The strength of gut microbiota transfer along social networks and genealogical lineages in the house mouse

Abstract The gut microbiota of vertebrates is acquired from the environment and other individuals, including parents and unrelated conspecifics. In the laboratory mouse, a key animal model, inter-individual interactions are severely limited and its gut microbiota is abnormal. Surprisingly, our understanding of how inter-individual transmission impacts house mouse gut microbiota is solely derived from laboratory experiments. We investigated the effects of inter-individual transmission on gut microbiota in two subspecies of house mice (Mus musculus musculus and M. m. domesticus) raised in a semi-natural environment without social or mating restrictions. We assessed the correlation between microbiota composition (16S rRNA profiles), social contact intensity (microtransponder-based social networks), and mouse relatedness (microsatellite-based pedigrees). Inter-individual transmission had a greater impact on the lower gut (colon and cecum) than on the small intestine (ileum). In the lower gut, relatedness and social contact independently influenced microbiota similarity. Despite female-biased parental care, both parents exerted a similar influence on their offspring’s microbiota, diminishing with the offspring’s age in adulthood. Inter-individual transmission was more pronounced in M. m. domesticus, a subspecies, with a social and reproductive network divided into more closed modules. This suggests that the transmission magnitude depends on the social and genetic structure of the studied population.


Introduction
The gut microbiota of animals represents a diverse consortium of micr oor ganisms dominated by Eubacteria and their viruses (Vemuri et al. 2020 ).The extensive functional potential encoded by micr obial meta genomes expands the metabolic ca pabilities of the host and influences various aspects of its phenotype (Heijtz et al. 2011, den Besten et al . 2013, Beura et al . 2016 ).Unsurprisingly, disruptions in the functions of the gut micr obiota, r eferr ed to as dysbiosis, often lead to negative health consequences.In modern human societies, the pr e v alence of various inflammatory and metabolic disorders associated with dysbiosis of the microbiota has significantly increased in recent decades (Cekanaviciute et al. 2017, Manfr edo Vieir a et al. 2018 ).Ther efor e, a crucial objectiv e of current research on host-associated gut microbiota is to compr ehend the mec hanisms that sha pe the micr obiota during host ontogeny and give rise to variation in its composition.
While a subset of bacteria invades the gut from the environment, a substantial proportion of the gut microbiota consists of bacterial species with highly specific ecological requirements, making them r ar el y found in envir onmental sources outside the host body (Li et al. 2016 ).Direct transfer of these bacteria between host individuals is belie v ed to facilitate their spread within host populations (Tung et al. 2015, Moeller et al. 2016b, Amato et al. 2017 ).In inv ertebr ates, the connections between bacterial adap-tation to obligatory symbiosis and their vertical transmission, often spanning host speciation e v ents, ar e well known (Moran et al. 2008 ).In v ertebr ates, the tightness of the host-micr obiota r elationship is still lar gel y unexplor ed.A fe w empirical studies hav e onl y r ecentl y uncov er ed bacterial linea ges that hav e co-div ersified with their v ertebr ate hosts, as well as changes in the phenotype and genome of these bacteria that facilitate their persistence in the host gut and their transmission between host individuals (Moeller et al. 2016a, Waskito and Yamaoka 2019, Karcher et al. 2020, Suzuki et al. 2022 ).
Vertebr ate embryos gener all y de v elop in a sterile or near-sterile environment, and the colonization of their bodies by symbiotic bacteria begins as early as hatching or birth (Perez-Muñoz et al. 2017, Kennedy et al. 2023, T ěšický et al. 2024 ).Parental transmission of bacteria is an important pathway that facilitates the establishment of a normal microbiota in newborns.Inoculation by maternal v a ginal micr obiota during deliv ery is likel y the most significant initial source of gut bacteria in therian (viviparous) mammals (Dominguez-Bello et al. 2010, Ferretti et al. 2018 ).Ho w e v er, other forms of physical contact with parents are also likely to facilitate the establishment of microbial populations in v ertebr ate juveniles .T his is supported by the similarity in microbiota between parents and offspring of oviparous vertebrates (Kreisinger et al. 2017, Kouete et al. 2023 ), where direct transfer of maternal microbiota during birth or breastfeeding is not possible.In addition to v ertical tr ansmission of gut microbiota along host genealogical lineages mediated by contacts with parents, social contacts with distantly related conspecifics represent another channel of gut microbiota transmission.This is supported by many studies, in which the intensity of social interactions between individuals predicts the similarity of their gut microbiota (Lax et al. 2014, Tung et al. 2015, Amato et al. 2017, Dill-McFarland et al. 2019, Sarkar et al. 2020, Raulo et al. 2021 ).
The bacteria transmitted and incorporated into the vertebrate gut immediately affect the community structure, but they can also exert indirect effects that manifest with a time delay.In this r egard, consider able attention has been paid to the role of acquired bacteria in the subsequent succession and maturation of the gut community in early life humans (reviewed in Wang et al. 2020 ).Ho w e v er, studies dir ectl y quantifying the c hanges in the str ength of tr ansmission effects ov er time ar e r ar e .For example , in humans, the effect of cesar ean section, whic h pr e v ents the transfer of v a ginal micr obiota to offspring, is strongest in neonates up to a few months old, but significant differences due to delivery mode persist until se v er al years of age (Roswall et al. 2021 ).In birds, social transmission of microbiota during copulations leads to a correlated composition of cloacal microbiota between sexual partners (White et al. 2010, Kreisinger et al. 2015 ).Ne v ertheless, the similarity decreases rapidly after a few days when copulation is experimentally restricted (White et al. 2010 ), suggesting that the effect of transmission is only transient.
