Global Microbiome Research in Diverse Populations: Why Cohort Diversity Changes the Science

For Research Use Only. Not for use in diagnostic procedures.

When researchers talk about the human microbiome, the word "human" deserves more scrutiny than it usually gets. A 2024 analysis published in the Journal of Clinical Investigation put a hard number on a problem many in the field had sensed but never quantified: 71% of all publicly available human microbiome samples come from just 14% of the global population. The countries that contribute the most data — the United States, the United Kingdom, Denmark, and a handful of other wealthy nations — are home to a narrow slice of human genetic, dietary, and environmental diversity. The rest of the world, including most of Africa, South Asia, and Latin America, remains systematically underrepresented. This is not just an equity problem. It is a scientific one — and it changes what we think we know about the human microbiome.

The Numbers Behind the Bias

The scale of the sampling imbalance has come into sharp focus only recently. Blake and colleagues (2024) analyzed metadata from nearly 700,000 samples in public repositories and found a pattern that mirrors broader research disparities: the United States alone accounts for roughly 40% of all deposited microbiome samples, while the entire African continent contributes less than 3%. European and North American samples together dominate every major body-site catalog, from gut to skin to oral cavity.

A second landmark study, led by the Blekhman lab and published in Cell (2025), integrated 168,464 human gut metagenomes from 68 countries — the largest meta-analysis of its kind. The authors found that geographic origin explained as much variation in microbiome composition as technical variables like DNA extraction kit or sequencing platform. In other words, where a sample comes from is not a nuisance variable to be corrected away; it is a primary biological signal. Yet the bulk of the samples came from just a few countries, compressing the signal into a narrow band of human experience.

The Segata lab's global analysis of 32,000 metagenomes (Cell, 2025) pushed this further by identifying strain-level geographic structure across 583 gut microbial species. Some strains appeared almost exclusively in specific regions, meaning that a microbiome "reference" built from European donors may miss entire branches of the microbial tree that are common elsewhere.

Where Geography Leaves Its Mark

The idea that geography shapes the gut microbiome is not new — but the resolution at which we can now detect it is. The Segata lab's strain-level atlas revealed that for many species, geography is a stronger predictor of which strain a person carries than any individual-level trait. Two people living in the same region are more likely to share a bacterial strain than two people with the same diet living on different continents.

This geographic fingerprint has practical consequences. Microbiome-based diagnostic models trained on European or North American cohorts often lose accuracy when applied to populations from other regions. A classifier built to detect colorectal cancer from gut microbiome signatures in a French cohort, for instance, may fail in a South African cohort — not because the biology of cancer differs, but because the baseline microbial landscape against which the signal is measured looks entirely different.

Africa's underrepresentation has been especially costly. The largest dedicated effort to fill this gap, the African Gut Microbiome Atlas led by Maghini and colleagues (Nature, 2025), profiled 1,801 women across four countries — Burkina Faso, Ghana, Kenya, and South Africa — and found bacterial taxa and functional pathways not represented in existing reference catalogs. Some of these novel taxa carried genes involved in carbohydrate metabolism that may reflect diets rich in fibrous plant foods uncommon in Western populations. If these taxa and their metabolic capabilities go uncharacterized, our understanding of what the gut microbiome can do remains incomplete.

One of the earliest and most vivid examples of population-specific microbial function comes from a 2010 study by Hehemann and colleagues, published in Nature. The researchers discovered that the gut bacterium Bacteroides plebeius in Japanese individuals carried a gene for porphyranase — an enzyme that breaks down the sulfated polysaccharides found in nori, the seaweed used in sushi. This gene, which had transferred horizontally from marine bacteria, was virtually absent from North American gut metagenomes. The finding was not just a curiosity; it demonstrated that diet can drive the acquisition and retention of microbial functions in a population-specific way.

Diet, Ancestry, and Microbial Signatures

Diet is the most intuitive driver of population-level microbiome differences, but it is not the only one. Host genetics also plays a role — and here too, diversity matters. A 2022 study by Boulund and colleagues, published in Cell Host & Microbe, analyzed gut microbiome and host genotype data from 4,117 participants across six ethnic groups in the Netherlands (the HELIUS study). The researchers found that host genetic variants associated with specific microbial taxa in one ethnic group often showed no association — or a different one — in another group. In effect, microbiome-genome associations do not travel well across populations.

• Dietary fiber: Populations consuming traditional high-fiber diets (e.g., rural Burkina Faso, the Yanomami) harbor microbial taxa and carbohydrate-active enzymes rarely seen in Western cohorts.
• Host genetics: Microbiome-genome associations identified in European populations often fail to replicate in South Asian, African, or admixed cohorts, as shown by Boulund et al. (2022).
• Environmental exposures: The Yanomami, an isolated indigenous group in the Venezuelan Amazon, carry the highest bacterial and functional diversity ever recorded in the human gut — including antibiotic resistance genes in a population with no known antibiotic exposure (Clemente et al., Science Advances, 2015).
• Medications and lifestyle: Prescription drug use, sanitation infrastructure, and cooking fuel type vary dramatically between populations and each influences the gut microbiome, as demonstrated by Vujkovic-Cvijin et al. (Nature, 2020).

