Materials and methods

Sample collection. In July 2022, we collected representative Utricularia jamesoniana plants from the Robert and Catherine Wilson Botanical Garden, part of Las Cruces Biological Station (LCBS) owned by the Organization for Tropical Studies (OTS). LCBS is in southern Costa Rica in the Coto Brus county of Puntarenas Province, ca. 5 km away in a straight line from the Panamanian border and 4 km in a straight line to the SW of the town of San Vito, coordinates 8° 47' 9.54" N 82° 57' 37.49" W (Figure 2). The Garden is at about 1,200 m on the Fila Cruces, part of the Fila Costeña Sur, located south of the Coto Brus Valley. The Garden is in the Premontane Rain Forest Life zone, according to ^{Tosi (1969)}, and is surrounded by young and mature secondary forests. The Utricularia jamesoniana specimens we collected were not planted; instead, they were found naturally growing as epiphytes (or facultatively epilithic) among dense bryophyte mats on a cultivated Socratea exorrhiza (Mart.) H.Wendl. palm. The mat containing three U. jamesoniana specimens, separated by a few centimeters, was carefully collected onto a polyethylene bag. Then, in order to keep the specimens as intact as possible, the entire mat was then transported to the laboratory for processing on the same day. Once in the laboratory, with gloves and under aseptic conditions, two specimens were separated for this study and the third was used as a voucher for the herbarium. The voucher was adequately preserved in 70 % ethanol solution in the spirit collection of the USJ herbarium (Acuña et al. 3183). The plant collection was carried out with the respective permits of the authorities (collection License ACLAP-077-2021) and the official permission of the Biodiversity Commission of the Universidad de Costa Rica (CBio-73-2022, resolution # 371). To augment the sample size, extensive efforts were undertaken to procure additional plant specimens, both from the original collection site and alternative locations. Regrettably, it was not possible to find more specimens. In consideration of the preservation of Utricularia jamesoniana populations, we opted to conduct the endosymbiont characterization exclusively with the two extant samples at our disposal.

Figure 2 The geographical location of sampling point in Las Cruces Biological Station in southern Costa Rica. Utricularia jamesoniana was found growing as an epiphyte among dense bryophyte mats. The collected plants grew on the mossy stilt roots of mature, cultivated Socratea exorrhiza palm. 

Molecular analyses. Plant samples were washed thoroughly in running water, and healthy and physically undamaged tissues (leaves and bladders) were selected and sterilized on the surface with 70 % ethanol and aseptically cut. The DNA was extracted from approximately 100 mg of the plant tissues using a DNA isolation kit (PowerSoil, Qiagen, USA) following the manufacturer's instructions. PCR, library preparation, and sequencing were performed by Novogene (Novogene Bioinformatics Technology Co., Ltd, CA, USA). Briefly, this process included the prokaryotic amplicon library of the V4 hypervariable region of the 16S rRNA gene using universal primers 515F and 806R (^{Caporaso et al. 2011}). The eukaryotic amplicon library was generated by amplifying the ITS2 region using primers ITS3-2024F (5' GCATCGATGAAGAACGCAGC) and ITS4-2409R (5' TCCTCCGCTTATTGATATGC). PCR amplification of targeted regions was performed using specific primers connecting with barcodes. The PCR conditions for both markers consisted of 95 °C for 3 min initial denaturation followed by 35 cycles at 95 °C for 45 s, 53 °C for 1 min, 72 °C for 1 min, and a final extension at 72 °C for 5 min. The PCR products with the proper size were selected by 1 % agarose gel electrophoresis. Each sample's exact amount of PCR products was pooled, end-repaired, A-tailed, and further ligated with Illumina adapters. Libraries were sequenced on a paired-end Illumina Novaseq-6000 platform to generate 250bp paired-end raw reads.

