leptothallum across the severe rocky desertification area. A–C: Bacterial diversity metrics: Chao1 index (A), Shannon index (B), and NMDS ordination plot (C); D–F: Fungal diversity metrics: Chao1 index (D), Shannon index (E), and NMDS ordination plot (F). Sampling sites: HJ (Huajiang), RL (Rongli Village, Wangmo County), ZF (Zhenfeng), and WZ (Wengzhai, Xingyi) represent four distinct rocky desertification areas

The compartment biomarker taxa and microbiome function prediction

To identify biomarker taxa specific to compartments and the ecological functions of microbial communities in H. leptothallum at four rocky desertification sites, we carried out LEfSe analysis (LDA > 2, P < 0.05) at different taxonomic levels (phylum, order, family, and genus). The significant biomarkers included: the bacterial phylum Actinomycetota, the order Solirubrobacterales, the class Thermoleophilia, family Micromonosporaceae, and genus norank_f__67-14; as well as the fungal order Hypocreales, class Sordariomycetes, family Trichosphaeriaceae, and genus Beauveria (Supplementary Fig. S1A–D). Each sample contained 3–10 unique bacterial taxa and 7–10 unique fungal taxa (LDA > 2; Supplementary Fig. S1A–D). Functional prediction using PICRUSt2 (bacteria) and FUNGuild (fungi) databases revealed 25 dominant bacterial functional groups (excluding unknowns), with high abundances in Energy production and conversion, Amino acid transport and metabolism, Transcription, Translation/ribosomal biogenesis, and Inorganic ion transport and metabolism, showing minimal inter-sample variation (Supplementary Fig. S2A). In contrast, fungal communities exhibited 17 functional groups dominated by pathogenic (Plant Pathogen, Animal Pathogen) and saprotrophic guilds (Litter Saprotroph, Soil Saprotroph, Undefined Saprotroph, Fungal Parasite), with minor contributions from Epiphyte and Endophyte guilds, displaying significant functional divergence across samples (Supplementary Fig. S1B).

Functional covariance networks in the microorganism of H. leptothallum

The correlation network analysis revealed that there were 380–400 nodes and 280–527 edges in the four sample bacterial networks. The positive influence rates ranged from 47.82 to 55%, with diameters of 10–14 edges. The mean degrees were between 1.474 and 2.703. Graph densities were measured at 0.04–0.007, while mean path lengths ranged from 4.327 to 6.015. The modularity (MD)was found to be 0.531–0.716, indicating a modular structure with MD > 0.5 (P < 0.05, R > 0.5; Fig. 3A; Supplementary Tab. S3). The fungal network consisted of 538–573 nodes and 330–666 edges with positive impact rates ranging from 51.9 to 56.16%. The mean degrees were between 1.227 and 2.593, diameters ranged from 11 to 18 edges, graph densities were 0.002–0.005, mean path lengths varied from 5.153 to 7.568, and modularities (MD) were between 0.596 and 0.786 (P < 0.05, R > 0.5; Fig. 3B; Supplementary Tab. S3). The primary microbiota observed in the four samples were the phyla Pseudomonadota, Actinomycetota (bacteria), and Ascomycota (fungi), showing the highest level of connectivity in the results. While there were slightly different positive and negative correlations in various samples, the overall trend showed more positive connections than negative ones. This indicates that microorganisms in the crusted ecosystem of H. leptothallum in a severe rocky desertification area were highly cooperative.

Fig. 3.

Microbial correlation networks in H. leptothallum crusts from geographically distinct severe rocky desertification areas. A Bacterial networks for HJ (Huajiang), RL (Rongli Village), WZ (Wengzhai), and ZF (Zhenfeng) samples; B Fungal networks. Node size corresponds to node degree (connectivity), with red edges indicating positive correlations (Spearman’s ρ > 0.5) and blue edges indicating negative correlations (ρ <  − 0.5; P < 0.05). Node colors represent taxonomic phyla (see legend). The number of nodes and edges in each network is indicated below the corresponding figure

Metabolic carbon sources utilization by H. leptothallum-associated microorganisms

The metabolic profiles of H. leptothallum-related microorganisms were analyzed to investigate their preferences for carbon source utilization using Biolog-ECO technology. The average well color development (AWCD) values of the moss-associated microbial communities showed continuous increases from 1–6 days of incubation. After 6 days, the rate of AWCD increase slowed significantly while carbon source utilization efficiency peaked (Fig. 4A). However, no significant differences in AWCD values were observed among the three moss microbial communities (P > 0.05; Fig. 4B), indicating comparable carbon source utilization capacities across communities. Further analysis of six carbon source categories using 6-day AWCD values revealed a significant difference between XYMH and both YHMH and HXMH in the utilization of carboxylic acids and amine (P < 0.05; Fig. 4B). Detailed examination of 31 individual carbon sources demonstrated that XYMH exhibited significantly higher utilization than YHMH and HXMH for seven specific substrates: α-D-lactose, 2-hydroxybenzoic acid, α-butyronic acid, D-malic acid, L-threonine, L-α-glycerol, and N-acetyl-D-glucosamine (P < 0.05; Supplementary Fig. S3A–F; Supplementary Tab. S4). These results identify these seven compounds as signature carbon sources preferentially utilized by microorganisms associated with H. leptothallum in severe rocky desertification environments.

