Fish skin microbiomes are rarely studied in inland water systems, in spite of their importance for fish health and ecology. This is mainly because fish species distribution often covaries with other biotic and abiotic factors, complicating the study-design. We tackled this issue in the northern part of the Jordan River system, in which a few fish species geographically overlap, across a steep gradients of water temperature and salinity. Using 16S rRNA metabarcoding, we studied the water properties that shape the skin bacterial communities, and their interaction with fish taxonomy. We found that considering the skin-community contamination by water microbial community is important, even when the water and skin communities are apparently different. With this in mind, we found alpha diversity of the skin-communities to be stable across sites, but higher in bentic loaches, compared to other fish.Beta diversity was found to be different among sites and to weakly covary with the dissolved oxygen, when treated skin-communities were considered. In contrast, water temperature and conductivity were strong factors explaining beta diversity in the untreated skin-communities. Beta diversity differences between co-occurring fish species emerged only for the treated skin-communities.Metagenomics predictions highlighted the microbiome functional implications of excluding the watercommunities contamination from the fish skin-communities. Finally, we found that human induced eutrophication promotes dysbiosis of the fish skin-community, with signatures relating to fish health. This finding was in line with recent studies, showing that biofilms capture sporadic pollution events, undetectable by interspersed water monitoring.