Female terminalia morphology and cladistic relations among Tok‐Tok beetles (Tenebrionidae: Sepidiini)

Tok‐tokkies are one of the most iconic lineages within Tenebrionidae. In addition to containing some of the largest darkling beetles, this tribe is recognized for its remarkable form of sexual communication known as substrate tapping. Nevertheless, the phylogenetic relationships within the group remain poorly understood. This study investigates the usefulness of female terminalia morphology for delimiting Sepidiini and reconstructing relationships among it. Data on the structure of the ovipositors, genital tubes and spicula ventrali have been generated for >200 species representing 28 Pimeliinae tribes. This dataset was used in a comparative analysis at the subfamilial level, which resulted in recognition of several unique features of tok‐tokkie terminalia. Additionally, new features linking phenotypically challenging tribes also were recovered (Cryptochilini + Idisiini + Pimeliini). Secondly, 23 characters linked to the structure of female terminalia were defined for tok‐tok beetles. Cladistic analysis demonstrates the nonmonophyletic nature of most of the recognized subtribes. The morphological dataset was analysed separately and in combination with available molecular data (CAD, Wg, cox1, cox2, 28S). All obtained topologies were largely congruent, supporting the following changes: Palpomodina Kamiński & Gearner subtr.n. is erected to accommodate the genera Namibomodes and Palpomodes; Argenticrinis and Bombocnodulus are transferred from Hypomelina to Molurina; 153 species and subspecies previously classified within Psammodes are distributed over three separate genera (Mariazofia Kamiński nom.n., Piesomera stat.r., Psammodes sens.n.). Psammodes sklodowskae Kamiński & Gearner sp.n. is described. Preliminary investigation of the ovipositor of Mariazofia basuto (Koch) comb.n. was carried out with the application of microcomputed tomography, illuminating the muscular system as a reliable reference point for recognizing homologous elements in highly modified ovipositors.


Introduction
The morphology of the female terminalia is considered essential in reconstructing phylogenetic relationships among darkling beetles (Doyen, 1994;Doyen and Tschinkel, 1982;Iwan and Kami nski, 2016;Tschinkel and Doyen, 1980), yet detailed comparative analyses of the ovipositors and/or genital tubes even at tribal levels are scarce (e.g. Banaszkiewicz, 2006;Bouchard and Yeates, 2001;Matthews and Bouchard, 2008). Furthermore, many recently published descriptions of new species and genera within Tenebrionidae lack data on female terminalia (Ferrer, 2002;Fouqu e, 2013;Kami nski, 2011;Kami nski et al., 2021). This discordance is probably driven by multiple factors, from practical ones such as the relative complexity of dissecting the female terminalia, which could often lead to partial specimen destruction, to the subjective opinions that the ovipositors/genital tubes are not informative among the studied groups (Kami nski pers. obs.). Furthermore, the morphology of the female genitalia of many species of darkling beetles is not easily interpreted under the current terminological framework . Namely, it is assumed that the plesiomorphic ovipositors of darkling beetles are composed of four coxite lobes arranged in a single line, with a large gonostylus situated on top of the apical lobe. Although this bauplan is the basis for defining homology among the whole family, it can be problematic in the case of some strongly modified groups (Doyen and Tschinkel, 1982;Tschinkel and Doyen, 1980). The Sepidiini Eschscholtz (Tenebrionidae: Pimeliinae), commonly known as the tok-tok beetles, are a good example of a group with strongly modified ovipositor (Kami nski et al., 2019). Most of the known species of this group possess 3-lobed coxites, whereas the gonostyli are absent . Although Sepidiini were included in the only phylogenetic study available for the subfamily (Doyen, 1994) that has considered data on female terminalia, the ovipositor structure of the tribe remained insufficiently investigated, and the paper lacked direct indication of homologous structures (figs 101 and 104 in Doyen, 1994). Only recently, Kami nski et al. (2021) had incorporated the 3-lobed ovipositor of tok-tok beetles into the tenebrionid terminological framework, and highlighted the potential value of some characters for intertribal classification promoting this study.
Tok-tok beetles are a morphologically diverse tribe of darkling beetles, composed of >1000 species distributed throughout the African continent and Southern Palaearctic (Kami nski et al., 2019). As a consequence of their unique tapping behaviour, which is assumed to be a form of sexual communication Lighton, 1987Lighton, , 2019, they are one of the most conspicuous beetle groups in the region (Matthews et al., 2010). To date, 59 genera of Sepidiini have been described . However, recently conducted molecular-based studies have revealed the need to designate several previously unrecognized phylogenetic groups at the genus level-suggesting that the current number of genera is an underestimation Kami nski et al., 2021). Owing to the interesting morphology and the large body size of some of the species, tok-tok beetles attract the attention of collectors; however, taxonomic contributions to the group are still scarce. The main reason for this is the lack of clear morphological definitions for the majority of taxa (Penrith, 1986), which concerns all taxonomic levels--from species to subtribes . Furthermore, the aforementioned phylogenetic studies of tok-tok beetles were focused mainly on Molurina Solier and left the remaining subtribes scarcely sampled or not included at all (i.e. Sepidiina Eschscholtz). As a result, Sepidiini remains a taxonomically challenging group.
The morphological variability of female terminalia of tok-tok beetles remains almost completely uninvestigated. Although Doyen (1994) analysed terminalia of the representatives of four of five currently recognized subtribes (i.e. Hypomelina Koch: Brinckia Koch and Iugidorsum Louw; Molurina: Moluris Latreille and Phrynocolus Lacordaire; Sepidiina: Sepidium Fabricius; and Trachynotina Koch: Cyrtoderes Dejean and Somaticus Hope), he did not mention any major differences among them, besides uniqueness of the ovipositors of Sepidium. He concluded that the following features can be used to distinguish Sepidiini from the remaining Pimeliinae: (1) oblique baculus of coxites (lateral view); (2) elongated paraprocts; (3) strongly sclerotized 4th lobes of coxites; and (4) reflexed arm of spiculum ventrale. This study aims to test these conclusions and evaluate the usefulness of the morphology of the female terminalia for phylogenetic studies within the tok-tok beetles by combining dense taxon sampling and the application of cladistic methods. Furthermore, a detailed investigation of the muscular system of the ovipositor with the usage of the microcomputed tomography (microCT) has been performed on representatives of tok-tok beetles (Mariazofia basuto comb.n.). The presented data can be implemented in future projects as an additional reference point for establishing homology within Pimeliinae, especially between strongly modified ovipositors.

