Fungi dominate eukaryotic microbial assemblages in hot spring biofilms of East and Southeast Asia Authors: Subhasri Thirumaran, Christaline George, Chananwat Kortheerakul, Nitthiya Khunthong, Chitrabhanu Sharma Bhunjun, Chayanard Phukhamsakda, Kok-Gan Chan, Ying Chang, Eunice Jingmei Tan, Maurycy Daroch, Kevin D. Hyde, Kian Mau Goh, Rungaroon Waditee-Sirisattha, Stephen B. Pointing Published: 2025 ABSTRACT Terrestrial hot springs are globally distributed extreme environments, and these systems have long served as natural laboratories for studying microbial life under thermal stress. While much of the research to date has focused on thermophilic bacteria and archaea, there is a growing appreciation for the diversity and ecological significance of eukaryotic microorganisms in these habitats. In this study, we used metagenomic sequencing to assess inter-domain microbial diversity in biofilms from 47 circumneutral hot springs across East and Southeast Asia, with a specific focus on resolving eukaryotic taxa and their ecology. Whilst all biofilm communities were dominated by bacteria, the microbial eukaryotes represented approximately 10% of the taxonomic diversity and accounted for 1.3% of overall taxa abundance, indicating a small but significant presence. We provide the first comprehensive inter-domain checklist of over 14,500 microbial taxa in hot springs. Patterns in diversity were significantly correlated with temperature, hydrogen sulfide, and pH in hot springs. Fungi emerged as the most abundant and prevalent eukaryotic group, indicating an important role as eukaryotic saprotrophs, with Ascomycota yeasts comprising the most individually abundant taxa. Among other microbial eukaryotic phyla, the photosynthetic Chlorophyta and Bacillariophyta were most abundant. Predatory/grazing microbial eukaryotes were relatively less diverse and abundant. Network co-occurrence analysis was used to indicate extensive and specific biotic interactions between eukaryotes and bacteria in biofilms. We further employed metatranscriptomics to identify putatively active taxa, revealing that most detected eukaryotes were transcriptionally active. While fungi accounted for most transcripts, the highest RNA:DNA ratios were observed among predatory and photosynthetic taxa, suggesting elevated activity in these functional groups. Overall, our findings highlight the diversity, interactions, and activity of eukaryotes in Southeast Asian hot spring biofilms, underscoring their potential importance in shaping microbial community structure and function in extreme environments.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40793-025-00816-0. Introduction Terrestrial hot springs are globally distributed and represent extreme microbial habitats characterised by elevated temperatures, fluctuating pH levels, and variable geochemical conditions [1]. These habitats have long fascinated microbiologists as natural laboratories for exploring the limits of life. While most previous research has focused on the thermophilic Bacteria and Archaea, there is growing recognition of the importance and diversity of eukaryotic microorganisms in these systems [2–4], since their early description as thermophiles with an upper limit of approximately 60 °C [5]. Some individual taxa such as the acidophilic genus Cyanidium (Rhodophyta) are well understood for their occupancy of a specific acidic hot spring niche [6]. The application of high-throughput sequencing of eDNA has progressively advanced a more comprehensive understanding of microbial eukaryote distribution [7]. Diverse microbial eukaryotes, including fungi, protists, and microalgae, have been shown to occupy a range of ecological niches in hot springs including water [8–10], sediment [11–13], and biofilms [14–16]. Yet they remain poorly understood in ecological terms relative to their bacterial and archaeal counterparts [15, 17]. Survival of eukaryotic microorganisms in hot springs requires adaptation to environmental stressors, including high temperature and oxidative stress [18, 19]. Thermotolerant fungi, are important saprotrophs of plant debris in waterways and thereby likely contribute to carbon and nutrient cycling in geothermal habitats [20]. The thermophilic protists may influence microbial food webs through predation and symbiotic interactions, serving as both grazers and hosts for bacterial and archaeal communities as demonstrated for mesophilic aquatic systems [21], and deep-sea marine hydrothermal habitats [22]. These traits indicate that microbial eukaryotes likely play an important role in structure–function relationships within geothermal microbiomes and in maintaining stability of ecosystem processes under extreme environmental pressures. Advances in molecular and computational tools have significantly expanded our understanding of thermotolerant and thermophilic eukaryotes including fungi, protists, and microalgae that were previously overlooked due to cultivation biases. An important early study in acidic hot springs at Lassen Volcanic Park used DGGE to separate rRNA gene phylotypes indicative of diverse fungi and protists at temperatures of up 68 °C, thus potentially expanding the thermal limits for microbial eukaryotes [23]. A number of PCR-based metabarcoding studies have built upon early findings and uncovered a wide range of thermotolerant and thermophilic eukaryotes including fungi, protists, and microalgae. Unspecified