Fig. 4.

Carbon source utilization profiles of H. leptothallum crust microbiomes and their effect on the separation rate of culturable bacteria. A Line graph of the temporal dynamics of average well color development (AWCD) over time. B Histogram of carbon source utilization efficiency. C Utilisation patterns of six carbon sources categories: I(sugars), II( acids), III( amino acids), IV( esters), V( alcohols), VI( amines). Growth response of Bryobacter sp. strains on gradient-supplemented media. D-F: Bacterial colonies cultured on α-D-lactose, D, L-α-glycerophosphate, and L-threonine amended mineral agar plates. Control groups with basal medium containing equivalent carbon/nitrogen loads through alternative compounds. All media were prepared by filter-sterilization (0.22 μm) and incubated at 28 °C for 3 d. (XYMH: H. leptothallum from Xingyi county; YHMH: H. leptothallum from Yanhe county, HXWH: H. leptothallum from Huaxi county)

The culturable microorganisms of H. leptothallum

We employed multiple basal media combined with specific formulations to isolate host-associated microorganisms from H. leptothallum. The results showed that bacterial isolation in 1/3 LB, 1/2 LB, LB, 1/3 TB, 1/2 TB, and TB broth was 23%, 20%, 15%, 17%, 19% and 16%, respectively (Supplementary Fig. S4). For fungal isolation, the efficiencies in 1/3 PDA, 1/2 PDA, PDA, 1/3 SDA, 1/2 SDA, SDA, 1/3 CDA, 1/2 CDA and CDA were: 14%, 16%, 13%, 9%, 9%, 6%, 10%, 12%, and 11%, respectively (Supplementary Fig. S5). LB and PDA were found to be the superior media for cultivating host-associated microorganisms of H. leptothallum during screening, with 1/3 LB and 1/2 PDA showing optimal performance for the bacteria and fungi, respectively.

Specific isolation media were created by analyzing the preference of carbon source utilization of H. leptothallum microorganisms. The diameters and densities of bacterial colonies were larger when grown in a specific medium supplemented with 2 g/L α-D-lactose, N-acetylglucosamine, L-threonine, and L-α-glycerol, as well as 0.2 g/L 2-hydroxybenzoic acid and D-malic acid, and 0.1 g/L α-butyric acid, compared to the control (1/3 LB) group and other additive concentrations. There was an increase in the number of bacteria in these media, but there was no significant increase in the number of bacterial species. The results demonstrated that the introduction of a particular carbon source improved the growth rate of cultivable bacteria in H. leptothallum; however, it did not induce an increase in the diversity of cultivable bacteria (Fig. 4D–F).

The taxonomic classification of these culturable microorganisms is further expanded. Molecular identification of 88 bacterial (including 6 endosymbionts) and 135 fungal isolates (including 4 endosymbionts) through 16S rRNA gene sequencing revealed taxonomic diversity at the family and genus levels. Culturable host-associated bacteria comprised 61 species spanning 23 genera within 16 families, while host-associated fungi included 90 species distributed across 50 genera from 31 families (Fig. 5D, E; Supplementary Tab. S5, S6). Among culturable bacteria, the top four families were Lysobacteraceae, Microbacteriaceae, Pseudomonadaceae, and Bacillaceae, with a relative abundance of over 7%. Eight bacterial families exhibited relative abundance exceeding 3.6%, with Bacillaceae (30.9%) and Pseudomonadaceae (21.8%) dominating the community (Fig. 5A; Supplementary Tab. S5). For fungi, the five most abundant families (relative abundance > 7%) were Nectriaceae, Sporocadaceae, Phaeosphaeriaceae, Cladosporiaceae, and Didymellaceae. A total of 12 families of fungi had a relative abundance greater than 2.4%, with Cladosporiaceae (11.9%) and Didymellaceae (14.3%) being the primary families of culturable fungi (Fig. 5B; Supplementary Tab. S5, S6).

Fig. 5.

