Table 1.

Limits of detection of microbial cells across extraction kits.

Round	Extraction kit	Threshold (%)	16S	ITS
			LOD	Read depth	Samples retained	Samples retained (%)	LOD	Read depth	Samples retained	Samples retained (%)
1	PowerSoil	50	1.40E+03	73	81	86	2.66E+00	3	94	100
		80	1.40E+05	449	55	59	2.66E+02	223	45	48
		90	1.40E+05	1812	43	46	>2.66E+07	199,755	0	0
		95	1.40E+08	8676	39	41	>2.66E+07	165,759,100,584	0	0
	PowerSoil Pro	50	1.40E+02	46	79	84	2.66E+01	49	70	80
		80	1.40E+02	692	61	65	2.66E+04	1094	39	45
		90	1.40E+06	7520	33	35	2.66E+05	18,669	14	16
		95	>1.40E+09	145,206	0	0	>2.66E+07	698,158	0	0
	Norgen Stool	50	1.40E+07	1960	18	19	2.66E+00	2	88	100
		80	1.40E+07	4705	12	13	2.66E+05	850	7	8
		90	1.40E+08	8224	10	10	>2.66E+07	314,049,798	0	0
		95	1.40E+08	14,215	9	10	>2.66E+07	>1.00E+12	0	0
2	PowerSoil	50	1.40E+02	88	93	97	2.66E+00	3	91	100
		80	1.40E+02	3224	80	83	2.66E+00	6	91	100
		90	>1.40E+09	88,894	0	0	2.66E+00	12	86	95
		95	>1.40E+09	6,228,705	0	0	2.66E+00	24	75	82
	MagMAX Microbiome	50	1.40E+03	1286	79	82	2.66E+00	5	92	100
		80	1.40E+04	4147	66	69	2.66E+00	479	64	70
		90	1.40E+04	8970	63	66	>2.66E+07	451,442	0	0
		95	1.40E+04	19,454	39	41	>2.66E+07	176,467,642,943	0	0
	NucleoMag Food	50	1.40E+05	8379	61	64	2.66E+00	1	88	100
		80	1.40E+05	14,121	45	47	2.66E+00	2	88	100
		90	1.40E+06	19,430	27	28	2.66E+02	74	69	78
		95	1.40E+07	26,321	18	19	>2.66E+07	26,901,301	0	0
	Zymo MagBead	50	1.40E+02	489	42	44	2.66E+00	1	90	100
		80	1.40E+02	1838	32	33	2.66E+00	4	90	100
		90	1.40E+06	4531	22	23	2.66E+01	466	38	42
		95	1.40E+06	11,445	10	10	>2.66E+07	5,179,906,841	0	0

Titrations of a mock community containing known numbers of cells of bacterial and fungal species (see Materials & Methods) were used to identify the number of per-sample reads needed to meet certain LOD thresholds (i.e., the percentage of reads mapped to expected taxa vs contaminants). For each dataset, the read depth corresponding to a threshold of 50% was used for filtering samples before community analyses, as recommended [29,33]. The number and percentage of samples retained after filtering based on the read depth for each threshold and LOD estimates for bacterial and fungal cells are shown for 16S- and fungal ITS data, respectively. Rounds 1 and 2 indicate different sequencing runs; because sampling effort was not normalized such to compare absolute read counts, comparisons should not be made across sequencing runs.

ITS: Internal transcribed spacer; LOD: Limits of detection.