Microbial colonization and community stability in a marine post-smolt RAS inoculated with a commercial starter culture

•Investigation of the microbiome during four early production cycles in the first marine post-smolt RAS in Norway.•Inoculating biofilter biofilm carriers with a commercial start-culture from brackish water.•Inoculums associated taxa were almost sole among dominating bacteria in the biofilter until the fourth production cycle, and microbes frequently identified over time showed higher average of relative abundance.•New and lost taxa between sampling times, was quantified and named OTUflow, and the flow was largest for microbes not originating from the start culture.•The establishment of the nitrite oxidating microbes in the biofilter was a delayed and dependent upon a lower salinity than provided with sea water. The performance of a commercial starter culture was investigated in a new marine post-smolt RAS, analyzing the microbial communities of 100 samples collected monthly over a year from biofilter biofilm carriers, tank wall biofilm, production water and fish skin. Totally 139 operational taxonomic units (OTUs) were defined in the starter culture, of which the classified members of Rhodobacterales, Bacteroidetes, Alteromonadales and Planctomycetes were largely the first colonizers of the biofilter carriers. Early colonizing OTUs that dominated biofilter biofilm carriers (> 5% relative abundance) were stably present over time, but the development went slowly from a few OTUs with very high relative abundance to several dominant ones with lower relative abundance. Operating taxonomic units not associated with the starting culture became prominent on the biofilter biofilm carriers only towards the end of the trial period. These were termed environmental OTUs. Comparing the two OTU quantitives in a ratio, where counts were based on all OTUs in the sample, the starter culture OTUs:environmental OTUs were 1.2 and 0.9 at the first and last sampling time for the biofilter biofilm carriers. Correspondingly, for all defined OTUs in the RAS sampling sites together, the ratio changed from 0.8 to 0.6 during experiment. Independent of origin, omniscient OTUs at a sampling site, did also have the highest relative abundances and were normally shared between biofilter biofilm carriers and the production water. New and lost OTUs between sampling times were on average 44 % of the OTUs defined, and this OTUflow was strongest for low abundant environmental OTUs. The maturation of the biofilter with respect to nitrification took long time, and the Nitrospira strain in the sta