Back from the Past: DNA Barcodes and Morphology Support Ablabesmyia americana Fittkau as a Valid Species (Diptera: Chironomidae)

Short, standardized gene fragments for species identification (DNA barcodes) have proven effective in delineating closely-related insect species, and can be critical characters to include in taxonomic studies. This is also the case for the species-rich and widely distributed fly family Chironomidae (non-biting midges). Inspired by observed genetic differences in partial COI gene sequences between North American and European populations of the chironomid Ablabesmyia monilis sensu lato, we investigated whether or not the morphology of male and female adults supported the distinction of more than one species. Our results support that the junior synonym Ablabesmyia americana is a valid species separate from A. monilis, and that A. monilis sensu stricto is distributed both in the Palearctic region and in North America. We provide re-descriptions of all of the major life stages of A. americana and of the adult female of A. monilis.


Introduction
Over the last ten years, molecular approaches to analyze species boundaries have seen rapid development.Genetic divergences in the partial cytochrome c oxidase subunit 1 (COI) sequences (so-called DNA barcodes [1]) have proven effective for species recognition in several animal groups, including non-biting midges belonging to the family Chironomidae (Insecta, Diptera) (e.g., [2][3][4][5][6]).In several cases, DNA barcodes have also aided the detection of presumably cryptic species, whose distinction a posteriori could be confirmed by both morphology and nuclear genetic markers (e.g., [7,8]).Furthermore, DNA barcodes have been shown suitable to associate life stages in the family Chironomidae [9][10][11].
Ablabesmyia is a large and ecologically diverse genus in the subfamily Tanypodinae.About 90 species are known worldwide ( [12] and citations within), many of them described from the Neotropical Region [13].These small to medium size dipterans (1.00-6.00mm) are found in a broad range of aquatic systems, from small streams and ponds to lakes and rivers.Larvae can live in freshwater sponge colonies [14].The genus was erected by Johannsen [15] with Tipula monilis Linnaeus, 1758, as the type species.Historically Ablabesmyia had a wider concept, with most of its members placed in the Pentaneura group A of Edwards [16] and Johannsen [17].Later, Freeman [18] and Roback [19] relegated Ablabesmyia to subgeneric status in Pentaneura (Philippi, 1865), while Fittkau [20] gave it full generic status.Ablabesmyia larvae are predators, and are recorded in many ecological studies (e.g., [21][22][23][24]).The genus is one of the most distinctive and well-defined genera within the tribe Pentaneurini, with adult males possessing a homologous complex of dorsomedial blades and lobes in their genitalia unique to the taxon [25,26].
The genus Ablabesmyia comprises four valid subgenera: Ablabesmyia Johannsen, Asayia Roback, Karelia Roback and Sartaia Roback.However, Oliveira, Silva and Gessner [13] reported inconsistencies in the establishment of these subgroups, and suggested at least one additional subgenus for Ablabesmyia, without promoting this taxonomic action.Identification of the adult Ablabesmyia is primarily based on features of the genitalia, as it is for many other members of Chironomidae.The morphological identification can be challenging, particularly for non-experts, and normally requires time-consuming dissections of the genitalia.In addition, some Ablabesmyia species lack diagnostic morphological traits in their adult life stage, and need associated larvae and pupae to be morphologically identifiable to species-level.Despite these obstacles, the taxonomy of Ablabesmyia has been investigated in multiple geographic regions, especially in the Neotropics (e.g., [13,14,[27][28][29][30][31]), providing identification keys to species-level, as well as extensive reference collections.This makes Ablabesmyia a good model taxon to investigate the congruence between molecular and morphological traits.
Ablabesmyia americana was erected by Fittkau [20] as a species separate from A. monilis (Linnaeus, 1758) based on the inner features of the male genitalia, but without specifying the exact characters.Subsequently, Roback [25] compared North American with European specimens, and was unable to find any valid differences to justify the distinction of a separate species for the New World specimens, synonymizing A. americana with A. monilis.Through the sampling and DNA barcoding of A. monilis populations in North America and Europe, we discovered considerable genetic divergence across the northern Atlantic region.We were therefore inspired to investigate if this divergence could be supported by the morphology of larvae, pupae and adult life stages, and if the synonymy of Ablabesmyia americana and A. monilis can be supported.