Biofinformatic analyses. We used the DADA2 version 1.21 to process the Illumina-sequenced paired-end fastq files and to generate a table of amplicon sequence variants (ASVs), which are higher-resolution analogs of the traditional OTUs (^{Callahan et al. 2016}). Briefly, we removed primers and adapters, inspected the quality profiles of the reads, filtered and trimmed sequences with a quality score < 30, estimated error rates, modeled and corrected amplicon errors, and inferred the sequence variants. Then, we merged the forward and reverse reads to obtain the full denoised sequences, removed chimeras, and constructed the ASV table. We assigned taxonomy to the ASVs with the function assignTaxonomy, of DADA2. For the prokaryotic assignment, we used as input the SILVA reference database version 138.1 (^{Quast et al. 2013}). We carried out a second taxonomic assignment of the prokaryotic ASVs using the tool IDTAXA of DECIPHER (Murali et al. 2018) with the same version of the SILVA database and using the RDP database version 18 (rdp.cme.msu.edu). For the fungal taxonomic assignment, we used the UNITE ITS database version 8.3 (https://doi.org/10.15156/BIO/1280049). We carried out a second taxonomic assignment of the fungal ASVs using the tool IDTAXA of DECIPHER with the same version of UNITE as a reference and a confidence threshold > 60 %, and additionally, a third classification using the Classifier tool (^{Wang et al. 2007}) implemented in the Ribosomal Database Project (rdp.cme.msu.edu) using as reference the Warcup Fungal ITS trainset V2 database (^{Deshpande et al. 2015}). We verified and manually curated the consistency between the taxonomic assignments of the different programs for prokaryotes and fungi. In cases of discrepancies, a comparison with the BLAST tool of NCBI Genbank was applied. Sequence data were deposited at the NCBI Sequence Read Archive under accession number PRJNA1005337 (https://www.ncbi.nlm.nih.gov/sra/PRJNA1005337). In the 16S rRNA prokaryotic dataset, we removed sequences assigned to Chloroplast and Eukarya. This process generated 518,491 sequences from the four samples. The average number of sequences per sample was 129,623 (ranging from 111,131 to 149,377). In the ITS2 eukaryotic dataset, we removed sequences assigned to vertebrates and plants. This process generated 283,135 sequences from the samples. The average number of sequences per sample was 70,784 (ranging from 57,195 to 94,867).

Statistical analyses. The statistical analyses and visualization of results were performed with the R statistical program (^{R Core Team 2023}) and the R studio interface. Package Vegan v. 2.6-4 (^{Oksanen et al. 2022}) was used to calculate alpha diversity estimators and non-metric multidimensional scaling analyses (NMDS). Data tables with the amplicon sequence variants (ASV) abundances were normalized into relative abundances and converted into a Bray-Curtis similarity matrix. To determine significant differences between the bacterial and fungal community composition according to the plant tissue, we used the non-parametric multivariate analysis of variance (Permanova) with adonis2 function and 999 permutations. To estimate differences in the relative proportion of microorganisms and the diversity indices, we used the non-parametric Kruskal-Wallis test, while pairwise comparisons were performed using Wilcox and Dunn's tests (with Bonferroni adjustment). The map of the sampling point was generated with ggplot2 and sf libraries using the world database.

Results

In this work, we compared the prokaryotic and fungal communities in the leaves and bladders of the carnivorous epiphytic plant Utricularia jamesoniana. The prokaryotic community of the samples was composed of 2,258 amplicon sequence variants (ASVs), according to the analysis of sequences of the V4 region of the 16S rRNA gene. All the prokaryotic sequences were assigned to 44 phyla and 87 classes. Proteobacteria were the most abundant, representing 66 % of all sequences and 37.5 % of the ASVs. Actinobacteriota represented 11.3 % of the sequences and 9.3 % of the ASVs, and Acidobacteriota represented 3.9 % of the sequences and 5.6 % of the ASVs. Other abundant groups were Firmicutes, Verrucomicrobiota, Planctomycetota, and Cyanobacteria. Archaea represented only 0.28 % of the sequences and 0.56 % of the ASVs.

The fungal community was composed of 1,620 amplicon sequence variants, according to the analysis of sequences of the ITS region. All the fungal sequences were assigned to 10 phyla and 32 classes. Ascomycota was the most abundant phylum comprising 79,6 % of the sequences and 78.9 % of the ASVs, while Basidiomycota represented 20.2 % of the sequences and 19.3 % of the ASVs. Other less abundant fungal phyla included Rozellomycota and Chytridiomycota.

The fungal classes with the highest relative abundance were Dothideomycetes, Sordariomycetes, Eurotiomycetes, and Agaricomycetes. Within Dothideomycetes, the most abundant ASVs were unclassified members of Mycosphaerellaceae and Cladosporiaceae, while Colletotrichum, Aspergillus, and Thanatephorus were the most abundant genera within Sordariomycetes, Eurotiomycetes, and Agaricomycetes, respectively.

The bacterial classes with the highest relative abundance were Alphaproteobacteria and Gammaproteobacteria. The most abundant genera within Alphaproteobacteria were Acidocella, Bradyrhizobium, and Rhizobium. The most abundant genera within Gammaproteobacteria were Ferritrophicum, Ferrovum, and Chromobacterium. In general, we observed differences in the composition of some particular groups of the prokaryotic communities between the leaves and traps (Figure 3A).

