Abstract
Terrestrial and aquatic ecosystems are interconnected through runoff and hydrological networks that facilitate the transfer of microbial communities across landscapes. While microbial transport along surface waters is well documented, the role of subsurface hydrological paths in shaping microbial community composition remains poorly understood, particularly in complex karst systems. Here, we studied bacterial communities under stable hydrological conditions across a peri-alpine karst landscape, where mixed limestone-sandstone catchments drain via both surface and subsurface hydrological networks into Lake Thun (Switzerland). We profiled 16S rRNA gene sequences from soils, sediments, surface and subsurface waters, and distinct lake strata. All environments except the lake exhibited high microbial diversity. We observed a clear transitional gradient in bacterial communities along the terrestrial-aquatic interface, with environment type explaining 19% of total variation. Core microbiome analyses revealed both environment-specific and shared taxa, with the strongest overlap between surface and subsurface hydrological networks (63.8%–84.6% shared core taxa). Co-occurrence network analysis identified six major modules. Three of them represented distinct metabolic assemblages tightly associated with specific environment types: peat soils, lake strata, and the subsurface network, respectively. One recurrent module spanned multiple environments and was linked to redox-driven processes, including the oxidation of nitrogen compounds, metals, and methane. Two additional modules comprised aquatic copiotrophs associated with streams and soil heterotrophs prone to export and short-term persistence within the hydrological network. Overall, our results demonstrate that specific environmental settings and hydrological connectivity jointly contribute to selection of microbial species within the karst landscape.
Introduction
Runoff and hydrological networks often facilitate the transfer of biotic and abiotic components between terrestrial and aquatic environments and shape the composition and distribution of microbial communities across multiple scales and interfaces (; Stadler and del Giorgio, 2022) including soils and headwater streams (; Ruiz-González et al., 2015; Niño-García et al., 2016, 2017), or sediment and the water column (Zhang et al., 2021). This influence is also detectable along the stream-river continuum (; , ; Savio et al., 2015). However, little is known about microbial community transfers and compartmentalisation within interconnected surface and subsurface environments (Zhu et al., 2021; Ruff et al., 2024).
Terrestrial and aquatic microbiomes are influenced by local physico-chemical parameters. Soil microbes respond to pH, moisture content, vegetation cover, type of parent bedrock, and temperature (; ; Terrat et al., 2017). Aside from pH (Ruiz-González et al., 2015) and water temperature (), aquatic microbial communities are also influenced by dissolved oxygen () and nutrient concentrations (Newton et al., 2011; Logue et al., 2012), water retention time (Ruiz-González et al., 2015; Niño-García et al., 2016), as well as streambed characteristics (Mendoza-Lera et al., 2017).
The downstream transport of dissolved organic matter and microbial taxa from terrestrial sources into aquatic environments is well known (, ; ). Conceptual frameworks such as the soil-headwaters continuum propose that microbial transport within terrestrial catchments proceed laterally through soils before entering aquatic networks (). Microbial community simplification in aquatic systems usually follows a downstream gradient and is reflected by declining alpha diversity (; Ruiz-González et al., 2015). However, in landscapes where streams divide into parallel surface and subsurface waterways, this process is expected to become increasingly complex and less predictable. How microbial communities are distributed across these parallel pathways, and the extent to which they exchange between surface and subsurface domains, remains poorly understood.
Studies of aquatic karst subsurface components, including springs, cave pools, and cave streams, indicate the presence of a locally persistent core microbiome predominant in these environments under low-flow conditions (; Pronk et al., 2006; Shabarova et al., 2014; Savio et al., 2019; Pedron et al., 2022). This persistence is usually linked to long water residence times and interrupted hydrological connections, which promote strong environmental filtering and taxa adapted to stable subsurface conditions. At the same time, these environments show high sensitivity to hydrological disturbance. Recharge events and increased flow rapidly alter microbial assemblages through surface-derived inputs. Such events are followed by successional recovery toward low-flow community states once hydrological stability is restored (Pronk et al., 2006; Shabarova et al., 2014; ; Savio et al., 2019). These observations identify hydrological regime and water residence time as primary drivers of microbial community composition in karst subsurface. In addition to hydrology, catchment elevation, water chemistry, and seasonality have been shown to influence bacterial communities in subsurface aquatic environments (; Shabarova and Pernthaler, 2010; Shabarova et al., 2014; Slabe et al., 2021).
While hydrological connectivity is recognized as a driver of microbial community assembly, studies integrating terrestrial environments with both surface and subsurface aquatic networks across landscapes remain rare. Here, using 16S rRNA gene amplicon sequencing, we studied bacterial community composition across a peri-alpine karst landscape on the northern shore of Lake Thun (Switzerland), encompassing soils from diverse parent-rock and vegetation types, sediments, surface waters, subsurface aquatic components, rivers, and two lake strata, sampled under hydrologically stable conditions. By focusing on periods of low and stable flow, we aimed to access baseline community patterns under minimal active dispersal and mixing. We hypothesized that under such conditions, microbial community structure would reflect environmental filtering and hydrological compartmentalization, resulting in distinct but partially connected microbial assemblages across terrestrial, and surface and subsurface aquatic systems. Consistent with this hypothesis, our results reveal varying degrees of connectivity among environments, both across and within terrestrial and aquatic boundaries, and identify environment type-specific, niche-based as well as transfer driven microbial signatures across the karst landscape.
Materials and methods
Site description
The sampling site (Supplementary Figure S1), located on the northern shore of Lake Thun (Switzerland), includes both surface and karst catchments dominated by subsurface discharge through an extensive underground drainage network. Both surface and subsurface aquatic systems ultimately drain into Lake Thun. This lake is a peri-alpine oligotrophic system situated on the northern edge of the Alps (46.6958° N, 7.7212° E), with a maximum depth of 216.5 m (). The karst areas at the study site host some of the most extensive cave systems in the world (), with a combined explored length of approximately 320 km, ranging in elevation from 480 to 2000 m a.s.l (; ). Two main catchments are recognized: the smaller is represented by the St. Beatus Caves system (12 km), and a larger one extends from the Bätterich and Gelbenbrunnen springs at Lake Thun through caves such as Bärenschacht, Faustloch, F1, K2, and others to the Siebenhengste-Hohgant cave system (, ). The surface flow is represented by the Lombach stream (the fourth largest inflow to Lake Thun), along with permanent streams such as the Grönbach and temporary ones such as the Sundbach (). The region features a patchy geological composition, dominated by sedimentary rocks, primarily limestone (LS), quartz-sandstone (QS), their mixtures (calcareous sandstone, CS), along with flysch (FL), and less-defined debris (schutt, ST) in the stream beds and on slopes (). The landscape exhibits a diverse mosaic of vegetation zones, including grasslands, forests (spruce and mixed), bogs (encompassing wetlands, flat bogs, and raised peat bogs), and areas with limestone pavements (see Supplementary Text for details).
Sampling
The main sampling campaign was conducted in September 2020 during stable hydrological conditions (n samples = 132, Supplementary Table S1) and included both surface and subsurface components of the hydrological network (pools, streams) as well as the two main rivers that drain into Lake Thun (Aare and Kander), along with different lake strata (epi- and hypolimnion) and soil samples encompassing distinct vegetation types and bedrock formations (mentioned above). Previous (2018, 2019) and subsequent (2021, 2022) sampling campaigns conducted under comparable hydrological and seasonal conditions provided additional samples along central hydrological flow paths, including surface and subsurface waters, lake environments, soils and sediments (n = 69). Additionally, samples from European karst and non-karst lakes (n = 19) (Salcher et al., 2025) were included to increase the size of the lake dataset and improve the robustness of within-group analyses of alpha diversity.
Water samples from the surface streams (n = 58), pools (n = 6), rivers (n = 4), and lake (n = 10) were collected in sterile bottles. Water was pre-filtered using 20 μm mesh and filtered on site through Sterivex polyethersulphone filter units (0.22 μm pore size, Millipore, USA) using a peristaltic pump until the filter was saturated with biomass. The filter units were heat sealed on both ends after adding 0.75 mL DNA–RNA shield (Zymo Research, USA), transported to the lab and stored at −80 °C until further processing. Eight water samples from Lake Thun’s deepest point were collected with a Niskin bottle from two depths (5 and 180 m) during four sampling campaigns. Two additional samples were taken next to the underwater spring Bätterich at 1 and 6.5 m depths. Water samples from subsurface pools (n = 25) and streams (n = 22) collected in the caves were taken with 50 mL syringes and manually filtered through Sterivex filters. A complete list of all aquatic samples is presented in the Supplementary Table S1.
