Good point about species differences -I work almost exclusively with human samples, so I forget that sometimes.
We have not had any success with 2x300 on MiSeq in at least two years, including a test of the newest kits just a couple of weeks ago (although we will be retesting that shortly). The quality is too poor to cover an amplicon of more than 450-500 bp. With the 500-550 bp amplicon from 5' RACE (of human BCR), we get low rates of merging R1/R2 and high error in the overlap region even for those reads which can be merged. We are already using primers close to the 5' end of CH1 (~25 nt in, if I remember correctly).