Hi 
Yes, you're totally right, some of my reads are less than 300bp. Here attached is my sequence length summary.
So the principle is that if my parameter is 300bp, it will remove all the reads less than 300bp and then merge them? I am not very clear about the reason.
In that case, I should set a number that less than the length of the shortest reads?
I tried command as follows:
qiime dada2 denoise-paired \
--i-demultiplexed-seqs /sbidata/lzhang/201911_hydractinia/RawData/16S_data/output/demux.qza \
--p-trunc-len-f 240 \
--p-trunc-len-r 220 \
--o-table table.qza \
--o-representative-sequences rep-seqs.qza \
--o-denoising-stats denoising-stats.qza
results show as below,
Trully thanks.
As @benjjneb mentioned, the high fraction of reads as chimeras should be worried. Next I would check this. Thanks:)