A limitation of our current knowledge is that all studies on gut micr obiota tr ansmission to date hav e focussed exclusiv el y on the communities of the lo w er intestinal segments (i.e.colon and cecum), while the impact of inter-individual transmission on microbiota in other gut sections remains completely unknown.There is compelling evidence that the gut microbiota outside the lo w er intestine differs from the lo w er gut microbiota in its composition and its effects on host physiology and immune functions (Zoetendal et al. 2012, Kastl et al. 2020 ).While ecological gradients along the gut appear to play a major role in these differences between gut communities (Stearns et al. 2011 ), the mode of bacterial transmission can represent yet another important factor affecting the assembly and turnover of the local microbiota, similar to what has been shown for gut vs. oral communities (Valles-Colomer et al. 2023 ).
Furthermor e, ther e is a lack of suitable model systems to study the effects of micr obiota tr ansmission.While studies in wild v ertebr ate populations are crucial, collecting data on the interindividual variability of the microbiota that allows discrimination of sources of microbiota transmission (i.e .parents , conspecifics , and environment) can be extremely challenging.Another caveat is that micr obiota tr ansmission itself is not the only factor affecting microbiota variation between individuals.Factors such as diet composition, geneticall y encoded r egulatory mec hanisms of the host, and the host's health and ontogenetic state can either restrict or promote the growth of transmitted bacteria in the gut (Hasan and Yang 2019 ).Typically, these factors show high variation in wild populations and are difficult to determine.
At the same time, ca ptiv e-br ed model animals may not be optimal for this type of r esearc h.A prime example is the laboratory house mouse, which is an important model in v arious ar eas of biomedical r esearc h, including the effects of gut microbiota on the host (Ericsson and F ranklin 2021 ).Ho w e v er, the specific pathogen-free mice used for this purpose have a highly altered micr obiota compar ed to their wild ancestors, with dr amatic effects on some of their phenotypic traits (Rosshart et al. 2017 ), raising concerns about the gener al r ele v ance of this model.The recent call for a mouse model with more realistic host-symbiont interactions has led to the concept of "r e-wildering" labor atory mice (Graham 2021 ).For the successful application of such a concept, it is necessary to understand the mechanisms that lead to the assembl y of micr obial comm unities in the house mouse under natural conditions.
We investigated the transmission of gut bacteria between individuals of two closel y r elated house mouse subspecies ( Mus musculus musculus and M. m. domesticus , hereafter referred to as MMM and MMD, r espectiv el y).We compar ed the tr ansmission patterns among three gut sections: the colon, the cecum, and the ileum.The mice used in this study were first generation offspring from free-living populations, possessing wild-type micr obiota, and wer e maintained for se v er al gener ations in seminatur al ar ena experiments that mimic ked fr ee-living populations (Bendová et al. 2022, Mikula et al. 2022 ).Unlike experiments conducted in laboratory settings, the mating of mice and their social contacts were not constrained by external factors .T he common garden setup ensured a homogeneous environment, including a uniform diet.This reduced the chances that microbiota variability was causally driven by factors that varied spatially and directly influenced microbiota composition.The absence of emigration and immigr ation fr om and into the studied population allo w ed detailed analyses of genetic parentage (based on microsatellite data) and social interactions (using radiofrequency identification micr otr ansponders).In this way, the transmission of the microbiota along social networks and genealogical host lineages (Mikula et al. 2022 ) could be tr ac ked with a degree of precision and completeness that would be nearly impossible to ac hie v e in fr ee-living populations of this species.
Notably, the two host subspecies in the experiment differed in their social and r epr oductiv e patterns.Specificall y, MMD displayed more "modular" social networks r epr esented by closer and mor e separ ated units compar ed to MMM.Since the "modules" were identified with re producti ve units (demes), it was concluded that the experimental MMD populations had more "demic" social structure than MMM.This finding was also consistent with a higher frequency of multiple paternity in the latter subspecies (Mikula et al. 2022 ).This provided a unique opportunity to assess the impact of this variation on the patterns of inter-individual transmission of gut microbiota.
Our main objectives were as follows: (i) to compare the strength of inter-individual transmission of gut bacteria between closely related and less related individuals in different gut segments and in two house mouse subspecies, (ii) to identify specific bacteria whose abundance variation is explained by social contacts and host genetic relatedness, and (iii) to analyze the temporal persistence of maternal and paternal microbiota transmission.

Data collection
We collected data from 198 house mouse individuals sampled as part of the study conducted by Mikula et al. ( 2022 ), which aimed to compare the social and re producti ve structure between two house mouse subspecies bred under semi-natural conditions in a common garden setup.Detailed information about the 2014 experiment, whic h is anal yzed in this work, can be found in Mikula et al. ( 2022 ).A subset of the adult individuals from this experiment was pr e viousl y anal yzed by Bendov á et al. ( 2022) to assess micr obiota div er gence betw een the tw o mouse subspecies.
