An interesting study and a fantastic example of well documented, reproducible code. Thank you, James et al!

re: contaminants. I took a look at your supplement and Rcode, and noticed lots of Halomonas. Is this sourced from the Phusion polymerase? I ask as I'm seeing similar trends in my initial wave of MiSeq data. Did you ever contact the manufacturer or look further into this?

Also, why do the 2-step PCR, as apposed to the 1-step outlined in the Fadrosh et al, 2014 Microbiome paper? Is this to limit effects of the inherent contaminants?

Thanks and great work!