The Informational Model of the Holobiont: Statistical Tests for Selection and Extension to a Theory of Variable Interactions

arXiv:2608.23504 · q-bio.PE · Submitted 2026-08-24 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Genomics Radio. Generated commentary on the latest computational biology and genomics papers.

Ines: We're looking at the paper: "The Informational Model of the Holobiont".

Marcus: Here's the gist of the paper. The Informational Model of the Holobiont: Statistical Tests for Selection and Extension to a Theory of Variable Interactions reviews and generalizes an evolutionary information-theoretic model for holobionts, developing statistical tests to partition informational change into contributions from host types, microbial components, and host-microbiome associations.

Ines: Core Model Formulation The paper introduces a framework where evolutionary change is quantified using Jeffreys divergence and partitioned into contributions from the host, microbial components, and host-microbiome associations Building on these partitions, it develops statistical tests to identify whether observed informational change is attributable to selection acting on host types, microbial-component states, or particular host-microbiome combinations.

Marcus: This framework extends to a general hierarchical formulation called the Theory of Variable Interactions (TVI), allowing informational change to be partitioned recursively across arbitrary organizational levels.

Ines: Informational Partitioning in the Holobiont The fitness of a holobiont is decomposed into host, microbial, and association components, defined by the relation logW ij=logW H (i)+logW M (j)+logaij.

Paper summary: Ines: So this paper, "The Informational Model of the Holobiont: Statistical Tests for Selection and Extension to a Theory of Variable Interactions," is reviewing and generalizing an evolutionary information-theoretic model for holobionts.

Marcus: The core idea is that they quantify evolutionary change using Jeffreys divergence and partition that total informational change into contributions from the host, microbial components, and host-microbiome associations.

Ines: Building on those partitions, they develop statistical tests to determine if the observed informational change is due to selection acting on host types, microbial component states, or specific host-microbiome combinations.

Marcus: They then extend this framework into a general hierarchical formulation called the Theory of Variable Interactions, which lets them partition that informational change recursively across arbitrary organizational levels.

Yuki: It sounds like they are trying to create a unified language for how selection works in these composite systems, moving beyond just looking at the whole system as one unit.

Ines: They define fitness by breaking it down into host, microbial, and association parts using that specific log relation involving log W H I plus log W M J plus log a I J.

Marcus: And the change due to selection is then partitioned into three separate terms: the change in host marginals, the change in microbial marginals, and the change in association structure.

Yuki: It shows that selection affects these different layers differently—the host and microbe marginally through their effective fitnesses, while the association changes are captured by something called J assoc.

Ines: For an aggregate model, they find that n JT asymptotically follows a chi squared distribution with K H K M minus one degrees of freedom <ref:2608.23504#pg1>.

Marcus: And they also test those marginal divergence components separately using chi squared distributions with KH minus one and KM minus one degrees of freedom respectively <ref:2608.23504#pg1>.

Yuki: When you move to the structured multicomponent extension, where you have K M equals the product of k from one to K K Mk possible microbiome configurations, the total statistic n JT follows a chi squared distribution with nu J equal to K H K M minus one degrees of freedom <ref:2608.23504#pg1>.

Ines: They also introduce TVI, or the Theory of Variable Interactions, which provides a unified framework for quantifying evolutionary change in both biological entities and their interactions.

Marcus: The TVI preserves local statistical dimensionality across hierarchical levels while higher levels introduce additional relational dimensions.

Yuki: This recursive structure means that units at one level can be made up of smaller units whose distributions can undergo further informational change, which is really key for understanding complex biology.

Ines: They show how the realized relative fitness decomposes into changes in marginal frequencies together with association factors operating at distinct organizational levels.

Marcus: This decomposition shows a 'G G represents association among the group classes composing the higher-level set; B G J B J represents association among the internal states of its constituent groups, and W H G L J L G L represents association among components within constituent group l.

Yuki: That way they’re mapping out exactly where all these different kinds of associations are happening in the system, whether it’s between host classes or between components inside a group.

Ines: They recover previous models by setting L equals one, and they get the aggregate holobiont model by replacing K microbial components with a single component for the microbiome.

Marcus: And they confirm that total Jeffreys information JT is exactly preserved when you perform an exact relabelling of the multicomponent microbiome.

Yuki: It’s a solid foundation because it shows that even when you simplify the structure, like merging microbes into one unit, the core information about selection remains intact.

Conclusion: Ines: So we’ve walked through how this paper, "The Informational Model of the Holobiont: Statistical Tests for Selection and Extension to a Theory of Variable Interactions," uses information theory to partition evolutionary change in holobionts.

Marcus: The main thing is that they provide a rigorous statistical test framework that separates the informational effects into component and association contributions, which is analytically separable across hierarchical levels.

Ines: This means researchers can now pinpoint whether the change they see in their data is driven by how hosts or microbes are changing individually, or if it’s specifically due to the way those parts are interacting.