Briefly, we used first-generation mice born in capti vity, re presenting the G1 pr ogen y of wild mice from two populations of each subspecies.At the beginning of the ∼9-month-long experiment, 12 MMM and 12 MMD individuals wer e r eleased in two adjacent enclosures (4 × 2 m each) equipped with six nest boxes.After ∼6 months, the two enclosures were interconnected with a short tunnel [see Fig. S1 in Mikula et al. ( 2022 )].The mice were k e pt under constant conditions, including a light-dark regime of 14:10 h and a temper atur e of ∼20 • C.They had access to standard food pellets and water ad libitum .The enclosures were regularly checked every 3 da ys , and ne wborns wer e marked by toe-clipping, with their birth dates r ecorded.Eac h mouse was injected with a r adiofr equency identification micr otr ansponder to monitor their movements .T he entrance of each nest box and the tunnel connecting the enclosures were each equipped with two transponder readers.At the end of the experiment, all mice were sacrificed by cervical dislocation and immediately dissected.Tissue samples were collected fr om thr ee gut sections: 1 cm of the distal part of the ileum, the entire cecum, and the colon.Each sample was placed in a sterile cryotube, sna p-fr ozen in liquid nitr ogen, and stor ed at −80 • C until analyzed.During the dissections, only flame-sterilized tools were used, and special care was taken to avoid potential sample contamination.

Intensity of social interactions and level of genetic relatedness
Mouse DN A w as extr acted fr om ethanol-pr eserv ed toe clips using the Qiagen DNeasy 96 Blood & Tissue Kit.A panel of 25 microsatellites (Mikula et al. 2022 ) was used for par enta ge assignment.The par enta ge anal ysis was performed using CERVUS v. 3.0.3(Kalinowski et al. 2007 ) at a 95% confidence le v el.Individuals who met the defined r equir ements, suc h as r eac hing sexual maturity at the estimated time of conception, were included in the analysis as candidate parents .T he parentage data were analyzed using the R pac ka ge AGHmatrix (Amadeu et al. 2016 ) to reconstruct MMM and MMD genealogies, and relatedness coefficients between individuals were estimated.Importantly, we did not detect any hybrid between MMM and MMD.
Matrices describing the intensity of social contacts between MMM and MMD individuals were constructed as described in Mikula et al. ( 2022 ).In brief, the social contact between two individuals was measured as the time they spent together in any of the nest boxes, regardless of other individuals potentially present in the nest boxes.A visit to a nest box was defined as the time between entering (recorded as a twofold signal: the first from the outer reader, follo w ed b y that fr om the inner r eader) and leaving the nest (recorded in reverse order, i.e. inner-outer reader).Consequently , in our study , we define social interactions based on the spatio-tempor al co-occur ance of two individuals.During such encounters , multiple beha vioral interactions and associated types of physical contact could take place.Ho w e v er, our experimental design did not allow us to observe and distinguish between them.

16S rRNA sequencing of gut microbiota
Meta genomic DNA fr om gut tissue samples was extracted using the Po w erSoil DN A Isolation Kit (QIAGEN).Sequencing libraries wer e pr epar ed using a two-step Pol ymer ase Chain Reaction (PCR) protocol as described in Čížková et al. ( 2021 ).In the first PCR, standar d metabar coding primers of Klindw orth et al. ( 2013) targeting the V3-V4 region of the bacterial 16S rRNA gene (S-D-Bact-0341-b-S-17: CCTA CGGGNGGCWGCA G and S-D-Bact-0785-a-A-21: GA C-TA CHV GGGTATCTAATCC) were used.These primers were ex-tended by inline barcodes and priming sites for the second PCR.In the second PCR, dual indexes were introduced, and Illuminacompatible Nextera-like sequencing adapters were added.Each metabar coding PCR w as performed in technical duplicate to account for PCR and sequencing stoc hasticity.The PCR pr oducts were pooled based on their concentration and sequenced using an Illumina MiSeq platform (v3 kit, 300 bp pair ed-end r eads) at CEITEC, Brno, Czech Republic.
The resulting fastq files were demultiplexed, and 16S rRNA primers were detected and trimmed using Skewer (Jiang et al. 2014 ).R pac ka ge dada2 (Callahan et al. 2016 ) was used to eliminate low-quality reads (expected error rate per paired-end read > 1), to denoise the quality-filter ed r eads, and to gener ate abundance matrices of read counts for each 16S rRNA sequencing v ariant (her eafter ASV) in eac h sample.Denoising was performed with the pool = "pseudo" option to increase the sensitivity of detection of r ar e ASVs in samples with low sequencing cov er a ge (Bardenhorst et al. 2022 ).Subsequentl y, uc hime2 (Edgar 2016(Edgar et al. 2016 ) was used to detect and eliminate chimeric ASVs with Silva database v.138 (Quast et al. 2013 ) as a reference for filtering the c himer as.Taxonomy assignment for non-c himeric ASVs was performed with 80% posterior confidence using the RDP classifier (Wang et al. 2007 ).The Silva database v.138 (Quast et al. 2013 ) was used for bacterial ASV annotations .T he consistency of ASV profiles between technical duplicates was c hec ked using Pr ocrustes anal ysis (Legendr e and Legendr e 1998 ).The duplicated data wer e then merged, eliminating any ASVs that were not detected in both duplicates.

Sta tistical anal yses
We emplo y ed linear mixed-effects models (LMMs) to assess the strength of the association between microbiota composition, genetic relatedness, and intensity of social contacts.To c har acterize the variation in microbiota composition among samples, we calculated Bray-Curtis dissimilarities (based on the r elativ e abundance of ASVs) and J accar d dissimilarities (based on the presence/absence of ASVs).Phylogenetically weighted distances (e.g.UniFr ac), whic h downweight compositional div er gence caused by closely related bacterial ASVs, were not considered.Such a feature is not desirable for our study, as phylogenetically related bacteria may have similar properties (e.g.oxygen tolerance or spoluration ability) that facilitate their dispersal (Galperin et al. 2012 ).
