Delimiting Species with Single-Locus DNA Sequences

Abstract

DNA sequences are increasingly used for large-scale biodiversity inventories. Because these genetic data avoid the time-consuming initial sorting of specimens based on their phenotypic attributes, they have been recently incorporated into taxonomic workflows for overlooked and diverse taxa. Major statistical developments have accompanied this new practice, and several models have been proposed to delimit species with single-locus DNA sequences. However, proposed approaches to date make different assumptions regarding taxon lineage history, leading to strong discordance whenever comparisons are made among methods. Distance-based methods, such as Automatic Barcode Gap Discovery (ABGD) and Assemble Species by Automatic Partitioning (ASAP), rely on the detection of a barcode gap (i.e., the lack of overlap in the distributions of intraspecific and interspecific genetic distances) and the associated threshold in genetic distances. Network-based methods, as exemplified by the REfined Single Linkage (RESL) algorithm for the generation of Barcode Index Numbers (BINs), use connectivity statistics to hierarchically cluster-related haplotypes into molecular operational taxonomic units (MOTUs) which serve as species proxies. Tree-based methods, including Poisson Tree Processes (PTP) and the General Mixed Yule Coalescent (GMYC), fit statistical models to phylogenetic trees by maximum likelihood or Bayesian frameworks.Multiple webservers and stand-alone versions of these methods are now available, complicating decision-making regarding the most appropriate approach to use for a given taxon of interest. For instance, tree-based methods require an initial phylogenetic reconstruction, and multiple options are now available for this purpose such as RAxML and BEAST. Across all examined species delimitation methods, judicious parameter setting is paramount, as different model parameterizations can lead to differing conclusions. The objective of this chapter is to guide users step-by-step through all the procedures involved for each of these methods, while aggregating all necessary information required to conduct these analyses. The "Materials" section details how to prepare and format input files, including options to align sequences and conduct tree reconstruction with Maximum Likelihood and Bayesian inference. The Methods section presents the procedure and options available to conduct species delimitation analyses, including distance-, network-, and tree-based models. Finally, limits and future developments are discussed in the Notes section. Most importantly, species delimitation methods discussed herein are categorized based on five indicators: reliability, availability, scalability, understandability, and usability, all of which are fundamental properties needed for any approach to gain unanimous adoption within the DNA barcoding community moving forward.

Keywords: ASAP; Assemble Species by Automatic Partitioning; BIN; Barcode Index Number; GMYC; General Mixed Yule Coalescent model; Multiple threshold; PTP; Poisson Tree Processes; Single threshold.