Breakdown is normally due to active sites in the sample path. You may just need to clean the inlet with the swab after running 515. Could be inlet, fused silica line from inlet to Y splitter, the splitter itself or the columns.
Let us know if you have more information on the setup you have.
I am using a Topaz drilled hole uniliner from Restek as my injection port liner, A 5m. 0.32mm ID guard column, and 0.32mm ID RTX-CLP, RTX CLPII 30m. column pair from Restek. I always replace the Uniliners with new ones before running any pesticides. Yesterday I had a 19 injection run of drinking water 505 pesticides, and the initial Endrin breakdown check was 27.6% on the front column and 19.3% on the back column. At the end of the 19 injection run (all clean drinking water standards and samples) I re-ran the breakdown check and it was 60% on the front column and 45% on the back column. My guess is there is either water in some of my extracts of gas impurities causing my Endrin breakdown to skyrocket after a short run of clean extracts.
Is switching to a split injection possible? We run 608/8081 on our 6890. We used to run splitless with the drilled hole uniliners and I frequently had trouble passing the endrin degradation check. I ended up switching to the Restek topaz precision liner with wool and a 10:1 split. My peak shape improved significantly and I haven't had difficulty passing a degradation check since then. In fact, I find I can run for a lot longer without doing maintenance than I used to be able to. Plus, I change my gold seal much less frequently. I know using wool seems like the worst thing for passing a degradation check, but it works for us. I don't know how "dirty" your sample matrices are as compared to ours, but it might be worth a try if you have enough response/signal-to-noise to try a split injection (my lowest calibration standards for 608/8081 are 5 ppb for pesticides, 50 ppb for toxaphene and 20 ppb for chlordane and I have acceptable responses at those levels).
For reference: We are using the same RTX-CLP column with no guard column, but we use helium as carrier gas. When I do find it necessary to do inlet maintenance I rarely cut more than ~8" from the column. We are also running 552 on the same instrument and I am usually able to pass a degradation check even after running a batch of 552 samples. Primarily, the instrument runs industrial waste samples for PCB/Tox/Chl screening and I am also usually able to pass degradation checks after running those samples (which are frequently quite dirty). I've also been doing some method development on a 7890/7000 triple quad for running pesticides by 625. I am using a Restek single taper gooseneck liner w/wool, HP-5MS UI column and a pulsed splitless injection. I have a bit more difficulty passing degradation tests on that instrument (maintenance is needed about every two extraction batches or any time I've run a raw influent extract). This suggests to me that the split vs splitless potentially makes a pretty big difference in the degradation.
(As an aside, if you're interested, I did a pretty comprehensive study of different liner types for our pesticides analysis when I switched from the uniliner to the precision liner. I wasn't evaluating endrin degradation, just peak shape and reproducibility. I'd be happy to send you my conclusions.)