Cross-Kingdom Microbiome Profiling: Bacteria, Fungi, and Archaea in One Study

Microbial communities rarely function as bacteria-only systems. In host-associated and environmental ecosystems, bacteria, fungi, and archaea can participate in shared nutrient cycles, metabolic cross-feeding, competition, physical associations, and other ecological processes. A study that measures only one microbial domain may therefore capture only part of the community structure relevant to the biological question.

Cross-kingdom microbiome profiling is not simply the addition of fungal or archaeal sequencing to an existing bacterial workflow. Different organisms vary in cell-envelope structure, genome size, marker-gene copy number, biomass, extraction efficiency, and representation in reference databases. These differences influence both laboratory recovery and downstream interpretation. This guide provides microbial ecologists, agriscience researchers, biotechnology teams, and biomedical researchers with an endpoint-first framework for designing studies that integrate bacterial, fungal, and archaeal communities using parallel marker-gene sequencing, whole-metagenome shotgun sequencing, or reduced-representation approaches such as 2bRAD-M.

Key Takeaways

• Cross-Kingdom Profiling Starts with the Research Question: Researchers should define which microbial domains, taxonomic resolution, functional outputs, and quantitative endpoints are required before selecting a sequencing method.
• Multiple Technical Routes Are Available: Parallel bacterial 16S, fungal ITS, and archaeal marker assays provide domain-specific profiles; shotgun metagenomics provides broader genomic information; reduced metagenomics can provide another route to species-level multi-domain taxonomy.
• Cross-Domain Read Percentages Are Not Automatically Cell Ratios: Marker copy number, genome size, extraction recovery, amplification efficiency, and reference representation can all influence observed abundance.
• Extraction Must Be Validated Across the Target Organisms: No single enzymatic or mechanical lysis recipe is universally optimal for bacteria, fungi, and archaea. Recovery should be evaluated using appropriate controls and representative samples.
• Species-Level Taxonomy and Functional Metagenomics Are Different Deliverables: A method that identifies bacteria, fungi, and archaea at species level does not automatically provide unrestricted genes, pathways, mobile elements, or genome assemblies.
• Network Associations Require Cautious Interpretation: Composition-aware statistical methods can identify candidate cross-domain associations, but a network edge does not by itself demonstrate direct biological interaction or causality.
• Published 2bRAD-M Studies Support Difficult-Sample Feasibility: Benchmark experiments have included total DNA inputs down to 1 pg, degraded DNA, and separate high-host-background tests. These are published feasibility observations rather than universal sample guarantees.

Defining Cross-Kingdom Microbiome Profiling

Moving Beyond Bacteria-Centric Community Profiles

16S rRNA sequencing has provided a foundation for bacterial microbiome research, but bacteria represent only one component of many microbial ecosystems. Fungal communities can contribute distinct metabolic capabilities and structural interactions, while archaea can occupy specialized ecological niches such as methanogenic environments. The importance of each domain varies substantially by specimen type and biological system.

Published multi-kingdom studies illustrate why these domains can provide complementary research information. Liu et al. analyzed bacterial and fungal features across colorectal cancer research cohorts and reported improved classification performance when multi-kingdom information was incorporated compared with selected single-domain models (Liu et al., 2022). Su et al. similarly incorporated bacterial, archaeal, fungal, viral, and functional microbiome features in an autism spectrum disorder research cohort and developed multi-kingdom research classification models (Su et al., 2024).

These studies demonstrate that multi-domain information can add value in selected datasets, but they do not establish that every microbiome project requires every microbial kingdom or that multi-kingdom profiling will always improve classification. The appropriate scope should be determined by the ecosystem, research hypothesis, and expected biological contribution of each domain.

Why Study Design Must Precede Sequencing

Running separate assays and attempting to integrate them only after sequencing can create interpretation problems if the assays were not designed to answer the same biological question. Differences in sample allocation, extraction, amplification, sequencing depth, taxonomic resolution, and data normalization can make apparent cross-domain relationships difficult to interpret.

A cross-kingdom study should therefore establish, before laboratory processing, which domains must be measured, whether relative or quantitative abundance is needed, how negative and positive controls will be incorporated, which taxonomic level is required, and whether the downstream analysis focuses on community composition, ecological associations, or direct functional genes. For background on marker-based workflows, see our 16S/18S/ITS amplicon sequencing overview.

Define the Research Endpoint Before Choosing a Method

The phrase "cross-kingdom microbiome" can refer to several different experimental goals. A method suitable for one goal may be poorly matched to another.

Clarifying this endpoint prevents a common design problem: selecting a technology because it reports multiple kingdoms and later discovering that it does not provide the functional, quantitative, or interaction-level information required by the hypothesis.

