The Afro–Oriental Genus Yaeprimus Sasa et Suzuki (Diptera: Chironomidae: Chironomini): Phylogeny, New Species and Expanded Diagnoses

Expanded generic diagnoses of all life stages of Yaeprimus Sasa et Suzuki, 2000 (Lunditendipes Harrison, 2000, syn. n.) are given. Yaeprimus tropicus comb. n. is redescribed as an adult based on type material. Additionally, a new species Y. balteatus sp. n. from Oriental China is described based on the adult male and pupa. The phylogenetic position of Yaeprimus within Chironomini and the validity of the new species are explored based on concatenated five genetic markers (18S, 28S, CAD1, CAD4, and COI-3P) through both mixed–model Bayesian inference and maximum likelihood methods. The results strongly support Yaeprimus as sister to Imparipecten Freeman, 1961, which counters a previously proposed systematical position based solely on morphology.


Introduction
Yaeprimus , was established based on adult males of Yaeprimus isigaabeus Sasa et Suzuki [1] collected from Ishigaki Island in Japan. Subsequently, it was revised in detail of all stages by Yamamoto and Yamamoto [2] based on reared associated material. The genus was stated to show close relationship to some Lauterborniella related genera, such as Apedilum Townes, Zavreliella Kieffer and Paralauterborniella Lenz which are characterised in the larva by having a six-segmented antenna and alternate Lauterborn organs (Microtendipes group sensu Cranston et al. [3]), although the adult male of Yaeprimus lacks the median volsella or enlarged superior volsella base characteristic of the group.
Simultaneously, Harrison [4] described four new Chironomidae genera from South Africa, amongst which males of Lunditendipes Harrison shared similar fore tibia and anal tergite setae with the Asian Yaeprimus. Noticeably, an important diagnostic character that appears to distinguish Lunditendipes from Yaeprimus is the absence of basal setae of the superior volsella according to Harrison's original description. However, this is a flawed observation according to the examination of the types materials deposited in ABM by Helen Barber-James. Actually, those specimens bear two inner basal setae clearly arising from tubercle-like setigers and two heavily sclerotized concavities each containing two strong setae in tergite IX, thus resembling Y. isigaabeus [5]. The same character states have been observed also on material deposited in ZSM, which were collected from Kruger National Park in north-eastern South Africa and identified as Lunditendipes by Martin Spies [6]. The two species of Y. isigaabeus and

