sure it doesn't look like a standard miscrosatellite gel. But it's not about the polymorphism here (it is obvious - banding patterns are different among accessions). It's about that you've somehow amplified multilocus fragmets.This looks more like an ISSR (inter-simple sequence repeats) banding pattern. You shoud check your primers. If they contain a miscrosatellite core sequence, then the products are ISSRs and they have to be analyzed as dominant multilocus markers.
As per previous suggestions by Mukta and Neha, you need to adjust the concentration of primer and polymerase along with increasing the annealing temp. However, this might also be due to contaminated primer. Try making new primer dilutions. Moreover, as per my experience, some SSR primers do give a lot of non-specific binding patterns and are not fit for studying introgression patterns. But these can be used to study diversity by scoring the bands on the basis of band size and finally developing a haplotypic coding.