Dryophthorinae weevils (Coleoptera: Curculionidae) of the forest fl oor in Southeast Asia: Three-marker analysis reveals monophyly of Asian Stromboscerini and new identity of rediscovered Tasactes

The nominal genus Tasactes Faust, 1894, consisting of two originally included nominal species from Myanmar, is rediscovered for the fi rst time since being erected. Adult weevils herein assigned to the taxonomically re-defi ned Tasactes were abundant in forest fl oor litter at fi ve localities in China (Yunnan and Sichuan), plus one specimen is available from Shaanxi and three from Nepal. Phylogenetic analysis of a 2,275 bp matrix concatenated from one mitochondrial (COI) and two nuclear markers (ITS2 and 28S) revealed that the monophyletic Tasactes consists of eight evolutionary signifi cant terminal clades, either allopatric (three) or sympatric (two on Cang Shan in Yunnan and three on Mount Emei in Sichuan). The genus Tasactes is nested within the monophyletic Stromboscerini, while the tribe is sister to monophyletic Dryophthorus. The two morphological diagnostic characters of Tasactes, which are unique within the tribe, are the transversely truncated antennal club and conically projecting velvety apex of the club. So defi ned, Tasactes renders the genus Orthosinus paraphyletic. Considering the taxonomic neglect and uncertainties surrounding nominal Stromboscerini, all herein reported members of this tribe, including the Tasactes, are not assigned to Linnaean species. This paper illustrates the “clogging taxonomy” phenomenon, in which obscure historical names render taxonomic assignment of newly sampled specimens precarious. All the data used herein (localities, sequences, specimen images) are available online in public datasets dx.doi.org/10.5883/DS-TASACT1 and dx.doi.org/10.5883/DS-TASACT2.


INTRODUCTION
This paper was motivated by the discovery of numerous minute and slow-moving Stromboscerini weevils (Fig. 1A) recorded (Figs 1B-E) when forest leaf litter in Southwest China and nearby areas was sifted (Fig. 5 and Supplementary Table S1).The original plan of producing an integrated phylogeny using DNA, morphology and other data sources was abandoned once the grossly unsatisfactory state of Stromboscerini taxonomy became apparent (Grebennikov, 2018).The nominal tribe lacks an underlying phylogenetic hypothesis and is likely to be multiply non-monophyletic (Grebennikov, 2018).The type genus from Madagascar was thought to be unrelated to the rest of the tribe distributed mainly in Southeast Asia (Grebennikov, 2018), potentially leaving all but the type genus without a tribal assignment.Furthermore, doubts remain about the monophyly of all non-monotypic Recent Stromboscerini genera.The widely accepted and century old synonymy of the type species of the genus Xerodermus is doubted (Grebennikov, 2018), which threatens the taxonomic identity of the genus.It became evident that any research on the numerous newly sampled Stromboscerini would be haunted by the phenomenon of "clogging taxonomy" (Grebennikov, 2016), in bidirectional sequencing) was performed in a commercial laboratory "Canadian Center for DNA Barcode" (CCDB, http://www.ccdb.ca/) at the University of Guelph, Ontario, Canada, following the standard laboratory protocol (Ivanova et al., 2006(Ivanova et al., , 2014)).A cocktail of two primer pairs was used to amplify the DNA barcoding fragment (Supplementary Table S2).Obtained sequences, electropherograms, gel images, specimen data and specimen images (Supplementary Fig. S1) can be seen online in the public BOLD dataset available at dx.doi.org/10.5883/DS-TASACT1.Alignment of these sequences was easy as they contained no insertions/deletions (= indels), while their NJ clustering was performed using the online BOLD engine.