Core Technical Routes for Multi-Kingdom Profiling

Route 1: Parallel Multi-Marker Amplicon Sequencing

A conventional multi-domain strategy uses separate marker assays for different microbial groups, such as bacterial 16S rRNA sequencing, fungal ITS sequencing, and an archaeal 16S assay. This approach can be efficient when the primary objective is community profiling within each domain. Relevant workflows include our bacterial 16S rRNA sequencing, fungal ITS sequencing, and archaeal 16S rRNA sequencing services.

The major limitation is that the assays are technically independent. Different primers, marker copy numbers, PCR efficiencies, sequencing depths, and reference databases mean that a bacterial 16S read and a fungal ITS read are not equivalent quantitative units. The resulting datasets can be analyzed jointly for patterns of presence, prevalence, or association, but raw percentages from separate amplicon assays should not be interpreted as direct bacteria-to-fungi cell ratios.

Route 2: Whole-Metagenome Shotgun Sequencing

Shotgun metagenomics sequences DNA across the complete extracted pool rather than amplifying a predefined marker. This allows one library to capture bacterial, fungal, archaeal, and other DNA-derived sequences while also providing direct information about microbial genes and pathways. DNA viruses may also be represented when viral DNA is present, although a standard DNA shotgun workflow should not be treated as a comprehensive method for RNA-virus profiling.

The major challenge is depth. If fungi or archaea constitute a small fraction of the total microbial DNA, or if host DNA dominates the library, a relatively small fraction of total reads may represent the organisms of interest. Required sequencing therefore depends on community composition, target prevalence, host background, genome size, and whether the intended endpoint is taxonomy, gene profiling, or genome reconstruction. There is no universal read count that guarantees comprehensive cross-kingdom recovery. For direct metagenomic analysis, explore our microbial metagenomics platform.

Route 3: Reduced Metagenomics via 2bRAD-M

2bRAD-M uses type IIB restriction enzymes to generate short genomic tags and analyzes a reduced representation of microbial genomes. Sun et al. introduced the method for species-resolved microbiome profiling and demonstrated simultaneous bacterial, archaeal, and fungal profiling using species-specific restriction-tag references (Sun et al., 2022).

The original study included low-input experiments using total DNA down to 1 pg, experiments with severely fragmented DNA, and separate high-host-background simulations. A later study specifically examined host-rich samples without prior host depletion and reported that differences in type IIB restriction-site representation between microbial and human genomes contributed to preferential microbial information recovery under the tested conditions (Jiang et al., 2025).

These results support 2bRAD-M analysis for microbiome research as a potential option when the main deliverable is species-level bacterial, fungal, and archaeal taxonomy from difficult material. The published low-input and high-host experiments should, however, be interpreted as feasibility benchmarks rather than universal acceptance specifications.

Pre-Experimental Design: Extraction Must Be Validated Across Kingdoms

There Is No Universal Cross-Kingdom Lysis Recipe

Cell-envelope structure varies substantially both among and within microbial kingdoms. Many bacteria differ in peptidoglycan structure and susceptibility to lysis. Fungal cells may contain chitin, glucans, and other resistant wall components. Archaeal envelopes are also diverse and can include S-layers or, in selected lineages, pseudomurein-containing structures.

For this reason, a cross-kingdom extraction workflow should be validated rather than assumed to recover every organism equally. Mechanical disruption such as bead beating can improve recovery of resistant cells, while enzymatic pretreatment may be useful for particular target communities. More aggressive disruption can also fragment DNA, so the optimal balance depends on downstream library requirements.

• Use representative biological material: Pilot samples should reflect the major matrices and expected microbial communities in the study.
• Evaluate resistant organisms: Positive controls or mock communities can help determine whether the extraction protocol systematically under-recovers particular cell types.
• Keep production extraction consistent: Changing extraction chemistry during a cohort can create artificial differences that resemble biological shifts.
• Include negative controls: Extraction blanks are especially important when one or more microbial domains are present at low biomass.
• Match disruption intensity to the downstream method: Efficient lysis should be balanced against excessive fragmentation when longer DNA molecules are required.

Cross-Kingdom Biomass Can Be Highly Unequal

The relative contribution of bacteria, fungi, and archaea varies widely among human-associated, plant-associated, soil, aquatic, fermented, and built-environment microbiomes. One domain may represent only a small fraction of the microbial DNA in a particular matrix. Study planners should therefore avoid assuming a fixed bacterial-to-fungal or bacterial-to-archaeal ratio.

For parallel amplicon assays, each domain is amplified independently, which can improve sensitivity for minority groups but prevents direct quantitative comparison between assays. In unified DNA-based approaches, sequencing or tag depth should be evaluated to determine whether minority-domain taxa are recovered reproducibly enough for the intended analysis.