signatures for ascomycete and basidiomycete fungi were recorded in waters of Croatian hot springs [9], and they have also been mentioned as minor components of sequence libraries from Southeast Asian hot springs [15, 20]. A comprehensive assessment of fungi in hot spring waters was recently conducted for the Tangchi Hot Spring in China [10], where the three most abundant genera encountered were yeast-like/unicellular taxa Rhizophydium (Chytridiomycota), Aureobasidium (Ascomycota), and Rhodotorula (Basidiomycota). With regard to the protists, an early clone-library based study of hot springs in Iceland indicated that Chlorophyta and Ciliophora were abundant protistan phyla in waters of circumneutral hot springs [14]. An extensive survey of protists in New Zealand hot spring waters revealed diverse protistan taxa across a broad gradient of pH and temperature [8]. The study concluded that Ciliophora taxa made up over half of all protistan assemblages, and observed that overall most protistan taxa occurred at very few locations [8]. In another large-scale study of hot spring sediments of the Himalayan geothermal belt in China, the findings indicated that Bacillariophyta were the most abundant protistan taxa, followed by Ciliophora > Ochrophyta > Chlorophyta [11]. The diversity and distribution of eukaryotic microorganisms in hot springs remain poorly understood, despite their potential ecological importance. Furthermore, a regional knowledge gap exists for hot springs in East and Southeast Asia. In previous studies many taxa have likely been overlooked or over-represented due to methodological biases, such as primer specificity issues and mismatches in PCR-based metabarcoding studies, and insufficient genomic databases. Furthermore, the central question of how abundant and active microbial eukaryotic phyla may be compared with each other, and with their bacterial and archaeal counterparts in geothermal springs remains unresolved. Greater understanding in this regard can improve inter-domain considerations in models of microbial community assembly, resilience, and succession under environmental stress, as demonstrated for other ecosystems [24]. In this study, we investigated the diversity, distribution, co-occurrence, and putative activity of microbial eukaryotes in hot springs of East and Southeast Asia. This was prompted by our recent findings that microbial eukaryotes comprise a small yet significant fraction of diversity in bacteria-dominated hot spring biofilms from waters that receive large inputs of allochthonous plant debris from the surrounding tropical landscapes [15]. We employed shotgun sequencing of metagenomes and metatranscriptomes to overcome the limitations of previous PCR-based studies, and advance understanding of microbial eukaryotes distribution in hot springs relative to their bacterial and archaeal counterparts. The findings add to a growing body of knowledge on the diversity and ecology of eukaryotic microorganisms under severe environmental stress. Materials and methods Hot springs sampling Photosynthetic microbial biofilms were sampled aseptically at 47 thermally defined tectonic hot spring features at 18 geothermal locations throughout East and Southeast Asia (Fig. 1). The circumneutral hot springs ranged in temperatures from 39 to 67 °C which was within the thermal range for photosynthetic activity (Supplementary Dataset SD1). Biofilm samples were aseptically collected into 500 µl RNA-later preservative solution and immediately stored on ice in the field and during transport, then frozen at -80 °C for laboratory storage. Negative sampling controls contained RNA-later only and were subjected to the same processing as biological samples. The major aqueous abiotic variables shown by a number of previous studies to correlate most highly with hot spring biofilm diversity were measured [15, 25–27], namely temperature using a digital thermocouple (Fluke), pH using an electrode with automatic temperature compensation (Mettler Toledo), and hydrogen sulfide using colorimetric assay (Hach). Geolocation was recorded using GPS in decimal degrees (Garmin). These data are available in Supplementary Dataset SD1.Fig. 1Schematic showing hot spring locations in East and Southeast Asia that were sampled for this study. All hot springs were of circumneutral pH and spanned a thermal gradient from 39–68 °C. For location metadata please refer to Supplementary Dataset SD1 Schematic showing hot spring locations in East and Southeast Asia that were sampled for this study. All hot springs were of circumneutral pH and spanned a thermal gradient from 39–68 °C. For location metadata please refer to Supplementary Dataset SD1 DNA and RNA extraction and sequencing All samples were rinsed in UltraPure DNAse-free distilled water (ThermoFisher) prior to extractions. DNA extraction was carried out for all samples using the Powerlyzer Powersoil Kit (Qiagen). RNA was extracted from a subset of samples spanning a temperature range 45–61 °C (from location SW, Supplementary Dataset SD1) after first flash-freezing in liquid nitrogen and homogenization in TRIzol reagent (Thermofisher), followed by RNA recovery using the Direct-zol RNA miniprep kit (Thermofisher), with a ribosomal RNA depletion step using the Ribo-Zero Plus rRNA depletion kit (Illumina). The recovered DNA and RNA were quality-checked using a Bioanalyzer (Agilent) prior to PCR-free library preparation and sequencing