Bray-Curtis and J accar d dissimilarities w ere then transformed into similarity indices using the formula (f1): where S and D represent the similarity and dissimilarity between the i -th and j -th samples, r espectiv el y.The logit-tr ansformed v alues of the similarity indices were used as response variables in LMMs, assuming a Gaussian error distribution.To account for une v en sequencing depth, the ASV abundance matrix was r ar efied befor e calculating micr obiota dissimilarities .T he rarefaction thr esholds wer e set to the minim um sequencing depth r eac hed (i.e.2374 reads).To address the non-independence of pairwise (dis)similarity values, we employed the R pac ka ge lmer-MultiMember (van Paridon et al. 2022 ), which allo w ed us to include pairs of mouse individuals associated with each similarity value as a multigroup membership random effect.The models were fitted separately for each gut section.As predictors in the LMMs, we considered the effect of host subspecies identity, the intensity of social contacts between host individuals, their r elatedness coefficients, and inter actions betw een the latter tw o variables and subspecies identity.Stepwise backw ar d elimina-tion of nonsignificant terms from the full model was performed with step.lmerModLmerTestfunction from the pac ka ge lmerTest (Kuznetsova et al. 2017 ), and likelihood ratio tests were used to determine the significance of the predictors .T he proportion of variance explained by the effects of relatedness or social contacts was calculated using the R pac ka ge v ariancePartition (Hoffman and Schadt 2016 ).
ASVs whose between-sample variation was predicted by the effect of relatedness or social interactions were identified using Pr ocrustes anal ysis .T he read counts in the ASV abundance matrix wer e tr ansformed to a center ed log r atio (clr) (Aitc hison 1982 ) using R pac ka ge compositions (v an den Boogaart and Tolosana-Delgado 2008 ).Pairwise absolute differences in clr values between all samples were calculated for each ASV included in the analyses .T he resulting dissimilarity matrices were scaled using principal coordinate analysis (PCoA) with Cailliez correction for negativ e eigenv alues using R pac ka ge a pe (P ar adis and Sc hliep 2019 ) and resulting PCoA scores were used as responses in Procrustes analyses (conducted using function from R pacakge vegan; Oksanen et al. 2020 ).Matrices containing relatedness coefficients and social interaction intensity were inverted using the f1 formula, scaled using PCoA, and the resulting PCoA scores were used as response variables in Procrustes analyses.ASV-level analyses were conducted separ atel y for eac h host subspecies and gut section, using a subset of ASVs present in at least 20 samples of a gut section.The false discov ery r ate (FDR) method by Benjamini and Hoc hber g ( 1995 ) was employed to account for false-positive detections due to multiple testing, and results with FDR values < 0.05 wer e consider ed significant.
We hypothesized that the strength of parental effects on the offspring microbiota is not constant o ver time , but gr aduall y decreases after the offspring achieve independence from parental care.To test whether the strength of parental effects on offspring micr obiota v aries with offspring a ge, we calculated micr obiota similarities between parents and their genetic offspring for all gut sections, and included them as a response in LMMs.Offspring age , sex, gut sections , subspecies , and their interactions were used as predictors.LMMs included parental and offspring identity as m ultigr oup membership random effects .T he nonlinear effect of offspring age on dissimilarity with parents was modeled using natural cubic splines.To determine the optimal complexity of the model fit, linear models (estimating the intercept and slope for the effect of offspring age) were compared with models including nonlinear splines with degrees of freedom ranging from two to four, using the Akaike information criterion.Tempor al c hanges in similarity in microbiota composition between offspring from the same litter wer e anal yzed in the same manner as described abo ve , but the effect of sex was not considered because it varied between sibling pairs.To compare the patterns of microbiota dissimilarity between parents and offspring, and between offspring from the same litter, with the ov er all dissimilarity of micr obiota composition, we estimated the av er a ge population-wide dissimilarity between genetically unrelated individuals (relatedness coefficients < 0.25) for each intestinal section using LMMs with m ultigr oup membership random effects.In all LMMs, microbiota dissimilarities wer e logit-tr ansformed (Warton and Hui 2011 ) to ac hie v e a normal distribution of residuals.
It is essentially unclear how similarity in microbiota composition, due to genetic relatedness or the intensity of social interactions, translates into alpha diversity at the individual level.Since we are unable to formulate plausible hypotheses for microbial alpha div ersity, we intentionall y did not conduct thor ough anal yses of alpha diversity in our study.

Gut microbiota v aria tion
We analyzed a total of 585 microbiota samples ( Table S1 , Supplementary Data ) from three gut sections (ileum, cecum, and colon) of 198 house mice belonging to two subspecies (52 males and 45 females of MMM, and 51 males and 50 females of MMD).Tw o colon, tw o cecum, and five ileum samples were not included into analyses due to missing sample or possible contamination during the DNA isolation.The sequencing data included 8 065 266 high-quality reads, with an av er a ge sequencing depth per sample of 13 787 (range 2374-45 661).We detected a total of 3136 nonchimeric bacterial ASVs.
