Problem with gaps and coverage in haploid assembly

[dtusso2020]

I am using Canu to assemble a yeast haploid genome from PacBio long reads, and I have encountered an issue related to the use of purge haplotigs.

After running purge haplotigs, the assembly improves in terms of continuity and number of contigs, but I have observed a significant reduction in the percentage of coverage when comparing the resulting assembly with the reference genome. When aligning the two generated haplotypes, I notice that regions not present in the main assembly are complete in the second haplotype. I have attached an image of the alignment visualized in Icarus (QUAST). Where the secondary haplotype contains sequences missing in the main assembly. There are gaps in the assembly that are partially covered by the other haplotype.

Is there a recommended strategy to merge these regions without losing haplotype separation?

I appreciate your time and look forward to any suggestions you can provide.