The Yanomami example is particularly instructive. When Clemente and colleagues characterized the gut microbiomes of this uncontacted Amerindian group, they found not only novel bacterial lineages but also functional genes absent from reference databases. The Yanomami gut carried nearly twice the microbial diversity of a typical U.S. resident. This finding raises a sobering question: if the Western lifestyle systematically depletes microbial diversity, and most of our reference data comes from Western populations, are we studying a diminished version of the human microbiome without realizing it?

Understanding these population-level differences is not just an academic exercise — it shapes how researchers design studies using tools like metagenomic shotgun sequencing and influences the interpretation of microbiome-disease associations. The field is learning that cohort composition is not a secondary consideration — it is a primary determinant of what a study can and cannot conclude.

What Underrepresentation Costs

The consequences of sampling bias ripple through every corner of microbiome science.

• Biomarker discovery is constrained. A microbial signature associated with type 2 diabetes in a Danish cohort may have little predictive value in an Indian cohort — not because the underlying biology is different, but because the microbial context in which the biomarker operates is population-specific. The practical result is that microbiome-based diagnostics developed in high-income countries may underperform everywhere else, reinforcing rather than reducing global health disparities.
• Therapeutic development starts from a narrow baseline. Probiotics, prebiotics, and microbiome-targeted drugs are typically tested in populations whose baseline microbiome composition reflects a Western, industrialized lifestyle. The absence of certain microbial lineages in these populations means that therapeutic strategies are optimized for a depleted ecosystem — and may behave unpredictably in populations with higher baseline diversity.
• Reference databases carry built-in blind spots. When a sequencing read cannot be mapped to any known genome, it is often discarded as noise. But reads that look like noise against a European-centric reference may represent real — and biologically important — microbial sequences from underrepresented populations. The African Gut Microbiome Atlas, for instance, recovered thousands of genes not present in the widely used Unified Human Gastrointestinal Genome (UHGG) catalog.
• Host-microbiome interactions are population-contingent. The Boulund et al. (2022) finding that mGWAS results vary by ethnicity means that genetic risk scores incorporating microbiome data — an area of active investigation — will require population-specific validation before they can be applied clinically.

Figure 1: Geographic distribution of human microbiome samples in public repositories. The United States and a small number of European countries contribute the vast majority of publicly available data, while Africa, South Asia, and Latin America remain severely underrepresented.

The field is beginning to reckon with these limitations. Researchers focused on microbiome study design now routinely ask whether their cohorts capture enough population diversity to support the conclusions they intend to draw. But asking the question is not enough — closing the gap requires deliberate effort.

Building a More Complete Map

Encouragingly, the map is expanding. Several large-scale initiatives are systematically filling the representation gap.

The African Gut Microbiome Atlas (Maghini et al., 2025), covering four countries and 1,801 women, is a foundational step — but its authors acknowledge that 1,801 individuals cannot represent the genetic, dietary, and environmental heterogeneity of an entire continent of 1.4 billion people. Scaling up is essential.

The microBiomap.org resource (Blekhman et al., 2025) provides a publicly accessible interface to 168,464 metagenomes from 68 countries, allowing researchers to query microbial prevalence and abundance by geography. This kind of resource transforms "where does this taxon live" from an anecdotal question into a data-driven one.

In Asia, large population cohorts are also under construction. The Chinese Gut Microbiome Reference, built from 247,134 metagenome-assembled genomes, represents a major contribution from a population that has historically been better sampled than many others but still lags behind European and North American cohorts in terms of per-capita representation.

Arif and Graham, writing in Trends in Microbiology (2025), offer practical guidance for researchers analyzing global microbiome data. They emphasize that statistical methods designed for well-balanced European cohorts may produce spurious results when applied to multi-population datasets with unequal representation. They also highlight the importance of data sovereignty — ensuring that communities whose microbiomes are sampled retain control over how their data is used and shared.

Closing the representation gap is not a one-time effort. As populations change — through migration, dietary shifts, urbanization, and antibiotic use — their microbiomes change with them. A diverse microbiome atlas built today may need updating a decade from now. The goal is not a static snapshot but a living, evolving reference that reflects the full spectrum of human microbial ecology.

Figure 2: Comparison of gut microbiome composition across diverse populations. Diet, host genetics, and environmental exposures produce distinct microbial signatures that challenge the generalizability of findings from single-population studies.

Terms That Shape the Debate

Some concepts recur throughout the global microbiome diversity literature. A shared vocabulary helps.

• WEIRD bias: An acronym — Western, Educated, Industrialized, Rich, Democratic — used to describe the overrepresentation of samples from wealthy, industrialized nations in biomedical research broadly, and microbiome research specifically.
• Strain-level geography: The observation that within a single bacterial species, different strains dominate in different geographic regions, as demonstrated by the Segata lab's global atlas.
• Microbiome-genome associations (mGWAS): The study of correlations between host genetic variants and gut microbiome composition. These associations are increasingly recognized as population-specific.
• Horizontal gene transfer (HGT): The movement of genes between bacteria, including between species. The B. plebeius–nori example demonstrates how HGT from environmental bacteria to gut microbes can confer diet-specific metabolic capabilities.
• Reference catalog bias: The systematic under-detection of microbial sequences from underrepresented populations when using reference databases built primarily from European and North American samples.
• Data sovereignty: The principle that communities contributing biological samples retain rights over how their data are stored, analyzed, and shared — a growing concern in global microbiome research.