Marcus: The Theory of Variable Interactions extends earlier work by treating evolutionary change as occurring through changes in biological entities and also through the structures connecting them.

Yuki: What this implies for us is that we can analyze complex evolutionary processes not just by looking at the final structure, but by understanding the dynamics of how those components are interacting over time.

Ines: It moves us toward a system where we can formally separate component-level evolution from interaction-level evolution during analysis.

Marcus: It’s about getting a formal way to separate those two forces when studying systems that involve both host and microbiome dynamics, which is incredibly useful for cohort studies too.

Yuki: So, by giving us this partition, we get better tools to understand the history of species by looking at how the system has adapted at different organizational scales.

Centro de Investigación Mariña · Universidade de Vigo

q-bio.PE

Submitted: 2026-08-24

Updated: 2026-11-04

Comments: 89 pages, 2 figures, 4 tables

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 88/100

The gist: The Informational Model of the Holobiont: Statistical Tests for Selection and Extension to a Theory of Variable Interactions reviews and generalizes an evolutionary information-theoretic model for…

Key concepts

Jeffreys Divergence
This is a mathematical tool used to quantify the informational difference between two probability distributions. In this model, it measures how much the distribution of fitness or configuration changes due to selection, allowing researchers to track specific informational shifts.
Theory of Variable Interactions (TVI)
TVI is a unified framework that treats evolutionary change as occurring simultaneously across different organizational levels. It allows for recursive partitioning where higher-level units can be composed of lower-level components, maintaining statistical structure throughout the hierarchy.
Informational Partitioning
This involves breaking down the total evolutionary change into distinct parts: changes in host characteristics, changes in microbial states, and changes in how they interact. This decomposition helps determine whether selection primarily acts on individual parts or on their combinations.

Terminology

Summary

The Informational Model of the Holobiont: Statistical Tests for Selection and Extension to a Theory of Variable Interactions reviews and generalizes an evolutionary information-theoretic model for holobionts, developing statistical tests to partition informational change into contributions from host types, microbial components, and host-microbiome associations. p. 0

Core Model Formulation

The paper introduces a framework where evolutionary change is quantified using Jeffreys divergence and partitioned into contributions from the host, microbial components, and host-microbiome associations Building on these partitions, it develops statistical tests to identify whether observed informational change is attributable to selection acting on host types, microbial-component states, or particular host-microbiome combinations. This framework extends to a general hierarchical formulation called the Theory of Variable Interactions (TVI), allowing informational change to be partitioned recursively across arbitrary organizational levels. p. 0

Informational Partitioning in the Holobiont

The fitness of a holobiont is decomposed into host, microbial, and association components, defined by the relation logW ij=logW H (i)+logW M (j)+logaij. The average change of Z due to selection is defined as J T=∑i, j (q'ij−qij)log q'ij/qij. This total change is partitioned into J T=J H +J M+J assoc+ Eholo. p. 5 Selection changes host and microbial marginal distributions through effective marginal fitnesses WH(i)Ai and WM(j)Bj, while changes in the host-microbiome association structure are captured by Jassoc through the post-selection association factor a'ij.

Statistical Testing and Hierarchy

For an aggregate-microbiome model, the statistic nJT asymptotically follows a chi-squared distribution with (KHKM − 1) degrees of freedom. The marginal divergence components can be tested analogously, following chi-squared distributions with KH − 1 and KM − 1 degrees of freedom. p. 6 In the structured multicomponent extension, the total number of possible microbiome configurations is K M=∏ k=1 K K Mk. p. 7 The total statistic nJT asymptotically follows a chi-squared distribution with ν J=K H K M−1 degrees of freedom. p. 6

Theory of Variable Interactions (TVI)

The Theory of Variable Interactions (TVI) provides a unified information-theoretic framework for quantifying evolutionary change in both biological entities and their interactions, with local statistical dimensionality preserved across hierarchical levels while higher levels introduce additional relational dimensions. The total informational change can be evaluated from the joint distribution of group-class identity and internal configuration. p. 6 This formulation reveals a recursive structure where units at one organizational level may themselves consist of multicomponent units whose marginal distributions and associations can undergo further informational change. p. 0

Hierarchical Structure and Association

The hierarchical formulation extends the same principle used throughout the preceding models, allowing for associations among group classes, among internal configurations of different groups, and among components within each group. The realized relative fitness across the complete hierarchy is decomposed into changes in marginal frequencies together with association factors operating at distinct organizational levels. p. 1 This decomposition shows that a 'G, g represents association among the group classes composing the higher-level set; a ' B, g, jg represents association among the internal states of its constituent groups; and a ' W, g,l, j l gl represents association among components within constituent group l.