Materials and Methods
The specimens included in this study were selected to represent taxa with a possible taxonomic conflict in the Ablabesmyia gr.monilis sensu Fittkau [20].Fieldwork was conducted near and in the streams, rivers, fens, ponds and lakes of eastern, central and northern Norway, and Churchill, northern Canada (Manitoba).The material was initially collected for inventory studies and part of the International Barcode of Life initiative.Fourth-instar larvae and pupae were collected with hand nets, and adults sampled with sweep nets, emergence and Malaise traps.Specimens from Japan identified to A. monilis were kindly provided by Hiromi Niitsuma through Bohdan Bilyj.A small piece of tissue for DNA extraction was sampled under a stereomicroscope and sent to the Canadian Centre for DNA Barcoding at the University of Guelph (CCDB, www.dnabarcoding.ca) for DNA isolation, polymerase chain reaction (PCR) and sequencing.The remnants of nearly all of the sampled specimens were then macerated in potassium hydroxide (KOH) and mounted on permanent microscope slides in Euparal for species identification with a compound microscope.Measurements were taken according to Epler [32].Morphological terminology and abbreviations follow Roback [25] and Saether [33], supplemented by Kowalyk [34] for larval cephalic setation (head hairiness) and Silva et al. [35,36] for larval terminology.One slide-mounted male and one slide mounted female specimen from the Johannsen collection out of Ithaca dated prior to his 1905 work were analyzed.Seven additional slides with larvae and two with females were also examined, but these either did not have their date or their locality written on the labels.Material from the Johannsen collection is deposited in the Cornell University Insect Collection, Ithaca, New York, USA (CUIC).Other examined material is deposited in the NTNU University Museum, Norwegian University of Science and Technology, Trondheim, Norway (NTNU-VM); or the Centre for Biodiversity Genomics, University of Guelph, Canada.
Sequence information along with an image and collateral information for each voucher specimen were entered into the Barcode of Life Data System (BOLD) [37].The detailed specimen records and sequence information, including trace files, are available in BOLD through the dataset 'DS-ABLA, Ablabesmyia americana and A. monilis: A review' with doi: 10.5883/DS-ABLA.GenBank accessions are given in Table 1.
Polymerase chain reaction (PCR) products were purified for bidirectional sequencing with BigDye 3.1 termination (Applied Biosystems, Foster City, CA, USA) at the CCDB.Alignments were implemented on amino acid sequences, and refined using the Muscle algorithm in MEGA X [41], and were trivial, since no indels were present.Only sequences > 300 bp were included in the final alignment.
Maximum likelihood (ML) tree searches were performed using the Tamura 3-parameter with gamma correction for rate heterogeneity (T92+G).This model was determined to be the most suitable model of nucleotide substitution for our dataset using the hierarchical likelihood ratio test in MEGA X. Bootstrap analyses were conducted with 1000 pseudoreplicates [42].Neighbor-joining tree searches and analyses of intra-and interspecific genetic divergences were conducted using the Kimura-2-parameter (K2P) model [43] for easier comparison with most published DNA barcode investigations.We used Ablabesmyia aspera (Roback, 1959) and A. longistyla Fittkau, 1962 as outgroups in the analysis, and rooted the resulting tree with A. longistyla, as this species is morphologically most divergent from A. monilis sensu lato.The Barcode Index Numbers [44] equivalent to genetic clusters in our resultant tree were analyzed in BOLD to investigate the documented distribution of each genetic lineage.