Taxonomic composition of culturable microbial communities in H. leptothallum Crusts. A, C: Relative abundance of culturable bacterial families and genera. B, D: Relative abundance of culturable fungal families and genera

At the genus level, Arthrobacter, Exiguobacterium, Stenotrophomonas, Pseudomonas, and Bacillus represented the top five bacterial genera (relative abundance > 3%). Among the 11 genera of bacteria with relative abundance greater than 3%, Bacillus (27%) and Pseudomonas (19%) stood out as the dominant genera of culturable bacteria (Fig. 5C; Supplementary Fig. S6; Supplementary Tab. S6). Trametes, Fusarium, Penicillium, Didymella, and Cladosporium were the top five genera in abundance of culturable fungi. The dominant genera among culturable fungi were Didymella (7.4%) and Cladosporium (12.3%), with fourteen fungal genera having a relative abundance greater than 4% (Fig. 5D; Supplementary, Fig. S7; Supplementary Tab. S7).

Screening of potential functional strains

Our comparative analysis of high-throughput sequencing data and culturable isolates revealed conserved microbial taxa at the genus level. Bacterial communities across samples contained 14, 13, 13, and 12 core genera (52%-61% of culturable bacteria), predominantly from the Actinobacteria, Bacillota, and Pseudomonadota phylum. Key bacterial taxa included Bacillus, Pseudomonas, Arthrobacter, Streptomyces, Rhodococcus, Microbacterium, and Agromyces (Fig. 6A, C; Supplementary Tab. S7). The total number of fungi genera was 36, 35, 37, and 37, accounting for 70%-74% of the culturable fungi. These core microorganisms included the genera Arxiella, Paramyrothecium, Phaeosphaeria, Phoma, Purpureocillium, Pestalotiopsis, Fusarium, Penicillium, Didymella, and Cladosporium, all belonging to the phylum Ascomycota (Fig. 6B, D; Supplementary Tab. S8). It was observed that shared microorganisms were prevalent across the different samples and culturable microorganisms, with fungi exhibiting a greater number of shared microorganisms than bacteria. The core potential functional taxa of microorganisms of H. leptothallum were the phyla Actinobacteria, Bacillota, Pseudomonadota, and Ascomycota.

Fig. 6.

Comparative analysis of high-throughput sequencing data and culturable isolates. A, B Representative bacterial and fungal colony morphologies screened by both high-throughput and culturable methods; C, D Venn diagrams comparing genus-level overlaps between culturable microorganisms and high-throughput sequencing data from four severe rocky desertification areas, C: bacterial community overlaps (blue: culturable strains) and D: fungal community overlaps (red: sequencing data). Shared and unique taxa illustrate core versus site-specific culturable microbiota. The strain codes ‘MHXX’ and ‘MHXZ’ correspond to bacterial and fungal isolates, respectively, derived from pooled H. leptothallum samples collected from the four severe desertification sites

To identify the functional strains of H. leptothallum, we compared the microbial nodes and culturable microorganisms within the top 50 genera. The core microbial nodes in the four samples (HJ, RL, ZF, and WZ) shared 4, 2, 3, and 3 microorganisms with the bacteria, accounting for 8.7%-17.4% of the culturable bacteria, respectively (Fig. 7A; Supplementary Tab. S9). These core microorganisms, including Sphingomonas, Bacillus, Microbacterium, and Pseudomonas, belong to the Actinobacteria, the Thic-walled Bacteria, and the Ascomycetes taxa. There were two microorganisms (Bacillus and Microbacterium) common to the core microorganisms and culturable bacteria of the four samples. For fungi, the core nodes of the four samples (HJ, RL, ZF, and WZ) had 9, 7, 10, and 6 genera in common, which accounted for 12–18% of the culturable fungi, respectively (Fig. 7B; Supplementary Tab. S10). The core fungi of H. leptothallum shared five common microorganisms: Fusarium, Dendrosporium, Purpureocillium, Pseudopithomyces, and Subcompartmental, all of which belonged to the Ascomycota taxa.

Fig. 7.

Comparative Venn diagrams of cultivable microorganisms versus core microbiota (top 50 genera) in four severe rocky desertification regions at the genus level. A Bacterial community overlap; B Fungal community overlap. HJ, RL, WZ, ZF: high-throughput sequencing data. CB/CF: Cultivable bacterial and fungal

Assessment of plant growth-promoting characteristics of culturable microorganisms