Molecular Work
DNA was extracted using MAGEN ® (Beijing, China)Tissue DNA kit in the Molecular Lab of Institute of Groundwater and Earth Science, Jinan University, and QIAGEN ® (Hilden, Germany) DNA Blood and Tissue kits at the Tianjin Agricultural University. Standard protocols were followed except for the lysis time and final elution volumes, and all the samples were lysed overnight at 55 • C and eluted with 40 µL of eluent. Universal primers were adopted following previous studies (Table A1). Processes for gene fragments amplification were followed as previous studies except for slight moderation of annealing temperature [18][19][20]. Polymerase chain reaction (PCR) products were electrophoresed in 1.0% agarose gel, then shipped to Majorbio Company, Guangzhou for purification and bidirectional sequence. One mitochondrial gene (COI-3P), two ribosomal genes (18S and 28S), and two sections of the nuclear protein-coding gene (CAD1 and CAD4) were chosen as in Cranston's work to match the comprehensive dataset of Chironomidae. Additionally, the standard barcode, one fragment of the mitochondrial gene (COI-5P) proposed by Hebert [21] was sequenced to explore cryptic species and calculate generic distance.
Forward and reverse sequences were assembled automatically and manually edited with Sequencher 4.8 (Gene Codes Corp.). Alignment of the sequences used Muscle algorithm [22] on nucleotides in MEGA X [23]. Some ambiguous bases were eliminated based on the results of alignments and trace file, while the remnants were adopted and showed in the International Union of Pure and Applied Chemistry (IUPAC) code. For protein-coding genes, introns were excised using the GT-AG Diversity 2020, 12, 31 3 of 21 rule [24] and an amino acid alignment was used as a guide to elucidate exon/intron boundaries. For 18S and 28S rDNA, ambiguous regions were excluded with GBlocks v0.91b using default setting except allowing half gap positions within the final blocks [25,26]. All selected genes except for the standard barcode (COI-5P) were concatenated with PhyloSuite v1.1.14 [27] to implement the maximum likelihood and Bayesian inference. In case of missing gaps, they were filled up by "?" to ensure that all sequences were in the same length. The optimal models for each subset were selected by Partition Finder 2 [28] based on the Bayesian information criterion (BIC) and corrected Akaike information criterion (AICc). The best scheme was as follows: GTR+I+G for the 18s, 28s and the first two codons for all protein-coding sequence, GTR+G for the third codon of COI-3P and GTR+I+G for third codon of CAD1 and CAD4. Maximum likelihood (ML) phylogenetic analysis was conducted using IQ-TREE 1.6.8 [29] with 1000 bootstrap replicates in a rapid bootstrap analysis and a "greedy search" for the best-scoring ML tree. Bayesian inference was performed in MrBayes v3.2.6 [30]. During the processes, Markov chain Monte Carlo (MCMC) iterations were run with four chains on two runs for 10 million generations, sampled every 100 generations with a burn-in of 0.25. Convergence among the runs was monitored using Tracer v1.6 [31], with the first 25% trees discarded as burn-in. The final average standard deviation of split frequencies was 0.003. Both analyses were completed using the best fitting scheme selected by Partitionfinder. Two species of tribe Tanytarsini were selected as outgroups for this has been considered to be the nearest neighbor of the tribe Chironomini.
In total, 235 sequences of 51 specimens were added to the molecular dataset, 112 of which were downloaded from GenBank (https://www.ncbi.nlm.nih.gov/genbank/) and five of which retrieved from BOLD (Table A2). Forty species were chosen to stand for four complexes, two groups and some ambiguous genera within Chronomini referring to a previous study [3]. Members of Microtendipes were enlarged particularly to test the Yamamoto hypothesis. Specimens with less than three markers were excluded from the dataset of concatenated genes. List of all species, specimens, individual images, georeferences, primers, sequences, trace files and other relevant laboratory data of sequenced specimens can be seen online through the publicly accessible dataset "Yaeprimus" on the BOLD website (www.boldsystems.org) [32,33].

Mapping
The distribution map ( Figure 1) was made using ArcGis™ software [34], with all possible GPS locations of 17 sites implanted into the vector of World Map (http://www.vectorworldmap. com). For older specimens without GPS data, estimates were made from the finest available detail (e.g., city/country) available from either specimens or publications.

Mapping
The distribution map (Figure 1) was made using ArcGis™ software [34], with all possible GPS locations of 17 sites implanted into the vector of World Map (http://www.vectorworldmap.com). For older specimens without GPS data, estimates were made from the finest available detail (e.g., city/country) available from either specimens or publications.  Terminology: Life stage, F, female; L, larva; M, male; P, pupa. Morphology, T II-X, tergite II to X; Ta 1 -Ta 5 , tarsus 1 to tarsus 5. Molecular, BS, bootstrap support; PP, posterior possibility.

DNA Barcode
Three COI DNA barcodes of new species from adult males clustered into the same BIN (BOLD:ADH0469), with a maximum intraspecific pairwise genetic distance of 0.64%, and 11.38% divergence to the nearest BIN (BOLD:ACT7861). The nearest neighbor (Sample ID: BIOUG28352-C08) from Guanacaste in Costa Rica, was unidentified in BOLD.