Figure 3 Relative abundance of the main taxonomic groups in bladders and leaves of Utricularia jamesoniana. A corresponds to the main prokaryotic groups, and B to the main fungal groups. 

We also observed variations in the composition of the fungal communities within samples and between tissues. Notably, we detected a higher proportion of Lecanoromycetes, Leotiomycetes, Orbiliomycetes, and Sordariomycetes in the bladder, while Dothideomycetes and Eurotiomycetes were more abundant in the leaves. We detected 25 and 30 classes of fungi in the bladder and leaf tissues, respectively. Chytridiomycetes were detected only in the bladder, where they have been found actively growing in aquatic Utricularia species (^{Sirová et al. 2018b}). Conversely, some classes were only present in the leaves, including Agaricostilbomycetes, Saccharomycetes, and Rhizophydiomycetes (Figure 3B).

We observed notable differences in the alpha diversity indices of prokaryotic communities in the tissues of Utricularia jamesoniana. The richness of prokaryotic ASVs has higher in the bladders with respect to the leaves, with mean richness values of 1,174 and 356, respectively. On the other hand, the mean value of the Shannon diversity index in the traps was also higher than that of the leaves, with mean values of 6.0 and 4.3, respectively (Figure 4). We also observed differences in the indices of fungal communities. However, the mycobiome was more diverse in the leaves than in the traps. The richness of fungal ASVs has higher in the leaves with respect to the traps, with mean richness values of 652 and 433, respectively. On the other hand, the mean value of the Shannon diversity index in the leaves was also higher than that of the traps, with mean values of 4.2 and 3.7, respectively (Figure 4).

Figure 4 Alpha diversity values of prokaryotic and fungal communities in the bladder and leaves of Utricularia jamesoniana. The upper panel represents the richness of Amplicon Sequence Variants (ASVs), while the panel below shows the values of the Shannon Index. 

The NMDS analysis of the bacterial communities showed that the trap samples clustered together, and their community composition separated from the plant leaves (Figure 5A). However, the Permanova showed no significant differences in the structure of prokaryotic communities between tissues (Permanova > 0,05). The analysis of the fungal communities showed that the composition in both bladders and leaves tissues in one of the plants clustered together, whereas they separated in the second plant (Figure 5B). The Permanova showed no significant differences in the structure of prokaryotic communities between tissues (Permanova > 0.05).

Figure 5 NDMS analysis of the microbial communities in bladders and leaves of Utricularia jamesoniana. A corresponds to the prokaryotic groups, and B to the fungal groups. 

The heat map shows the variations in the relative abundances of the sample's most abundant genera of prokaryotes. In general, most genera presented relative abundances of less than 5 %. However, some of the most abundant genera in the bladder were Bradyrhizobium, Ferritrophicum, Acidocella, Methylocapsa, and Mycobacterium, while in the leaves, they were Acidocella, Ferrovum, Methylocapsa, Rhizobium, and Gallionella. Acidocella and Methylocapsa (Figure 6A).

Figure 6 Heatmap of the most abundant genera in the bladder and leaves tissues of Utricularia jamesoniana. A corresponds to the prokaryotic genera determined according to the relative abundance of 16S rRNA gene sequences, and B to the fungal genera determined according to the relative abundance of ITS sequences. 

Our results confirm the previously observed differences in fungal diversity within and between samples regarding the most prevalent fungal genera across the tissue samples (Figure 6B). In particular, we identified several fungal genera that were highly abundant in traps, including Thanatephorus, Clavatospora, and Phaeothecoidiella. In contrast, leaves' most prevalent fungal genera were an unclassified member of the Mycosphaerellaceae, Aspergillus, and Phaeococcomyces.

Discussion

We elucidate the distinctive differences in prokaryotic and fungal communities between the traps of Utricularia jamesoniana and its leaves. The specialized bladders or traps constitute a singular ecological niche that harbors a multitude of microbial communities. These communities potentially exert a significant influence on plant fitness by participating in prey digestion and facilitating environmental adaptation (^{Caravieri et al. 2014}, ^{Sirová et al. 2018a}). Notably, the resident microbiome within these traps may undergo modifications due to the introduction of microbial communities from captured prey, as well as the assimilation of metabolizable compounds derived from the prey. It is pertinent to recognize that each prey item might harbor its own distinct microbial diversity, as well as inherent chemical characteristics, thereby introducing an additional layer of complexity to the trap's microbiome composition.

