This is the peer reviewed version of the following article: Lyons, P. P., Turnbull, J. In this study, high throughput 16S rRNA sequencing was used to investigate the effect of a
171Samples were allowed to settle, and total genomic DNA was extracted and purified using the
192Illumina libraries were prepared following the method described by Caporaso, Lauber, Walters, All fish consumed both diets readily and upon conclusion of the trial, the weighed individuals 273 from the treatment group had a higher mean weight and condition factor than the control group.
274The final mean weight and condition factor (± SE) for the treatment group was 136.6 ± 12.1g 275 and 1.44 ± 0.06 whereas these values for the control group were 116.5 ± 9.3g and 1.33 ± 0.04 276 respectively ( Figure S1). A t-test was performed using the Minitab 15 statistical software to 277 test for significant differences between the performance parameters for both groups, however 278 no such differences were found (p = 0.107).
Sequence data and diversity analyses 280After quality filtering of sequences, a total of 18,282,541 sequences remained for analysis, The overall microbial community composition was similar in both the control and treatment 296 populations of fish. The distribution of OTU's at the phylum level of both the control and 297 treatment libraries is illustrated below (Figure 1). The vast majority of reads were assigned to 298 nine separate bacterial phyla, although an overall total of 13 phyla were recorded. Within these 299 phyla, 13 microbial classes dominated (Figure 2 is still yet to be fully characterized.
329Principal coordinate analyses, when visualized based on the thetaYC distance matrix 330 comparing similarities in community structure, showed that samples were broadly
412The Tenericutes were the dominant microbial phylum in the vast majority of samples, followed