Hello Marycarmen,
I took a look at your files, and I agree. Most reads make it through filtering and denoising, but then are removed during merging. 
q2-DADA2 will throw away any reads that are not able to merge. So we have to make sure you have enough area of overlap to merge your reads successfully.
Here is how I calculate area of overlap for paired end reads. 
(I'm not really choosing the overlap, I'm trying to calculate how long the overlap could be based on the length of the region amplified and the length of the reads).
|----------------------------------------------- 16S amplicon ~1500 bp
314f |> <| 806r primer
|---------------------| regioned amplified 806-314 = 492 bp
250 bp f |----------->
<---------| 250 bp r read
the area of overlap ^ (250+250-492 = 8 bp)
If you trimmed your reads:
|---------------------| regioned amplified 806-314 = 492 bp
250 bp f |----------->
<-------| 230 bp r read
Now there is no overlap^ (250+230-492 = -12 bp)
I'm not sure if I'm explaining this well... 
Do you think your reads will overlap?