Per sequence GC content still not good after trimming

I am analyzing NGS data for a prokaryotic genome and FastQC initially failed “Per sequence GC content”. I then ran Trimmomatic on my Fastq files and after running FastQC on the trimmed paired dataset, the GC content is still not good. could someone tell me what is the reason for this?

The first .jnp shows the result I got before trimming and the second one is for after trimming.

Thank you