Phylogenetic Analysis
The structure of the maximum likelihood (ML) tree was basically similar with that of the Bayesian inference (BI) tree except for some weakly supported clades. As expected, both approaches strongly supported (node A, BS = 100, PP = 1) that Y. isigaabeus and new species group together. Yaeprimus together with Imparipecten Freeman forms a new clade in both trees (node B, BS = 82, PP = 0.98), then the clade is sister to the assemblage of Chironomus complex, Harnischia complex and Nilothauma Kieffer (node C) in ML tree, while shifts to the assemblage of Polypedilum Kieffer, Endochironomus Kieffer and Stenochironomus Kieffer (node C) in BI tree, but either connections to Chironomus or Polypedilum clades without support. The positions of remaining genera mostly conform to previous results [3]. Besides Yaeprimus, something interesting has been discovered in our study. After the inclusion of Chinese populations, Nilothauma is verified as sister to the Chironomus complex + Hanischia complex (node D, BS = 83, PP = 0.99), which showed some tendency in previous work [3] but without robust support. In a Microtendipes group (Node E, BS = 90, PP = 0.99), the Chinese populations of Paratendipes (node G) are paraphyletic although lack of support. Saether's hypothesis [35], based on characters of female adults that Patatendipes is sister to Microtendipes + Nilothauma, is rejected in both analyses. Our results show that Patatendipes (node F, BS = 100, PP = 1) is close to Paralauterborniella Lenz, while Microtendipes is close to Australian Paucispinigera Freeman (Node H, BS = 99). The positions of Apedilum Townes (node I) and Paraborniella Freeman (node J) vary within acceptable range between two analyses, tough both nodes are weakly supported (Figures 2 and 3).
Terminology: Lifer stage, F, female; L, larva; M, male; P, pupa. Morphylogy, T II-X, tergite II to X; Ta1-Ta5, tarsus 1 to tarsus 5. Molecular, BS, bootstrap support; PP, posterior possibility.  Three COI DNA barcodes of new species from adult males clustered into the same BIN (BOLD:ADH0469), with a maximum intraspecific pairwise genetic distance of 0.64%, and 11.38% divergence to the nearest BIN (BOLD:ACT7861). The nearest neighbor (Sample ID: BIOUG28352-C08) from Guanacaste in Costa Rica, was unidentified in BOLD.  Three species conformed to most generic diagnosis given by Yamamoto and Yamamoto [2], except for the following emendations.
Legs ( Figure 4F, Figure 8B,C). Tibial combs of mid and hind tibiae nearly or completely fused; if separated narrowly (Y. balteatus, Y. tropicus), the large (inner) comb usually bears 0-2 straight spurs, the small (outer) comb always has a long apically hooked spur. If fused (Y. isigaabeus), there is only one hooked spur, arising more proximally on the outer surface of the comb base. The number of spurs per tibia variable even within a single specimen. Pulvilli present.
Abdomen ( Figure 5A). Anterolateral areas of first segment slightly sclerotized with two distinct patches bearing several concentrated setae; T II-VIII with two rows of regular transverse setae centrally. Tergite VIII slightly tapered anteriorly ( Figure 4G).
Hypopygium. Tergite IX with a regular row of median anal seta, usually grouped laterally, arising from the distinct pigmented oval field ( Figure 4H) or blank oval pits ( Figure 8F). Anal tergite band absent or weak. Superior volsella ( Figure 6C,D, Figure 8F) with bare base, 1-3 basal setae, arising from the distinct tubercle base, digitus slender distally or with a slightly elongated ventro-lateral ledge distally (hooked), covering a partial or the whole width of digital apex, lacking any outer seta on digitus. Gonostylus was normal ( Figure 4H) or reduced ( Figure 8F), with several distal-medial setae of different sizes, the seta on the distal-inner corner being the longest and thickest.
Pupa Tergite spinulation of III-V split into anterior and posterior patches or completely fused. All spinules were nearly uniform sized. Conjunctival bands present on T III and T IV, continuous or medially interrupted. Posterolateral corner of T VIII with 'comb' of 1 main tooth and 3-4 small side teeth. Taeniae pattern ( Figure 7A) on A IV-VIII, 4, 4, 4, 4, 4. Uniserial fringe with 20-30 taeniae. Dorsal seta of anal lobe present.
Distribution. Yaeprimus was known only from two small Japanese islands for Y. isigaabeus. Our study has expanded the genus distribution to south China, South Africa, and Zimbabwe. All specimens have been collected from subtropical and tropical regions.
Remarks. The integrated systematical work of Yaeprimus has not been conducted completely before this study. The previous suggested phylogenetic placement was based solely on selected distinctive morphological character states of three life stages, rather than through a formal data matrix. Previously-argued conclusions were not fully reliable lacking rigorous parsimony analysis. Some important characters were ignored, for example, the anteriorly tapered tergite VIII, the inner setal arrangement of gonocoxite, the condition of humeral pit, and the abdominal setae and the pulvilli status. The current inclusion of an additional two species expands the variation within the genus, Head. Antenna ( Figure 4C) with pale brown (approximal and distal) or brown (middle) flagellomeres, with almost dark plume.   Thorax ( Figure 4D). Antepronotals 1-2; dorsocentrals 6-9, 6, usually alternately accessorized with 3-5 tiny pits; acrostichals 10-16, 14 (3) arranged in robust two rows, ending in anterior 1/3 before the hump; prealars 3-4, 3, supraalars 1. Scutellum had 7-8 setae, in single row. Tiny trans-oval humeral pits present.
Pupa (n = 1) (Figure 7)    Figure 4A,B and Figure 5A). Generally brown with some pattern in legs and abdomen. Legs were yellow, except for dark brown all femur and complete mid tibia, and pale brown apex of fore and hind tibia. Fore-tarsus with brown apex and gradually brown in Ta 2 -Ta 5 , others tarsi all pale yellow ( Figure 4A,B for colorful photo). Anterolateral corners of A I, posterior half of A III and A V, and almost entire A VI-VIII dark brown. A IX was dark brown, the hypopygium with brown gonocoxite and pale yellow gonostylus.
Pupa (n = 1) (Figure 7) Total length ca. 3.0 mm. Cephalic tubercle absent ( Figure 7B), frontal setae small, 40 µm long, subequal to the gap between two frontal setae. The thoracic setation as in Figure 7C, thoracic horn invisible. The abdomen ( Figure 7A) with dense spinulations in T II-IV, no clear delimitation between the anterior patch and median patch. Continuous conjunctival spinule bands present in T III and IV. The tergite II hook row continuous, short, 30% of the width of segment II, comprising ca. 20 hooks. A V-VII distorted. Comb VIII with one main tooth and three small accessory teeth. The anal lobe 140 µm long and 150 µm wide, with 20-24 taeniae, dorsal seta present.
Remarks. The new species shares the same comb pattern with Y. tropicus and reduced gonostylus with Y. isigaabeus. It can be distinguished from the other two species by the distinct scutal tubercle and anal median tergal seta arising from the common pale pit rather than sclerotized concavities. For pupa, the new species can be separated from Y. isigaabeus by the fused sub-rectangular spinulations and small point-free area in the middle area of T II-IV and conjunctives continuously.
Distribution: China (Guangdong and Hainan). Anal tergite with two pairs or three pairs of median setae, arising from a heavily pigmented field, grouped laterally. Superior volsella base with 1-2 inner seta, without microtrichia, digitus with a basal inner seta, and distal elongated, with a ventro-lateral ledge apically.

