The global demand for seafood is expected to double by 2050, necessitating efficient and sustainable aquaculture. Microbiome-directed dietary interventions, such as MDFs, offer a promising approach to improve animal production by targeting specific gut microbes. While MDFs have shown success in terrestrial animals, their efficacy in aquaculture remains largely unexplored. This study focuses on the effects of α-mannans and β-mannans, plant-derived glycans with established prebiotic effects in mammals, on the gut microbiome and physiology of Atlantic salmon. Previous studies have suggested positive growth effects from carbohydrate supplementation in salmonid feed, with α-mannooligosaccharides (α-MOS) showing particular promise. In vitro studies using a salmon gut simulator have also demonstrated that α-MOS supplementation alters the microbial community composition, increasing lactic acid-producing *Carnobacterium* and enhancing the production of propanoic and formic acids, both beneficial to animal health. However, the lack of genomic information for the salmon gut microbiome has hindered in-depth evaluations of MDF effectiveness. Recently developed resources, such as the Salmon Microbial Genome Atlas (SMGA), provide a valuable tool for functional omics analysis, allowing for a more comprehensive understanding of dietary effects and distinguishing between unaffected and metabolically activated microbial populations. This study uses the SMGA to systematically investigate the functional effects of mannan supplementation on the salmon holobiont (gut microbiome-salmon system) via a series of feeding trials with varying mannan inclusion levels and across different life stages. The primary objectives are to understand the mechanistic link between salmon gut microbes, their metabolic functions, and host physiology, and to evaluate the potential of mannans as MDFs in salmon aquafeed. This will involve 16S rRNA gene profiling, metatranscriptomics, targeted metabolomics, and transcriptomics of salmon organs.

Existing literature highlights the potential of microbiome-modulating feed ingredients to enhance aquaculture sustainability. Studies have shown positive growth effects from supplementing salmonid feed with carbohydrates, particularly α-MOS and fructo-oligosaccharides. In vitro research using salmon gut simulators indicates α-MOS supplementation leads to shifts in microbial communities, increasing lactic acid bacteria and beneficial short-chain fatty acid (SCFA) production. MDFs, such as β-mannans, have also shown promise in terrestrial animal production, selectively targeting specific microbial species and altering SCFA profiles. However, a significant gap exists in understanding the salmon gut microbiome's functional potential. Prior research was limited by a scarcity of genomic information, hindering the in-depth assessment of MDF effectiveness. The recent development of the SMGA, a comprehensive resource of salmon gut metagenomes and genomes, provides the necessary data to overcome this limitation and perform thorough functional analyses.

The study involved two feeding trials. The first, the "low dose mannan trial," used Atlantic salmon (initial weight 29 ± 0.97 g) fed either a control diet or diets supplemented with 0.2% α-mannan (MC1 and MC2) or acetylated β-galactoglucomannan (MN3). Sampling occurred at four time points across different life stages (parr, pre-smolts, smolts, and post-smolts). The second trial, the "high dose β-mannan trial," used salmon (initial weight 130 g) fed either a control diet or a diet supplemented with 4% MN3. Sampling was conducted at two time points. Both trials employed various -omics techniques: * **16S rRNA gene sequencing:** To profile bacterial community composition in feeds and salmon gut contents. * **Host transcriptomics:** To analyze gene expression in salmon gut tissue. * **Metatranscriptomics:** To analyze microbial gene expression in gut contents. * **Targeted metabolomics:** To analyze SCFAs in gut contents (high-dose trial only). Detailed protocols for DNA and RNA extraction, library preparation, sequencing, bioinformatic processing (including quality control, ASV inference, taxonomy assignment, and differential expression analysis), and statistical analysis are described in the supplementary methods. The Salmon Microbial Genome Atlas (SMGA) was utilized for genome-resolved metatranscriptomics analysis. Phenotypic scoring included measurements of weight, length, and organ integrity. Statistical analyses involved alpha and beta diversity estimations, differential abundance analyses, differential gene expression analyses, and correlation testing.

The low-dose mannan trial (0.2% MDF inclusion) showed no significant effects on key performance indicators (KPIs), gut microbiome structure, or host gene expression. Although 16S rRNA sequencing identified 839 bacterial genera, no MDF-driven structural changes were observed. Metatranscriptomic analysis showed only 208 differentially expressed genes (DEGs) between MDF-fed fish and controls, none metabolically linked to mannan. Similarly, host transcriptomics revealed no significant differences in gene expression between MDF diets and the control. The high-dose β-mannan trial (4% MN3) did result in significant changes in microbiome composition (Shannon diversity p=0.045, Bray-Curtis distance p<0.05), with increased abundance of *Burkholderia-Caballeronia-Paraburkholderia* (BCP) and *Pseudomonas*, and decreased *Limosilactobacillus*. However, this did not translate into changes in host phenotype or metabolism. Only 6 DEGs in the hindgut and 2 in the pyloric caeca were significant, and none were linked to galactoglucomannan utilization. Despite the lack of mannan-degrading capacity, analysis of metatranscriptomic data revealed active utilization of various carbohydrates (pectic galactans, xylans, chitin, etc.) by gut microbes, including *Lactobacillus* and *Limosilactobacillus*. Salmon RNA sequencing confirmed expression of transporters involved in SCFA uptake. Targeted metabolomics indicated an increase in acetate in fish fed 4% MN3. While the BCP group showed abundance shifts, this did not translate to metabolic interactions with the diet. The analysis highlighted the active roles of *Lactobacillus* and *Limosilactobacillus*, pointing towards potential substrates (pectin-derived galactans or (arabino)xylo-oligosaccharides) that might effectively promote their growth and beneficial outputs.

This study's findings challenge the initial hypothesis that mannan-based MDFs would effectively modulate the salmon gut microbiome and improve host physiology. The lack of significant effects at both low and high doses suggests that these specific MDFs may not be suitable for stimulating beneficial microbes under normal rearing conditions. The limited metabolic response to the mannans highlights the importance of high-resolution microbiome characterization prior to designing and implementing MDF strategies. The identification of active metabolic pathways and the potential for targeting endogenous lactic acid bacteria such as *Lactobacillus* and *Limosilactobacillus* provides valuable insights for future MDF development. Focusing on naturally present beneficial microbes and understanding their metabolic capacity is crucial for identifying effective pre- and probiotic candidates. Future research could explore alternative MDFs, such as pectin-derived galactans or (arabino)xylo-oligosaccharides, tailored to the enzymatic capabilities of identified beneficial microbes. Additionally, studies exploring microbiota responses under stress conditions or in disease states would identify additional targets for MDF interventions. The integration of the SMGA resource demonstrated its value in providing functional insights beyond taxonomic characterization.

This comprehensive multi-omics study revealed that the tested mannan-derived MDFs had negligible effects on salmon gut microbiome function and host physiology, highlighting the critical need for high-resolution, host-specific microbiome characterization before designing microbiome intervention strategies. The study underscores the importance of focusing on naturally present beneficial microbes and their metabolic capabilities for developing effective MDFs. Future research should explore alternative MDFs targeted at specific microbial groups and investigate microbiome responses under disease or stress conditions.

The study's limitations include the use of industry-standard inclusion levels that may not have been optimal, and the focus on specific mannan types. A larger-scale study with a wider range of MDFs and different inclusion levels would further strengthen the findings. Furthermore, the study focused on normal rearing conditions, and the response of the microbiome to environmental stressors or disease was not fully explored.