Soil samples (n = 53) were collected from the following parental rock environments: quartz-sandstone, calcareous sandstone, limestone, flysch and schutt; encompassing several vegetation types, including forest and grassland (F + GL), limestone pavement (LP), and bog (PB). Sampling sites included locations situated within the lateral expansion zone of streams as well as sites without stream contact; stream presence or absence at each location was recorded. At each site, a composite sample of five soil cores - four from the corners and one from the center - of a randomly selected 5 m × 5 m plot were taken from the A-horizon (depth varying from 5 to 40 cm). The collected fractions were pooled, sieved through a 5 mm mesh to remove plant debris and small stones and a homogenized subsample was transferred to a sterile 50 mL polypropylene vial and stored at 4 °C until further processing. In the laboratory, duplicates of 200 mg of soil were aliquoted into sterile 2 mL microcentrifuge tubes, amended with 500 μL of DNA/RNA Shield and stored at −80 °C until DNA extraction. Water content of soil samples was determined gravimetrically at 65 °C until a constant weight was achieved (modified from ). Soil pH was measured by mixing 10 g of soil with Milli-Q water at a 1:5 (w/v) ratio. The suspension was shaken for several minutes, allowed to settle, and the pH was measured using a YSI pH probe (). Composite samples representing the upper layer of the stream sediment were collected at five points across the stream bed, using a plastic tube for surface (n = 8) and cave streams (n = 13). Pooled samples were frozen within 24 h and stored at −20 °C until delivered to the lab where they were processed with same protocol as the soil samples. The complete metadata of soil and sediment samples is present in the Supplementary Table S2.
Water chemistry
Water filtered through Sterivex filters was collected in plastic amber bottles (250 mL) for the analysis of physico-chemical parameters as described before (Park et al., 2023). This included ion chromatography for cations (Mg2+, Na+, K+, Ca2+, NH4+) and anions (Cl−, SO42−, F−, NO2−, NO3−), pH, conductivity and alkalinity and concentrations of dissolved reactive phosphorus (DRP), dissolved phosphorus (DP), dissolved organic carbon (DOC), and dissolved nitrogen (DN). Additionally, vertical profiles for pH, temperature, conductivity, dissolved oxygen, and chlorophyll a concentration were taken during lake sampling using a submersible multiparametric probe (YSI EXO3, Yellow Springs Instruments, Yellow Springs, OH, USA). All chemistry measurements for water samples are provided in Supplementary Table S1.
DNA extraction and amplification of 16S rRNA gene fragments
Genomic DNA was extracted using the Quick-DNA Faecal/Soil Microbe Kit (D6012, Zymo Research, USA) with slight modifications. Briefly, the Sterivex filter units were opened, the filters were cut on sterile Petri dishes and added to BashingBead™ Lysis Tube along with the DNA–RNA shield. The soil and sediment samples were directly transferred to BashingBead™ Lysis Tube. Bead beating was executed on a vortex adapter fitted with tube holder assembly (Vortex-Genie 2, Mo Bio Labs, USA) for 15 min at maximum speed. The collected filtrate was amended with a Genomic Lysis buffer (400 μL of 95% ethanol was added to soil and sediment samples). DNA extraction was done according to the kit manufacturer’s instructions, and concentrations were quantified using a Qubit 3 Fluorometer using the Qubit dsDNA HS Assay kit (Q32850, Thermo-Fischer Scientific, USA). Primers 515f and 926r (Parada et al., 2016), flanked by CS1 and CS2 adapters (Fluidigm Corporation, USA) were used to amplify the V4-V5 region of 16S rRNA genes. PCR was carried out using Q5 High Fidelity Mastermix (New England Biolabs, USA) under the following conditions: initial denaturation at 98 °C for 30 s, followed by 28 cycles of denaturation (98 °C/15 s), annealing (50 °C/60 s) and extension (72 °C/90 s); final extension step lasted for 7 min at 72 °C.
Library preparation, amplicon sequencing and sequence analyses
Sequencing libraries were prepared from PCR products using unique labelling (Fluidigm Corporation, USA) and equimolar aliquots were sequenced on an Illumina MiSeq platform at Genomics and Microbiome Core Facility of Rush University (V2 chemistry for 2 × 250 paired reads), with an average sequencing depth of 15,000 raw reads per sample. Cutadapt version 4.7 (Martin, 2011) was used for removing primer sequences. The trimmed sequences were merged and dereplicated using R package dada2 version 1.28 (R version 4.3.0) () with default parameters and reads’ trimming adjusted according to the observed quality. Forward reads were truncated at 230 bp, reverse reads at 200 bp and reads with expected error rates exceeding 3 were removed. Additionally, chimeric sequences were removed using UNOISE3 (minsize = 8) in USEARCH11 (, ). Taxonomic assignment was done using SILVA database v138.1 (https://www.arb-silva.de/, Quast et al., 2013). The obtained taxonomic assignments were mapped to the Genome Taxonomy Database (GTDB) nomenclature where possible (Parks et al., 2022). To validate sequencing accuracy and assess potential contamination, 12 blank DNA extractions and nine mock communities were included in the sequencing runs alongside the environmental samples (ZymoBIOMICS® Microbial Community DNA Standard and Standard II, ZymoResearch, USA). Mock communities served as reference standards to evaluate efficiency of chimera removal and to detect contaminations (). Amplicon sequence variants (ASVs) were ranked by relative abundance within each mock community and compared to the expected composition as provided by the manufacturer’s reference datasheet. To ensure data quality, samples with reads lower than 15,000, post merging of replicates, were excluded from the analysis, and then data set was normalized to the same value using the rarefy function in R package vegan version 2.6–8 (Oksanen et al., 2018).
Statistical analysis
Statistical analysis was conducted using the R environment version 4.2.3 (R Core Team, 2020). Alpha diversity metrics (observed richness and Shannon’s index) among the different environment types were calculated using R package vegan and compared using the Kruskal–Wallis test along with Dunn’s post-hoc pairwise comparisons (with Benjamini-Hochberg adjustments) using R package FSA version 0.10.1 (Ogle et al., 2026). Nonmetric multidimensional scaling (NMDS) based on Bray-Curtis dissimilarities was used to visualize distances between bacterial communities across the landscape using R packages phyloseq version 1.40.0 (McMurdie and Holmes, 2013) and vegan. Differences in microbial communities was evaluated by permutational multivariate analysis of variance (PERMANOVA) using the adonis2 function. Pairwise comparisons between sample types were performed using the pairwiseAdonis R package version 0.4.1 (). Multivariate homogeneity of group dispersions was tested using PERMDISP (betadisper function) to evaluate whether significant PERMANOVA results were due to differences in group centroids or dispersions. An additional NMDS was performed on phylum-level data. Further multivariate analyses were conducted in CANOCO version 5.1 (ter Braak and Smilauer, 2012) to assess the influence of physicochemical parameters on phyla distribution in soil and aquatic environments. Samples from the 2020 campaign supplemented by four epilimnion and hypolimnion samples from summer and autumn campaigns in 2018–2019 were used for this purpose. Environment type was included as a supplementary variable, while geographic coordinates and altitude were used as covariates for soil samples only. Gradient length was evaluated to guide method selection. Identification of core taxa was done using a prevalence-abundance filtering approach adapted from R package Microbiome (). Within each environment, two metrics were calculated: (i) prevalence, i.e., the proportion of samples in which an ASV was present, and (ii) mean relative abundance across samples. A core microbiome subset of ASVs present in ≥50% of samples with a mean relative abundance ≥0.1% were prepared (Neu, 2021). A different filtering approach (relative abundance ≥0.5% and prevalence in ≥4 samples) was used for the whole karst dataset. For both subsets, co-occurrence network analysis was performed using modified R scripts available at https://github.com/RichieJu520/Co-occurrence_Network_Analysis () using Spearman’s correlation coefficient (|ρ| > 0.6) and significance level <0.01. Networks were visualised using Gephi version 0.10.1 (). Functional profiles for the network members across the whole karst dataset were predicted with the FAPROTAX database using R package microeco (1.2.4) () and summarized for every metabolic module. Additionally, ggplot2 (version 3.5.2; Wickham, 2016) and ComplexHeatmap () were used for data visualization.
