If you refer to http://onlinelibrary.wiley.com/doi/10.1111/j.1574-6968.2009.01648.x/full, it seems to be very difficult to have on single set of specific primers to cover all. Thus, the option is to use universal archaeal primers, amplicons subjected to NGS and then extracting info on ANME subgroups. The sample suspected to have that content before http://www.ncbi.nlm.nih.gov/pubmed/21676272?dopt=Abstract&holding=npg analysed y 16S ideally before going to full metagenomes and metatranscriptomes http://www.nature.com/ismej/journal/v8/n5/full/ismej2013212a.html