Jackknife environment clustering in UniFrac, by sample. (a) An unweighted UniFrac algorithm and (b) a weighted UniFrac algorithm were used, and were not normalized as different evolutionary rates of gene did not need to be accounted for. Jackknife counts for each are provided for each node. The weighted UniFrac algorithm takes into account abundance of sequences, and is better suited to analysis of mixed bacterial samples. Samples are labeled by individual moose (1–8) and sample type (rumen, R or colon, C), and gender, weight and age information is provided in the legend.