Three-marker matrix formation and Maximum Likelihood (ML) analysis
A subset containing 45 Dryophthorinae terminals was selected and sequenced for two additional nuclear DNA markers, ITS2 and 28S (Table 1), using the primers in Supplementary Table S2.A total of 21 Tasactes and 15 other Stromboscerini terminals were selected (Table 2) to best represent clusters recovered in the NJ analysis (Supplementary Fig. S2).To place Stromboscerini in a wider phylogenetic context and test its taxonomically-implied monophyly, nine non-Stromboscerini terminals were incorporated in the analysis, including two terminals representing the likely closely related genus Dryophthorus Germar, 1824 (Gunter et al., 2016, Fig. 2 and Table 2).In the absence of a phylogenetic hypothesis for Stromboscerini and Dryophthorinae (and to avoid an excessive number of indels in noncoding and fast evolving ITS2 when a distant organism is added to the matrix), no non-Dryophthorinae terminals were added to root the topology.All reported and newly generated DNA sequences, as well as their electropherograms, gel images, specimen data and specimen images can tribe, and if monophyletic, then to hypothesise its sistergroup and shed light on its internal phylogenetic structure.The second goal is to anchor one of the inadequately known historical genus-group names (Tasactes) to a newly detected clade.Overall, this paper was designed as another example (Riedel, 2017) of the work required on taxonomically and phylogenetically orphaned groups, when newly hypothesized clades may for the fi rst time be linked with old and inadequately known nominal taxa.

Specimen sampling and handling
A total of 145 sifted samples of forest litter were collected in mainland China (122), Vietnam (7) and Taiwan (16); their twoletter two-digit codes are explained in Supplementary Table S1.Litter was sifted using a hand-held sifter (Figs 1B-D) fi rst through 12 × 12 mm mesh, then through an insert (Fig. 1B) with a fi ner 7 × 7 mm mesh.Live specimens (Fig. 1A) were extracted using suspended Winkler funnels (Fig. 1E) and preserved in 96% ethanol.A few specimens were opportunistically collected by means other than litter sifting (mainly by hand), while specimens from Nepal (clade E in Fig. 3) were received from Joachim Schmidt; these specimens do not have litter sample codes (Table 2) and are denoted in Fig. 3 by country names.All the specimens have a label with the code CNCCOLVG0000XXXX serving as a Sample ID in the Barcode of Life Database (= BOLD, Ratnasingham & Hebert, 2007, www.bold.org),while the last four digits serve as unique specimenidentifi ers (Figs 2-4 and Table 2).All the specimens studied were adults and are deposited in the Canadian National Collection of Insects, Arachnids and Nematodes in Ottawa, Canada (CNC).

Specimen selection for DNA barcoding and Neighbour Joining (NJ) clustering
Sixty specimens of Dryophthorinae (36 of Tasactes and 15 of four other Stromboscerini genera) were DNA barcoded (= sequencing of 658 bp of COI-5', Hebert et al., 2003a, b).All laboratory work (including DNA extraction, purifi cation, PCR and     be seen online in the public BOLD dataset DS-TASACT2 available at dx.doi.org/10.5883/DS-TASACT2.Alignment of the ITS2 and 28S sequences was done using the MAFFT 7 online platform (Katoh et al., 2002, Katoh & Toh, 2008a) with the Q-INS-i algorithm (Katoh & Toh, 2008b) utilising the information on the secondary structure.No parts of the three alignments were excluded from the analysis.Three aligned single-fragment datasets were concatenated using Mesquite 3.11 (Maddison & Maddison, 2011).Analysis of the 2,275 bp concatenated dataset was done using the ML phylogenetic method and the RAxML 7.2.7 (Stamatakis et al., 2008) algorithm on a computer cluster at the Cyberinfrastructure for Phylogenetic Research (CIPRES) (Miller et al., 2010) with 100 non-parametric bootstrap (Felsenstein, 1985) replicates.The topologies were visualized using FigTree v1.4 (Rambaut, 2014).

