Persistent members, bounded conclusions
A core microbiome across life: what cohort stability does and does not mean
Shared genera can illuminate an ecosystem. They cannot turn a small cohort into your personal normal range.

Short answer
A core microbiome is a research definition of microbes repeatedly detected across selected samples. In a small Mediterranean cohort, several bacterial genera met strict shared-presence criteria across three age groups. That does not establish a universal healthy checklist, an individual normal range or a biological-age test. Crucially, this was not a study following the same people from childhood to old age.
Thirty people, many observations, three different clocks
The phrase “across life” sounds like a decades-long diary. The January 2026 Scientific Reports paper by Susana Ruiz-Ruiz and colleagues asks a related but different question. Researchers recruited thirty healthy volunteers in the Valencian Region of Spain, divided into three groups of ten. At recruitment, the youngest participants were aged 2–5 years, adults were 27–44, and older adults were 69–81. Although the paper calls its youngest group “Infants,” these were not newborn babies. Here we call them young children to keep the population clear.
Researchers attempted repeated stool sampling within each group. They collected 219 samples out of a planned 240. Those observations are not 219 independent volunteers: several samples came from the same person. After quality control, there were 218 samples for bacterial 16S analysis, 156 for metagenomics and 79 for metatranscriptomics. Each analytical layer therefore has a different denominator. Treating every result as though it came from an equally complete dataset would hide an important limitation.
The maximum sampling periods differed too: 723 days for the young-child group, 754 for adults and 524 for older adults. These are useful repeated observations, not a complete life history. The study combines a longitudinal component within people with comparisons between different age groups. The distinction matters whenever a headline suggests that a particular person’s ecosystem has remained unchanged for eighty years.
Imagine three family photo albums, each covering roughly a year or two, belonging to people of very different ages. Comparing the albums can reveal recurring features and differences between groups. It cannot supply the missing decades between one person’s childhood and later life. That analogy describes the design, not a measurement or a new result.
“Core” changes when its definition changes
A shared core is not simply a list of the most abundant bacteria. The investigators defined cores at the genus level using prevalence: how often a genus was detected across samples in an age group. They examined thresholds of 100%, 80% and 50%, with a relative-abundance detection threshold of 0.0001. A stricter prevalence rule selects fewer shared genera; a looser rule admits more. Neither automatically becomes the medically correct rule for a person.
Under the strict 100% rule, six genera or genus-level classified groups were shared across the three age-group cores. These included Anaerostipes, Faecalibacterium and Bacteroides, together with less fully characterized groups. With the 80% rule, 32 genera were shared across all three age-group cores; at 50%, the overlap was 65. Those are different answers to different operational questions, not contradictory estimates of how many bacteria a healthy gut must contain.
Prevalence also differs from abundance. A genus can be detected frequently without making up a large fraction of the community. Conversely, a genus can be abundant in some samples but fail a requirement to appear in all samples. “Core” therefore tells you about a defined pattern of detection. It does not by itself tell you the amount, activity, strain identity or effect of every member.
This is also where a population pattern must be separated from a personal reference interval. A reference interval would require a defensible framework for interpreting an individual measurement, including the population, measurement method and intended use. This paper’s shared-presence analysis is not that framework. For a related editorial explanation of why diversity is not a personal health grade, keep the same boundary in mind: a compact ecological summary cannot answer every clinical question.
Tables scroll horizontally on narrow screens; focus the table region and use arrow keys.
| Group at recruitment | Observed sampling frame | Core definition | Interpretation limit |
|---|---|---|---|
| Young children, 2–5 years | 10 participants; maximum 723 days | Genus prevalence at 100%, 80% or 50% | Not newborns; cannot define a universal pediatric normal range |
| Adults, 27–44 years | 10 participants; maximum 754 days | The same three operational thresholds | Not a reference interval for all adults or all laboratories |
| Older adults, 69–81 years | 10 participants; maximum 524 days | The same three operational thresholds | Not proof that an individual retained the same microbes since childhood |
Stability means a comparison, not a frozen ecosystem
The authors compared genera in repeat samples from the same individuals using Jaccard similarity, a measure based on shared membership. Shared genera generally declined as the time between samples increased. The paper describes approximately 60% conservation over longer observed intervals, with the youngest group showing greater variability. That summary is about overlap in detected genera under the analysis; it is not a statement that a person is “60% healthy.”
Importantly, the within-person comparison is not the same calculation as comparing a sample with an age-group core. To explore a broader age span, researchers also used fixed reference cores defined from the age groups. The paper explicitly notes that they did not have microbiota information from a healthy individual throughout their lifespan. A difference between a young child’s sample and an older-adult reference core cannot fill that missing history.
A separate model-based stability analysis examined fluctuations in genus abundance over time. Its results suggested greater fluctuation in the youngest group. The adult reference region in that analysis should not be converted into a health verdict for children or older adults. The study did not validate a diagnostic rule, and being farther from an adult model reference is not equivalent to being ill.
For the reader, the useful habit is to ask “stable relative to what?” Membership, relative abundance, reference-core similarity and expression are different comparison objects. A community can preserve some members while changing their proportions. The study’s multiple analytical layers help make that question visible; they do not remove the need to specify it.