Results
Mitochondrial DNA COI (mtDNA COI) sequences were recovered from 56 specimens of four Ablabesmyia species (Table 1).Most of the aligned sequences were 658 base pairs long (94.6%) with 164 variable sites (24.9%), and included 137 sites (83.5%) being possibly parsimony informative.The majority of the variable sites were in the third codon-position (Table 2).The sequences were highly AT-biased, particularly in the third position, with a combined average AT-composition of 89.2% (Table 2).A hierarchical likelihood ratio test of aligned sequences in MEGA X returned the Tamura 3-parameter with the gamma correction for rate heterogeneity (T92+G) as the best model (−lnL = 2435.952,BIC = 6045.744,AIC = 5096.616).Average intra-and interspecific K2P-distances for the examined Ablabesmyia species were 3.7% and 12.5%, respectively (Table 3).Maximum intraspecific divergence was found in A. aspera (8.0%).The lowest interspecific divergences were observed between Ablabesmyia americana and A. monilis (average 10.0%).Intraspecific and interspecific distances generated by the ML model yielded similar results (data not shown).There were no identical mtDNA COI sequences between species, and all species were distinguishable by genetic deviation and character state variances (Figure 1).The lowest interspecific divergences were observed between Ablabesmyia americana and A. monilis (average 10.0%).Intraspecific and interspecific distances generated by the ML model yielded similar results (data not shown).There were no identical mtDNA COI sequences between species, and all species were distinguishable by genetic deviation and character state variances (Figure 1).The ML tree shows large barcode divergence between all sampled Ablabesmyia species.Ablabesmyia americana and A. monilis comprised well-supported monophyletic clusters, completely concurring with the morphological identifications (Figure 1).For comparison, a Neighbor-joining tree was also generated using the K2P model, and produced identical trees (data not shown).Bootstrap support showed minor deviation between analyses, but were always 90% or higher for all species groups, except for the principal Ablabesmyia aspera group, which was endorsed by 84% of the bootstrap pseudoreplicates in the NJ-analysis.Nucleotide sequences from specimens recognized morphologically as Ablabesmyia americana and A. monilis differed by a minimum of 10.0%.The adult male of Ablabesmyia americana presents a shorter, less curved aedeagal blade, and slightly weaker, inferior volsella in the male hypopygium (Figure 2A,B), whereas A. monilis exhibits a longer, more strongly curved dorsal lobe, and more conspicuous inferior volsella.Regarding adult females, Ablabesmyia americana has less than one-fourth dark sclerotized seminal capsules as opposed to about one-half to two-thirds sclerotization in A. monilis (Figure 2C,D The ML tree shows large barcode divergence between all sampled Ablabesmyia species.Ablabesmyia americana and A. monilis comprised well-supported monophyletic clusters, completely concurring with the morphological identifications (Figure 1).For comparison, a Neighbor-joining tree was also generated using the K2P model, and produced identical trees (data not shown).Bootstrap support showed minor deviation between analyses, but were always 90% or higher for all species groups, except for the principal Ablabesmyia aspera group, which was endorsed by 84% of the bootstrap pseudoreplicates in the NJ-analysis.Nucleotide sequences from specimens recognized morphologically as Ablabesmyia americana and A. monilis differed by a minimum of 10.0%.The adult male of Ablabesmyia americana presents a shorter, less curved aedeagal blade, and slightly weaker, inferior volsella in the male hypopygium (Figure 2A and 2B), whereas A. monilis exhibits a longer, more strongly curved dorsal lobe, and more conspicuous inferior volsella.Regarding adult females, Ablabesmyia americana has less than one-fourth dark sclerotized seminal capsules as opposed to about one-half to two-thirds sclerotization in A. monilis (Figure 2C and 2D).The specimens determined as Ablabesmyia monilis by morphology build at least two distinct barcode clusters (Figure 1).However, we found no evident morphological characters that aid separation the species in these groups.Specimens of Ablabesmyia monilis were collected in Japan and Norway (Table 1, Figure 1), but one of the clusters (BIN BOLD:ABY9333) includes 327 records from eastern and western Canada in BOLD.Finally, specimens of Ablabesmyia aspera, used as an outgroup in our analysis, were grouped into three separate barcode clusters (BINs BOLD:AAF3626, BOLD:AAF3627, BOLD:AAF3638).Specimens of Ablabesmyia aspera of these three clusters were collected in Canada and Norway (Table 1), and exhibited a clear geographical structuring of genetic diversity across the distribution range, with one of the group composed only by species from Norway (BIN BOLD:AAF3626).Legs.Foreleg: Width at apex of tibia 66-70 µm, tibia with single, apical and pectinate spur 50 (1) µm long, with 9 (1) lateral teeth; ta 1-3 with preapical pseudospurs.Mid leg: Width at apex of tibia 67-70 µm, tibia with two apical spurs 36-37; 61-67 µm long, with 4-6 lateral teeth; ta 1-3 with preapical pseudospurs.Hind leg: Width at apex of tibia 65-71 µm, tibia with two apical spurs 33-36; 70-71 µm long, with 3-6 lateral teeth; comb not indistinct; ta 1-3 with preapical pseudospurs.Claws slender, distally recurved and pointed and with large basal protuberance.Lengths and proportion of leg segments as in Table 4.