using PE150 kits on the Novaseq 6000 sequencing platform (Illumina). Bioinformatic analysis The metagenomic and metatranscriptomic data were subjected to adapter removal and quality filtering with a quality score threshold of Q30 using fastp v. 0.23.2 [28]. To remove human contaminants, Bowtie2 version 2.4.5 [29], was employed to align reads against the human reference genome (GRCh38_no-alt). The resulting clean reads (metagenomes (N = 47): 597 million reads, metatranscriptomes (N = 4): 21 million reads) were classified taxonomically using Kraken2 v. 2.1.3 [30] with the NCBI nt database, a large collection inclusive of GenBank, RefSeq, Third Party Data, and Protein data Bank approximately 900 GB in size and created in May 2024. The sequencing quality filetring, read processing metrics, and rarefaction steps are shown in Supplementary datasets SD2-4 and SD7. Following the classification with Kraken2 classification, species level abundance for taxa with read count ≥ 10 was estimated in all samples using Bracken v.2.8 [31]. This approach combines taxonomic assignments made using Kraken2 with information about the genomes, to estimate species and/or genus level abundance more precisely based on a Bayesian probability algorithm. The species abundance across all samples were combined and their taxonomic lineages were assigned using Bioinformatics Tools (bit) v. 1.9.7 [32]. A Phyloseq object was then created using the taxonomic and abundance data using Phyloseq v. 1.48.0 package in R v. 4.0 [33] for use in ecological and statistical analysis. After removing taxa identified as incompletely taxonomically assigned, unknown, or derived from higher plants and animals, the final dataset comprised 14,528 microbial taxa across all domains. This dataset is available in Supplementary Dataset SD5. Statistical analysis Alpha diversity metrics were estimated using Phyloseq v. 1.48.0 [33], with p-values calculated using the general linear model (GLM). Beta diversity was analysed using non-metric Multidimensional Scaling (NMDS) based on Bray–Curtis distances using Vegan v. 2.6.10 [34]. The influence of abiotic variables on microbial community structure was assessed by fitting environmental vectors using the envfit function. Co-occurrence network plots were generated using the NetComi package v.1.2.0 [35], and viruses were removed before plotting. A stringent semi-parametric rank-based approach, the SPRING [8] method, was employed to measure associations among taxa. Putative activity of taxa was estimated by calculating the log-transformed RNA:DNA ratio from metatranscriptomes and metagenomes for all reads of a given assigned identity. Results and discussion Inter-Domain microbial diversity of hot spring biofilms Our metagenomic approach enabled the estimation of relative abundances for taxonomically assigned reads across the domains of life (Fig. 2A, Supplementary Dataset SD5). From our extensive dataset of 14,528 taxa, we identified 12,337 bacteria (accounting for 98.3% of total read abundance), 1,441 microbial eukaryotes (1.3%), and 495 archaea (0.3%). To our knowledge, this represents the first direct comparative analysis of relative abundances across bacteria, archaea, and eukarya in hot springs, and crucially, captures the full spectrum of eukaryotic microbial diversity, addressing a key gap in previous studies that primarily targeted specific groups such as bacteria and archaea [15, 17], protists [8, 11] or fungi [10]. In addition, we detected 255 viral taxa spanning 14 phyla (Supplementary Dataset SD5); however, viruses were excluded from downstream analyses as they were beyond the scope of this study. While k-mer based classifiers such as Kraken + Bracken [30], have limitations related to potential over-classification, this method has also demonstrated high sensitivity and low false-negative rates in microbiome analyses [36–38]. Importantly, it circumvents the primer biases and amplification mismatches inherent in PCR-based methods and allows simultaneous profiling of the entire microbial community. Notably, although eukaryotic microorganisms comprised only 1.3% of quality-filtered, taxonomically assigned reads, they represented approximately 10% of the overall taxonomic diversity, thus underscoring their ecological relevance in these extreme environments (Supplementary Dataset SD5).Fig. 2Relative abundance of metagenomic reads assigned by Domain and Phylum. Data plotted for A all Domains, B Bacteria, C Eukarya, D Archaea. All phyla encountered at < 1% relative abundance are grouped as others. Domain-level distribution plots for individual samples is shown in Supplementary Dataset SD6 Relative abundance of metagenomic reads assigned by Domain and Phylum. Data plotted for A all Domains, B Bacteria, C Eukarya, D Archaea. All phyla encountered at < 1% relative abundance are grouped as others. Domain-level distribution plots for individual samples is shown in Supplementary Dataset SD6 The Phylum-level taxonomic composition for bacteria (Fig. 2B) illustrated the dominance of phototrophic cyanobacteria in the photosynthetic biofilms, and this was consistent with our recent description for Southeast Asian hot springs using both 16S rRNA gene sequencing [15]. The overall taxonomic composition was also broadly consistent with that recorded for other hot spring locations including Africa [39, 40], China [41–43], Europe [25, 44, 45], India [46–48], Japan [49, 50], Australasia [51, 52], North America [26, 53–55], South America [56, 57], and Southeast Asia [15, 42, 58–61]. Our metagenomic approach to eukaryotic microbial diversity estimation revealed that fungi were overwhelmingly the most abundant and prevalent eukaryotic taxa relative to other phyla (Fig. 2C). This provides suggests the potential importance of fungi in hot springs relative to other eukaryotic microorganisms, and there are no comparable studies that have concurrently targeted the breadth of eukaryotic microorganisms in hot spring biofilms. The relatively low abundance and diversity of archaea (Fig. 2D), further validates our earlier estimates that archaea comprise a minor component of photosynthetic biofilms in Southeast Asian hot springs [15]. Despite this the Archaea are ecologically important in other geothermal habitats [62]. All domains displayed negative correlations for Species Richness (Fig. 3A) and Shannon’s Diversity Index (Fig. 3B) with temperature, pH, and hydrogen sulfide, although these were strongest for eukaryotes. Distance-based correlations for beta diversity indicated no clear separation by country of origin, and for all domains temperature had the greatest correlation with observed taxonomic distributions (Fig. 3C). For the eukaryotes the second most important variable was pH whereas for bacteria and archaea it was hydrogen sulfide, potentially reflecting the relatively low sulfide levels in Southeast Asian hot springs indicating its role as a metabolite for bacteria and archaea rather than an inhibitor of metabolism as can occur with highly sulfidic springs [63, 64]. The estimation of abiotic influence is intended here only as a broad indicator of the influence from major abiotic variables known to be most important to hot spring biofilms [15, 25–27]. We cannot discount influence from other unmeasured variables but a large number of studies have shown that temperature, pH, and hydrogen sulfide are by far the most important, and other variables seldom add significant additional explanatory power [15, 17]. It is also worth mentioning that Southeast Asian hot springs span a strong thermal gradient but very little pH variability in the circumneutral springs that support photosynthetic biofilms, and thus we did not observe a primary influence for pH as observed in a New Zealand study of planktonic diversity where springs spanned a large gradient from highly acidic to alkaline pH [8].Fig. 3Cross-Domain alpha and beta diversity for microbial taxa. A Species Richness shown as Chao 1, B Shannon’s Diversity Index, C PCoA of Bray Curtis Dissimilarities between communities overlayed with vectors showing the influence of temperature, pH, and hydrogen sulfide using envfit, where the arrow direction denotes influence and length of arrow denotes the strength of association Cross-Domain alpha and beta diversity for microbial taxa. A Species Richness shown as Chao 1, B Shannon’s Diversity Index, C PCoA of Bray Curtis Dissimilarities between communities overlayed with vectors showing the influence of temperature, pH, and hydrogen sulfide using envfit, where the arrow direction denotes influence and length of arrow denotes the strength of association Eukaryotic microorganisms comprised a small yet significant part of the community The intriguing discovery that eukaryotic microorganisms comprised approximately 10% of microbial taxa in hot spring biofilms led us to focus in more detail on this aspect of the data. The eukaryotic diversity spanned 17 phyla (Fig. 4A). This complements a recent large-scale survey of eukaryotic microbial diversity in genomic datasets from diverse environments that uncovered significant novel diversity [65]. To facilitate comparisons, we employ the recently updated super-phylum level classification system in our study [66].Fig. 4Eukaryotic microbial diversity in hot springs. A Taxonomic diversity for the Eukaryota are indicated by a Sankey Plot. Bar thickness denotes relative abundance of taxa in each taxonomic group. B Heatmap illustrating relative abundance of the 100 most abundant eukaryotic taxa by sample. Abundance is shown as the fraction of overall Eukaryotic taxa abundance in a given sample. Taxa distribution by sample location is shown in Supplementary Dataset SD7 Eukaryotic microbial diversity in hot springs. A Taxonomic diversity for the Eukaryota are indicated by a Sankey Plot. Bar thickness denotes relative abundance of taxa in each taxonomic group. B Heatmap illustrating relative abundance of the 100 most abundant eukaryotic taxa by sample. Abundance is shown as the fraction of overall Eukaryotic taxa abundance in a given sample. Taxa distribution by sample location is shown in Supplementary Dataset SD7 The fungi (Amorphea) made up almost three quarters of all eukaryotic taxa encountered, with Ascomycota displaying the highest phylum level abundance and prevalence. This is consistent with a recent meta-analysis of fungi from metagenomes in diverse extreme environments, which concluded this phylum is prevalent in stressful habitats [67]. We considered that the high abundance of fungal taxa may partially reflect the predominance of forested environments at our sampling sites, where visible allochthonous inputs of plant debris were common in the hot springs. Given that many fungi are lignocellulolytic saprotrophs, their presence may result from passive introduction via plant material [68]. However, the taxonomic identity