Cross-Kingdom Quantification: Relative Abundance Is Not Absolute Abundance

A central challenge in multi-domain microbiome research is determining what an abundance value actually represents. Raw sequencing proportions should not automatically be interpreted as cell-number ratios across kingdoms.

Why Amplicon Counts Are Not Directly Comparable Across Domains

Bacterial 16S, archaeal 16S, and fungal ITS assays use different primers and amplification conditions. Marker copy numbers also differ among organisms, and PCR efficiency varies according to primer-template matching and amplicon properties. As a result, a sample containing 70% bacterial reads in one library and 30% fungal reads in a separate ITS library does not imply a 70:30 bacterial-to-fungal cell ratio.

Why Unified DNA Sequencing Still Requires Caution

Shotgun sequencing removes some marker-specific PCR effects, but DNA abundance is still influenced by genome size, cellular DNA content, extraction recovery, ploidy, host DNA, and sequencing efficiency. Reduced-representation approaches introduce additional dependence on the number and specificity of informative restriction tags represented in each genome.

Cross-kingdom relative abundance can therefore be highly useful for comparing the same taxon or domain across consistently processed samples, but direct biological interpretation of DNA read fractions as organism counts requires additional validation.

Options for More Quantitative Microbiome Measurements

Several complementary strategies can support quantitative interpretation:

• qPCR or Digital PCR: Domain- or taxon-specific assays can estimate microbial gene-copy abundance independently of sequencing relative abundance.
• Cell Counting or Flow Cytometry: Appropriate samples may permit direct estimates of total microbial cell concentration, which can be combined with sequencing composition.
• Whole-Cell Spike-In Controls: Adding a known quantity of an exogenous organism before extraction can capture some variation introduced during lysis, extraction, library preparation, and sequencing.
• Synthetic DNA Spike-Ins: DNA standards can help normalize downstream DNA recovery or sequencing processes, but a DNA standard added after cell lysis cannot measure cell-lysis efficiency and should not automatically be converted into absolute cell counts.

No single quantitative strategy corrects every source of cross-kingdom bias. The appropriate method depends on whether the desired unit is DNA copies, genome equivalents, cells, biomass, or relative ecological composition.

Analytical Architecture: Designing Cross-Kingdom Data for Valid Comparison

Do Not Treat Cross-Kingdom Tables as Ordinary Independent Measurements

Microbiome abundance data are compositional: increasing the measured proportion of one feature can mathematically reduce the apparent proportion of others even when their underlying quantities do not change. This problem becomes especially important when bacterial, fungal, and archaeal features with very different prevalence distributions are combined into one analysis.

Log-ratio approaches such as Centered Log-Ratio (CLR) or Isometric Log-Ratio (ILR) transformations can support compositional analysis, but they do not automatically solve every problem. Zero values, detection limits, sparsity, prevalence filtering, batch effects, and covariate structure still require explicit consideration.

Association Methods Have Different Statistical Meanings

Simple Pearson or Spearman correlations applied directly to relative abundance tables can generate misleading associations. More specialized methods address parts of the compositional problem in different ways. For additional background on simpler paired bacterial-fungal analyses, see our guide to correlation analysis of 16S and ITS sequencing.

• SparCC: Estimates pairwise correlations from compositional microbial data under assumptions about network sparsity. It should be interpreted as an association method rather than a direct conditional-dependence model.
• SPIEC-EASI: Uses sparse graphical-model approaches to estimate conditional relationships among microbial features under its modeling assumptions.
• FlashWeave: Uses graphical-model approaches designed to infer candidate direct associations and can incorporate heterogeneous metadata in appropriate workflows.

The choice among these methods should reflect sample size, sparsity, repeated measures, covariates, and the network question rather than being based solely on software popularity. More extensive network methodology is covered in our microbial community network analysis guide.

A Network Edge Is Not Proof of Biological Interaction

A positive association between a fungus and bacterium may arise because they directly interact, respond similarly to the same environmental variable, occupy the same niche, or are indirectly linked through other community members. A negative association may likewise reflect competition, environmental filtering, or compositional structure rather than direct antagonism.

Highly connected taxa can be described as candidate network hubs, but statistical centrality alone does not establish that an organism is an ecological keystone or that it stabilizes the community. Such hypotheses require follow-up evidence from longitudinal analysis, perturbation experiments, co-culture systems, spatial measurements, metabolite data, or other mechanistic approaches. The interpretive limitations of cross-kingdom co-occurrence networks are also discussed by Lee et al. in their plant microbiome review (Lee et al., 2022).

