Borrowing the Pentatomomorpha tome from the DNA barcode library: Scanning the overall performance ofcox1as a tool

Abstract

Pentatomomorpha (Hemiptera: Heteroptera) occurs worldwide with almost 19,000 species within six superfamilies. Their wide morphological, behavioral, and ecological diversity is remarkable, making them subject of basic and applied studies. The mitochondrial cytochrome c oxidase subunit I (cox1) proposed as a DNA barcode for Metazoa is used in species discovery and identification, relying on threshold values to split intra‐ and interspecific sequences and a comprehensive library to accurate identification. Here, we scanned all the pentatomomorphan cox1 sequences on Barcode of Life Data System library aiming to provide an overview of available data; verify barcoding gaps at generic level; infer individual empirical threshold values to distinguish congeneric species; and test efficiency of cox1 based on the Probability of Correct Identification (PCI) analysis. Our final dataset comprised 12,189 sequences, covering five superfamilies, 32 families, 460 genera, and 1068 species. The dataset abundance and composition were biased to families with economic importance, that is, Pentatomidae, Lygaeidae, Scutelleridae, Coreidae, and Rhyparochromidae. Barcode gaps were detected for most of the analyzed genera, reaffirming the efficiency of cox1 for Pentatomomorpha. We inferred threshold values for 131 genera and found a global PCI of 74.33%, suggesting that one out of four analyzed species suffer from operational biases or hide cryptic species. We brought examples to illustrate how cox1 can be used to flag inconsistencies, refine, and shed light onto future studies on Pentatomomorpha. We emphasize the efficiency of cox1 as DNA barcode for these true bugs, advocating for its combined use with, for example, morphology in integrative approaches.

Exploring Large-Scale Patterns of Genetic Variation in the COI Gene among Insecta: Implications for DNA Barcoding and Threshold-Based Species Delimitation Studies

“…Virgilio et al (2010) compared the performance of DNA barcoding across six insect orders by using 15,948 COI sequences of 1995 species [ 45 ]. Bianchi and Gonçalves (2021) explored the genetic variation in the COI genes of 1068 species from Pentatomomorpha [ 46 ]. Nevertheless, the numbers of insect species and the taxon coverage that are reported in previous studies are still limited, which makes the evaluation of the genetic variation in COI among Insecta inaccurate.…”

Identifying the Genetic Distance Threshold for Entiminae (Coleoptera: Curculionidae) Species Delimitation via COI Barcodes

“…Standard thresholds have been investigated in many insect groups. For ladybirds and cryptorhynchine weevils, 3% can be used as the standard threshold for COI barcoding [ 43 , 46 ], 3.3% for pentatomomorphans [ 18 ], 5% for fruit flies [ 50 ], and a threshold value above 7% for species of Cleus [ 21 ]. 2.20% is the universal threshold applied in BOLD, and a new OTU (Operational Taxonomic Unit) is formed when the threshold of sample sequence is more than double the standard (>4.40%) [ 38 ].…”

“…Future research on Entiminae may benefit from these empirical values as they are more reliable than arbitrary, fixed thresholds. As new sequences and taxonomic knowledge amass, re-evaluating threshold values will improve their accuracy for species delimitation and hence applicability [ 4 , 18 ].…”

“…In DNA barcoding, taxonomic names are assigned to query sequences after comparing their sequences with those of reference databases. Although it is more accurate to use multiple loci for species delimitation analyses [17], single-locus species delimitation method is still widely applied in DNA barcoding studies of bacteria, fungi and invertebrates [18,19]. Over the last 17 years, 9395 Entiminae COI-5 sequences from 534 species were available in the Barcode of Life Data Systems (BOLD, http://www.boldsystems.org/, 15 October 2020), providing valuable resources for species delimitation in this subfamily.…”

“…Instead of depending on fixed thresholds, it is better to directly generate an optimized threshold from the data [ 20 , 26 ]. We also need to reassess these threshold values when taxonomic knowledge is updated and new sequences are generated, to improve its reliability and practical applications [ 18 ].…”