of abundant fungal taxa suggest that this was not the case and that a unique hot spring assemblage occurred, as seen for cellulolytic bacteria in tropical hot springs [69]. The most prevalent and abundant individual taxon was a yeast (Saccharomyces cerevisiae) and overall the hot springs supported no less than 50 different Saccharomyces species (Supplementary Dataset SD5). We considered the possibility that S.cerevisiae may have arisen as a reagent contaminant but it was not recovered in negative controls. The second most abundant yeast was Moesziomyces antarcticus, a common yeast in freshwater sediments worldwide. Numerous other yeast taxa were identified across both Ascomycota and Basidiomycota. We propose that these, along with other unicellular fungi such as zoosporic Chytridiomycota, may reflect a selective advantage for unicellular over filamentous growth forms within hot spring biofilm habitats, an adaptation similarly observed for yeasts in other extreme environments [70, 71]. The diversity of fungi in hot springs was unexpectedly high and spanned six phyla, and this supports the recent postulation of thermophily as an ancestral state for fungal lineages [72]. In addition to the yeasts and zoosporic taxa, a number of relatively low-abundance fungal signatures were affiliated to known wood-decay and plant pathogenic filamentous macrofungi, which may reflect utilisation of lignocellulosic inputs at high temperature in hot springs as observed for thermophilic composting microbiomes [73, 74]. Other filamentous fungal taxa belonged to the insect pathogenic/saprotrophic Ophiocordyceptaceae family for Southeast Asian samples, which aligns with the high insect populations in tropical habitats. Filamentous fungi are generally viewed as having a maximum growth temperature of approximately 55 °C [75]. However, it must also be considered that widely dispersed fungal spores may be at least partially responsible for such taxonomic signatures since macrofungi are also recorded as components of habitats where fungi are unexpected, e.g., marine animal microbiomes and anoxic habitats [76, 77]. The protistan phyla identified in this study were taxonomically affiliated within the renamed super-phyla Archaeplastida, Amorphea, Discoba, Provora, and the TSAR. The photosynthetic Chlorophyta and Bacillariophyta comprised the vast majority of diversity (405/581 taxa) and abundance for these taxa (Fig. 4B). Overall, the phototroph Chlorella sorokiniana (Chlorophyta) and related taxa were the most abundant phototrophic eukaryotes and were prevalent and abundant at temperatures ≤ 60 °C in our study. This taxon is previously recorded from hot springs in Japan growing at temperatures of up to 42 °C [78]. Most of the remaining abundant taxa belonged to freshwater and epilithic taxa within the Chlorophyta, e.g., Chlamydomonas reinhardtii, and Bacillariophyta, e.g., Achnanthes inflata. Other phyla supporting photosynthetic taxa included the Cercozoa, Euglenozoa, and Rhodophyta albeit with relatively low abundance (Supplementary Dataset SD5). The prevalence of photosynthetic taxa among the protistan phyla may reflect that the biofilms were primarily composed of phototrophic bacteria and so the conditions may have been favourable for photosynthesis across Domains. Heterotrophic predatory/grazing protists were less diverse and abundant/prevalent overall compared with phototrophic eukaryotes. At the phylum level the Euglenozoa were most abundant with much of the diversity in this phylum arising from taxa with plant, e.g. Phytomonas, and insect, e.g. Strigomonas, hosts, and so we concluded these were likely introduced allochthonously (Fig. 4B, Supplementary Dataset SD5). Among other predatory phyla the Ciliophora and Evosea were most abundant. The common grazing taxon Paramecium tetraurelia (Ciliophora) known for feeding on microbial biofilms was frequently encountered. Within the Evosea, predatory slime moulds such as Cavendaria and Dictystelium were commonly observed. Other low abundance predatory taxa included representatives of the Nebulidia, from the relatively newly described super-phylum of microbial predatory taxa the Provora [79]. These taxa likely feed on the phototrophic and other microorganisms in these biomass-rich biofilms. The detection of predatory/grazing microbial eukaryotes adds to our understanding of trophic complexity in hot spring biofilms, and future studies to unravel the role of these higher trophic states on community turnover are warranted. This should be considered in tandem with other potential drivers such as viral lysis which have been identified as potentially important in hot springs where predation and grazing are absent [80, 81], as well as potential interactions between protists and viruses [82]. Many of the eukaryotic taxa across all phyla were new records for the hot springs habitat (Supplementary Dataset SD2). Despite the bacterial and archaeal dataset displaying large numbers of known thermophilic taxa, assigning eukaryotes as thermophiles from taxonomic data is not so straightforward given the paucity of information on this Domain in hot springs. The widely studied and cultivated thermophilic amoeba model organism Echinamoeba thermarum was not recovered [83], nor were any other members of the Echinamoebidae. Distributions of eukaryotic microbiota in other hot spring studies have reported a somewhat stochastic occurrence. For example, diatoms and