3.1- Data summary 
We generated an average of 78,809 PITS (range: 9,352-282,57; Table S1) and 88,987 FITS (range: 15,409-382,888; Table S2) sequences for each of our 288 amplicon libraries (24 PCR replicates for each of six extracts, two primer sets). Following adapter removal and quality trimming, we retained an average of 37,640 PITS reads (range: 6,148- 166,279; Table S1) and 63,436 FITS reads (range: 12,360-323,310; Table S2 ) per PCR replicate. We were unable to assign to taxa an average of 2.3% of PITS and 76.1% of FITS quality trimmed reads. The large number of unassignable reads in the FITS data are probably the result of reference database incompleteness, which is a known problem for this taxonomic group and one that we attempted to mitigate by increasing the target depth of sequencing for our FITS libraries.
We found little evidence of contamination introduced during sample processing, and no evidence of index hopping between libraries during sequencing. Of the eight extraction negative control libraries and four PCR negative control libraries per primer that each had between 2,684 and 169,767 reads (Table S1 and S2), species were not shared between control and true samples. We found no evidence of contamination in our PITS data set. The FITS extraction negative control libraries contained a maximum of 11 reads that matched an “unidentified environmental” fungus. We removed all reads from the PCR amplicon libraries that were assigned to this “unidentified environmental” fungus. The PCR negative control libraries generated using both the PITS and FITS primers included only primer dimer chimeras. The three spiral ginger libraries used to track index hopping generated 52,675-299,400 sequences. All ginger sample sequences assigned to plant taxa aligned to Costus pulverulentus , and no sequences assigned to barcodes for libraries other than the ginger sample libraries were assigned to Costus , indicating there was no index hopping.