Rhizobial bacteria have been reported as one of the most abundant groups in some aquatic Utricularia species, while Acidocella and Bradyrhizobium have also been reported in the traps of different carnivorous plants (^{Grothjan & Young 2022}, ^{Sickel et al. 2019}). The predominance of Bradyrhizobium, a nitrogen-fixing bacterium (^{Zahran 1999}), was particularly conspicuous within the traps. Nevertheless, uncertainty persists regarding the nitrogen-fixing capability of this species within a non-root nodule, such as the bladder. On the other hand, it is plausible that Bradyrhizobium might serve additional functions, such as metabolizing different carbon sources (^{VanInsberghe et al. 2015}).

Furthermore, the traps exhibited a notable abundance of metal-oxidizing bacteria such as Acidocella, Ferritrophicum, Ferrovum, and Chromobacterium which suggests the occurrence of these processes within this plant tissue (^{Johnson et al. 2014}, ^{Okamoto et al. 2017}, ^{Weiss et al. 2007}). However, only Chromobacterium has been documented previously as a constituent of the microbiome associated with Utricularia, but in aquatic species (^{Grothjan & Young 2022}). This assemblage of microbial species suggests an environment conducive to stimulating microbial metabolism, leading to the release and recycling of nutrients (^{Sirová et al. 2003}, ^2018a).

The differences in the relative abundances of some bacterial and fungal groups between the trap samples may be potentially related to both variations in the preys captured over time and the current stage of prey decomposition. In addition, the interior of the trap is completely sealed and maintains an anoxic environment (^{Adamec 2007}) and hosting a complex microbial food web responsible for a significant proportion of enzymatic activity associated with prey digestion (^{Caravieri et al. 2014}, ^{Sirová et al. 2009}). The preservation of anoxic conditions within the traps potentially serves as the mechanism for incapacitating captured prey (^{Adamec 2007}).

Given this context, it stands to reason that the microbial community dwelling within these traps has adapted to this unique environment, which experiences intermittent spikes of elevated oxygen concentration subsequent to firing events, followed by periods of anoxia (^{Adamec 2007}). These fluctuating conditions create a niche favoring the coexistence of organisms that exhibit heightened tolerance to anoxia (^{Richards 2001}, ^{Sirová et al. 2003}), consequently contributing to prey degradation (^{Caravieri et al. 2014}). These conditions could also be responsible for the changes in the composition of the fungal communities in this tissue and the lower values of diversity observed.

Proteobacteria emerged as the predominant inhabitants within the Utricularia jamesoniana tissues. The prevalence and roles of this taxonomic group are intricately linked to the microenvironment provided by the plant tissue, its genetic composition, physiological attributes, as well as the biotic and abiotic factors inherent in the surrounding environment (^{Afzal et al. 2019}, ^{Brader et al. 2017}, ^{Hardoim et al. 2015}). Prior microbiome investigations associated with other carnivorous plants have consistently pinpointed Proteobacteria as the most abundant phylum (^{Alcaraz et al. 2016}, ^{Chan 2019}, ^{Sickel et al. 2019}).

Taxonomic parallels with earlier studies conducted on other aquatic Utricularia indicate the prevalence of Rhizobium as one of the most abundant genera within certain species. Likewise, Acidocella and Bradyrhizobium have been noted in the traps of diverse carnivorous plants (^{Grothjan & Young 2022}, ^{Sickel et al. 2019}). Among Gammaproteobacteria, Chromobacterium has also been described as part of the microbiome across several Utricularia species (Grothjan & Young 2022). To date, the majority of research efforts have focused on aquatic Utricularia species, including U. gibba L., U. australis R.Br., and U. vulgaris L. (^{Alcaraz et al. 2016}, ^{Sirová et al. 2018b}) characterized by their larger size and accelerated growth rates. Notably, this study represents a pioneering investigation into the composition of the microbiome and mycobiome associated with an epiphytic species such as Utricularia jamesoniana.

One of the most revealing results of this study was the elevated levels of prokaryotic species richness and Shannon diversity detected within the traps, constituting a digestive ecosystem (^{Adamec 2007}, ^{Sirová et al. 2018a}, ^b) where microorganisms face challenging living conditions typical of a digestive ecosystem. This intriguing observation prompts speculation that the increased richness and diversity of communities within the traps stem from the dynamic nature of this environment. In contrast, the conditions within the leaves can be more stable.