Yaeprimus isigaabeus Sasa et Suzuki
Pupa Cephalic tubercle absent, frontal setae reduced, subequal to the gap between two setae. Dorsal seta of anal lobe present. This species has been described by Harrison [4]: some emendations and additional characters are given here.
Male (n = 4) ( Figure 8) AR 1.2-1.5, LR 2.1-2.3; thorax ( Figure 8A) slight hump, without scutal tubercle, small pale humeral pit present. Mid ( Figure 8B) and hind tibia ( Figure 8C) with two separated combs, the small combs with long-hooked spurs, the large comb without spur in the mid tibia, with 1-2 outstanding straight spurs in the hind tibia. Pulvilli present. Abdomen II-VIII with two regular rows of setae, T VII ( Figure 8D-F) slightly tapered anteriorly. Location of anal tergite median setae ( Figure 8D-G) as that in Y. isigaabeus, two pairs of strong setae arising from the heavily pigmented areas, with variation one side two setae, another side three. Apart from those two pigmented areas, an isolated additional seta may present, arising directly from the cuticle. Superior volsella basal ( Figure 8H) with two inner basal setae arising from tubercles, digitus bare, with a weak ventro-lateral ledge apically. Gonostylus is not reduced, normal ( Figure 8D-G).
Remarks. Y. tropicus was characteristic by having a normal gonostylus and a weak tergal band. Distribution. Zimbabwe (Lundi River); South Africa (KwaZulu-Natal). This species has been described by Harrison [4]: some emendations and additional characters are given here.