What Each Cross-Kingdom Strategy Cannot Tell You

This distinction helps prevent taxonomic, functional, quantitative, and ecological claims from being conflated. A study can successfully produce species-level multi-kingdom profiles without directly measuring metabolic function, absolute cell counts, or physical interactions between organisms.

A Practical Cross-Kingdom Study Design Workflow

Step 1: Define the Domains and Primary Endpoint

Decide whether the project needs bacteria plus fungi, bacteria plus archaea, all three domains, or an even broader virome or eukaryotic component. Specify whether the primary outcome is diversity, species abundance, candidate associations, quantitative abundance, functional genes, or ecological networks.

Step 2: Evaluate Sample Biology

Consider expected microbial biomass, host background, DNA integrity, inhibitors, and whether one domain is expected to occur at substantially lower abundance than the others. These factors can influence extraction and sequencing sensitivity.

Step 3: Select the Sequencing Architecture

Use parallel marker assays when separate domain-specific community profiling is sufficient. Consider shotgun sequencing when direct genes and broader genome information are required. Consider reduced-representation species profiling when taxonomy is the principal deliverable and sample properties make unrestricted whole-metagenome sequencing inefficient.

Step 4: Validate Extraction and Controls

Use representative pilot samples and appropriate controls to evaluate whether the extraction workflow consistently recovers the target domains. Incorporate negative controls, relevant positive controls, and batch tracking into the production design.

Step 5: Predefine the Quantification Strategy

Determine whether the analysis will use domain-specific relative abundance, unified DNA sequence proportions, qPCR/dPCR measurements, cell-based normalization, or spike-in standards. This decision should be made before interpreting cross-domain abundance differences.

Step 6: Design the Statistical Analysis Before Network Construction

Establish prevalence filtering, zero handling, normalization, covariate adjustment, and multiple-testing strategies before exploring cross-kingdom associations. Network modeling should follow the statistical properties of the data rather than being added after results are observed.

Summary: Build One Study Design, Not Three Independent Microbiome Assays

Cross-kingdom microbiome profiling is most informative when bacterial, fungal, and archaeal measurements are designed as components of the same biological experiment. Simply generating three sequencing datasets does not guarantee that they can be quantitatively compared or interpreted as one ecosystem.

• Start with the endpoint: Decide whether the study requires taxonomy, quantitative abundance, direct functional genes, or association networks.
• Select the method around the endpoint: Parallel marker assays, shotgun metagenomics, and reduced metagenomics provide different types of information.
• Validate recovery across organisms: Extraction bias can distort multi-domain profiles before sequencing begins.
• Separate relative abundance from absolute abundance: Raw sequencing proportions are not automatically microbial cell ratios.
• Use network analysis as hypothesis generation: Statistical associations should be followed by biological validation when mechanistic interaction is central to the conclusion.
• Recognize method boundaries: Species-level multi-kingdom taxonomy does not automatically provide strain genomes, unrestricted genes, metabolic activity, or causality.

The strongest cross-kingdom study therefore aligns biological scope, extraction strategy, sequencing architecture, quantitative framework, and statistical interpretation before the first sample enters the laboratory. For broader community-level sequencing options, explore our microbial diversity analysis solutions.

Frequently Asked Questions (FAQ)

References

1. Liu NN, Jiao N, Tan JC, et al. Multi-kingdom microbiota analyses identify bacterial–fungal interactions and biomarkers of colorectal cancer across cohorts. Nature Microbiology. 2022;7(2):238–250. [DOI: 10.1038/s41564-021-01030-7]
2. Su Q, Wong OWH, Lu W, et al. Multikingdom and functional gut microbiota markers for autism spectrum disorder. Nature Microbiology. 2024;9(9):2344–2355. [DOI: 10.1038/s41564-024-01739-1]
3. Sun Z, Huang S, Zhu P, et al. Species-resolved sequencing of low-biomass or degraded microbiomes using 2bRAD-M. Genome Biology. 2022;23:36. [DOI: 10.1186/s13059-021-02576-9]
4. Jiang et al. High-resolution microbiome analysis of host-rich samples using 2bRAD-M without host depletion. npj Biofilms and Microbiomes. 2025. [DOI: 10.1038/s41522-025-00851-2]
5. Lee KK, Kim H, Lee YH. Cross-kingdom co-occurrence networks in the plant microbiome: Importance and ecological interpretations. Frontiers in Microbiology. 2022;13:953300. [DOI: 10.3389/fmicb.2022.953300]
6. Kurtz ZD, Müller CL, Miraldi ER, et al. Sparse and Compositionally Robust Inference of Microbial Ecological Networks. PLoS Computational Biology. 2015;11(5):e1004226. [DOI: 10.1371/journal.pcbi.1004226]

For Research Use Only (RUO). Not for use in diagnostic procedures.