To investigate the probiotic function of culturable microbial strains, we initially examined phosphate-solubilizing and protease-producing abilities. Subsequently, we reassessed the NH₃-, auxin-like compounds- and iron-carrier-producing capacities of the strains with positive results in the initial evaluation. Phosphate solubilisation and protease production capacity were evaluated in 88 bacterial strains and 135 fungal strains isolated. The results indicated that 6.8% of the bacteria and 15.6% of fungi exhibited organic phosphorus solubilization capabilities. The ratio of D (hyaline circle diameter) to d (colony diameter) ranged from 1.06 to 2.17 for bacteria, while D/d ranged from 1.12 to 3.27 for fungi. The proportion of bacteria able to solubilize inorganic phosphorus was 19.3%, with a D/d range of 1.1–2.69. Over 90% of the bacteria had a D/d between 1.0 and 2.0, while the fungal strains showed little to no capability in solubilizing inorganic phosphorus. The proportion of protease-producing bacterial strains and fungi was 53.4% and 40%, respectively. Bacterial protease-producing D/ d ranged from 1.11 to 6.10, with about 90% falling between 1.0 and 5.0 (Fig. 8A; Supplementary Fig. S10; Supplementary Tab. S11). Fungal D/d values for protease production varied from 1.04 to 4.77, with approximately 90% falling between 1.0 and 4.0 (Fig. 9A; Supplementary Fig. S11; Supplementary Tab. S12). The bacterial strains demonstrated a weaker ability to solubilize organic phosphorus, yet a stronger ability to solubilize inorganic phosphorus and a greater capacity for protease production compared to the fungal strains.

Fig. 8.

Functional characterization of 55 bacterial strains isolated from H. leptothallum crusts. A–D: Quantitative assays for protease production (A), NH₃ production (B), Iron carrier production (C), and auxin-like compounds production (D). CK1: negative control; CK2: positive control; the number on the bottom of the tube indicates the strain number. E: Radar map of six high-performance strains, (a–f) evaluating organophosphorus/organic phosphorus solubilization, protease activity, NH₃ production, siderophore synthesis, and auxin-like compounds production. The strain codes ‘MHXX’ and ‘MHXZ’ correspond to bacterial and fungal isolates, respectively, derived from pooled H. leptothallum samples collected from the four severe desertification sites

Fig. 9.

Functional characterization of 69 fungal strains isolated from H. leptothallum crusts. A–D: Quantitative assays for protease production (A), NH₃ production (B), Iron carrier production (C), and auxin-like compounds production (D). CK1: negative control; CK2: positive control; the number on the bottom of the tube indicates the strain number. E: Radar map of six high-performance strains, (a–f) evaluating organophosphorus/organic phosphorus solubilization, protease activity, NH₃ production, siderophore synthesis, and auxin-like compounds production. The strain codes ‘MHXX’ and ‘MHXZ’ correspond to bacterial and fungal isolates, respectively, derived from pooled H. leptothallum samples collected from the four severe desertification sites

Following the initial assessment, 55 bacterial strains and 69 fungal strains were reviewed again to determine their capability in producing NH₃, auxin-like compounds, and iron carriers. The results indicated that bacteria had a 51% abundance of strains with strong NH₃-producing capacity (Fig. 8B; Supplementary Fig. S10; Supplementary Tab. S11), while fungi had less than 6% abundance. Among fungi, around 4% were unable to produce NH₃, but the majority had weak or strong NH₃-producing capacity, with an abundance exceeding 43% (Fig. 9B; Supplementary Fig. S11; Supplementary Tab. S12). The relative abundance of Fe-producing carriers in bacteria and fungi varied from 0 to 92.44%. Bacteria had a relative abundance of 76.4% capable of producing Fe-carriers, while fungi had a relative abundance of 88.4%. Among bacteria, 3.6% produced Fe-carriers with a relative abundance of 50% or more, compared to fungi, where more than 24.6% produced Fe-carriers (Figs. 8C, 9C; Supplementary Figs. S10, S11; Supplementary Tabs. S11, S12). Absorbance measurements were conducted for auxin-like compounds at concentrations of 0.5 mg/L, 5 mg/L, 10 mg/L, 15 mg/L, and 20.5 mg/L, resulting in the generation of a standard curve. The standard curve formula is Y = 18.96X + 0.0285, with an R² = 0.9929, where Y represents absorbance and X represents the concentration of auxin-like compounds in g/L. Approximately 40% of the bacterial strains were able to produce auxin-like compounds at a concentration of 10 mg/L or higher, whereas only 1.4% of the fungal strains exhibited the same capability (Figs. 8D, 9D; Supplementary Figs. S10, S11; Supplementary Tabs. S11, S12). The findings indicated that bacterial strains exhibited higher ability in NH₃ and auxin-like compounds production but lower ability in iron carrier production compared to fungal strains. The bacterial genera Pseudomonas and Arthrobacter, as well as fungal genera such as Ascochyta, Clonostachys, Penicillium, and Cladosporium, exhibit superior plant growth-promoting traits (Figs. 8E, 9E).