Ileum exhibited the lo w est alpha diversity and displayed a distinct microbiota composition compared to the cecum and colon.Furthermor e, the ileum differ ed in terms of its taxonomic content.PCoA r e v ealed clear differ ences in micr obiota composition betw een the tw o house mouse subspecies ( Fig. S1 , Supplementary Data ).

Effect of social contacts and relatedness on microbiota similarity
The effects of social contacts and relatedness varied significantly in their r elativ e importance between the two host subspecies, as indicated by their significant interactions with subspecies identity ( Table S2 , Supplementary Data ).Subsequent LMMs conducted on data subsets corresponding to the three gut sections in each mouse subspecies r e v ealed significant influences of both social contacts and relatedness on the similarity of gut microbiota in the cecum and colon ( Table S3 , Supplementary Data ).These two factors explained a gr eater pr oportion of the variation in gut microbiota in MMD compared to MMM.The combined effect of relatedness and social contacts accounted for 8%-10% of the total variation in MMD, whereas it was < 3% in MMM (Fig. 1 ).In the case of the ileum microbiota, the effect of relatedness was only significant for MMD (for both Bray-Curtis and J accar d similarities) and the effect of social interaction was only significant for MMM and the model with J accar d but not Bray-Curtis similarities ( Table S3 , Supplementary Data ).
To explore the possibility that the effects of inter-individual transmission of the microbiota might be restricted to close family members that are genetically highly related, we conducted additional analyses by excluding dissimilarities related to full siblings and mother-offspring or father-offspring pairs.Subsequently, we r e-r an the afor ementioned anal yses .T he results sho w ed that the micr obiota v ariation explained by both social contacts and relatedness decreased, but remained significant for all gut sections of MMD.In contrast, no significant effect of relatedness was found in MMM ( Fig. S2 and Table S3 , Supplementary Data ).This finding shows that in MMD, the transmission of microbiota occurs along genealogical lineages, while in MMM the social transmission dominates in less related individuals.
The analyses at the level of individual bacteria included 446, 471, and 81 ASVs for the cecum, colon, and ileum, r espectiv el y, that were detected in at least 20 individuals.A significant effect of genetic relatedness was found in 29.1% of ASVs from the cecum, 24.0% of ASVs from the colon, and 11.1% of ASVs from the ileum.Social contacts had an effect in 7.2% of ASVs from the cecum, 1.2% from the colon, and 0% from the ileum.
In both house mouse subspecies, social contacts exhibited a congruent effect on ASVs v ariation.Specificall y, we observ ed a significant correlation between the Procrustes sums of squares (inv ersel y r elated to the str ength of the corr elation between the in-Figure 1. Proportion of variation in microbiota composition explained by social contacts and relatedness.Estimates were performed separately for each gut section (ileum, cecum, colon), each mouse subspecies (MMD, MMM), and for two microbiota similarity indices (Bra y-Curtis , J accar d) using linear mixed-effects models.tensity of social contacts and the similarity of abundances of individual ASVs) calculated separ atel y for the two mouse subspecies (Spearman correlation: rho = 0.2972, P < 0.0001 for colon, rho = 0.3215, P < 0.0001 for cecum, and rho = 0.2650, P = 0.0171 for ileum).These findings suggest that individual gut bacteria consistently exhibited dependence or independence on social contacts regardless of the host's genetic background.In the colon and cecum, we also found a significant correlation in the Procrustes sums of squares between MMM and MMD for the effect of genetic r elatedness (Spearman corr elation: rho = 0.1986, P < 0.0001 for the colon and rho = 0.2416, P < 0.0001 for the cecum), although the strength of these correlations w as lo w er compared to social contacts.Ho w e v er, ther e was no significant a gr eement between the two mouse subspecies in the case of ileum ASVs correlated to host genetic relatedness (Spearman correlation: rho = −0.0496,P = 0.6593).
A greater number of ASVs were significantly associated with genetic relatedness and social interactions in MMD than in MMM (Fig. 2 ).In the case of genetic relatedness, the difference between subspecies can be explained by the ov er all lower effect of relatedness on the variation in ASVs abundance in MMM (assessed by paired Wilcoxon test for Procrustes sums of squares; P < 0.0001 for all gut sections).In the case of the social inter action, Pr ocrustes sums of squar es v alues for eac h ASV did not differ between MMD and MMM for the colon ( P = 0.1094) and cecum ( P = 0.3278), while they wer e slightl y lower for the MMM ileum microbiota ( P = 0.0221).This suggests that the lo w er number of significant associations in the MMM cannot be explained by the reduced role of social contacts in this subspecies and is ther efor e pr obabl y due to lo w er statistical po w er.
ASVs whose abundance correlated with the genetic relatedness of their hosts were found across a wide range of taxa, includ-ing primarily Clostridia, Bacteroidia, Campylobacteria, and to a lesser extent Bacilli.On the other hand, ASVs associated with social tr ansmission wer e pr edominantl y fr om the genus Helicobacter , with 18 ASVs showing a significant association with social contact intensity (Fig. 2 ).In addition to Helicobacter , social contacts wer e corr elated with the abundance of 17 other ASVs belonging to the genera Mucispirillum , Lactobacillus , Alistipes , Odoribacter , Lachnoclostridium , and Rikenella , as well as the Lac hnospir aceae, the Muribaculaceae, and the Desulfovibrionaceae family.