Matching names and clades, specimen visualization and the limitations of this study
To link the Stromboscerini clades with fi ve nominal genera, primary types of all Recent valid genera of the tribe were studied (Grebennikov, 2018), including the syntypes of both nominal Tasactes species stored in the "Senckenberg Naturhistorische Sammlungen Dresden" in Dresden, Germany.Morphological characters given in the key to Stromboscerini genera (Morimoto, 1985) were also used.To visualize the adult morphology of Stromboscerini and link it with the DNA topology, habitus and antennae of specimens of all 14 evolutionary signifi cant terminal clades (= ESTC, = candidate species) were imaged and superimposed on the phylogenetic tree (Figs 2, 3).Since sexual dimorphism has never been reported or noted in the Stromboscerini, no attempt was made to sex specimens.Lacking comparative data on the genitalia, no attempt was made to use this source of information.No new taxonomic acts (such as the description of new taxa or designation of the type species of Tasactes) are herein performed, since such actions would require an effort far exceeding the scope of this paper, which aims to provide the very fi rst phylogenetic glimpse of the old and poorly known nominal taxa associated with the vaguely defi ned family-group name "Stromboscerini".

RESULTS
The ML tree of 45 terminals analyzed using the concatenated 2,275 bp matrix recovered the monophyletic Stromboscerini as a sister to the monophyletic Dryophthorus; all three clades have 100% bootstrap support (Fig. 2).All 36 Stromboscerini terminals are grouped in 14 ESTC (clades A-N in Fig. 3), each with high bootstrap support (99-100%); eight of these clades (D-K in Fig. 3, in red) form the monophyletic Tasactes with bootstrap support of 89%.Relationships among eight Tasactes clades are well-resolved with high bootstrap support (97-100%),

Monophyly and sister group of Stromboscerini
Recovery of the strongly supported clade formed by reciprocally monophyletic Dryophthorus and Stromboscerini is one of the two most signifi cant results of this study.Even though this relationship was recently, and for the fi rst time hypothesized (Gunter et al., 2016) in a study also using three DNA markers (28S, 16S and COI), the reported clade's statistical support was notably lower (83-85%).Even more signifi cantly, in Gunter et al. (2016) the tribe Stromboscerini is represented by a single terminal of Dryophthoroides Roelofs, 1879, thus leaving this tribe's monophyly untested.The herein reported analysis utilizing representatives of fi ve nominal Stromboscerini genera is the fi rst adequately designed and successful attempt to test tribe monophyly.Both "aberrant" genera of Stromboscerini (Nephius Pascoe, 1885 andStromboscerus Schoenherr, 1838) are uniquely characterized by having ocular lobes on the anterior edge of the prothorax (Morimoto, 1985: 74) and thought to be unrelated to the rest of the tribe (Grebennikov, 2018), were not included in the present analysis, thus postponing the decisive test of the tribe's monophyly (and of its name, since the latter is the type genus).

Monophyly of Tasactes
The second most signifi cant result is that all morphological diagnostic characters of nominal Tasactes (Morimoto 1985: 74: elongate and ventrally non-contiguous eyes, 6-segmented antennal funicle, transversely truncated antennal club and conically projecting velvet apex to the club) occur in the specimens forming clade D-K (Fig. 3).Most signifi cantly, all specimens in this clade clearly display both club characters (Figs 4O, P), which among all Stromboscerini are unique to this genus.Remarkably, both club characters are not mentioned by Johannes Faust in the original generic description, but were fi rst reported nearly a century later by Morimoto (1978Morimoto ( , 1985)).It might also be noted that four antennomers immediately proximal to the club in all specimens herein attributed to Tasactes are nearly subquadrate, while in all other herein examined Stromboscerini they are about 1.5 times as wide as long (Fig. 3).This unique morphological match, as well as the relative similarity in habitus (Fig. 10 in Grebennikov, 2018) and geographical distribution (Fig. 5) of the type specimens of both nominal Tasactes species and those forming clade D-K (Fig. 2), are the decisive factors supporting the linkage between the genus-group name and the herein so named clade.It should be added that within the herein re-defi ned genus Tasactes, clade D-I might be separated from clade J-K by two diagnostic characters of the former: larger body size (Fig. 3) and evenly bent rostrum (Fig. 4).

