It's because I use special quantum-based supersymmetric nuetrino tachyon techonology in my crossovers. The govt won't let me say more about it.

Part of it's the smoothing, but look at the scales on the datasheet graph. That 'mess' at 400 to 600 is only about 3dB deep.
Also, I'm not sure those response measurements are quite right -- they weren't measured by Dayton, but by the manufacturer (NOT using OmniMic!) overseas. I've talked to the people at Dayton about redoing the plots themselves. That peak the plot shows at 150Hz makes no sense at all, I don't know how a low fs woofer could do that. Here's what I measured on that DS12 woofer (1/48th octave res, about 20" away):

For comparison, here's what I measured on a TD15M under similar conditions:

I do tend to focus on the 6th octave type stuff, particularly for dips.