The diversity of the endophytic fungal community within the traps exhibited lower values compared to that observed in the leaves. This trend stands in contrast to the patterns observed within bacterial communities. Notably prevalent within this tissue were the genera Thanatephorus, Leohumicola, Clavatospora, Orbilia, and Phaeothecoidiella. However, the distinctive attributes and mechanisms underlying these genera's apparent compatibility within the host traps fluctuating anaerobic microenvironment demand further exploration. Particularly intriguing is the prevalence of Thanatephorus, known as the teleomorph of Rhizoctonia, a recognized plant pathogen (^{Anderson 1982}, ^{Donk 1956}), which was present even when no disease symptoms were observed on the plants examined.

In general, the phylum Ascomycota predominated in the tissues of Utricularia jamesoniana, with higher abundances of Dothideomycetes, Sordariomycetes, Eurotiomycetes, a pattern consistent with earlier findings in other carnivorous plants belonging to the same botanical family (^{Rueda-Almazán et al. 2021}). Particularly noteworthy is that a substantial proportion, approximately 45 %, of the fungal genera were present in both the traps and leaves. This observation underscores the notion that certain endophytic fungi can colonize diverse tissues within the same plant organism effectively (^{Behie et al. 2015}, Rueda-Almazán et al. 2021, ^{Wu et al. 2013}).

The interactions between plants and endophytes have been recognized for their myriad benefits to plants, particularly species that thrive in challenging or nutrient-deficient environments, as is the case with carnivorous plants (^{Ellison & Adamec 2018}, ^{Jobson et al. 2018}, ^{Rueda-Almazán et al. 2021}). However, it is crucial to highlight that not all identified fungi necessarily qualify as pure endophytes, as some could originate from external and potentially serve as nutrient sources (^{Sirová et al. 2018b}).

In the present investigation, we identified distinct groups of genuine endophytes, such as Aspergillus, Cladosporium, Colletotrichum, and Mycosphaerellacea. These groups have previously been isolated from both leaves and traps in other carnivorous plant species (^{Glenn & Bodri 2012}, ^{Lee et al. 2014}, ^{Naseem & Kayang 2018}, ^{Quilliam & Jones 2010}, ^{Rueda-Almazán et al. 2021}). However, some genera were exclusive to either the traps or leaves. Pseudochaetosphaeronema, Periconia, Remotididymella, and Phragmotaenium have also been described for the first time in an epiphytic carnivorous plant species. These findings underscore the distinctive biodiversity ecosystem hosted by Utricularia jamesoniana, justifying comprehensive further investigation.

Finally, the exploration of plant-microbial interactions serves as a window into the intricate web of ecological relationships spanning diverse lineages and their evolutionary trajectories. The dynamic composition of these microbial communities can fluctuate in response to a spectrum of biotic and abiotic factors that shape these interactions, where microorganisms undertake an array of functions, encompassing enzymatic activities such as prey digestion, fortification against pathogens, facilitation of growth, and protection (^{Hung & Rutgers 2016}, ^{Lee et al. 2014}, ^{Rueda-Almazán et al. 2021}). Examining the microbiome and mycobiome of carnivorous plants provides a captivating model that advances our understanding of plant ecology, evolution, and the intricate interplay between plants and their microbial inhabitants.

We contribute novel insights that enhance the comprehension of the microbiological community associated with Utricularia jamesoniana, shedding light on previously undocumented groups present within plant leaves and traps. For future investigations, it is imperative to incorporate a larger sample size to facilitate robust statistical analyses while exercising caution to prevent any adverse impact on the vulnerable populations of this delicate species. Furthermore, it is necessary to explore in more detail the physiological and biochemical nuances characterizing the traps. This entails discerning the intrinsic microbiome residing within the traps and disentangling it from the influx of nutrient-rich biological material. Additionally, there is a need to meticulously document any discernible shifts in microbial populations triggered by the plant's climatic seasonality and phenological state.

This study offers insight into the intricate microbial and fungal communities associated with traps and leaves of Utricularia jamesoniana. The distinctive biodiversity composition suggests plausible ecological connections between the plant tissues and the microbiome. These interactions and their potential entanglement with the fungal communities hold promise for illuminating the intricate host-endophyte dynamics. A more exhaustive microbiome characterization can unveil the plant's dietary preferences, nutrient cycling potential, and the spectrum of associated microorganisms. The description of several previously unreported fungal groups highlights the specific and under-explored nature of this microbial and fungal habitat. We emphasize the need for further investigation to unravel the microbiome and mycobiome of this epiphytic carnivorous plant, its intricate relations, and interaction dynamics.

Data accessibility

The datasets presented in this study can be found in the NCBI Sequence Read Archive under accession PRJNA1005337: https://www.ncbi.nlm.nih.gov/sra/PRJNA1005337.