Drought resistance of culturable microorganisms of H. leptothallum

Among the 55 bacterial strains tested for drought tolerance, only 16 strains (MHXX1, MHXX7, MHXX17, MHXX19, etc.) exhibited no significant difference in absorbance in 10% PEG from the control (0% PEG), demonstrating their drought tolerance. The 39 bacterial strains left showed significantly lower absorbance in 10% PEG compared to the control (P ≤ 0.05) and were not drought-tolerant (Fig. 10A; Supplementary Tab. S13). Among the 69 fungal strains screened for drought tolerance, only 19 (MHXZ7, MHXZ88, MHXZ105, etc.) were found to be tolerant, with no significant difference in colony diameter in 10% PEG compared to the control (0% PEG). The remaining 50 fungal strains had colony diameters significantly lower in 10% PEG than the control (P ≤ 0.05) and were not drought-tolerant (Fig. 10B, C; Supplementary Tab. S14). The results indicated that 29% of the bacterial strains and 27.5% of the fungal strains exhibited drought tolerance.

Fig. 10.

Screening of drought resistance of culturable bacterial strains. A and B, the results of differential evaluation of 10 strains of bacteria and fungi, respectively; C, the schematic diagram of the drought resistance experiment of fungi. The strain codes ‘MHXX’ and ‘MHXZ’ correspond to bacterial and fungal isolates, respectively, derived from pooled H. leptothallum samples collected from the four severe desertification sites

Culturable microorganisms stimulate gametophyte growth in H. leptothallum co-cultures

To identify the most effective microbial partners for gametophyte growth stimulation, we selected 16 strains from a library of 55 bacteria and 69 fungi. The strains were chosen through a comprehensive functional screening process, prioritizing those demonstrating a combination of key plant growth-promoting traits, including auxin biosynthesis, phosphate solubilization, siderophore production, and drought tolerance. Since fungal isolates showed significantly weaker performance in these assays, all 16 finalists were bacterial. Statistically significant growth enhancements were noted in plant height and fresh weight of the root-inoculated gametophyte of H. leptothallum after 7 days of co-culture with these beneficial strains. Eleven strains, closely related phylogenetically to species such as Stenotrophomonas bentonitica, Agromyces terreus, Streptomyces pratensis, S. rhizophila, Curtobacterium oceanosedimentum, Paenarthrobacter nitroguajacolicus, Bacillus safensis, Pseudomonas fluorescens, Peribacillus muralis, and Rhodococcus qingshengii (specifically, MHXX12, MHXX22, MHXX30, MHXX44, MHXX58, MHXX62, MHXX64, MHXX68, MHXX80, MHXX3, MHXX35), demonstrated a significant increase in plant height (p < 0.05). Notably, strain MHXX9, closely related to Rhodococcus erythropolis, uniquely enhanced the fresh weight and plant height of moss group-cultivated seedlings (Fig. 11A–C; Supplementary Figs. S8, S9; Supplementary Tab. S15).

Fig. 11.

Promotion effects of functional strains on H. leptothallum gametophyte growth. A Schematic diagram of strain inoculation on tissue-cultured seedlings; B Microscopic visualization of root colonization, MHXX68 and others denote the strain number; C Statistical analysis of plant height post-inoculation (7 days). Different letters indicated a significant difference compared to the control group. The strain codes ‘MHXX’ and ‘MHXZ’ correspond to bacterial and fungal isolates, respectively, derived from pooled H. leptothallum samples collected from the four severe desertification sites.

Discussion

Microbial community structure of H. leptothallum crusts in Severe karst rocky desertification areas

H. leptothallum is coated with various host-associated microorganisms, which assist in its ability to withstand the challenging conditions of the karstic rocky desertification area by offering protective effects and aiding its defense mechanisms. Our findings provide valuable insights into the interaction between H. leptothallum and its microorganisms. The bacterial and fungal compositions of the samples showed a high degree of similarity, indicating that H. leptothallum in the severe karst desertification area hosts a specialized host-associated microbiota. The bacterial communities consisted primarily of Pseudomonadota, Actinobacteria, Bacillota, Acidobacteria, and Chloroflexota at the phylum level, in line with the core phyla identified in moss-associated microbiomes [4, 49]. Pseudomonadota, particularly those enriched with nitrogen-fixing taxa, play a key role in organic decomposition, nutrient cycling [4, 50, 51], and the secretion of extracellular polysaccharides for sand stabilization and the formation of biological crust [52]. Furthermore, the Chloroflexota phylum is composed of oligotrophic bacteria that can produce energy through solar radiation and 3-hydroxypropionate double cycling [51]. The Actinomycetota and Acidobacteriota phyla enhance host nutrition through complex organic carbon degradation [51, 53].