Discussion
The analysis of partial COI sequences supports the separate species status of Ablabesmyia americana and A. monilis.Moreover, we could associate the species A. aspera and A. longistyla to molecular markers.The ML tree revealed that the majority of the analyzed species are distinctly separated by mtDNA COI sequences, and there are indications of morphologically cryptic lineages within A. monilis and A. aspera.
The presence of a barcode gap is a key concept in DNA barcoding [47].When the amount of intraspecific genetic divergence is substantially smaller than the amount of interspecific genetic divergence, a barcode gap exists [48].In our study, the pairwise interspecific genetic distances in the sampled Ablabesmyia species were clearly higher than the intraspecific divergences, demonstrating a barcode gap.
The average K2P distance values found for conspecific comparisons (3.7%) was discordant from those found in the literature, which range from 0.1% to 0.8% [49].However, this disparity can be a result of cryptic diversity among our sampled species (see below).The mean cutoff value for species delimitation is likely higher in Chironomidae than in many other insect groups, and previous studies have found that morphological species fit poorly with Barcode Index Numbers [50].For example, Lin, Stur and Ekrem [6] suggest a 4-5% threshold for genetically separate species of Tanytarsus (Chironominae), while Song, et al. [51] considered a 5-8% cutoff appropriate to discriminate species of Polypedilum (Chironominae).Herein, we refrain from suggesting an average threshold to delineate the species of Ablabesmyia.Yet any cutoff should be used with caution, since unusual deep intraspecific divergences can indicate cryptic species.Broader sampling throughout the distribution range of A. monilis would provide valuable data for the further analysis of the species boundary for this species.Nevertheless, data currently in BOLD shows that one genetic lineage of A. monilis sensu stricto (BOLD:ABY9333) is widely distributed in western Canada.Other BINs within A. monilis have records from Europe (BOLD:AAU0774, BOLD:AAF3633), Japan (BOLD:ADB8891), and central Norway (BOLD:ACE5418).
Ablabesmyia moniliformis was established by Fittkau [20] on the basis of Japanese species treated as Pentaneura monilis (Linnaeus, 1758) by Tokunaga [52], which he considered different from the European species [53].However, Fittkau proposed this new name without providing a diagnostic description, neither made he reference to one, nor designated a holotype.Yet the proposed name satisfies the provisions of Article 13.1.2 of the International Code of Zoological Nomenclature, and the name is available and considered a nomen dubium [29].Moreover, it seems that the Pentaneura monilis Tokunaga [52] type series may, in fact, comprises two or perhaps three species, including Ablabesmyia monilis, A. prorasha Kobayashi and Kubota, 2002 and A. jogancornua Sasa and Okazawa, 1991 [29].Furthermore, Tokunaga [52] did not describe nor illustrate the form of the aedeagal complex, a crucial feature for the diagnosis of the Ablabesmyia species.This has prevented the identification of these species unequivocally, and is likely the reasons that the name moniliformis has been overlooked (or disregarded) by Japanese peers [53].Based on our results, there is evidence supporting a separate cluster of A. monilis including records from Japan and Norway.This could represent one of the species in Tokunaga's Ablabesmyia monilis specimens, and thus A. moniliformis.However, no morphological differences could be determined between these two barcode clusters.In addition, the material used by Tokunaga [52], deposited in the Kyushu University Museum, is deteriorated by the long-term storage in alcohol [29].This prevents the comparison of our material with the Ablabesmyia moniliformis type series.The observed pattern could also be explained by the restricted geographical sampling of A. monilis throughout its distributional range.Therefore, additional material is needed to understand the variation between the specimens of A. monilis and closely-related species throughout their current distribution.Lastly, the female of Ablabesmyia alba, described by Chaudhuri, et al. [54] for India, resembles A. monilis by having seminal capsules with about one half to two thirds dark sclerotized, and may not be readily separable from this species.Further investigations using molecular data and morphological characters from all life stages will likely reveal the relationship of A. alba with A. monilis.
In addition to the divergences observed within A. monilis, there are two well-separated genetic lineages of A. aspera in Canada (BOLD:AAF3627 and BOLD:AAF3628), and a slightly divergent group with Norwegian specimens (BOLD:AAF3626), indicating cryptic species or divergence caused by geographical isolation in this group.Ablabesmyia aspera belongs to the monilis group sensu Roback, [19] and can be distinguished from all the species of that group by the longer, heavier aedeagal blades and the dorso-mesally approximated dorsal lobes, in natural position [19].The species is widely distributed in Canada and the United States [25].In Europe, Ablabesmyia aspera is only recorded from Norway [55].
DNA-aided species determination has become regular practice in various fields of study, including agriculture [56], bioengineering [57] and conservation biology [58].Moreover, DNA barcodes are recognized as a powerful tool to separate species in a range of chironomid groups [5,6,10,11,51,[59][60][61][62].We found mtDNA COI valuable in exploring sequence diversity and distinguishing morphologically similar species within the genus Ablabesmyia.Furthermore, DNA barcodes indicate further cryptic species in Ablabesmyia, or deep intraspecific divergence caused by geographical isolation.We believe that broader geographical sampling and the inclusion of nuclear genes will strengthen our findings and provide additional evidence for the geographical structuring of genetic lineages in Ablabesmyia.