Flight ability of Stromboscerini
An inability to fl y is a widespread phenomenon in Pterygota and weevils (Arzanov & Grebennikov, 2017).Even though the fl ight ability of Stromboscerini was not studied and the herein analyzed specimens were not systematically Fig. 5. Distribution of Tasactes.Map was prepared using online software SimpleMappr (Shorthouse, 2010); the northeastern record is based on a single specimen that was not included in the concatenated analysis (Fig. 3), but DNA barcoded and clusters with other Tasactes as a sister to clade G (Fig. S2).
examined for the presence of full-sized hind wings, members of Allaeotes Pascoe, 1885 differ from the rest of the tribe by having non-effaced elytral shoulders (= humeri present, Fig. 3) and being trapped by fl ight intercept traps (specimens from Vietnam provided by Adam Brunke, not included in the present analysis), both of which indicate that members of this genus can fl y.The sister-group relationship between the supposedly volant genus Allaeotes (clade A in Fig. 3) and the supposedly fl ightless rest of the tribe (clade B-N in Fig. 3) indicates that fl ight capacity in Stromboscerini was irreversibly lost by the most recent common ancestor of clade B-N.

Elusive identity of Orthosinus and Xerodermus
The recovered Stromboscerini topology (Fig. 2) indicate that the genus Orthosinus currently containing seven nominal species from Sri Lanka, India, Myanmar, China, Indonesia and Japan (Grebennikov, 2018) is not monophyletic.Not only is it represented by two clades (clade B-C and clade N in Fig. 3) that are distinctly different in eye shape (Figs 4B, C, N), but there are numerous additional specimens attributable to this genus according to the key to genera (Morimoto, 1985), which were not included in the present analysis.The mystery is compounded by the existence of the nominal genus Xerodermus Lacordaire, 1866, with four nominal species in Sri Lanka and India (Grebennikov, 2018).This nominal genus was a synonym of Orthosinus when the most recent key to the tribe's genera was published (Morimoto, 1985), but was then resurrected as a valid genus (Alonso-Zarazaga & Lyal, 2002), even though the taxonomic identity of its type species is doubtful (Grebennikov, 2018).Both genera seem to represent the tribe's "garbage bin", accommodating all species not assignable to other easier-to-recognize nominal genera characterized, likely, by apomorphic character states (such as 4-or 5-segmented antennal funicle or by exceedingly narrow vertical eyes).Identifying and determining the phylogenetic limits of these two relatively large nominal genera is, therefore the main challenge to matching the taxonomy and phylogeny of the tribe.

Fig. 1 .
Fig. 1.Stromboscerini weevils.A -weevils extracted from a sample of typical forest fl oor litter collected on Mount Emei, Sichuan, China; arrows indicate Stromboscerini; B -sample of litter in a bag prior to sifting through a 12 × 12 mm mesh and then through a 7 × 7 mm mesh; C -sifter with 12 × 12 mm mesh; D -forest fl oor litter habitat of Stromboscerini weevils; E -Winkler funnel with litter suspended in mesh bags.

Fig. 2 .
Fig. 2. Unrooted Maximum Likelihood inference phylogram of Dryophthorinae weevils based on the analysis of the concatenated 2,275 bp matrix.Digits at internodes are bootstrap values.Images are to scale and illustrate corresponding tree terminals.

Fig. 3 .
Fig. 3. Rooted Maximum Likelihood inference phylogram of monophyletic Asian Stromboscerini based on the analysis of the concatenated 2,275 bp matrix.Digits at internodes are bootstrap values.Fourteen evolutionary signifi cant terminal clades are labelled A-N; eight of them belonging to the monophyletic genus Tasactes are in red.Blue eye signs indicate illustrated terminals; habitus images are to scale.

Table 1 .
DNA fragments used in analyses.

Table 2 .
GenBank accession numbers of Dryophthorinae weevils; for sample code see TableS1.