Table 2.

Assessment of factors influencing microbial community beta-diversity in this study.

Data type	Distance metric	Factor	Adjusted R2	df	AIC	F	p-value
16S	Unweighted UniFrac	Host identity	0.93	60	−979.83	152.03	0.0002
	Unweighted UniFrac	Extraction kit	0.001	6	−988.95	3.2	0.0004
	Weighted UniFrac	Host identity	0.84	60	−428.1	58.62	0.0002
	Weighted UniFrac	Extraction kit	0.02	6	−533.75	19.27	0.0002
	Jaccard	Host identity	0.96	60	−1234.51	231.05	0.0002
	Jaccard	Extraction kit	0.002	6	−1254.66	4.92	0.0002
	RPCA	Host identity	0.89	60	−644.89	86.15	0.0002
	RPCA	Extraction kit	0.004	6	−663.81	4.73	0.0002
ITS	Jaccard	Host identity	0.8	59	−191.58	33.9	0.0002
	Jaccard	Extraction kit	0.02	6	−235.28	8.51	0.0002
	RPCA	Host identity	0.72	59	−30.05	22.32	0.0002
	RPCA	Extraction kit	0.02	6	−62.07	6.64	0.0002
Metagenomic	Unweighted UniFrac	Host identity	0.92	65	−1063.58	141.03	0.0002
	Unweighted UniFrac	Extraction kit	0.005	6	−1107.59	8.79	0.0002
	Weighted UniFrac	Host identity	0.87	65	−672.15	81.26	0.0002
	Weighted UniFrac	Extraction kit	0.02	6	−807.08	24.46	0.0002
	Jaccard	Host identity	0.93	65	−1195	168.79	0.0002
	Jaccard	Extraction kit	0.003	6	−1225.27	6.57	0.0002
	RPCA	Host identity	0.85	65	−579.25	70.95	0.0002
	RPCA	Extraction kit	0.01	6	−633.1	10.4	0.0002

Results are from forward, stepwise model selection, following Shaffer et al. (2021) [33]. Values are based on permutation tests of variation explained by redundancy analysis (n = 5000 runs), done separately for unique distance metrics for 16S, the fungal ITS, and shotgun metagenomic data. The full model included extraction round (i.e., Round 1 vs 2), sample biomass (i.e., high vs low biomass), sample type, host subject identity and extraction kit as model variables. 16S data were rarefied to 10,000 quality-filtered reads per sample or had samples with fewer than 10,000 reads excluded when using RPCA distances (n = 640 samples). Fungal ITS data were rarefied to 630 quality-filtered reads per sample or had samples with fewer than 630 reads excluded when using RPCA distances (n = 978 samples). Shotgun metagenomic data were rarefied to 2100 host- and quality-filtered reads per sample or had samples with fewer than 2100 reads excluded when using RPCA distances (n = 1044 samples). Rarefaction depths were selected to maintain at least 75% samples (50% for fungal ITS data) from both high- and low-biomass datasets.

AIC: Akaike information criterion; df: degrees of freedom; ITS: Internal transcribed spacer; RPCA: Robust principal components analysis (i.e., Robust Aitchison distance).