The research, led by Christian Maltecca and Chiara Gini of North Carolina State University together with colleagues at Purdue University, Utah State University, the University of Missouri, and the University of Florence, tackled a problem that has dogged microbiome science for years. In small mechanistic studies, the number of molecular features — genes, transcripts, metabolites — vastly outnumbers the number of animals, making per-feature statistical analysis nearly meaningless. Signals drown in noise, and false discoveries proliferate. The team’s solution was to aggregate features into functional guilds defined by the KEGG Orthology database, grouping molecular players by the biological functions they perform rather than by their taxonomic identity or individual names. This aggregation, they showed, captured temporal structure that random groupings could not, passing a permutation test with a p-value of 0.012 and reproducing the results of standard KEGG-module profiling.

The experimental design was as rigorous as it was demanding for the animals involved. From a reference population of 1,381 commercial Landrace × Yorkshire sows, the researchers selected animals from the extreme top and bottom 1 percent of genomic estimated breeding values for thermotolerance. Thirteen thermotolerant sows and twelve thermosensitive sows underwent controlled cyclic heat stress from day 4 to day 18 of lactation, a window chosen because it coincides with peak milk production and maximal metabolic demand. Fecal samples were then profiled across three complementary omics layers: metagenomics on days 4, 8, and 14, capturing the genetic potential of the microbial community; metatranscriptomics on days 0 and 18, capturing which genes were actively being expressed; and metabolomics on days 0, 8, and 18, capturing the chemical end products of microbial metabolism. All animal procedures were approved by the Purdue University Institutional Animal Care and Use Committee.

When the team ran linear mixed-effects models on the guild-level data under false-discovery-rate control, a clear pattern emerged. The gut’s functional response to heat stress was dominated by temporal reorganization — changes over time — rather than by differences between the genetic lines. In the metatranscriptome, 13 of 22 functional guilds showed significant temporal changes that survived FDR correction. In the metabolome, 14 of 31 guilds did, with the sphingolipid guild standing out at a q-value of 2.3 × 10⁻⁷, one of the strongest signals in the entire dataset. Sphingolipids are lipid signaling molecules with roles in barrier function and inflammation, and their dramatic temporal shift suggests that heat stress reshapes the chemical conversation between gut microbes and the host intestinal lining.

Perhaps the most striking finding was what the researchers did not see. At baseline and at every subsequent timepoint, the two genetic lines were statistically indistinguishable in their functional guild profiles, with PERMANOVA p-values of 0.21 or higher. Genetic line explained at most 1.2 percent of the variance in guild abundance, while timepoint explained 16 to 22 percent. In other words, the clock — the physiological progression of lactation under heat stress — mattered roughly twenty times more than genotype. The functional response to heat stress was shared across thermotolerant and thermosensitive animals, suggesting that the gut microbiome’s behavior tracks the host’s physiology rather than its breeding value.

This does not mean genetics is irrelevant to heat tolerance. The authors’ interpretation is subtle but consequential: host genotype shapes gut function only indirectly, through physiology, and the divergence between thermotolerant and thermosensitive lines is expressed taxonomically rather than functionally. Different microbial species may populate the guts of the two lines, but those species appear to perform overlapping functional roles, so the overall functional output converges. For breeding programs, this implies that selecting for thermotolerance may not require the gut microbiome to be fundamentally different; instead, the microbiome may simply follow the physiological state of the host, whatever that state happens to be.

The methodological contribution may prove as important as the biological one. Multi-omics experiments in livestock are expensive, so sample sizes are typically small, and the statistical machinery developed for human cohort studies often breaks down when features outnumber subjects by orders of magnitude. By collapsing thousands of molecular features into a manageable number of KEGG-defined functional guilds, the researchers restored statistical power while preserving biological interpretability. They validated the framework against random feature groupings, demonstrating that the guild structure captured real, non-random temporal organization, and confirmed that it reproduced the results of conventional module-level profiling. The approach also lends itself to integration across layers: because metagenomic, metatranscriptomic, and metabolomic features can all be mapped to shared functional categories, the framework allows DNA potential, RNA activity, and chemical output to be compared on common ground.

The temporal dimension of the findings deserves particular attention. Heat stress was imposed from day 4 through day 18 of lactation, and samples were timed to bracket that window. The fact that so many guilds changed over time — robustly, reproducibly, and in both genetic groups — suggests that the gut microbiome undergoes a stereotyped trajectory under heat stress, a choreographed sequence of functional shifts rather than a random perturbation. Whether this trajectory represents adaptation, stress response, or a byproduct of reduced feed intake and altered metabolism remains an open question, but its consistency across animals hints at underlying regulatory logic that could eventually be targeted nutritionally, perhaps through dietary interventions that support beneficial microbial functions during the most vulnerable days of lactation.

For a pork industry grappling with rising temperatures, the practical implications are twofold. First, genomic selection for thermotolerance remains viable, and this study strengthens the case that it can proceed without needing to engineer a specific gut microbial profile. Second, interventions aimed at the microbiome — probiotics, prebiotics, or feed additives — may be most effective if they are timed to the temporal dynamics the study revealed, supporting the gut’s functional community during the phases when it is most actively reorganizing. The shared response across genetic lines is, in this sense, good news: a single intervention strategy might work across the whole herd rather than needing to be tailored to individual genotypes.

The study also carries a broader lesson for microbiome science beyond swine. As multi-omics datasets grow in complexity, the field is increasingly recognizing that function, not taxonomy, may be the more stable and meaningful unit of analysis. Two microbial communities can look entirely different species-by-species yet perform nearly identical functional work, a phenomenon this study documents elegantly in a livestock setting. By demonstrating that functional guild analysis can tame the statistical curse of small-n, high-dimensional studies, the researchers have offered a template that could be applied to heat stress in cattle, poultry, and even human studies of environmental stress. The gut, it seems, keeps its own calendar — and understanding that calendar may matter more than knowing which microbes are on the roster.

Subject of Research: Gut microbiome functional responses to heat stress in thermotolerant and thermosensitive lactating sows

Article Title: Functional guild analysis characterizes gut microbiome responses to heat stress in thermotolerant and thermosensitive lactating sows across three omics layers

Article References: Maltecca, C., Gini, C., Wen, H., Byrd, M. H., van Vliet, S., Brito, L. F., Johnson, J. S., Jiang, J., & Tiezzi, F. (2026). Functional guild analysis characterizes gut microbiome responses to heat stress in thermotolerant and thermosensitive lactating sows across three omics layers. BMC Genomics. https://doi.org/10.1186/s12864-026-13410-6

Image Credits: AI Generated

DOI: 10.1186/s12864-026-13410-6

Keywords: heat stress, gut microbiome, lactating sows, thermotolerance, multi-omics, metagenomics, metatranscriptomics, metabolomics, functional guilds, genomic selection, swine, sphingolipids