fungi have been shown by a combined microscopy and sequencing approach to occur as patchy and localised populations in microbialite structures of Himalyan hot springs [84]. In a study of hot spring waters across a wide pH gradient in New Zealand hot springs occurrence of protists was also patchy, with over 90% of taxa occurring in less than 10% of locations and many specific to individual locations [8]. Our metagenomic approach revealed that most taxa are not as patchily distributed in biofilms, although it is unclear if this reflects in part the higher sequencing coverage in our study, or if micro-habitat differences between high-biomass biofilms relative to open water, sediment, or microbialite geothermal habitats drive different diversity patterns as observed for bacteria in hot springs [85]. A significant negative correlation of Shannon’s Diversity Index with temperature was observed for abundant phyla, i.e., those present at ≥ 1% relative abundance (Fig. 5A). This included three fungal phyla Ascomycota, Basidiomycota, and Mucoromycota; as well as the Bacillariophyta and Euglenozoa. In general terms the decrease in Shannon’ Diversity Index with increasing environmental stress may be suggestive of a trend towards a more specialised community with fewer and more abundant taxa, i.e., less evenness. No significant correlations with Shannon’s Diversity Index were observed for pH or hydrogen sulfide. There was no clear geographic separation of eukaryotic assemblages between locations and instead temperature appeared to be the most important driver of beta diversity overall (Fig. 5B). An exception was seen for Ascomycota and Chlorophyta where hydrogen sulfide was more highly correlated, likely due to taxa occurrence at a small number of high-sulfide locations. Hydrogen sulfide is known to be toxic to primary metabolism in Ascomycete yeasts [86], and chlorophyte algae [87], although in contrast the cyanobacteria that dominate hot spring biofilms can be adapted to sulfide exposure [88]. A study of planktonic protists in hot spring waters also revealed that pH, temperature and sulfide were significant correlates of diversity [8].Fig. 5Patterns of distribution for eukaryotic microbial taxa. A Shannon’s Diversity Index for phyla present with overall relative abundance ≥ 1%, plotted against location temperature. No significant correlations were observed with either pH or hydrogen sulfide. B PCoA of Bray Curtis Dissimilarities was used to visualise patterns among communities overlayed with vectors showing the influence of temperature, pH, and hydrogen sulfide using envfit, where the arrow direction denotes influence and length of arrow denotes the strength of association Patterns of distribution for eukaryotic microbial taxa. A Shannon’s Diversity Index for phyla present with overall relative abundance ≥ 1%, plotted against location temperature. No significant correlations were observed with either pH or hydrogen sulfide. B PCoA of Bray Curtis Dissimilarities was used to visualise patterns among communities overlayed with vectors showing the influence of temperature, pH, and hydrogen sulfide using envfit, where the arrow direction denotes influence and length of arrow denotes the strength of association Overall, the data are consistent with a Eukaryotic assemblage where relatively few taxa are abundant and thermotolerant, but a large number of rare and low abundance taxa display little relationship with the thermal gradient and may therefore represent functionally less active or inactive taxa. The importance of rare taxa in microbial communities has nonetheless been recognised for their role as a reservoir of diversity that assists with resilience to environmental change [89, 90]. In the case of hot springs where turnover of cells is likely to be high due to the thermal environment, we interpret signatures from more labile cells that do not form resistant resting structures as genuine, whilst for others such as spore-producing fungi we believe further triangulation is required, which we addressed using metatranscriptomics. Bacteria and microbial eukaryotes displayed putative biotic interactions We sought to estimate the potential for microbial interactions between microbial eukaryotes and the bacteria and Archaea in the biofilms, as they may reveal putative predator–prey relationships or symbioses. We employed a stringent network co-occurrence analysis to indicate potential associations between all taxa, which were first visualized at phylum level for the large taxonomic dataset (Fig. 6A). Overall there were seven major interaction modules for the metacommunity and five of these included eukaryotic taxa. Several eukaryotic phyla displayed significant positive co-occurrence with each other indicating potential inter-dependence, e.g., the grazing Cercozoa (formerly Rhizaria) positively correlated to photosynthetic Chlorophyta, Rhodophyta and Streptophtyta taxa. The Cercozoa have been shown to regulate microbial turnover via grazing in soil habitats [91], and we envisage a similar role in hot spring biofilms. Other eukaryotic phyla with taxa sharing similar lifestyle also co-occurred, e.g., Ascomycota and Basidiomycota, Ciliophora and Evosea. All putative cross-domain interactions were between bacteria and eukaryota or bacteria and archaea. No archaea-eukaryota correlations were observed and this likely reflects that Archaea comprise a minor component of photosynthetic