The fungal communities were primarily composed of Ascomycota, Basidiomycota, Mortierellomycota, unclassified_k__Fungi, and Chytridiomycota. Consistent with previous findings, the Ascomycota and the Basidiomycota were identified as the most prevalent phyla in numerous mosses [4, 54]. The Ascomycota phylum degrades cellulose to decompose plant apoplasts [55], the Basidiomycota phylum aids plants in nutrient acquisition [37, 55], and the Mortierellomycota phylum enhances soil fertility by increasing soluble carbon levels and promoting effective phosphorus and phosphorus-related enzyme activities [56, 57]. Moreover, the Pseudomonadota, Actinobacteria, and Ascomycota phyla play a crucial role as fundamental microorganisms in the bacterial-fungal network of the four samples. They exhibit more positive associations than negative ones in the network and display a modular structure consistent with previous findings [58]. It indicates a strong level of collaboration among H. leptothallum microorganisms, potentially boosting the stability and resilience of the crust ecosystem. Overall, H. leptothallum microorganisms may aid in host adaptation to arid rocky desertification habitats and contribute to biological crust formation through various metabolic processes and ecological functions.

β-Diversity analyses revealed that the local environmental conditions strongly influence the composition and spatial distribution of H. leptothallum-associated microbial communities in analogous ecosystems. Functional predictions for bacterial and fungal communities further corroborate their critical role in host environmental adaptation. Bacterial functional profiles displayed remarkable consistency across samples, with energy production/conversion and amino acid transport/metabolism showing a notably high abundance, consistent with prior research [59]. These metabolic priorities likely enable bacteria to support H. leptothallum growth by facilitating carbon and nitrogen acquisition and organic matter decomposition. Fungal communities were dominated by Saprotrophs and Pathotrophs, with minor contributions from epiphytes and endophytes, consistent with reports of moss-associated fungi [4]. This functional partitioning implies dual nutritional strategies: saprotrophic degradation of dead host tissues and apoplastic substrates, coupled with endophyte-mediated nutrient exchange, collectively enhancing host growth.

Core microorganisms of H. leptothallum in severely desertified regions

Through the analysis of microbial carbon source utilization preferences and the development of culture media enriched with these specific substrates, microbial proliferation was significantly enhanced. This aligns with the known patterns of moss-associated microbiomes, which tend to favor sugars, carboxylic acids, and amino acids as their main carbon sources [60, 61]. Mechanistically, α-D-lactose (a disaccharide) and metabolic intermediates 2-hydroxybenzoic acid, α-butyric acid, and D-malic acid likely fuel central metabolism, while N-acetyl-D-glucosamine (a structural polysaccharide precursor) and L-threonine (an essential amino acid) may contribute to cell wall synthesis and protein biosynthesis. These carbon sources collectively support microbial growth by energizing metabolic pathways, modulating biosynthetic activities, and potentially suppressing competitor microbiota through niche-specific resource partitioning.

In the culturable data, Bacillus and Pseudomonas were identified as the dominant bacterial genera, with Didymella and Cladosporium being the predominant fungal genera. These findings align with studies on forest rhizosphere soils [62] and Eucommia ulmoides phyllosphere fungi [63], but differ from mangrove soils dominated by Penicillium and Trichoderma [64], highlighting the species-specific microbial composition of B. argenteum. Integrated high-throughput and culturable analyses identified Bacillus, Microbacterium, Fusarium, Cladosporium, Purpureocillium, Pseudopithomyces, and Didymella as core microbial taxa. Cladosporium, consistently identified as a core genus across studies [4], likely plays a critical role in moss microecosystems due to its broad environmental tolerance (e.g., salinity, osmotolerance) and diverse secondary metabolites [65]. Bacillus, with its genetic and metabolic diversity, contributes to nutrient cycling, stress resistance, and plant growth promotion through phosphate solubilization, nitrogen fixation, and production of auxin-like compounds, siderophores, and antimicrobial compounds [62, 66]. For instance, Bacillus amyloliquefaciens alleviates salt stress in cotton [67], while B. velezensis suppresses root rot in Lycium barbarum [68]. Microbacterium foliorum enhances arsenic uptake in Melastoma malabathricum [69], and Fusarium and Cladosporium sp. produce bioactive secondary metabolites (e.g., chromanones, naphthoquinone dimers) with biocontrol potential [70, 71]. The core taxa identified may aid in the adaptation of H. leptothallum to karst rocky desertification by engaging in phosphate solubilization, producing auxin-like compounds, and synthesizing secondary metabolites, highlighting their potential in the cultivation of artificial crusts.