Age-group evidence explorer
Choose a study group and a core definition to see what that comparison can support. This displays published study facts, not an analysis of your own microbiome. It does not ask for your age, symptoms or test results.
Assumptions: recruitment ages, main-text maximum sampling windows and Figure 3 core overlaps only. Overlap counts refer to all three groups, even when one group is selected. No interpolation between age groups, biological-age score or diagnostic threshold. Entries remain in this page; download creates a local text file.
Who is there, what could they do, and what is expressed?
The three analytical layers have distinct jobs. The bacterial 16S analysis supports taxonomic comparisons, including genus-level core definitions. Metagenomics examines microbial DNA and functional gene potential. Metatranscriptomics examines RNA expression. Finding a taxon is not the same as demonstrating that a particular function is being expressed at the same level, and RNA expression is not itself a proven health outcome.
The authors discuss possible functional uncoupling with age. That is an interpretation of their multi-omics observations, not a demonstration that replacing a named bacterium will restore a person’s health. This article does not import intervention findings from the paper’s reference list. Its reader task is narrower: understand what cohort stability can and cannot tell an individual.
| Publication and design | Population / denominator | Measurement | Boundary |
|---|---|---|---|
| Ruiz-Ruiz et al., 9 January 2026; repeat sampling | 30 healthy volunteers in Valencia; 219 collected stools | Sampling across three recruited age groups | Repeated samples do not increase the participant count |
| Same study; taxonomic layer | 218 quality-controlled 16S samples | Genus prevalence, overlap and within-person comparisons | Shared genera are not a clinical checklist or strain-level identity proof |
| Same study; DNA functional layer | 156 quality-controlled metagenomes | Functional gene potential | Potential is not direct proof of expression or benefit |
| Same study; RNA layer | 79 quality-controlled metatranscriptomes | Expression profiles | Incomplete series and abundance-related protocol bias constrain interpretation |
What to ask before calling a result “normal”
Start with the comparison population. A healthy volunteer cohort from one Spanish region is informative about that cohort. It does not automatically represent every geography, diet, age or clinical situation. Even within the cohort, the analysis found substantial differences between individuals. A group-level pattern does not erase those differences or give a unique explanation for an individual’s symptoms.
Next ask what “missing” means. A genus that fails a core rule may fall below a detection threshold or fail the chosen prevalence criterion. That is not equivalent to proving that a person lacks an essential organism. Likewise, a detected genus does not tell you that every strain within it behaves identically. Do not turn a genus list into a shopping list for supplements.
Then inspect time. A single stool sample cannot reproduce a within-person stability analysis requiring repeat observations. Two results obtained by different analytical pipelines may also be difficult to compare directly. Before interpreting a change, identify the method, reference group, analytical level and time window. This paper alone cannot supply a universal amount of change that should trigger treatment.
Finally, distinguish an open question from a result. The authors acknowledge the manageable but small cohort of thirty people, incomplete time series and metatranscriptomic biases toward abundant features. The study helps formulate questions about persistent microbial communities and functional differences. It does not establish whether testing these patterns improves care, whether changing them improves outcomes, or which intervention a person should choose.
A useful reading note can therefore be very short: “Three age groups, repeated samples, operational core definitions, no personal reference interval.” That preserves the finding without asking it to do a clinical job it was not designed to perform.
Frequently asked questions
Were the same people followed from childhood to old age?
No. Thirty volunteers belonged to three different recruited age groups. They provided repeated samples over limited periods. The broad age comparisons use different people and reference cores, not an individual lifetime record.
Does core mean a bacterium must be present in every healthy person?
No. Core is an operational definition based on genus detection and prevalence in this study’s samples. Changing the prevalence rule changes the core. The study does not establish a universal checklist for individual health.
Were the youngest participants newborn infants?
No. Although the paper calls this group Infants, participants were aged 2–5 years at recruitment. Newborn microbiome interpretations would go beyond the studied population.
Does approximately 60% overlap mean a 60% health score?
No. It summarizes shared detected genera in a defined comparison. It is not a diagnostic score, a percentage of beneficial bacteria or a treatment target.
Can the explorer estimate my biological age?
No. It filters published group descriptions and core-definition facts. It accepts no personal age or test measurements and does not calculate biological age, diagnosis or treatment.
What happens to my explorer selections?
Selections remain in the current page and are not saved in browser storage or sent to a server. Download creates a local evidence note. Changing either selection clears the previous output; reset clears both selections and the note.
Original source and editorial transparency
Ruiz-Ruiz S, Piquer-Esteban S, Pérez-Rocher B, et al. Lifetime existence of a core of mutualistic symbionts and functionally uncoupled taxa in the gut of a Mediterranean cohort. Scientific Reports (2026), published 9 January 2026. Original publication · 10.1038/s41598-026-35033-3.
We read the published original full-text XML: methods, results, discussion and main figure legends. Numerical statements here are confined to main-text findings; no reference-list study is treated as independently reviewed evidence. The approved source register and current Crossref metadata showed no indexed correction or update notice in this bounded check. The authors declare no competing interests.
The raster illustration was generated with AI and depicts conceptual ecosystems, not microscopy or measured microbial abundances. The explanatory diagram and explorer are original editorial aids, not study instruments. The optional contextual diversity link is further editorial reading, not the evidence source for this study.