Cryptic Diversity, but to What Extent? Discordance Between Single-Locus Species Delimitation Methods Within Mainland...

Single-locus molecular barcoding is a useful method for identifying overlooked and undescribed biodiversity, providing the groundwork for further systematic study and taxonomic investigation. A variety of methods for delimiting species from barcoding libraries have been developed and applied, allowing for rapid estimates of species diversity in a broad range of taxa. However, tree-based and distance-based analyses can infer different group assignments, potentially over- or underestimating the number of putative species groups. Here, we explore diversity of mainland species of anole lizards from the Chortís Block biogeographical province of northern Central America using a DNA barcoding approach, generating and analyzing cytochrome oxidase subunit I (COI) sequences for over 400 samples assignable to 33 of 38 (86.8%) native and one introduced mainland species. We subsequently tested the effects different models of nucleotide substitution, different species-delimitation algorithms, and reducing our dataset had on species delimitation estimates. We performed of two distance-based (ABGD, RESL) and three tree-based (bPTP, mPTP, GMYC) analyses on both the full dataset and a dataset consisting only of unique halotypes. From 34 nominal taxa, analyses of the full dataset recovered between 34 and 64 operational taxonomic units (OTUs), while analyses of the reduced dataset inferred between 36 and 59. Reassigning individuals to either mPTP-inferred or ABGD clustered (7.2% threshold) groups improved the detection of a barcoding gap across three different models of nucleotide substitution, removing overlap between intra- and interspecific distances. Our results highlight the underestimated diversity of mainland Chortís Block anoles, but the lack of congruence between analyses demonstrates the importance of considering multiple analytical methods when dealing with single-locus datasets. We recommend future studies consider the effects of different models of nucleotide substitution on proposed barcoding gaps, as well as the effect reducing a dataset to unique haplotypes may have on proposed diversity estimates.