Plant growth-promoting traits and potential functional strains

The evaluation of six plant growth-promoting (PGP) traits among culturable microorganisms revealed distinct functional patterns between bacterial and fungal strains. Bacterial strains exhibited stronger capabilities in inorganic phosphate solubilization, protease production, ammonia (NH₃) release, and auxin-like compounds biosynthesis, whereas fungal strains showed superior organic phosphate solubilization and siderophore production [72–74]. Substantial research has documented the roles of these bacterial strains in alleviating abiotic stresses and enhancing plant growth through secondary metabolites. Prior research has shown that Pseudomonas sp. (e.g., P. brassicacearum, P. polyglossia, P. antarctica) enhance plant growth by fixing nitrogen, solubilizing phosphate, producing IAA, and synthesizing siderophore [75–77], which is consistent with our results. P. putida MTCC5279 has been shown to mitigate drought stress in chickpea [75] and salt stress in Arabidopsis under phosphorus limitation [76]. Meanwhile, P. fluorescens REN1 enhances root elongation in rice [77]. Multiple strains of P. oryzihabitans exhibit stress-alleviating properties: Ep4 improves root growth in cadmium stress [78], NYCS1-5 enhances maize yield in salinity [79], Antg-12 reduces melon disease incidence while increasing fruit yield [80], and LP1 shows ACC deaminase activity to relieve abiotic stress [81, 82]. Arthrobacter globiformis A152 produces hyaluronidase, an important enzyme for plant–microbe interactions [83].

For fungi, there are limited but significant studies on secondary metabolites. Clonostachys rogersoniana, an endophyte of Selaginella delicatula, synthesizes alkaloids, phenolic acids, fatty acids, sterols, and polyketides like TMC154 [84, 85]. Penicillium expansum GY618 produces quinones, triterpenoids, and steroidal compounds [86], with most prior reports emphasizing its phytopathogenic role [87]. Cladosporium tenuissimum MT-35, an endophyte of Huperzia serrata, yields nitrogen-containing compounds, sterols, and polyketides with acetylcholinesterase inhibition and anti-inflammatory activities [88]. However, there is a lack of research on Ascochyta manawaorae, Phaeosphaeriaceae sp. (SRC1lsM3a), and Phaeosphaeria podocarpi. The potential use of these 12 microbial strains as bryophyte growth promoters to improve moss crust development remains largely uninvestigated. The strategic use of microbial resources could facilitate the extensive growth and cultivation of moss crusts in degraded ecosystems, offering potential for ecological restoration.

Interaction between microbial strains and H. leptothallum tissue

The co-cultivation experiments demonstrated a significant increase in the shoot height of tissue-cultured H. leptothallum seedlings upon inoculation with bacterial strains closely related to Stenotrophomonas bentonitica, Agromyces terreus, Streptomyces pratensis, Stenotrophomonas rhizophila, Curtobacterium oceanosedimentum, Paenarthrobacter nitroguajacolicus, Staphylococcus hominis, Pseudomonas fluorescens, Pseudomonas parafulva, Peribacillus muralis, Rhodococcus erythropolis, Rhodococcus qingshengii, and Brevundimonas bullata.

Extensive studies have documented the growth-promoting and biocontrol functions of these strains. Stenotrophomonas rhizophila DSM14405T is resistant to chromium and can reduce it [89], while also acting as a biocontrol agent against mango anthracnose caused by Colletotrichum gloeosporioides [90]. Curtobacterium oceanosedimentum DG-20 promotes plant growth by producing ammonia (51–73 μg/mL), indole-3-acetic acid (116–183 μg/mL), siderophore, and phosphate solubilization, and can alleviate cadmium stress in pepper plants [91]. Streptomyces pratensis S10 enhances wheat growth by producing siderophores, 1-aminocyclopropane-1-carboxylate (ACC) deaminase, auxin-like compounds, and through phosphate solubilization and nitrogen fixation, while also suppressing Fusarium graminearum hyphal invasion and toxin biosynthesis [92, 93]. Paenarthrobacter nitroguajacolicus promotes pistachio seedling growth by improving chlorophyll, carotenoid, phenolic, anthocyanin, and proline levels [94]. Pseudomonas fluorescens inoculation enhances blueberry biomass and root development [95], while Bacillus velezensis increases rice yield by suppressing Fusarium graminearum [96]. Rhodococcus erythropolis facilitates chromium tolerance in Agropyron cristatum [97]. The rapid shoot elongation observed in H. leptothallum seedlings is likely supported by these functional traits.