Results
Overview of data processing and quality control
We sequenced a total of 403 samples from soil (n = 164), sediment (n = 50), water (n = 164), and controls (n = 25) (including duplicates, Supplementary Figure S1, Supplementary Tables S1, S2). Following dada2 processing and removal of additional chimeric sequences with UNOISE3, 66,678 ASVs (92.3% of total ASVs) were retained. As Bray–Curtis dissimilarity analysis (Supplementary Table S3) revealed small distances between duplicates, confirming high reproducibility, replicates were merged for further analysis. A total of 280 ASVs (read abundance >10) were identified as contaminants from mock communities and negative controls and hence removed from the dataset (Supplementary Table S4). Samples with low reads were also removed, resulting in a final dataset of 208 samples containing 63,460 ASVs (soil: n = 53; sediment: n = 21; water: n = 142).
Alpha diversity variation across environment types
Alpha diversity was significantly higher in soils and streams (Shannon Index > 6.00, observed richness > 862.72) than in all other habitats (Shannon Index = 4.04–6.00, observed richness = 223.2–843.93; Supplementary Figure S2). Quartz-sandstone (QS) soils had comparatively lower Shannon and observed diversity than other soil types [calcareous sandstone (CS), limestone (LS), schutt (ST) and flysch (FL)]. Pools and lake samples showed the lowest diversity values. Kruskal–Wallis tests indicated highly significant differences in alpha diversity among the sampled environments for both Shannon (χ2 = 103.73, p = 3.136 × 10−16) and observed richness (χ2 = 102.86, p = 4.628 × 10−16). Pairwise comparisons confirmed that community structure and diversity differed significantly between soil and lake environments, whereas subsurface and surface sediments, pools and streams exhibited intermediate characteristics, showing no significant distinction from both systems (Supplementary Figures S2A,S2B). Patterns in observed richness generally mirrored those of Shannon diversity (r = 0.84, p < 0.001).
Community transitions within the karst landscape
Nonmetric multidimensional scaling (NMDS) of ASVs based on Bray–Curtis dissimilarity revealed a clear transition of bacterial communities along the terrestrial-aquatic interface, characterized by gradual compositional shifts throughout the environment types, except for a clearly defined lake cluster that included several river samples (Figure 1). Environment type explained 19% of the total variation in the bacterial community composition (PERMANOVA; F = 3.19, R2 = 0.19, p = 0.001). Despite heterogeneous dispersions (PERMDISP; F = 54.41, p = 0.001), pairwise PERMANOVA comparisons showed significant community differences among most studied environment types, especially between lake epilimnion and soil systems, though some pairs (e.g., surface/subsurface pools vs. river and lake epilimnion/hypolimnion) showed no significant separation (Figure 2; Supplementary Table S5) and closer distances as remaining environments.
Figure 1
Figure 2
A phylum-level NMDS based on the distribution of 72 phyla (including 7 Archaea) reproduced the ASV-level community patterns (stress = 0.145; Mantel r = 0.73, p = 0.001; Supplementary Figure S3), indicating that general phylum-level structure captures the environmental differentiation observed at higher taxonomic resolution (Figure 3). Pseudomonadota (35.9%), Planctomycetota (7.9%), and Verrucomicrobiota (6.5%) occurred across all environment types, whereas the relative contributions of other phyla varied markedly.
Figure 3
QS soils represented the most distinct terrestrial environment, showing overlap only with CS (Figures 1, 2). Acidobacteriota (28.6% ± 10.4%) dominated these soils and were primarily represented by Terriglobales and Subgroup-2 (SILVA, release 138.1) (Figure 3A). A further characteristic of QS soils was the enrichment of Desulfobacterota (5.19% ± 4.17%), Nitrospirota (1.53% ± 2.14%), and Thermoproteota (1.87% ± 2.09%), whose occurrence was additionally associated with bog vegetation and stream presence, as revealed by RDA analysis (Figure 3B). Acidobacteriota showed a strong negative correlation with pH (Figure 3B) and gradually declined along the soil gradient CS → LS → ST → FL, decreasing from 26.1% ± 6.7% in CS to 12.8% ± 5.9% in ST. This decline was accompanied by a taxonomic shift from Terriglobales (15.1% ± 8.7% in QS to 0.2% ± 0.2% in FL) to Vicinamibacterales (1.8% ± 2.5% in CS and 8.7% ± 2.1% in FL). A similar negative relationship with pH was also observed for the Binatota phylum. In contrast, Gemmatimonadota and Methylomirabilota were associated with higher pH values, whereas Verrucomicrobiota, Actinomycetota, and Myxococcales were more prevalent in soils with lower moisture content. Altogether, pH, moisture content, vegetation type, and stream presence accounted for 42.76% of the total phylum-level variation (adjusted explained variation = 37.04%) based on forward-selected RDA models (Figure 3B).
Sediments occupied transitional positions between terrestrial and aquatic environments; however, selected cave sediment communities were clearly separated in the NMDS plot (Figure 1). Aquatic environments were characterized by higher relative abundances of Bacteroidota (water = 19.2 ± 7.5% vs. soil = 5.4 ± 4.4%) and Actinomycetota (water = 4.7 ± 7.1% vs. soil = 2.1 ± 1.7%), the latter emerging as the dominant phylum in rivers and Lake Thun (19.9 ± 11.6% in rivers, 23.3 ± 3.2% in the epilimnion, and 15.3 ± 5.6% in the hypolimnion; Figure 3C). Surface waters from rivers and Lake Thun epilimnion showed the highest relative abundances of Cyanobacteriota (3.9 ± 3.4% and 9.1 ± 5.6%, respectively). These taxa, together with Bacteroidota, Pseudomonadota, and Bacillota, were positively associated with increased temperature and elevated phosphorus and DOC concentrations (Figure 3C). Patescibacteriota, Nanobdellota, Myxococcota, and Nitrospirota reached their highest abundances in surface and subsurface streams and were associated with alkaline conditions (up to 18.4, 10.9, 6.9, and 4.7%, respectively). Nitrospirota were additionally more abundant in lake hypolimnion samples and cave streams (Figure 3C). Verrucomicrobiota showed slight enrichment in cave pools (10.1 ± 7.5%) and streams (9.2 ± 5.3%), whereas Chloroflexota and Thermoproteota were more prevalent in lake hypolimnia (10.5 ± 4.5% and 4.7 ± 5.2%, respectively). Taken together, the selected physicochemical variables explained 32.1% of the total phylum-level variation in aquatic environments (adjusted explained variation = 25.85%) based on forward-selected RDA models (Figure 3C).
Core microbiome distribution and network analysis revealed high connectivity between the environments
The core microbiome (defined as ASVs present in ≥50% of samples with a mean relative abundance ≥0.1%) comprised both environment-specific ASVs (unique core) and ASVs shared among two or more environments (shared core) (Figures 4A,B). Cave sediments exhibited the lowest shared core fraction (33.8%), restricted to subsurface environments. QS soils and the lake hypolimnion, located at opposite ends of the terrestrial-aquatic transition path (Figure 1), also showed limited shared cores (~39%), overlapping mainly with other soils and the lake epilimnion, respectively. In contrast, LS, ST, and FL soils showed extensive overlap not only with other soils, but also with sediment and aquatic environments. Surface sediments displayed broad connectivity, sharing 62.1% of their core ASVs. The highest levels of overlap were observed among surface and subsurface aquatic networks, which shared between 63.8 and 86.5% of their core ASVs. At the same time, cave and surface pools contained the smallest absolute numbers of core ASVs (38 and 36, respectively; Figure 4C).
Figure 4
Although Acidobacteriota (soils), Actinomycetota (rivers and lake strata), and Bacteroidota and Pseudomonadota dominated both core types across environments, the taxonomic composition remained site-specific even at phylum level. Fibrobacterota were detected exclusively in the shared cores of subsurface pools and streams and in the unique core of the lake hypolimnion. The unique core exhibited greater taxonomic diversification and was frequently dominated by low-abundance phyla, including SZUA-79 and Spirochaetota (exclusive to QS), Tectomicrobia (exclusive to FL), Gemmatimonadota (exclusive to cave sediments and hypolimnion), Armatimonadota (exclusive to epilimnion and hypolimnion), and Babelota and SAR324 clade (exclusive to the hypolimnion). Multiple ASVs affiliated with Desulfobacterota and Bdellovibrionota were present in the unique core; Desulfobacterota occurred in QS, CS, LS, and ST soils, while Bdellovibrionota were found in FL soils, cave sediments, surface pools and sediments, and hypolimnion.
