AD sum about half of DP in vcf file

Question

[MH85]

In my vcf file, the sum of total allelic depth (AD) is about half of the raw read depth (DP) for all SNPs. Is it common? If not, any thoughts on what might be the reason(s) behind this?

I use a common pipeline

bwa mem 
bcftools mpileup -Ov -Q 20 -R target.txt | bcftools call -mv -Ov

The issue remains even if I generate vcf directly from mpileup and sum allelic depth of all alleles not only reference and alternative.

Thanks a lot in advance