Effects of environmental translocation and host characteristics on skin microbiomes of sun-basking fish

Variation in the composition of skin-associated microbiomes has been attributed to host species, geographic location, and habitat, but the role of intraspecific phenotypic variation among host individuals remains elusive. We explored if and how host environment and different phenotypic traits were associated with microbiome composition. We conducted repeated sampling of dorsal and ventral skin microbiomes of carp individuals (Cyprinus carpio) before and after translocation from laboratory conditions to a semi-natural environment. Both alpha and beta diversity of skin-associated microbiomes increased substantially within and among individuals following translocation, particularly on dorsal body sites. The variation in microbiome composition among hosts was significantly associated with body site, sun-basking, habitat switch, and growth, but not temperature gain while basking, sex, personality, or colour morph. We suggest that the overall increase in the alpha and beta diversity estimates among hosts were induced by individuals expressing greater variation in behaviours and thus exposure to potential colonizers in the pond environment compared to the laboratory. Our results exemplify how biological diversity at one level of organization (phenotypic variation among and within fish host individuals) together with the external environment impacts biological diversity at a higher hierarchical level of organisation (richness and composition of fish-associated microbial communities).

Methods
Estimation of alpha and beta diversity
All statistical analyses were performed in Rstudio v1.3.1093 (53) with R v3.6.0. We included three alpha diversity estimates. Observed number of ASVs (100% identical ‘amplicon sequence variants’) was used to illustrate the partitioning of ASVs according to sample type and environment. Statistical analysis of richness was based on estimates generated with default settings in the breakaway function (package breakaway v4.6.11). To incorporate an alpha diversity measurement that takes abundance and evenness into account we used Shannon-Weaver diversity index estimated from data subsampled to the smallest sample size (3775 reads per sample) using the diversity function in the vegan package (v2.5-6).

Apart from the subsampling prior to the Shannon index, raw data was used throughout the analyses. However, to explore the contribution of rare ASVs, we conducted a filtering step for comparison with the results based on raw data. To this end, we followed the method described in Bokulich, Subramanian: ASVs with a total count <10 within each sample and total abundance <0.01% across all samples were considered “rare”. This step decreased the total number of ASVs found in fish skin microbiomes from 16,881 to 1883 representing 11% of the total number of ASVs. The results based on data subsets are reported in Table S2 and S3.

For beta diversity, the data was transformed by centred log ratio (clr) allowing it to be used as input for linear regressions. Distances to group centroid for samples from each of the two environments were estimated from Euclidean distance matrix on clr-values, using the function betadisper (type = centroid) in the vegan package.