Correct, the PSHELL cards seemed like they were located correctly. After HyperFEA pre-processed the FEM, and created all of the files that it uses, I manually attempted to run the *_i00.bdf file, which created a fatal error in NASTRAN. Then, I went to the *_UA.PM1 file and added the missing MAT2 definitions, and the bdf file ran successfully (and HyperFEA successfully ran through all of the iterations that I was asking it to do).
I don't know if it matters or not, but the specific information that I had on the PSHELL/MAT2 was set up so the PSHELL had one MAT2 for the MID1 field (membrane) and a different MAT2 for the MID2 field (bending). Each of the PSHELLS (my model had 15 like this) referenced a different 2 different MAT2 definitions, so there were a total of 30 MAT2 definitions that were omitted from the *_UA.PM1 file.