2.4 Putative species estimation
The two single-locus DNA barcoding methods consisted of two fundamentally different approaches. First, we implemented three distance-based approaches: (1) the Automatic Barcode Gap Discovery – ABGD (Puillandre et al., 2012) performed using the online version of the software (https://bioinfo.mnhn.fr/abi/public/abgd/abgdweb.html); (2) the Assemble Species by Automatic Partitioning – ASAP (Puillandre et al., 2021), an updated implementation of the ABGD hierarchical clustering algorithm, performed with the ASAP web version (https://bioinfo.mnhn.fr/ abi/public/asap/). Both methods used the MUSCLE aligned matrix as the input file and adopted the Kimura model, following the default settings for all parameters. (3) The Barcode Index Number (BIN), a method implemented in BOLD, in which newly submitted and already available sequences clustered in unique BINs using a refined single linkage analysis in which records with high sequence similarity and connectivity are clustered and separated from those with lower similarity and sparse connectivity (Ratnasingham & Hebert, 2013).
Second, we applied four tree-based approaches, which models speciation along the branches of an inferred phylogenetic: (1) the Single Poisson Tree Processes – sPTP (Zhang et al., 2013), implemented using the PTP online version (http://species.h-its.org/ptp); (2) the Bayesian Poisson tree process – bPTP (Zhang et al., 2013), also conducted on the PTP web-server. Both analyses were conducted with 500,000 MCMC generations and other parameters as default. (3) the Multi-rate Poisson Tree Processes – mPTP (Kapli et al., 2017), performed with the mPTP web-server (https://mcmc-mptp.h-its.org/mcmc/), using the multi-rate Poisson tree process model and following default settings. All PTP analyses (sPTP, mPTP and bPTP) used the ML trees calculated with Iq-Tree. (4) the Generalized Mixed Yule Coalescent – GMYC (Pons et al., 2006), performed by submitting the single ultrametric MCC tree resulting from BI obtained from BEAST to the online version of the GMYC software (https://species.h-its.org/gmyc/), following default single-threshold (sGMYC). We also tested the multiple-threshold model (mGMYC); however, it did not perform well, overestimating putative species (data not shown). Similar species estimates with the mGMYC algorithm were also observed in previous studies (Fujisawa & Barraclough, 2013; Schwarzfeld & Sperling, 2015).