Temporal persistence of parental transfer and offspring microbiota similarity
The similarity of microbiota composition between siblings from the same litter exhibited nonlinear changes with a ge, as e videnced by comparing models that considered linear versus nonlinear trends ( Table S4 , Supplementary Data ).Moreover, there was a significant interaction between mouse age and gut section, indicating that the temporal dynamics of similarity varied along the gut ( Table S5 , Supplementary Data ).Consequentl y, separ ate models were fitted for the data from each gut section.
In terms of Bray-Curtis and J accar d similarity, the microbiota of the colon and cecum sho w ed an initial similarity among siblings at a young age ( < 20 days old) that was comparable to the population-wide av er a ge of unr elated individuals.Subsequentl y, the microbiota similarity gradually increased until ∼70-80 days of age and then converged back to the population-wide baseline, although it did not r eac h the baseline e v en in older mice ( > 200 da ys old).T he tempor al v ariation in the ileum microbiota was less dramatic, with older siblings showing only slightly increased similarity compared to the population-wide baseline (Fig. 3 , Fig. S3 , Supplementary Data ).T he gra y line and gra y shaded ar ea corr espond to the av er a ge similarity between genetically unrelated individuals (relatedness coefficient ≤ 0.25) and their 95% confidence intervals.
As with siblings, the microbiota similarity between parents and their genetic offspring also exhibited nonlinear changes with offspring age ( Table S4 , Supplementary Data ), and a significant interaction between offspring age and gut section indicated differ ential tempor al patterns of variation between gut sections ( Table S5 , Supplementary Data ).Age-dependent changes of Bray-Curtis and J accar d similarity sho w ed consistent patterns, and the results were congruent when analyzing the similarity of offspring to their genetic mothers or fathers (Fig. 4 and Fig. S4 , Supplementary Data ).The similarity of cecum and colon microbiota between offspring and their parents was highest between 60 and 70 days of a ge, gr aduall y conv er ging to the population-wide baselines later.In contrast to the lo w er gut, the confidence intervals for the parent-offspring similarity in ileum microbiota always ov erla pped with the confidence intervals for the population-wide similarity baseline, regardless of offspring age .T he interaction between subspecies and age was significant only in the case of Bray-Curtis similarity between mothers and their offspring, with MMM showing a greater maternal effect than MMD, particularly during the 60-70 day peak ( Fig. S5 , Supplementary Data ).When analyzing the similarities between the microbiota of mother and offspring, significant interaction between the sex and age of the offspring w as also found.Ho w e v er, the subsequent model pr edictions did not r e v eal a clear pattern ( Fig. S5 , Supplementary Data ).

Discussion
Our anal ysis r e v ealed that inter-individual tr ansmission had a significantly smaller effect in the small intestine (ileum) compared to the lo w er gut (cecum and colon).This difference was observed both at the whole comm unity le v el and for individual bacterial ASVs .T he small intestine exhibited lo w er alpha diversity and alter ed taxonomic composition, whic h has been observed not only in mice (Kreisinger et al. 2015, Suzuki and Nachman 2016, Bendová et al. 2020 ) but also in other mammalian species (Gresse et al. 2019 ).These v ariations ar e attributed to differ ent envir onmental conditions determined by the host immune system, high concentrations of bile acids, and differences in metabolic substrates (Zoetendal et al. 2012, Kastl et al. 2020 ).The microbiota of the small intestine mainly utilizes high-energy substrates of low complexity, whereas the utilization of dietary fiber entering the lo w er gut favors the de v elopment of div ersified comm unities ric h in v arious cr oss-feeding inter actions (Louis et al. 2021 ).
Indeed, the ileum samples in this study exhibited low taxonomic complexity, with bacteria of only three genera ( Candidatus Sav a gella, Lactobacillus , and Mycoplasma ) av er a ging 80% of the total community and being the dominant component of the ileum microbiota in almost all individuals .T he low strength of interindividual transmission observed for ileum bacteria may be partly due to the low variation in ileum communities.On the other hand, ileal bacteria (including the three genera found in our study) can gener all y toler ate r elativ el y high oxygen concentr ations (Zoetendal et al. 2012, Kastl et al. 2020 ), and C .Sav a gella forms spores that ar e excr eted in host faeces (Hedblom et al. 2018 ).While these traits may serve as preadaptations for bacterial transmission through social contact, they can also enable bacterial survival in the external en vironment.T he absence of the transmission signal in the ileum could ther efor e be a consequence of indir ect tr ansmission between individuals , for example , through copr opha gy of leftov er feces or ingestion of contaminated food or nest material.
Although inter-individual transmission explains only a small portion of the variation in the lo w er gut micr obiota, r elatedness and the intensity of social contacts exerted significant independent effects on the microbiota similarity between individuals .T he significance of social tr ansmission c hallenges the seminal study of Moeller et al. ( 2018 ), which sho w ed that vertical transmission along host genealogy is by far the most important mechanism Figure 4. Similarity in microbiota composition between parents and their offspring.The Bray-Curtis index, which considers the relative abundances of ASVs, was used to examine how the similarity in microbiota between parents and offspring varies on the time scale representing offspring age .T he colored line and shaded area correspond to the mixed model predictions and 95% confidence intervals for each gut section.The gray line and gray shaded area correspond to the average similarity between genetically unrelated individuals (relatedness coefficient ≤ 0.25) and their 95% confidence intervals.
sha ping gut comm unities in house mice .T his contradiction can be explained by the experimental design of Moeller et al. ( 2018 ), whic h suppr essed tr ansmission thr ough social contacts (no interindividual contacts except for mother and offspring, sibling mating scheme) and through environments (sterile cages), creating a strong a priori bias to w ar d vertical transmission.T herefore , it is not surprising that social transmission is significant when studied under more natural conditions.