Co-occurrence network analysis of the core microbiome yielded a network comprising 468 ASVs connected by 7,951 significant edges, whereas analysis on the complete karst dataset yielded a network of 889 ASVs and 11,997 edges (Figures 5A,B). Modularity analysis identified six major modules in both networks. Full-network modules (MMs) represented expansions of their corresponding core modules (CMs) through the incorporation of additional ASVs with more restricted environmental distributions (Supplementary Table S7). Overall, node positions were largely conserved within all the modules in both the networks: 47.8% of ASVs retained hub status in both analyses. Only 1.9% of ASVs gained hub status exclusively in the full network, while 1.3% were hubs only in the core network. Additionally, a clear separation was observed between modules 1–2 and 5–6 in both the networks (Figures 5A,B), with MM1b and MM6 both showing high degree nodes (Supplementary Table S7). Several widely distributed core taxa, including Rhodoferax, SAR11 clade IIIb, and Pseudomonas, were not assigned to any co-occurrence module and were absent from both networks.
Figure 5
Module 1 (CM1/MM1a-b) was primarily distributed in QS and CS soils, with limited representation in LS soils. In the full network, it expanded and was subdivided into MM1a and MM1b (Figure 3C). MM1a comprised ASVs largely restricted to QS soils and was enriched in Desulfobacterota and Pseudomonadota, including Methylocapsa and Methylocystis, as well as spatially restricted taxa from the expansion, such as Gallionella, Sideroxydans, and the methanogenic archaeon Methanoregula. FAPROTAX annotations indicated both aerobic and anaerobic heterotrophy, including fermentation. Additional functions included nitrogen fixation, sulfate respiration, methanogenesis, as well as methylotrophy and methanotrophy (Supplementary Table S8). MM1b consisted of more broadly distributed soil ASVs shared across QS, CS, and LS soils and was dominated by Acidobacteriota [Terriglobales and Subgroup-2 (SILVA, release 138.1)], Pseudomonadota, and Planctomycetota. FAPROTAX annotations predominantly indicated aerobic heterotrophy. Additional functions included hydrogen oxidation, with minor contributions from fermentation, nitrate reduction, and nitrogen fixation (Supplementary Table S8). Module 2 (CM2/MM2) consisted of two tightly connected subclusters present in both networks: one restricted to soils and the other additionally spanning sediments, cave pools, and cave and surface streams. Expansion in MM2 only incorporated 26 additional ASVs primarily with terrestrial distributions and included Acidobacteriota, Pseudomonadota, Planctomycetota, and Verrucomicrobiota. FAPROTAX indicated dominance of aerobic heterotrophy, with the presence of anaerobic chemoheterotrophy. Nitrogen cycling processes were also prominent in this module, with a strong contribution from nitrification (ammonia oxidation) (Supplementary Table S8).
Module 3 (CM3/MM3) was distributed mainly across surface streams and pools and, to a lesser extent, subsurface streams and sediments, and showed the lowest average node degree (6.06) and betweenness centrality (206.13) (Supplementary Table S7). CM3 included unclassified Rhizobiales and Rhodoferax (Pseudomonadota) as hubs. MM3 showed the largest expansion among all modules, recruiting 105 additional ASVs spanning rivers, surface streams, cave streams, and surface sediments, while remaining absent from soils. It was mainly driven by additional Pseudomonadota, Bacteroidota, Cyanobacteriota, and nitrogen-cycling taxa such as Nitrospira and Nitrosomonadaceae. FAPROTAX-based functional assignment showed dominance of aerobic heterotrophy and phototrophy. Additional functions included nitrate reduction, nitrogen fixation, ureolysis, and fermentation (Supplementary Table S8).
Module 4 (CM4/MM4) was associated primarily with subsurface environments, particularly cave streams and sediments, with limited representation in ST and FL soils. Nodes in CM4 were typically characterized by low degree but moderate to high betweenness centrality. The core module included Nitrosomonadaceae, Nitrospiraceae, Nitrosopumilales, Ferrovibrionales, Gallionellaceae, Anaerolineales, and Methylomirabilota. The expanded module also incorporated Gammaproteobacteria PLTA-13 and Burkholderiales TRA3-20 (SILVA, release 138.1). FAPROTAX indicated a combination of aerobic and anaerobic heterotrophy. Among all modules, this one showed the strongest signal for nitrification (ammonia oxidation), with additional contributions from iron and hydrogen oxidation (Supplementary Table S8).
Module 5 (CM5/MM5) was primarily linked to subsurface pools and streams, with key ASVs including Methylotenera, Methylobacter, unclassified Methylomonadaceae, Sediminibacterium, Limnohabitans, and Flavobacteriaceae. Expansion in MM5 occurred through the inclusion of additional ASVs affiliated with Verrucomicrobiota and Gammaproteobacteria, i.e., Crenothrix. FAPROTAX annotations indicated both aerobic and anaerobic heterotrophy, together with significant contributions from methylotrophy and methanotrophy (Supplementary Table S8).
Module 6 (CM6a-b/MM6) encompassed lake-associated communities, where the core network consisted of two submodules, with CM6b restricted to the lake hypolimnion. Module six showed the smallest expansion, with MM6 incorporating only eight additional ASVs, which nevertheless increased connectivity between CM6a and CM6b. Highly connected taxa in this module included members of the CL500-29 marine group (Planctomycetota), Sediminibacterium (Bacteroidota), Pedosphaeraceae SH3-11 (SILVA, release 138.1, Verrucomicrobiota), and Methylopumilus (Pseudomonadota). FAPROTAX annotations revealed aerobic heterotrophy and phototrophy (cyanobacteria), with a distinct contribution from methylotrophic taxa, as well as nitrogen cycling (Supplementary Table S8).
Notably, ASVs contributing to the closer taxonomic similarity between cave pools and Lake Thun strata (Figure 2) were almost absent within the detected modules (only few within MM5 and MM6). Shared taxa between cave pools and the hypolimnion included members of Gammaproteobacteria, namely families Nitrosomonadaceae, Methylophilaceae, Methylomonadaceae, as well as genera Limnohabitans and Rhodoferax, and ASVs affiliated with Nanopelagicales (Actinomycetota), Verrucomicrobiota, and Bacteroidota. Compositional similarity between cave pools and the epilimnion was associated with an additional contribution of Alcaligenaceae (Gammaproteobacteria) and Alphaproteobacteria.
Discussion
Our study provides a comprehensive view of bacterial community composition across an interconnected karst landscape spanning 14 distinct environment types and steps beyond the scope of cross-environmental studies that are usually restricted to spatially adjacent sites (; ; Zhu et al., 2021). This allows us to distinguish bacterial communities shaped by local physicochemical conditions from those influenced by hydrological transfer and connectivity within aquatic networks.
Alpha-diversity trends observed in our study confirm previous reports, with terrestrial environments showing higher diversity than aquatic systems (; Yang et al., 2025). Within terrestrial habitats, the most distinct environment identified by NMDS and RDA analyses, QS soils, also displayed the lowest average diversity and richness among soil types. A similar pattern was detected within aquatic systems, where lake strata showed the lowest average alpha diversity. In contrast, soil environments with higher connectivity, as well as aquatic environments positioned at the top of the hydrological network and associated with shorter water retention times, exhibited higher diversity indices. Thus, both species selection and dispersal enhancement shape diversity within the studied landscape. A comparable decrease in diversity, coinciding with increasing water retention time, has previously been reported within hydrological networks (Ruiz-González et al., 2015; Savio et al., 2015). Notably, alpha-diversity did not differ between the surface and subsurface components of the hydrological network.
Modules detected within co-occurrence networks revealed clear environmental signatures and distinct metabolic strategies but also indicated the transfer of associated community members across the landscape. Comparable modular patterns have been described in both soil and aquatic microbial networks, where co-occurring taxa often exhibit complementary metabolic roles beside co-variation in abundance (; ; Mukherjee et al., 2024).
In the soil-associated modules identified here, MM1a (distributed within QS soils) showed a clear association with peat bog vegetation. This was indicated by the prevalence of taxa adapted to anaerobic or microaerobic metabolisms, including sulphate-reducing Desulfobacterota, methanogenic archaea, aerobic methanotrophs, and microaerophilic iron oxidizers (Pester et al., 2010; ; Tan et al., 2019). Bog environments appear to promote syntrophic interactions among taxa with highly specific metabolic capabilities, which together can cope with heterogeneous redox conditions and patchy substrate availability (e.g., sulphate and metals).
