Generally speaking, the scientific community's standards for species descriptions, at least for macroscopic organisms, include morphological or other non-molecular characters by which the species may be distinguished from its close relatives based on observation. One can describe species based only on fixed nucleotide differences, but doing so means that all future specimens may only be identified by sequencing them, as well. So in my opinion, you need to think about why you want to describe species based on barcodes before you start doing it. It is a slippery slope.

DNA barcodes must be used with caution, as Andrew V Z Brower said. However, they can be useful to support/elucidate morphology-based delimitation.

I suggest the paper of Collins & Cruickshank (2013) if you're not familiarized with DNA barcoding or the interpretation of the plenty of methods that exist out there. Find the paper attached.

Dear Vabeiryureilai Mathipi , the straightforward answer to your question is yes, the parameters you mentioned (Baysean Inference Probability , bootstrap support, p distance, k2p distance) should be considered during the description of an organism (species) when you are dealing with molecular data (DNA barcoding). But you should choose either p-distance or the k2p distance, not both together. For the other question, generally, we do not use transition and transversion mutation for the species description. As you mentioned you are a beginner, I hope the following content may help you- https://www.megasoftware.net/mega1_manual/Phylogeny.html

However, though I guess you know, only for your information, it is the best way to identify a species with an integrated method of both morphology-based and molecular (DNA barcoding) based taxonomy not to rely solely on a specific technique. I hope this answer will help you. Best wishes.