biofilms in hot springs [15]. The strongest inter-domain interaction was indicated between the heterotrophic bacterial Actinomycetota with eukaryotic Euglenozoa and yeasts (Ascomycota and Basidiomycota). This may reflect that these groups have been implicated as important in biofilm formation [92, 93]. Overall we interpret the interactions between eukaryotes as largely reflecting shared habitat or feeding preferences. The inter-domain interactions are more suggestive of a trophic interaction between producers and consumers, as envisaged for soil [94], and marine habitats [95].Fig. 6Putative biotic interactions inferred from network co-occurrence analysis. A Network for all taxa visualized at Phylum level, where N = 75, E = 76, Density = 0.027, Average degree = 2, Modularity = 0.617, Clustering coefficient = 0.184. Sub-networks for each module that illustrate specific interactions named according to the most abundant eukaryotic group: B Euglenids, where N = 330, E = 2475, Density = 0.046, Average degree = 15, Modularity = 0.269, Clustering coefficient = 0.088. C Yeasts, where N = 500 (cutoff at 500 for better visualization, actual N = 1200), E = 5025, Density = 0.040, Average degree = 20, Modularity = 0.278, Clustering coefficient = 0.114. D Diatoms, where N = 136, E = 484, Density = 0.042, Average degree = 7.1, Modularity = 0.514, Clustering coefficient = 0.180. E Algae, where N = 314, E = 1903, Density = 0.039, Average degree = 12.1, Modularity = 0.329, Clustering coefficient = 0.173. F Ciliates / slime-molds, where N = 34, E = 52, Density = 0.091, Average degree = 3.1, Modularity = 0.594, Clustering coefficient = 0.436. Larger node sizes indicate greater number of connections, network hubs are highlighted in black edging, green connectors indicate positive correlations and red connectors indicate negative correlations Putative biotic interactions inferred from network co-occurrence analysis. A Network for all taxa visualized at Phylum level, where N = 75, E = 76, Density = 0.027, Average degree = 2, Modularity = 0.617, Clustering coefficient = 0.184. Sub-networks for each module that illustrate specific interactions named according to the most abundant eukaryotic group: B Euglenids, where N = 330, E = 2475, Density = 0.046, Average degree = 15, Modularity = 0.269, Clustering coefficient = 0.088. C Yeasts, where N = 500 (cutoff at 500 for better visualization, actual N = 1200), E = 5025, Density = 0.040, Average degree = 20, Modularity = 0.278, Clustering coefficient = 0.114. D Diatoms, where N = 136, E = 484, Density = 0.042, Average degree = 7.1, Modularity = 0.514, Clustering coefficient = 0.180. E Algae, where N = 314, E = 1903, Density = 0.039, Average degree = 12.1, Modularity = 0.329, Clustering coefficient = 0.173. F Ciliates / slime-molds, where N = 34, E = 52, Density = 0.091, Average degree = 3.1, Modularity = 0.594, Clustering coefficient = 0.436. Larger node sizes indicate greater number of connections, network hubs are highlighted in black edging, green connectors indicate positive correlations and red connectors indicate negative correlations The five modules displayed taxa-specific eukaryota-bacteria interactions with contrasting patterns of interaction (Fig. 6). The “euglenids” module was dominated by interactions between Actinobacteria and photosynthetic Euglenozoa (Discoba), with both groups contributing hub taxa to the network (Fig. 6B). Other euglenids in the network were trypanosomatids that infect insect and plant hosts, making it unclear whether they represented authentic network participants. The “yeasts” dominated module (Fig. 6C) was the most complex and included the most taxa. Hub taxa comprised a number of heterotrophic bacteria and also included the basidiomycete yeasts Rhodotorula and Moesziomyces. All eukaryotic taxa in the network were fungi (Amorphea) and were mostly ascomycetous or basidiomycetous yeasts including some known from other extreme environments, e.g. Cutanitrichosporum, Exophilia, Rhodotorula, Talaromyces [19, 70, 96]. The “diatom” dominated module network (Fig. 6D) included a large number of known thermophilic heterotrophs and autotrophic sulfur-dependent bacteria. All module hubs were eukaryotic diatoms (Bacillariophyta, TSAR): Acanthoceras, Craticula, Encyonema, Petrodictyon, Psammodictyon, Stauroneis, and the nano-diatom Minutocelus. Other eukaryotes in the network included diverse diatoms and several Mucroromycota fungi (Amorphea) that displayed positive co-occurrences with diatoms. The “algae” dominated module (Fig. 6E), included a large number of taxa and included both bacterial and eukaryotic hub taxa. The bacteria hubs included several sulfur-dependent thermophiles, and the eukaryotes were Chlorophyta algae (Archaeplastida): Aegagropila, Coelastrella, Micrasterias Pedobesia, Phacotus; plus the phototrophic Cercozoan Paulinella (Cercozoa, TSAR) that has endosymbiotic chromatophores, and the Rhodophyta alga Rhodella (Rhodophyta, Archeaplastida). The ciliates/slime molds module (Fig. 6F) comprised predatory Ciliophora (TSAR) and Evosea (Amorphea). Strong positive co-occurrences were observed for three slime mould-ciliate pairings of taxa, e.g. Dictyostelium with Tetrahymena, Diderma with Halteria, and Rostrostelium with Blepharisma, Cavendaria, and Protostelium. Another interacting group also comprised several ciliates and the slime moulds Didymium. Interactions between eukaryotes and bacteria in