Discussion
It is a great challenge to include Y. tropicus and Y. balteatus into a single genus since the two species show great divergence comparing to species Y. isigaabeus, especially in the often-diagnostic tibial comb pattern. Although molecular results well support (BS = 100, PP = 1) the great affinity between Y. balteatus and Y. isigaabeus (see red clade), the monophyly of Yaeprimus could not be validated until the availability of molecular data of Y. tropicus. Here, the reasons why we allocate the three species into one genera are as follows. Y. isigaabeus and Y. balteatus lack tergite bands and bear a relatively short gonostylus, whereas Y. tropicus has a weak tergite band and normal gonostylus. The divergence is distinct yet can be also observed in other genera as well. For example, Pontomyia Edwards also contain two kinds of gonostylus, reduced in P. oceana Tokunaga, while, normal in other two species [36]. More examples can also be found in Chironomus (C. crassiforceps Kieffer) [10], Polypedilum (P. minimus Lin et al.) [37], Riethia (R. phengari Cranston) [16], Sticotochironomus (S. crassiforceps Kieffer) [10] and Orthocladius v. d. Wulp (O. brevistylus Yamamoto, Yamamoto et Tang) [38]. Actually, the shortened gonostylus has been assumed to relate to the convergent mating behavior since a range of species sharing this character has been found in some extremely habitats, such as marine, karst cave and alpine fauna [39]. Variation on tergite IX band from normal to absent can be treated as a continuously varying trait in a single genus because such divergence can also be found in Apedilum [40][41][42], Paralauterborniella [8] and Beardius [43].
The conical scale bearing a slender apical spur in the fore tibia is an important character state allowing us to allocate the three species into one genus, but the differences in mid and hind tibial combs are noteworthy. The two different patterns of tibial comb in Yaeprimus seem to represent two different evolutional trends. The pattern of fused comb with one curved spur will go to some non-core Microtendipes group such as Nilodosis Kieffer and Kribiocosmus Kieffer. Separated combs with 1-2 hooked spurs are typical in the core members of Microtendipes group and in the Polypedilum complex. Normally, species bearing two different kinds of tibial combs cannot be allocated into one genus, but some special cases can be also found in Parachironomus Lenz [6] and Synendotendipes Grodhaus [44].
Our molecular analysis indicates that Yaeprimus is sister to Imparipecten and distant from the Microtendipes group. Yaeprimus shares similar characters with Imparipecten, such as the superior volsella formed as a digitus in male adults, alternate apically-located Lauterborn organ in the larval antenna, and pattern of pupal taeniae of T V-VIII is 4, 4, 4, 4, yet the two genera can be easily separated in all life stages. The conflict between Yamamoto's hypothesis and our molecular analysis is likely a result of some subjective weighting of some morphological characters, such as female genitalia and larval antenna. Actually, some emphasized characters by Yamamoto and Yamamoto are common in a broader range of genera.
In larvae, alternate Lauterborn organs in Yaeprimus share the synapomorphy with most members in the Microtendipes group. But this trend is not constrained to this group, as similar Lauterborn organ pattern can be also found in Polypedilum nubifer group [50], Imparipecten [51] and Sticotochironomus. Yamamoto and Yamamoto [2] misinterpreted that larva bears a five-segmented antenna with two large Lauterborn organs on segment two which led them to regard it as an apomorphic condition.
Pupa of Yaeprimus shows apparent similarity to Paralauterborniella both in tergal spinulation and taeniae pattern. The two genera mainly differ in the condition of cephalic tubercle, which is absent in Yaeprimus while present in Paralauterborniella [9]. In this case, other pupal characters should be evaluated to balance the conflict between molecular analysis and morphology.
Meanwhile, we should notice that the position of clade Yaeprimus plus Imparipecten is unstable in both trees, which may be caused by an insufficient sample. Given the morphological divergence between Imparipecten and Yaeprimus, we hypothesized that there were still some other unknown genera showing great affinity with the above clade, linking the two genera and establishing their position within Chironomini.
In conclusion, we redefine genus Yaeprimus based on morphological and molecular evidence. Currently, there are three species included in the emended genus. Our molecular result supports Yaeprimus is close to Imparipecten rather than to the Microtendipes group, but some uncertainty remains due to limitations in sampling. To bridge the gap between morphologic and molecular results, more relevant genera are in demand for further studies.
* sequences retrieved from Bold, 'Y' means available, 'N' means not available. Accession number of # sequence was renewed by Cranston for the same entries listed for Paraborniella as for Paralauterborniella in Table A1 from their work in 2012.