In contrast, the ecological roles of bacterial strains that are most closely related to Stenotrophomonas bentonitica, Agromyces terreus, Streptomyces pratensis, Staphylococcus hominis, Pseudomonas parafulva, and Rhodococcus qingshengii have not been extensively investigated. This highlights the necessity for further research to elucidate their potential as plant growth-promoting rhizobacteria (PGPR). Interestingly, the strains identified in this study as promoting H. leptothallum growth were notably absent from previous screenings of 12 candidate functional microbes. Seedlings inoculated with these strains exhibited rapid microbial colonization followed by moss mortality, possibly due to an excess of nutrients in the culture medium, which could lead to pathogenic behavior in otherwise beneficial microbes. This conflicting outcome highlights the importance of optimizing inoculation protocols and nutrient conditions in artificial crust cultivation. Further investigations are necessary to clarify the dual roles of these strains as both growth promoters and potential stressors in H. leptothallum microecosystems.

The limited growth promotion seen in fungal isolates in our 7-day co-culture assay contrasts with the significant effects caused by bacteria. This difference is mainly due to the varying growth rates and the longer time needed for fungi to establish a functional interaction with their host. Bacterial strains like Pseudomonas and Rhodococcus are quick colonizers that can release growth-stimulating metabolites (such as auxins and siderophores) immediately, resulting in observable plant responses within days. In contrast, the establishment of a functional fungus-moss interaction is a slow process that takes a longer timeframe to show measurable growth effects. Previous evidence indicates that a significantly extended co-cultivation period, ranging from four to five weeks to as long as 72 days, is required before any observable change—either unchanged or slightly increased—in moss growth rates becomes apparent [98, 99]. Even a fungus associated with auxin-related gene expression did not immediately stimulate growth. The study revealed that some fungi displayed alterations in transcription patterns as they shifted from endophytism to saprotrophism, a process influenced by the vitality of the host, thus emphasizing the intricate and gradual establishment of their ecological role [97]. The process of recognition, colonization, and potential trophic mode switching cannot be captured in a short-term experiment.

According to this precedent, our 7-day experimental timeframe was likely too short for the fungi to progress past the initial colonization phase and establish a stable, beneficial symbiotic interaction necessary to produce a measurable growth advantage for H. leptothallum. Consequently, the absence of a noticeable fungal influence in our test does not suggest an absence of functional potential, but rather underscores a critical temporal mismatch. This highlights the importance of experimental designs that take into account the slower colonization and interaction dynamics of fungi when studying their contributions to plant growth, especially in the case of slow-growing bryophytes. Furthermore, our study primarily investigates the potential benefits of the moss-associated microbiota for the host. Further research is required to comprehensively elucidate the impact of H. leptothallum on its microbial community through root exudates, surface properties, or other physiological mechanisms to establish a symbiotic relationship. This paves the way for promising avenues of investigation in upcoming studies.

Conclusions

This study presents a comprehensive method for profiling in situ microbial communities and isolating validated strains for bioremediation of karst rocky desertification. We found that H. leptothallum, a key moss species in this ecosystem, harbors a consistent core microbiota across degraded sites. By combining high-throughput sequencing with carbon substrate-guided cultivation, we successfully bridged the gap between community analysis and culture collection. A synergistic interaction between bacterial and fungal symbionts was revealed through functional screening, leading to the identification of superior strains. These strains include drought-resistant bacteria and growth-promoting organisms closely related to Rhodococcus erythropolis. This approach illustrates the efficacy of employing a function-driven strategy for microbial resource identification. The resulting strain library and microbial functional profiles offer a solid foundation and inoculants for enhancing artificial biological crusts in karst rocky desertification control.

Author contributions

Shasha Wu: conceptualization, writing – original draft, methodology, validation, visualization, writing – review and editing, formal analysis, data curation, investigation. Yihong Xiang: formal analysis, data curation, resources, investigation. Xiaofan Chen: investigation, formal analysis, data curation, resources. Yueyan Lu, Yubi Li, Zheyu He, Qianya Zuo: investigation, formal analysis, data curation, resources. Wei Cao: conceptualization, investigation, funding acquisition, writing – original draft, writing review and editing, project administration, supervision. Jiaojiao Qu: conceptualization, investigation, funding acquisition, writing original draft, writing review and editing, project administration, supervision, data curation, validation.

Data availability

The data that support the findings of this study are available in the Supporting Information of this article. The authors confirm that all data underlying the findings are fully available without restriction. All sequencing data (i.e. generated either by the Sanger sequencing or by Illumina Miseq technology) are deposited at the NCBI database under the following accession numbers: PRJNA1257050, PRJNA1257061. The link for the Supporting Information is https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1257050, https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1257061

Declarations

Ethics approval and consent to participate

The authors have nothing to report.

Consent for publication

The authors have nothing to report.

Competing interests

The authors declare no competing interests.