this network were largely negative. Microbial eukaryotes are putatively metabolically active in hot spring biofilms As with all eDNA-based surveys, the detection of some taxa that produce spores or other dormant or inactive life stages may confound interpretations of ecological distribution. To address this, we performed metatranscriptomic sequencing on a subset of samples from the Singapore location (reflecting a broad temperature range from 46 to 61 °C) to identify microbial eukaryotic taxa with evidence of putative metabolic activity (Fig. 7). We compared total abundance and relative abundance (i.e., RNA:DNA ratio) of taxonomically assigned transcripts as a broad approximation of taxa that were metabolically active [97]. The data largely mirrored our metagenomic diversity estimates although some key differences were observed. Overall, the Ascomycota and Basidiomycota accounted for the vast majority of transcripts (Fig. 7A). This adds further evidence that fungi are important heterotrophic components in hot spring biofilms, and this is consistent with their proposed role as key players in extreme ecosystems [19]. Less abundant but nonetheless metabolically active zoosporic fungi from the Chytridiomycota represent new records for geothermal habitats, also the Zoopagomycota (formerly Zygomycota) indicated an unusual group of fungal taxa that can exhibit a predatory lifestyle on amoebas (Supplementary Dataset SD5) [98].Fig. 7Relative abundance of metabolically active taxa. A Relative abundance of taxa aggregated at phylum level in metagenomes and metatranscriptomes from a subset of samples (Singapore, 46–61 °C). B Potentially metabolically active taxa were estimated by calculating the RNA:DNA ratio for taxa at each temperature-defined interval. Metatranscriptomes and the corresponding metagenomes read processing data is shown in Supplementary dataset SD8 Relative abundance of metabolically active taxa. A Relative abundance of taxa aggregated at phylum level in metagenomes and metatranscriptomes from a subset of samples (Singapore, 46–61 °C). B Potentially metabolically active taxa were estimated by calculating the RNA:DNA ratio for taxa at each temperature-defined interval. Metatranscriptomes and the corresponding metagenomes read processing data is shown in Supplementary dataset SD8 Fourteen of the seventeen eukaryotic phyla recovered in metagenomes were present in transcriptomes and this suggest that many of them may be metabolically active. The highest RNA:DNA ratios occurred for photosynthetic algae (Rhodophyta), although the heterotrophic Cercozoa, Ciliophora, and Endomyxa (formerly Rhizaria), and to a lesser extent the Evosea, comprised a group of particularly active taxa that potentially represent a trophic level that predates/grazes other biofilm taxa, as observed for biofilms in other habitats [94, 95]. Activity as indicated by RNA:DNA was highest at temperatures ≤ 55 °C, and at 61 °C the dramatic drop in RNA:DNA suggests several microbial eukaryotes may have reached their thermal maximum for activity, which is consistent with other studies [5]. It is important to recognise that the approach we used to compare relative abundance of transcripts without mapping to individual genes yielded only an approximation of activity and so we make claims about relative activity between taxa with caution. Nonetheless the detection of significant levels of transcribed mRNA relative to DNA for a given taxon may be taken as a general proxy for metabolically active taxa. Conclusion This study provides a broad characterization of microbial eukaryotic diversity, abundance, and putative activity in hot spring biofilms across East and Southeast Asia, revealing that although they comprise a small proportion of overall abundance, eukaryotic microorganisms contribute notably to taxonomic richness. Metagenomes indicated diverse microbial eukaryotes that spanned putative photosynthetic, saprotrophic, and predatory lifestyles. Overall, the yeasts (Ascomycota and Basidiomycota) were the most prevalent and abundant eukaryotic taxa. The recovery of some taxa not typically associated with high-temperature environments highlights the need for careful interpretation of DNA-derived data in thermally dynamic systems, particularly where passive introduction or resting stages may contribute to detected diversity. Temperature emerged as a key environmental correlate shaping both diversity and inferred activity of microbial eukaryotes, with transcript-based analyses suggesting a general decline in activity above ~ 55 °C. While fungi accounted for most eukaryotic transcripts, photosynthetic and predatory protists also exhibited elevated RNA:DNA ratios, implying active participation in primary production and grazing. Co-occurrence analyses further suggested potential interactions between eukaryotic and bacterial taxa, particularly between yeasts, algae, and heterotrophic bacteria, though these require experimental validation. Overall, our findings underscore that microbial eukaryotes represent an underexplored but potentially functionally relevant component of hot spring biofilms. Further genome-resolved and experimentally focused investigations will be essential to elucidate their ecological roles, physiological limits, and contributions to microbiome structure and function under environmental stress. Supplementary Information Additional file1 (XLSX 3434 KB) Additional file1 (XLSX 3434 KB)