In contrast, the MM1b module lacked any associations with peat bogs or anoxic conditions and mainly included taxa commonly reported from mineral soils structured according to pH and substrate quality (; Rousk et al., 2010; ). This module was also more widely distributed across the environments and showed pronounced lineage turnover within Acidobacteriota along pH gradients, in line with previous reports (Sait et al., 2006; ). The dichotomy between MM1a and b shows how hydrology-influenced vegetation effects can reorganize microbial interactions even in soils originated from the same parent rock.
MM2 appeared strongly expanded across environments, highlighting a soil-derived community that seemed to partially persist following hydrological export into sediments and aquatic environments. The MM2 module mainly brought together representatives of the Burkholderiaceae, Nitrosomonadaceae, Sutterellaceae, Sphingomonadaceae, and Chitinophagaceae families. Members of these taxa are well known for their metabolic flexibility and high tolerance to disturbances, and ability to adapt well to fluctuating resources and physicochemical conditions (; Newton et al., 2011; ; ). This pattern suggests that MM2 members detected in aquatic environments and sediments may originate from hydrologically driven community coalescence, a process known to favor taxa with high physiological tolerance in terrestrial-aquatic transition zones (Rillig et al., 2016; Mansour et al., 2018). However, the absence of this module from aquatic systems with longer water retention times suggests its reduced capacity for long-term establishment.
Modules 3–6 were primarily associated with aquatic environments. MM3 was defined by surface and subsurface streams, which are characterized by short water retention times, frequent disturbance, and recurrent inputs of organic matter. The main representatives of this module were affiliated with Bacteroidota, particularly the orders Flavobacteriales and Cytophagales, together with diverse Cyanobacteriota and members of the Burkholderiaceae family. These taxa are core members of freshwater stream microbiomes, where they sustain by a balance between local growth and continuous replacement driven by hydrological mass effects (; Ruiz-González et al., 2015; Shabarova et al., 2021). Despite the presence of closely related, disturbance-tolerant taxa in both MM2 and MM3, the two modules showed markedly different distributions across environments, reflecting a separation between soil-derived and freshwater-associated communities.
The MM4 module was associated primarily with subsurface aquatic habitats and additionally detected in a subset of soil samples and surface streams. The taxonomic composition of this module pointed toward metabolisms characteristic of redox interfaces and microoxic conditions. It consisted of nitrifying and iron-oxidizing prokaryotes (i.a., Nitrosomonadaceae, Nitrospiraceae, Nitrosopumilales, Gallionellaceae), and nitrite-dependent methane oxidizers Methylomirabilota (; ; Shen et al., 2020). Assemblages with similar taxonomic and functional composition have been described in groundwater, sediments, and hyporheic zones (Mosley et al., 2022; Shen et al., 2022; ; Vuillemin et al., 2024). Overall, MM4 appears to represent a redox-driven assemblage shaped by a specific physicochemical niche rather than by discrete environmental boundaries.
MM5 represented a largely cave-restricted module associated primarily with cave pools, although a few individual ASVs were also detected in streams, rivers, and lake samples. The central feature of the module was represented by C1 metabolism and brought together obligate methanotrophs affiliated with Methylococcales (e.g., Crenothrix, Methyloglobulus, Methylobacter), with methylotrophic taxa such as Methylotenera and members of the Beijerinckiaceae (Orata et al., 2018; Taubert et al., 2019; ), and aerobic decomposers. Previous studies on a smaller subset of cave pools from this region have shown that under stagnant, oxygenated conditions only methylotrophic taxa persist over time, while methanotrophs are detectable only during flooding events and fail to maintain populations for extended periods thereafter (Shabarova and Pernthaler, 2010; Shabarova et al., 2013, 2014). Still, it remains uncertain whether such dynamics apply uniformly to all cave pools studied here. Localized methane availability remains a possibility, as cave pools could receive methane via groundwater inflow, sedimentary anoxic niches, or via diffusion from more profound karst conduits, despite overall oxygenated conditions (Webster et al., 2018; ). This module represented a distinct subsurface community, indicating an unexpected degree of compartmentalization within the hydrological network.
Finally, MM6 illustrated a classical freshwater plankton module defined by recurrent freshwater bacterial lineages such as Planktophila and other Nanopelagicales (hgcI clade, Actinomycetota), Polynucleobacter, Limnohabitans, and Methylopumilus, together with diverse Bacteroidota and Planctomycetota (; ; ; Salcher et al., 2019; ). This association of taxa is very similar to the types of microbial communities often reported in pelagic lake waters, where long water retention times, stable stratification, and sustained primary production favor the proliferation of specialized, slow-growing freshwater taxa with high substrate affinity and streamlined genomes (Newton et al., 2011; Salcher et al., 2015; Pernthaler, 2017; ). The fact that dominant members were represented by small-celled Actinomycetota (hgcI clade), together with oligotrophic Alphaproteobacteria and Verrucomicrobiota clearly indicates adaptations to low nutrient concentrations and the efficient utilization of dissolved organic matter, a hallmark of lake plankton communities. Bacteroidota likely contribute to the processing of polymeric organic matter resulting from phytoplankton, while Cyanobacteriota such as Cyanobium indicate coupling to primary production (Xie et al., 2024).
Conclusion
This study provides a unique system-level perspective on microbial community assembly across an interconnected karst landscape. By integrating hydrologically linked terrestrial, surface, and subsurface environments, we revealed how multiple factors interact to drive community assembly across environmental boundaries. Lake samples represented the most isolated assemblages, likely indicating underlying long-term selection processes promoted by extended water retention times. Distinct physicochemical conditions and hydrologically structured vegetation led to the establishment of heterogeneous yet metabolically well-defined communities in terrestrial environments. Less specialized and presumably more stress-tolerant soil fractions showed evidence of coalescence into streams and sediments, where hydrological forcing sustained the dominance of copiotrophic taxa. While surface and subsurface streams shared both of these features, microbial communities in cave pools displayed a distinct subsurface signature centered on C1-compound metabolism and showed the closest compositional similarity to those found in lakes. Additionally, we identified a cluster of taxa associated with nitrification and metal-oxidation processes which was crossing several environmental boundaries. Altogether, we showed how hydrology and physicochemical gradients structure microbial communities across environments and provided an integrative overview of the selective forces driving both compartmentalization and connectivity within a karst landscape.
Statements
Data availability statement
The datasets generated for this study have been deposited in the European Nucleotide Archive (ENA) and the National Centre for Biotechnology Information (NCBI) under BioProject accession number PRJEB106094 (https://www.ebi.ac.uk/ena/browser/view/PRJEB106094; https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJEB106094).
Author contributions
AS: Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing – original draft, Writing – review & editing. P-AB: Investigation, Methodology, Supervision, Validation, Writing – review & editing. VL: Investigation, Methodology, Writing – review & editing. PC: Investigation, Methodology, Writing – review & editing. MS: Data curation, Methodology, Validation, Writing – review & editing, Investigation. TS: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing.
Funding
The author(s) declared that financial support was received for this work and/or its publication. This work was supported by the Czech Science Foundation (Grant no. 20-23718Y to TS), covering material costs, services, and salaries of TS, AS, and P-AB.
Acknowledgments
We thank the Swiss Society for Speleology, Bern Section (SGH Bern), especially Diego Sanz, Rolf Siegenthaler, and Martin Achtman, for assistance with sampling caves and surface environments in Switzerland. We also thank Fran Kostanjšek and Kateřina Kabeláčová for support during field trips and laboratory work, Jakob Pernthaler for providing laboratory facilities during our stays in Switzerland, and the Laboratory for Water and Soil Protection of the Canton of Bern for help with sampling Lake Thun. We are grateful to Cecilia Chiriac and Rohit Ghai for valuable feedback on the manuscript.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors MS and TS declared that they were an editorial board member of Frontiers, at the time of submission. This had no impact on the peer review process and the final decision.
Generative AI statement
The author(s) declared that Generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb.2026.1813326/full#supplementary-material
References
1
ArbizuP. M. (2020) pairwiseAdonis: pairwise multilevel comparison using adonis. R package version 0.4 2020. Available online at: https://github.com/pmartinezarbizu/pairwiseAdonis (Accessed May 1, 2026).