Recovery of Previous Models

The preceding model of multicomponent sets under within- and between-group selection is recovered from this hierarchy by setting L = 1, so that each higher-level observational unit contains a single multicomponent group. The aggregate holobiont model is obtained by replacing the K microbial components by a single component representing the microbiome as a whole. p. 4 The total Jeffreys information JT is exactly preserved when an exact relabelling of the multicomponent microbiome is performed. p. 0

Conclusion and Significance

The framework provides a direct means of identifying the hierarchical source of the informational effects generated by selection. This approach partitions realized informational change into changes in host and microbial marginal distributions and changes in their association structure. p. 1 TVI extends the information-handling and holobiont framework developed by Carvajal-Rodríguez (2026) by treating evolutionary change as occurring simultaneously through changes in biological entities and in the structures connecting them. p. 0 This provides a formal partition of evolutionary change into component and association contributions, making their informational contribution analytically separable and statistically testable across hierarchical levels. p. 0

References

Bapteste, E., Papale, F., 2021. Modeling the evolution of interconnected processes: It is the song and the singers

Bordenstein, S.R., The Holobiont Biology Network, 2024. p. 1 The disciplinary matrix of holobiont biology

Carvajal-Rodríguez, A., 2026. Life as a Categorical Information-Handling System: An Evolutionary Information-Theoretic Model of the Holobiont

Carvajal-Rodríguez, A., 2025. QInfoMating: sexual selection and assortative mating estimation software

Carvajal-Rodríguez, A., 2024. Unifying quantification methods for sexual selection and assortative mating using information theory

Carvajal-Rodríguez, A., 2020. Multi-model inference of non-random mating from an information theoretic approach

Frank, S.A., 2012. Natural selection. p. 5 V. How to read the fundamental equations of evolutionary change in terms of information theory

Frank, S.A., 2025. Natural selection at multiple scales

Hartl, B., Levin, M., 2025. What does evolution make? p. 0 Learning in living lineages and machines

Levins, R., Lewontin, R., 1985. The Dialectical Biologist

Lewontin, R.C., 2000. The Triple Helix: Gene, Organism, and Environment

Lewontin, R.C., 1983. The Organism as the Subject and Object of Evolution

Lewontin, R.C., 1974. The genetic basis of evolutionary change

Margulis, L., Fester, R. p. 1 (Eds.), 1991. Symbiogenesis and symbionticism, in: Symbiosis as a Source of Evolutionary Innovation: Speciation and Morphogenesis

Mitchell, K.J., Cheney, N., 2025. The Genomic Code: the genome instantiates a generative model of the organism

Papale, F., Not, F., Bapteste, É., Haraoui, L.-P., 2024. The evosystem: A centerpiece for evolutionary studies

Roughgarden, J., 2020. Holobiont Evolution: Mathematical Model with Vertical vs. p. 1 Horizontal Microbiome Transmission

Sagan, L., 1967. On the origin of mitosing cells

Theis, K.R., Dheilly, N.M., Klassen, J.L., Brucker, R.M., Baines, J.F., Bosch, T.C.G., Cryan, J.F., Gilbert, S.F., Goodnight, C.J., Lloyd, E.A., Sapp, J., Vandenkoornhuyse, P., ZilberRosenberg, I., Rosenberg, E. p. 1 p. 1, Bordenstein, S. p. 1R., 2016. Getting the Hologenome Concept Right: an Eco-Evolutionary Framework for Hosts and Their Microbiomes

Week, B., Ralph, P.L., Tavalire, H.F., Cresko, W.A., Bohannan, B.J.M., 2025. Quantitative genetics of microbiome-mediated traits

Zilber-Rosenberg, I., Rosenberg, E., 2008.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved AI systems can perform hierarchical informational decomposition of evolutionary change by partitioning it into marginal changes, association changes among entities, and a finite-change residual across arbitrary organizational levels. This allows for testing whether observed informational change is attributable to selection acting on host types, microbial components, or specific host-microbiome associations.

  2. Improved AI systems can explicitly model and quantify variable interactions using the Theory of Variable Interactions (TVI), which provides a unified information-theoretic framework for biological entities and their interactions. This framework allows for recursive analysis where higher levels introduce additional relational dimensions beyond the marginals at lower levels, preserving local statistical dimensionality while capturing global structure.

  3. Improved AI systems can perform statistical tests on observed data by calculating Jeffreys divergence statistics corresponding to distinct organizational levels, such as the total change in complete holobiont configurations or specific association components like Jassoc. These statistics allow researchers to rigorously test hypotheses regarding the source of evolutionary signal, for example, determining if selection is caused by changes in host-microbiome associations versus changes in host or microbial marginal distributions.

  4. Improved AI systems can distinguish between different types of selection effects within a multicomponent system by analyzing the components of the total informational change JT, which is partitioned into J G (between-group), J I (within-group), and EGI (between-within coupling). This allows for the separation of selection acting on group classes from selection acting on internal configurations or component states.

  5. Improved AI systems can analyze complex hierarchical structures by utilizing the recursive formulation of TVI, which treats higher-level sets containing constituent groups as new entities subject to their own informational partitioning rules. This enables the analysis of nested organizational levels where changes in marginal frequencies and associations can occur simultaneously across all scales, quantified by association factors like a 'G, g.

Related papers