I do not think that dss being a pcr inhibitor matters much because you are amplifying something very well and sequencing the small band may give a good idea what your primers are doing and if the small band is related to the expected band. It might be that the strong amplification of small is inhibiting the right amplimer so a hemi nested or nested re pcr might work and it might be worth amplifying one sample in the presence of 6%dmso in case the shortmer is a looped out amplimer of the correct thing and dmso minimises secondary structure so helps amplification. As a control amplify normal mouse fecal dna to see if you get the small band. You have probably done this but Blast the primers against both mouse dna and gut floral dna.

If this band is non specific amplification try an annealing temperature gradient to well above the expected Ta and try running a sample with touchdown pcr to see if that improves the amplification. When everything has failed try titrating the dna of a single sample in the pcr...often with pcr inhibitors using less dna ( and less inhibitor) works better than more and yes this does contradict my earlier statement about pcr inhibition

Mr. Paul has already answered you very well with almost all possible ways to troubleshoot the problem. Could you share your primer sequences here?

A bit lateral thinking approach would be to seed the fecal mater to non-selective agar plates one in the normal incubator (aerobic microbiome) and one sealed in the bags perfused with nitrogen (anaerobic microbiome). Then you will have colony growth and much more bacterial material to work with. Then you can profile DNA extracts from the plates. You can even sample the colonies for the specific bacteria strains.