I have 4 samples of whole metagenome shotgun sequencing results (soil communities). Illumina, single end, read length ~76 bp. The raw reads were preprocessed using trimmomatic, adapters, short and low-quality reads were removed. The remaining reads were assembled using Spades (not --meta, because --meta doesnot work when reads are single end). Now I'm trying to map raw trimmed reads to my assemblies (bowtie2, default settings) and the mapping percent is 8-11%. Why so, any ideas? What I did wrong?