https://letmegooglethat.com/?q=What+is+the+difference+between+OTU+and+ASV+analysis%3F

OTU are merged based on homology, most of the time set to 95% for genus and 97% for species. OTU will then merge different reads into a same entity based on homology.

ASV (dada2) or other exact sequence variants tools considers each reads being different unless proven otherwise. For dada2, a Poison law is applied to the reads, based on the quality of lecture, to compare the probability of a read being different because of an error in sequencing or because of biology. In other words, the tool will assess based on quality of a given base if two reads are different or not.

The conceptual difference is quite important here. First of, it seems that OTU generates actually more taxa than ASV. Secondly, ASV will be able to track reads more precisely than OTU, especially in long read sequencing were dada2 authors are claiming that they can track allele differences. Finally, ESV is usually less consuming in term of computation.

OTU is now almost not used anymore, most labs are using ESV tools.