BWA mem vs. Sampe alignment

Question

[kirannbishwa01]

I have done BWA alignment before using - mem but need to rethink over it again. My main objective is to call polymorphisms from my genome resequenced data for two different populations for a plant model which has its reference genome sequenced.

I have illumina PE reads but after trimming the reads now range from 50 to 101 bp but majority above 80 bp.

1) why/which, -mem vs. -sampe would be the best choice for calling polymorphisms using the data I have?

2) I have done bwa mem with - M which flags the short split hits as secondary. why do we need this flag if we are aligning genomic reseq data? isn't it for data like RNAseq. If it can be used for genomic reads, how is it going to be useful.

Thanks,