A substantial difference in transmission patterns was observed between the two house mouse subspecies.Relatedness and social contacts explained a greater portion of variability in the composition of lo w er gut micr obiota and corr elated with mor e bacterial ASVs in MMD compared to MMM.Although this difference could be attributed to variations in the gut microbiota composition between the two subspecies, we find this explanation unlikely.Rather than being the result of a major taxonomic r earr angement involving phylogenetically dissimilar bacterial taxa, the microbiota differ ences wer e primaril y due to c hanges in the abundance of closely related bacterial variants (Bendová et al. 2022 ), which lik ely de pend on similar dispersal strategies.
In this r egard, differ ences in the patterns of social interaction and r epr oduction between the subspecies (Mikula et al. 2022 ) appear to be more relevant.Specifically, MMD exhibits a more "demic" social structur e, wher e most social inter actions, including mating, occur within cohesive social modules identified with basic r epr oductiv e units or demes .T his stronger genetic and social structuring of the MMD population also implies a more structur ed micr obiota tr ansmission.Consequentl y, limited tr ansmis-sion opportunities may pr edominantl y influence micr obiota dissimilarity in MMD.In contrast, the ability of bacteria to utilize inter-individual transmission conduits may be mor e pr onounced in the more panmictic population of MMM.In other w or ds, limited inter-individual interactions between demes may restrict microbiota transmission and serve as a major axis of microbiota differentiation in MMD, but not in MMM.
In this experiment, we demonstrated the effects of transmission over the short time scale of se v er al host gener ations.Ho w e v er, the observ ed tr ansmission patterns can explain the co-div er gence of certain gut bacteria with their host populations/species (Moeller et al. 2016a, Waskito and Yamaoka 2019, Karcher et al. 2020, Suzuki et al. 2022 ) as a consequence of longterm inter-individual transmission of bacteria through structured mating and social transmission conduits.
At the le v el of individual bacterial lineages, a notable pattern was the association between multiple ASVs of the genus Helicobacter and the intensity of social interactions or relatedness .T he dependence of this genus on inter-individual transmission has been observed in human Helicobacter pylori , where population genetic variation at the geographic level aligns with ancient human migr ation r outes (Falush et al. 2003, Linz et al. 2007 ).In wild house mouse populations, the highly efficient transmission of intestinal Helicobacter is evident from its extremely high prevalencevirtually all individuals we examined harbored re presentati ves of this genus in their gut microbiota (Wasimuddin et al. 2012, Bendov á et al. 2020, Čížk ová et al. 2024 ).We have also experimentall y demonstr ated the easy tr ansmissibility of Helicobacter through social contact by co-housing adult laboratory mice lacking Helicobacter with individuals from wild populations (Moudra et al. 2021 ).Giv en the efficiency of Helicobacter tr ansmission, it is likely that colonization of the intestinal mucosa under natural conditions occurs early in the life of mice, which may lead to a priority effect (Debray et al. 2022 ).This implies that the already occupied m ucosal nic he can pr e v ent the subsequent incor por ation of other Helicobacter v ariants, whic h is consistent with our pr e vious finding that two closely related Helicobacter ASVs exhibit near-perfect specificity for host subspecies under common garden conditions (Bendová et al. 2022 ).T hus , it appears that efficient inter-individual transmission can maintain microbiota divergence in closely related hosts even in syntopic populations.Ho w ever, the specificity of mouse subspecies for their "own" Helicobacter may also contribute to this pattern.
In addition to Helicobacter , numerous other bacterial ASVs from various taxonomic groups have been associated with intensity of social contact or host relatedness, and these associations were consistent between host subspecies .T his suggests that some gut bacteria ar e systematicall y tr ansmitted between individuals and that these bacteria hav e differ ent life-history str ategies, including both sporulating species (e.g. from the classes Bacilli and Clostridia) and non-sporulating species (e.g. from the class Bacteroidia).