2
BanerjeeS.SchlaeppiK.van der HeijdenM. G. A. (2018). Keystone taxa as drivers of microbiome structure and functioning. Nat. Rev. Microbiol.16, 567–576. doi: 10.1038/s41579-018-0024-1,
3
BastianMHeymannSJacomyM. “Gephi: an open-source software for exploring and manipulating networks,” in: Proceedings of the International AAAI Conference on Web and Social Media. (2009); 361–362. doi: 10.1609/icwsm.v3i1.13937
4
BechT. B.HellalJ.BadaxwiN.JakobsenR.AamandJ. (2023). Linking denitrification and pesticide transformation potentials with community ecology and groundwater discharge in hyporheic sediments in a lowland stream. Water Res.242:120174. doi: 10.1016/j.watres.2023.120174,
5
BesemerK.PeterH.LogueJ. B.LangenhederS.LindströmE. S.TranvikL. J.et al. (2012). Unraveling assembly of stream biofilm communities. ISME J.6, 1459–1468. doi: 10.1038/ismej.2011.205,
6
BodelierP. L. E.DedyshS. N. (2013). Microbiology of wetlands. Front. Microbiol.4, 2–5. doi: 10.3389/fmicb.2013.00079,
7
Brannen-DonnellyK.EngelA. S. (2015). Bacterial diversity differences along an epigenic cave stream reveal evidence of community dynamics, succession, and stability. Front. Microbiol.6:729. doi: 10.3389/fmicb.2015.00729,
8
CaillonF.BesemerK.PeduzziP.SchelkerJ. (2021). Soil microbial inoculation during flood events shapes headwater stream microbial communities and diversity. Microb. Ecol.82, 591–601. doi: 10.1007/s00248-021-01700-3,
9
CallahanB. J.McMurdieP. J.RosenM. J.HanA. W.JohnsonA. J. A.HolmesS. P. (2016). DADA2: high-resolution sample inference from Illumina amplicon data. Nat. Methods13, 581–583. doi: 10.1038/nmeth.3869,
10
ChauJ. F.BagtzoglouA. C.WilligM. R. (2011). The effect of soil texture on richness and diversity of bacterial communities. Environ. Forensic12, 333–341. doi: 10.1080/15275922.2011.622348
11
ChiriacM.-C.HaberM.SalcherM. M. (2023). Adaptive genetic traits in pelagic freshwater microbes. Environ. Microbiol.25, 606–641. doi: 10.1111/1462-2920.16313,
12
ChiriacM.-C.LayounP.FernandesC.Szőke-NagyT.KasalickyV.OkazakiY.et al. (2025). Ecological success in freshwater lakes: insights from novel cultivated lineages of the abundant Nanopelagicales order. Microbiome14:27. doi: 10.1186/s40168-025-02272-x
13
CojeanA. N. Y.KremerK.BartosiewiczM.FabbriS. C.LehmannM. F.WirthS. B. (2021). Morphology, formation, and activity of three different pockmark systems in peri-alpine Lake Thun, Switzerland. Front. Water3:666641. doi: 10.3389/frwa.2021.666641
14
CrumpB. C.AdamsH. E.HobbieJ. E.KlingG. W. (2007). Biogeography of bacterioplankton in lakes and streams of an arctic tundra catchment. Ecology88, 1365–1378. doi: 10.1890/06-0387,
15
CrumpB. C.Amaral-ZettlerL. A.KlingG. W. (2012). Microbial diversity in arctic freshwaters is structured by inoculation of microbes from soils. ISME J.6, 1629–1639. doi: 10.1038/ismej.2012.9,
16
DedyshS. N.IvanovaA. A. (2019). Planctomycetes in boreal and subarctic wetlands: diversity patterns and potential ecological functions. FEMS Microbiol. Ecol.95:fiy227. doi: 10.1093/femsec/fiy227,
17
DettmannU.KraftN. N.RechR.HeidkampA.TiemeyerB. (2021). Analysis of peat soil organic carbon, total nitrogen, soil water content and basal respiration: is there a ‘best’ drying temperature?Geoderma403:115231. doi: 10.1016/j.geoderma.2021.115231
18
EdgarR. C. (2010). Search and clustering orders of magnitude faster than BLAST. Bioinformatics26, 2460–2461. doi: 10.1093/bioinformatics/btq461,
19
EdgarR. C. (2016). UNOISE2: improved error-correction for Illumina 16S and ITS amplicon sequencing bioRxiv 081257 doi: 10.1101/081257
20
EttwigK. F.ButlerM. K.Le PaslierD.PelletierE.MangenotS.KuypersM. M. M.et al. (2010). Nitrite-driven anaerobic methane oxidation by oxygenic bacteria. Nature464, 543–548. doi: 10.1038/nature08883
21
FariaM.BertoccoT.BarrosoA.CarvalhoM.FonsecaF.Delerue MatosC.et al. (2023). A comparison of analytical methods for the determination of soil pH: case study on burned soils in northern Portugal. Fire6:227. doi: 10.3390/fire6060227
22
FarnleitnerA. H.WilhartitzI.RyzinskaG.KirschnerA. K. T.StadlerH.BurtscherM. M.et al. (2005). Bacterial dynamics in spring water of alpine karst aquifers indicates the presence of stable autochthonous microbial endokarst communities. Environ. Microbiol.7, 1248–1259. doi: 10.1111/j.1462-2920.2005.00810.x,
23
FaustK.RaesJ. (2012). Microbial interactions: from networks to models. Nat. Rev. Microbiol.10, 538–550. doi: 10.1038/nrmicro2832
24
Fernandez-CortesA.CuezvaS.Alvarez-GallegoM.Garcia-AntonE.PlaC.BenaventeD.et al. (2015). Subterranean atmospheres may act as daily methane sinks. Nat. Commun.6, 7003–7011. doi: 10.1038/ncomms8003,
25
FiererN.BradfordM. A.JacksonR. B. (2007). Toward an ecological classification of soil bacteria. Ecology88, 1354–1364. doi: 10.1890/05-1839.,
26
FusiM.RigaudS.GuadagninG.BarausseA.MarascoR.DaffonchioD.et al. (2023). Ideas and perspectives: the fluctuating nature of oxygen shapes the ecology of aquatic habitats and their biogeochemical cycles - the aquatic oxyscape. Biogeosciences20, 3509–3521. doi: 10.5194/bg-20-3509-2023
27
GounandI.HarveyE.LittleC. J.AltermattF. (2018). Meta-ecosystems 2.0: rooting the theory into the field. Trends Ecol. Evol.33, 36–46. doi: 10.1016/J.TREE.2017.10.006,
28
GriffinJ. S.LuN.SangwanN.LiA.DsouzaM.StumpfA. J.et al. (2017). Microbial diversity in an intensively managed landscape is structured by landscape connectivity. FEMS Microbiol. Ecol.93, 1–12. doi: 10.1093/femsec/fix120,
29
GriffithsR. I.ThomsonB. C.JamesP.BellT.BaileyM.WhiteleyA. S. (2011). The bacterial biogeography of British soils. Environ. Microbiol.13, 1642–1654. doi: 10.1111/j.1462-2920.2011.02480.x,
30
GuZ.EilsR.SchlesnerM. (2016). Complex heatmaps reveal patterns and correlations in multidimensional genomic data. Bioinformatics32, 2847–2849. doi: 10.1093/bioinformatics/btw313,
31
HahnM. W.HoetzingerM. (2019). “Polynucleobacter,” in Bergey’s Manual of Systematics of Archaea and Bacteria, (), 1–24.
32
HäuselmannP. (2002). Cave genesis and its relationship to surface processes: investigations in the Siebenhengste region BE -Switzerland. Karstologia43, 61–61.
33
HäuselmannP. (2018). Le réseau Siebenhengste- Hohgant (Suisse). Karstologia: Rev. Karst. Speleol. Phys.71, 33–44. doi: 10.3406/karst.2018.3135
34
HermansS. M.BuckleyH. L.CaseB. S.LearG. (2020). Connecting through space and time: catchment-scale distributions of bacteria in soil, stream water and sediment. Environ. Microbiol.22, 1000–1010. doi: 10.1111/1462-2920.14792,
35
HuA.JuF.HouL.LiJ.YangX.WangH.et al. (2017). Strong impact of anthropogenic contamination on the co-occurrence patterns of a riverine microbial community. Environ. Microbiol.19, 4993–5009. doi: 10.1111/1462-2920.13942,
36
HullarM. A. J.KaplanL. A.StahlD. A. (2006). Recurring seasonal dynamics of microbial communities in stream habitats. Appl. Environ. Microbiol.72, 713–722. doi: 10.1128/AEM.72.1.713-722.2006,
37
JeanninP. Y.HäuselmannP. (2012). “Siebenhengste cave system, Switzerland,” in Encyclopedia of Caves, eds. CulverD.WhiteW. (Burlington: ElsevierAcademic Press), 698–707.