Frequently Asked Questions

Why does population diversity matter for microbiome research?

A microbiome signature identified in one population may not replicate in another because diet, host genetics, environment, and baseline microbial composition all vary between populations. Without diverse cohorts, the field risks building diagnostic tools, therapeutic strategies, and reference databases that work well for a narrow slice of humanity but fail for most of the world's population.

What is the evidence that microbiome composition differs between populations?

Multiple lines of evidence now converge on this point. The Blekhman lab's integration of 168,464 metagenomes from 68 countries found that geography explains as much variation in microbiome composition as technical variables like DNA extraction kit. The Segata lab's strain-level analysis of 32,000 metagenomes showed that for many bacterial species, the strain a person carries is better predicted by geography than by any individual-level factor. Classic examples like the Japanese seaweed-digesting B. plebeius and the high-diversity Yanomami gut microbiome provide vivid illustrations.

How severe is the underrepresentation problem?

Blake et al. (2024) reported that 71% of publicly available microbiome samples come from just 14% of the global population. The United States alone accounts for roughly 40% of deposited samples, while the entire African continent contributes less than 3%. These numbers come from an analysis of nearly 700,000 samples in public repositories.

What initiatives are addressing the representation gap?

Several large-scale projects are working to close the gap. The African Gut Microbiome Atlas profiled 1,801 women across four African countries. The microBiomap.org resource provides queryable access to 168,464 metagenomes from 68 countries. The Chinese Gut Microbiome Reference adds 247,134 MAGs from Asian populations. The Segata lab's global strain atlas covers 32,000 metagenomes with geographic annotations. These are important steps, but the field acknowledges that much more work is needed — particularly in South Asia, Southeast Asia, the Middle East, and Latin America.

What should researchers consider when designing a microbiome study with diverse cohorts?

Researchers should consider several factors: selecting sampling sites that capture the intended diversity, standardizing sample collection and processing to avoid introducing site-specific technical artifacts, accounting for local dietary and lifestyle confounders, using analytical methods that handle unbalanced population representation, and engaging with local communities around data sovereignty. Practical guidance for these considerations is available in the review by Arif and Graham (Trends in Microbiology, 2025).

Figure 3: Timeline of global microbiome diversity research, from early population-specific studies to current large-scale atlas initiatives. Each milestone represents a step toward a more complete picture of human microbial ecology.

Related CD Genomics Microbioseq Services

• Microbiome Sequencing Services — End-to-end microbiome profiling for research cohorts of any size
• Full-Length 16S/18S/ITS Sequencing — High-resolution taxonomic profiling for diverse population studies
• Microbial Whole Genome Sequencing — Complete genome assembly for novel strain discovery in underrepresented populations
• Microbial Diversity Analysis — 16S/18S/ITS Sequencing — Cost-effective community profiling for large-cohort comparisons

For Research Use Only. Not for use in diagnostic procedures.

References

1. Blake KS, et al. Missing microbiomes: global underrepresentation restricts who research will benefit. Journal of Clinical Investigation. 2024;134(20):e183884. doi:10.1172/JCI183884
2. Abdill RJ, Graham SP, Rubinetti V, et al. Integration of 168,000 samples reveals global patterns of the human gut microbiome. Cell. 2025;188(4):1100-1118.e17. doi:10.1016/j.cell.2024.12.017
3. Andreu-Sánchez S, Blanco-Míguez A, Wang D, et al. Global genetic structure of human gut microbiome species is related to geographic location and host health. Cell. 2025;188(15):3942-3959.e9. doi:10.1016/j.cell.2025.04.014
4. Arif S, Graham C, et al. Analyzing human gut microbiome data from global populations. Trends in Microbiology. 2025. doi:10.1016/j.tim.2025.05.008
5. Maghini DG, et al. Expanding the human gut microbiome atlas of Africa. Nature. 2025;637(8046):674-683. doi:10.1038/s41586-024-08485-8
6. Hehemann JH, et al. Transfer of carbohydrate-active enzymes from marine bacteria to Japanese gut microbiota. Nature. 2010;464(7290):908-912. doi:10.1038/nature08937
7. Clemente JC, et al. The microbiome of uncontacted Amerindians. Science Advances. 2015;1(3):e1500183. doi:10.1126/sciadv.1500183
8. Vujkovic-Cvijin I, et al. Host variables confound gut microbiota studies of human disease. Nature. 2020;587(7834):448-454. doi:10.1038/s41586-020-2881-9
9. Boulund U, et al. Gut microbiome associations with host genotype vary across ethnicities and potentially influence cardiometabolic traits. Cell Host & Microbe. 2022;30(10):1464-1480.e6. PMID: 36099924