Testing the neutral mutation hypothesis by distribution of single locus heterozygosity