38
KarstensL.AsquithM.DavinS.FairD.GregoryW. T.WolfeA. J.et al. (2019). Sequencing experiments. mSystems4, 1–14. doi: 10.1128/mSystems.00290-19,
39
KasalickýV.JezberaJ.HahnM. W.ŠimekK. (2013). The diversity of the Limnohabitans genus, an important group of freshwater bacterioplankton, by characterization of 35 isolated strains. PLoS One8:e58209. doi: 10.1371/journal.pone.0058209,
40
KielakA. M.BarretoC. C.KowalchukG. A.van VeenJ. A.KuramaeE. E. (2016). The ecology of Acidobacteria: moving beyond genes and genomes. Front. Microbiol.7, 1–16. doi: 10.3389/fmicb.2016.00744,
41
LahitiL.ShettyS. (2017). Tools for microbiome analysis in R. Microbiome package version 1.31.2. Available online at: http://github.com/microbiome (Accessed May 1, 2026).
42
LauberC. L.HamadyM.KnightR.FiererN. (2009). Pyrosequencing-based assessment of soil pH as a predictor of soil bacterial community structure at the continental scale. Appl. Environ. Microbiol.75, 5111–5120. doi: 10.1128/AEM.00335-09,
43
LauferK.NordhoffM.RøyH.SchmidtC.BehrensS.JørgensenB. B.et al. (2016). Coexistence of microaerophilic, nitrate-reducing, and phototrophic Fe(II) oxidizers and Fe(III) reducers in coastal marine sediment. Appl. Environ. Microbiol.82, 1433–1447. doi: 10.1128/AEM.03527-15,
44
LiuC.CuiY.LiX.YaoM. (2021). Microeco: an R package for data mining in microbial community ecology. FEMS Microbiol. Ecol.97:fiaa255. doi: 10.1093/femsec/fiaa255,
45
LogueJ. B.LangenhederS.AnderssonA. F.BertilssonS.DrakareS.LanzénA.et al. (2012). Freshwater bacterioplankton richness in oligotrophic lakes depends on nutrient availability rather than on species-area relationships. ISME J 6, 1127–1136. doi: 10.1038/ismej.2011.184
46
LuetscherM.MoseleyG. E.FestiD.HofF.EdwardsR. L.SpötlC. (2021). A last interglacial speleothem record from the Sieben Hengste cave system (Switzerland): implications for alpine paleovegetation. Quat. Sci. Rev.262:106974. doi: 10.1016/j.quascirev.2021.106974
47
MansourI.HeppellC. M.RyoM.RilligM. C. (2018). Application of the microbial community coalescence concept to riverine networks. Biol. Rev.93, 1832–1845. doi: 10.1111/brv.12422,
48
MarínI.ArahalD. R. (2014). “The family Beijerinckiaceae,” in The Prokaryotes, eds. RosenbergE.DeLongE. F.LoryS.StackebrandtE.ThompsonF. (Berlin: Springer), 115–133.
49
MartinM. (2011). CUTADAPT removes adapter sequences from high-throughput sequencing reads. EMBnet J17, 10–12. doi: 10.14806/ej.17.1.200
50
McMurdieP. J.HolmesS. (2013). Phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS One8:e61217. doi: 10.1371/journal.pone.0061217,
51
Mendoza-LeraC.FrossardA.KnieM.FederleinL. L.GessnerM. O.MutzM. (2017). Importance of advective mass transfer and sediment surface area for streambed microbial communities. Freshw. Biol.62, 133–145. doi: 10.1111/fwb.12856
52
MosleyO. E.GiosE.CloseM.WeaverL.DaughneyC.HandleyK. M. (2022). Nitrogen cycling and microbial cooperation in the terrestrial subsurface. ISME J.16, 2561–2573. doi: 10.1038/s41396-022-01300-0,
53
MukherjeeI.GrujčićV.SalcherM. M.ZnachorP.SeďaJ.DevetterM.et al. (2024). Integrating depth-dependent protist dynamics and microbial interactions in spring succession of a freshwater reservoir. Environ. Microbiome19:31. doi: 10.1186/s40793-024-00574-5,
54
NeuA. T. (2021). Defining and quantifying the core microbiome: challenges and prospects. PNAS118, 1–10. doi: 10.1073/pnas.2104429118,
55
NewtonR. J.JonesS. E.EilerA.McMahonK. D.BertilssonS. (2011). A guide to the natural history of freshwater Lake Bacteria. Microbiol. Mol.Biol. Rev.75, 14–49. doi: 10.1128/mmbr.00028-10,
56
Niño-GarcíaJ. P.Ruiz-GonzálezC.del GiorgioP. A. (2016). Landscape-scale spatial abundance distributions discriminate core from random components of boreal lake bacterioplankton. Ecol. Lett.19, 1506–1515. doi: 10.1111/ele.12704,
57
Niño-GarcíaJ. P.Ruiz-GonzálezC.Del GiorgioP. A. (2017). Exploring the ecological coherence between the spatial and temporal patterns of bacterioplankton in boreal lakes. Front. Microbiol.8:636. doi: 10.3389/fmicb.2017.00636,
58
OgleD. H.DollJ. C.WheelerA. P.DinnoA. FSA: Simple Fisheries Stock Assessment Methods. (2026). doi:10.32614/CRAN.package.FSA
59
OksanenA. J.BlanchetF. G.FriendlyM.KindtR.LegendreP.Mcglinnet al (2018). Package ‘vegan’. Available online at: http://CRAN.Rproject.org/package=vegan (Accessed May 1, 2026).
60
OrataF. D.Meier-KolthoffJ. P.SauvageauD.SteinL. Y. (2018). Phylogenomic analysis of the gammaproteobacterial methanotrophs (order Methylococcales) calls for the reclassification of members at the genus and species levels. Front. Microbiol.9:3162. doi: 10.3389/fmicb.2018.03162,
61
ParadaA. E.NeedhamD. M.FuhrmanJ. A. (2016). Every base matters: assessing small subunit rRNA primers for marine microbiomes with mock communities, time series and global field samples. Environ. Microbiol.18, 1403–1414. doi: 10.1111/1462-2920.13023,
62
ParkH.ShabarovaT.SalcherM. M.KosováL.RychteckýP.MukherjeeI.et al. (2023). In the right place, at the right time: the integration of bacteria into the plankton ecology group model. Microbiome11, 112–115. doi: 10.1186/s40168-023-01522-0,
63
ParksD. H.ChuvochinaM.RinkeC.MussigA. J.ChaumeilP.-A.HugenholtzP. (2022). A complete domain-to-species taxonomy for Bacteria and Archaea. Nat. Biotechnol.38, 1079–1086. doi: 10.1038/s41587-020-0501-8,
64
PedronR.EspositoA.CozzaW.PaolazziM.CristofoliniM.SegataN.et al. (2022). Microbiome characterization of alpine water springs for human consumption reveals site- and usage-specific microbial signatures. Front. Microbiol.13:946460. doi: 10.3389/fmicb.2022.946460,
65
PernthalerJ. (2017). Competition and niche separation of pelagic bacteria in freshwater habitats. Environ. Microbiol.19, 2133–2150. doi: 10.1111/1462-2920.13742,
66
PesterM.BittnerN.DeevongP.WagnerM.LoyA. (2010). A “rare biosphere” microorganism contributes to sulfate reduction in a peatland. ISME J.4, 1591–1602. doi: 10.1038/ismej.2010.75,
67
PronkM.GoldscheiderN.ZopfiJ. (2006). Dynamics and interaction of organic carbon, turbidity and bacteria in a karst aquifer system. Hydrogeol. J.14, 473–484. doi: 10.1007/s10040-005-0454-5
68
QuastC.PruesseE.YilmazP.GerkenJ.SchweerT.YarzaP.et al. (2013). The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res.41, 590–596. doi: 10.1093/nar/gks1219
69
R Core Team (2020). R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing.