Knowing the age of the individuals in the population studied allo w ed us not only to quantify the ov er all effect of parental transmission on the gut microbiota, but also to assess how its magnitude c hanges fr om birth thr ough the period of par ental car e to adulthood.On the timescale corresponding to the age of the offspring, the similarity of the microbiota between parents and offspring sho w ed non-linear tr ajectories that differ ed significantl y between gut sections.Compared to the lo w er gut, the similarity of the ileum microbiota between parents and offspring displayed a less pr onounced a ge-dependent pattern.The tw o lo w er gut sections sho w ed compar able tr ajectories.During the br eastfeeding [i.e.before the age of ∼23 days; (König and Markl 1987 )], the microbiota of the colon and the cecum exhibited a low similarity with that of the par ents, whic h was e v en lo w er than the similarity between unrelated individuals .T his can be attributed to the influence of breast milk, which shapes the juvenile microbiota thr ough pr e-and pr obiotic compounds, antimicr obial factors or imm unoglobulins (v an den Elsen et al. 2019 ), r esulting in a distinct composition compared to that of adults (Wernroth et al. 2022 ).After weaning, the lower gut microbiota of the offspring gradually conv er ged to w ar d that of their parents .T he similarity was highest at 60-70 days of a ge, whic h corr esponds a ppr oximatel y to the time when offspring leave parental territory (van Zegeren 1979 , Groó et al. 2013 ), after r eac hing full independence fr om par ental car e and sexual maturity (at ∼35-40 days of age, [Latham and Mason 2004et al. 2004, Hiadlovská et al. 2015 )].Thereafter, the similarity of the lo w er gut microbiota between parents and offspring began to decline.At ∼100 days of age, it became statistically indistinguishable fr om micr obiota similarity between unr elated individuals, although it was still higher on av er a ge than the population-wide baseline .T hese patterns are consistent with the results of a recent meta-analysis of human gut metagenomes (Valles-Colomer et al. 2023 ), which sho w ed a decrease in the effects of parental transmission with offspring age.Some what sur prising was the compar able influence of both parents on the microbiota of their offspring.Considering the maternal transmission during birth and the closer contact with the offspring during parental care (König and Markl 1987 ), this pattern suggests the role of inherited genetic factors regulating the pro-lifer ation of tr ansmitted bacteria in the gut.Ho w e v er, it a ppears that these effects were less important compared to the impact of inter-individual transmission, as it is unlikely that genetic inheritance explains the peak of microbiota similarity at a specific age of the offspring.We found partial support for subspecies-specific v ariation in par ental effects, as MMM offspring exhibited gr eater similarity of microbiota with their mother between ∼60-70 days of life compared to MMD.
The transmission of microbiota from parents to offspring should also be reflected in the similarity of microbiota between siblings.Consistent with this expectation, the age-dependent patterns of microbiota similarity between parents and offspring and between siblings were largely congruent.The most striking difference was a persistent increase in sibling similarity relative to the population baseline .T his pattern suggests the pr esence of famil yspecific effects that have a long-term impact on the composition of the microbiota of the entire litter.For example, factors such as maternally or paternally induced stress in the offspring (Kemp et al. 2021 ) or other epigenetic factors such as a higher frequency of copr opha gy in young mice compared to adults (Ebino et al. 1987 ), could contribute to these effects.Ho w e v er, it is also possible that the high similarity of the siblings compared to the baseline is partly due to the fact that the sampled siblings were the same a ge, whic h was not the case in the baseline population or the parent-offspring pairs.
In conclusion, our findings demonstrate that inter-individual transmission, mediated either by the transmission of microbiota along genealogical lineages or by social contacts between conspecifics, has significant effects on the composition of the gut microbiota of the house mouse bred under semi-natural conditions.Ho w e v er, these effects ar e not consistent thr oughout the gastr ointestinal tr act, with inter-individual tr ansmission being notably weaker in the small intestine than in the lower gut.Importantl y, the r elativ e str ength of micr obiota tr ansmission fr om parents v aries gr eatl y depending on the a ge of the offspring.Failur e to account for these temporal dynamics can introduce strong biases in the quantification of parental microbiota transmission and affect consistency in replicated studies conducted with the same species.
Furthermore, we identified bacterial taxa that exhibited consistent transmission patterns between the two mouse subspecies, indicating their general ability for inter-individual transmission.Despite the similarity in the tr ansmitted bacteria, we observ ed clear differences in the strength of microbiota transmission between the two subspecies, likely influenced by variations in their social and r epr oductiv e networks .T his suggests that the transmission of bacteria among individuals is not only determined by their dispersal abilities, but also by the transmission opportunities mediated by interactions between individuals .T hese findings ha ve important implications for experiments conducted under laboratory conditions where mating and social contacts between mice are restricted.Although we were able to demonstrate the importance of social interactions in shaping the gut microbiota, our study was not designed to distinguish the r elativ e importance of different types of social encounters .T hese questions should ther efor e be clarified by further r esearc h.

Ac kno wledgements
Computational resour ces w ere supplied by the project "e-Infr astruktur a CZ" (e-INFRA CZ LM2018140) supported by the Ministry of Education, Youth, and Sports of the Czech Republic.We are grateful to all our colleagues who participated in the common gar den experiment.Finally, w e w ould like to thank the r e vie wers for their constructive comments, which helped us to impr ov e our manuscript.Computational resources were supplied by the pr oject.Car e of all mice mentioned in this manuscript, all pr ocedur es and experiments were performed according to the National Institutes of Health and the Office of Laboratory Animal Welfare guidelines for appropriate animal husbandry (OLAW assurance A3292-01), and all live mice were handled by an authorized person.

Figure 2 .
Figure 2. Effect of social interactions and genetic relatedness on abundance variation of bacterial ASVs .T he figure shows ASVs whose variation between individuals was significantly (FDR < 0.05) correlated with genetic relatedness or social interactions in red, nonsignificant associations in white, and ASVs not included in the analyses due to their low incidence in specific gut segments ( N < 20) in gr ay.Separ ate anal yses wer e performed for each gut segment (ileum, cecum, and colon) and mouse subspecies (MMM, MMD).

Figure 3 .
Figure3.Similarity in microbiota composition between siblings from the same litter.The Bray-Curtis index, which considers the relative abundances of ASVs, was used to examine how the similarity in microbiota between siblings varies on the time scale represented by sibling age .T he color ed line and shaded ar ea corr espond to the mixed model predictions and 95% confidence intervals for each gut section.T he gra y line and gra y shaded ar ea corr espond to the av er a ge similarity between genetically unrelated individuals (relatedness coefficient ≤ 0.25) and their 95% confidence intervals.