a third of the hifi reads not present in gfa

[zhenzhenyang-psu]

Dear Haoyu,
I would like to extract hifi reads that contribute to assembly of contigs in asm.dip.hap1.p_ctg.noseq.gfa and asm.dip.hap2.p_ctg.noseq.gfa.
However, to my surprise, less than two thirds of the total hifi reads (5,874,368) are present in the asm.dip.r_utg.gfa file. And only ~1/3 of the reads comprise each of the haplotype p_ctg.noseq.gfa file.
Is this expected?

See below:
awk '$1=="A"' hifiasm.0.19.9.asm.dip.hap1.p_ctg.noseq.gfa |wc -l
3404927
awk '$1=="A"' hifiasm.0.19.9.asm.dip.hap2.p_ctg.noseq.gfa |wc -l
3310774

awk '$1=="A"' hifiasm.0.19.9.asm.dip.r_utg.gfa |wc -l
5874368

9082038 all.hifi.read.ids

Does this mean that 9082038-5874368 reads (which are around 1/3 of the reads) are not used by hifiasm to construct assembly?

Thanks very much,
best,
zhenzhen