70
RilligM. C.LehmannA.Aguilar-TriguerosC. A.AntonovicsJ.CarusoT.HempelS.et al. (2016). Soil microbes and community coalescence. Pedobiologia59, 37–40. doi: 10.1016/j.pedobi.2016.01.001
71
RouskJ.BååthE.BrookesP. C.LauberC. L.LozuponeC.CaporasoJ. G.et al. (2010). Soil bacterial and fungal communities across a pH gradient in an arable soil. ISME J.4, 1340–1351. doi: 10.1038/ismej.2010.58
72
RuffS. E.de AngelisI. H.MullisM.PayetJ. P.MagnaboscoC.LloydK. G.et al. (2024). A global comparison of surface and subsurface microbiomes reveals large-scale biodiversity gradients, and a marine-terrestrial divide. Sci. Adv.10:eadq0645. doi: 10.1126/sciadv.adq0645,
73
Ruiz-GonzálezC.Niño-GarcíaJ. P.del GiorgioP. A. (2015). Terrestrial origin of bacterial communities in complex boreal freshwater networks. Ecol. Lett.18, 1198–1206. doi: 10.1111/ele.12499,
74
SaitM.DavisK. E. R.JanssenP. H. (2006). Effect of pH on isolation and distribution of members of subdivision 1 of the phylum Acidobacteria occurring in soil. Appl. Environ. Microbiol.72, 1852–1857. doi: 10.1128/AEM.72.3.1852-1857.2006,
75
SalcherM. M.LayounP.FernandesC.ChiriacM.-C.BulzuP.-A.GhaiR.et al. (2025). Bringing the uncultivated microbial majority of freshwater ecosystems into culture. Nat. Commun.16:7971. doi: 10.1038/s41467-025-63266-9,
76
SalcherM. M.NeuenschwanderS. M.PoschT.PernthalerJ. (2015). The ecology of pelagic freshwater methylotrophs assessed by a high-resolution monitoring and isolation campaign. ISME J.9, 2442–2453. doi: 10.1038/ismej.2015.55,
77
SalcherM. M.SchaefleD.KasparM.NeuenschwanderS. M.GhaiR. (2019). Evolution in action: habitat transition from sediment to the pelagial leads to genome streamlining in Methylophilaceae. ISME J.13, 2764–2777. doi: 10.1038/s41396-019-0471-3,
78
SavioD.SinclairL.IjazU. Z.ParajkaJ.ReischerG. H.StadlerP.et al. (2015). Bacterial diversity along a 2600 km river continuum. Environ. Microbiol.17, 4994–5007. doi: 10.1111/1462-2920.12886,
79
SavioD.StadlerP.ReischerG. H.DemeterK.LinkeR. B.BlaschkeA. P.et al. (2019). Spring water of an alpine karst aquifer is dominated by a taxonomically stable but discharge-responsive bacterial community. Front. Microbiol.10:28. doi: 10.3389/fmicb.2019.00028,
80
ShabarovaT.PernthalerJ. (2010). Karst pools in subsurface environments: collectors of microbial diversity or temporary residence between habitat types. Environ. Microbiol.12, 1061–1074. doi: 10.1111/j.1462-2920.2009.02151.x,
81
ShabarovaT.SalcherM. M.PorcalP.ZnachorP.NedomaJ.GrossartH. P.et al. (2021). Recovery of freshwater microbial communities after extreme rain events is mediated by cyclic succession. Nat. Microbiol.6, 479–488. doi: 10.1038/s41564-020-00852-1,
82
ShabarovaT.VilligerJ.MorenkovO.NiggemannJ.DittmarT.PernthalerJ. (2014). Bacterial community structure and dissolved organic matter in repeatedly flooded subsurface karst water pools. FEMS Microbiol. Ecol.89, 111–126. doi: 10.1111/1574-6941.12339,
83
ShabarovaT.WidmerF.PernthalerJ. (2013). Mass effects meet species sorting: transformations of microbial assemblages in epiphreatic subsurface karst water pools. Environ. Microbiol.15, 2476–2488. doi: 10.1111/1462-2920.12124,
84
ShenJ.SmithA. C.BarnettM. J.MorganA.WynnP. M. (2022). Distinct microbial communities in the soils, waters, and speleothems of a hyperalkaline cave system. J. Geophys. Res. Biogeo.127:e2022JG006866. doi: 10.1029/2022JG006866
85
ShenL.TianM.ChengH.LiuX.YangY.LiuJ.et al. (2020). Different responses of nitrite- and nitrate-dependent anaerobic methanotrophs to increasing nitrogen loading in a freshwater reservoir. Environ. Pollut.263:114623. doi: 10.1016/j.envpol.2020.114623,
86
SlabeM. O.DanevčičT.HugK.FillingerL.Mandić-MulecI.GrieblerC.et al. (2021). Key drivers of microbial abundance, activity, and diversity in karst spring waters across an altitudinal gradient in Slovenia. Aquat. Microb. Ecol.86, 99–114. doi: 10.3354/AME01956
87
StadlerM.del GiorgioP. A. (2022). Terrestrial connectivity, upstream aquatic history and seasonality shape bacterial community assembly within a large boreal aquatic network. ISME J.16, 937–947. doi: 10.1038/s41396-021-01146-y,
88
TanS.LiuJ.FangY.HedlundB. P.LianZ. H.HuangL. Y.et al. (2019). Insights into ecological role of a new deltaproteobacterial order Candidatus Acidulodesulfobacterales by metagenomics and metatranscriptomics. ISME J.13, 2044–2057. doi: 10.1038/s41396-019-0415-y,
89
TaubertM.GrobC.CrombieA.HowatA. M.BurnsO. J.WeberM.et al. (2019). Communal metabolism by Methylococcaceae and Methylophilaceae is driving rapid aerobic methane oxidation in sediments of a shallow seep near Elba, Italy. Environ. Microbiol.21, 3780–3795. doi: 10.1111/1462-2920.14728,
90
ter BraakC. J. F.SmilauerP. (2012). Canoco Reference Manual and User’s guide: Software for Ordination, Version 5.0. Ithaca: Microcomputer Power.
91
TerratS.HorrigueW.DequietdS.SabyN. P. A.LelièvreM.NowakV.et al. (2017). Mapping and predictive variations of soil bacterial richness across France. PLoS One12:e0186766. doi: 10.1371/journal.pone.0186766,
92
VuilleminA.Ruiz-BlasF.YangS.BartholomäusA.HennyC.KallmeyerJ. (2024). Taxonomic and functional partitioning of Chloroflexota populations under ferruginous conditions at and below the sediment-water interface. FEMS Microbiol. Ecol.100, 1–15. doi: 10.1093/femsec/fiae140,
93
WickhamH. (2016). ggplot2: Elegant Graphics for Data Analysis. New York: Springer-Verlag. doi: 10.1007/978-3-319-24277-4
94
WebsterK. D.DrobniakA.EtiopeG.MastalerzM.SauerP. E.SchimmelmannA. (2018). Subterranean karst environments as a global sink for atmospheric methane. Earth Planet. Sci. Lett.485, 9–18. doi: 10.1016/j.epsl.2017.12.025
95
XieG.SunC.LuoW.GongY.TangX. (2024). Distinct ecological niches and community dynamics: understanding free-living and particle-attached bacterial communities in an oligotrophic deep lake. Appl. Environ. Microbiol.90:e0071424. doi: 10.1128/aem.00714-24,
96
YangY.ChenC.GrossartH. P.LiuY. (2025). Community assembly characteristics of abundant and rare bacterial taxa in water, sediment and riparian soil of Wujiang River, China. Ecotoxicol. Environ. Saf.297:118262. doi: 10.1016/j.ecoenv.2025.118262,
97
ZhangL.Delgado-BaquerizoM.ShiY.LiuX.YangY.ChuH. (2021). Co-existing water and sediment bacteria are driven by contrasting environmental factors across glacier-fed aquatic systems. Water Res.198:117139. doi: 10.1016/j.watres.2021.117139,
98
ZhuH.-Z.ZhangZ.-F.ZhouN.JiangC.-Y.WangB.-J.CaiL.et al. (2021). Bacteria and metabolic potential in karst caves revealed by intensive bacterial cultivation and genome assembly. Appl. Environ. Microbiol.87:e02440. doi: 10.1128/AEM
Summary
Keywords
core microbiome, karst, microbial community composition and diversity, network analysis, terrestrial-aquatic interface
Citation
Singh AP, Bulzu P-A, Lanta V, Chaloupský P, Salcher MM and Shabarova T (2026) From soils to lake: interplay between hydrology and local environmental settings drives species selection across a karst landscape. Front. Microbiol. 17:1813326. doi: 10.3389/fmicb.2026.1813326
Received
18 February 2026
Revised
16 April 2026
Accepted
23 April 2026
Published
13 May 2026
Volume
17 - 2026
Edited by
Danny Ionescu, Technical University of Berlin, Germany
Reviewed by
Björn Brindefalk, Stockholm University, Sweden
Arnelyn D. Doloiras-Laraño, Ehime University, Japan
Updates
Copyright
© 2026 Singh, Bulzu, Lanta, Chaloupský, Salcher and Shabarova.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Tanja Shabarova, shabrik@gmail.com
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.