Figure 1 .
Figure 1.Maximum likelihood (ML) tree for species of Ablabesmyia, based on partial COI sequences (DNA barcodes) and using the Tamura 3-parameter model with gamma correction for rate heterogeneity.Numbers on branches are bootstrap values > 80%.

Figure 1 .
Figure 1.Maximum likelihood (ML) tree for species of Ablabesmyia, based on partial COI sequences (DNA barcodes) and using the Tamura 3-parameter model with gamma correction for rate heterogeneity.Numbers on branches are bootstrap values > 80%.

Author Contributions:
Conceptualization, E.S., F.L.d.S. and T.E.; methodology, E.S, F.L.d.S. and T.E.; analysis, E.S., F.L.d.S. and T.E.; data curation, E.S. and T.E.; writing-original draft preparation, F.L.d.S.; writing-review and editing, E.S., F.L.d.S. and T.E.; visualization, E.S., F.L.d.S. and T.E.; project administration, E.S. Funding: DNA barcode data in this publication were in part generated in collaboration with the Norwegian Barcode of Life Network (NorBOL), funded by the Research Council of Norway and the Norwegian Biodiversity Information Centre.F.L. Silva was supported by fellowships from the São Paulo Research Foundation (FAPESP -2016/07039-8 and 2018/01507-5).Acknowledgments: Thanks to the team at the Canadian Centre for DNA Barcoding for help with DNA barcode analysis, and to Paul Hebert for initiating the chironomid barcoding of Arctic midges in Churchill through a

Table 2 .
Variable and informative sites, and average nucleotide composition in the aligned COI gene sequences.

Table 3 .
Intra-and interspecific Kimura 2-parameter distances between Ablabesmyia species defined by morphology.N/A denotes species with only one specimen analyzed.