The reads in the .fna file are pre-joined by the facility. They do not contain the primers or adapters, and they have faux barcodes appended to them that are present in my mapping file for demultiplexing. They are also chimera checked, denoised, and quality checked (but not filtered).
I have the raw fastq paired end reads (unjoined) for each sample. I just haven't played around with them yet. That's on my to-do list.
I was able to get the data into QIIME 2. I demultiplexed (but not quality filtered) using QIIME 1's demultiplex_fasta.py script. I converted to fastq and filtered each sample individually. After creating a manifest, I got everything loaded into QIIME 2. Type was: SampleData[JoinedSequencesWithQuality] and format was: SingleEndFastqManifestPhred33. Everything uploaded fine and the analysis is proceeding without any hurdles. I used deblur, following similar approaches to the forum post here and the Moving Pictures Tutorial.
In the meantime, I will take your advice so I can enter into QIIME 2 from the beginning.
Thank you!
*First edit was to finish the post and the second edit was to add this note.