Workflow design
In order to accomplish the above stated goals, the three following levels of analyses were conducted.
Because one of the main goals of this study is to examine the diversity of female terminalia forms in Sepidiini, this tribe was the most intensively sampled group. Specimens representing 35 genera and all five currently recognized subtribes were included in the analysis (Appendix S1). Sepidiina is the most under-represented subtribe--two of eight known genera are included (Kami nski et al., 2019). This is mainly due to the lack of unambiguous diagnostic characters for the majority of currently recognized genera and the absence of reliable identification tools for species-level taxa. However, at the same time Sepidiina is the most characteristic and welldefined lineage within tok-tok beetles (Koch, 1958). For the remaining subtribes, the majority of the genera, which were not included in this study, currently are represented by species known only from relatively small series, with no female specimens present in the studied materials (e.g. Miripronotum Louw, Psammoryssus Kolbe, Stridulomus Koch, Triangulipenna Louw and Uniungulum Koch). For diverse genera such as Ocnodes F ahraeus, Psammodes Kirby and Somaticus Hope, species that spanned the morphological variability of these genera were chosen to test the monophyly of these groups. (Appendix S1).
Dissections were conducted on material from the following collections: California Academy of Sciences, San Francisco, USA (CASC); Ditsong National Museum of Natural History, Pretoria, South Africa (TMSA); Museum and Institute of Zoology of the Polish Academy of Sciences (MIZ PAS); Museo Nacional de Ciencias Naturales, Madrid, Spain (MNCN-CSIC); Purdue University, West Lafayette, USA (PERC); and United States National Museum of Natural History, Washington DC, USA (USNM).
In order to isolate the terminalia and associated structures, the abdomen of specimens to be studied were removed and cleared overnight in 10% potassium hydroxide. Reproductive systems were dissected using entomological needles and scalpels. To increase the visibility of certain features, some samples were stained with Chlorazol Black dissolved in glycerin or water. In the majority of cases the dissected structures then were transferred to a drop of glycerin on a microscope slide for imaging. However, in some cases the ovipositors were dried and later photographed in order to increase contrast and visualize sutures between subsequent coxite lobes. The analysed material was preserved in microtubes filled with glycerin and pinned beneath the specimens. Dissecting procedures are in Kami nski (2021a). Images of morphological details were taken using different imaging systems at MIZ PAS, TMSA and PERC. Raw images can be downloaded from Harvard Dataverse (Kami nski, 2021b).
Nomenclature follows that of Bouchard et al. (2011Bouchard et al. ( , 2021. Cladistic analysis of female terminalia features within Sepidiini. In order to assess the phylogenetic informativeness of female terminalia morphology within tok-tok beetles, a cladistic analysis based solely on features linked to ovipositor, genital tubes, spiculum ventrale and proctiger was performed. Dissections described in the previous section were used to code data into a character matrix (MorphoBank http://morphobank.org/permalink/? P4101). Morphological terminology follows Doyen (1966), Tschinkel and Doyen (1980) and Kami nski et al. (2021). To avoid potential inconsistency problems associated with analysis of large numbers of taxa (Kim, 1996), a preselection process to incorporate species or operational taxonomic units (OTUs) representing the morphological diversity of Sepidiini was undertaken. The process was driven by the observed variability of female terminalia and available molecular data Kami nski et al., 2021). For example, all analysed representatives of the Dichtha clade (see Kami nski et al., 2021) were observed to possess similarly structured ovipositors (3-lobed; merged lobe 1 and 2 strongly emarginate basally), female genital tubes (multibranched spermatheca), spicula ventrali (Yshaped) and proctigers (widely indented medially), and therefore only two of 11 studied species of this group--Amiantus octocostatus P eringuey and Dichtha cubica (Gu erin-M eneville)--were included in the cladistic analysis. Additionally, as molecular data for selected species already were available, the selected OTU also could be included in the combined phylogenetic analysis of morphological and molecular data (see next section). Stips gebieni (Hesse) of Adelostomini was used as an outgroup. According to Doyen (1994), adelostomid beetles share some unique features of the female terminalia with Sepidiini (e.g. spiculum ventrale with reflexed arms), which is used here as a justification for the outgroup selection. Moreover, preliminary phylogenomic data renders Adelostomini sister to Sepidiini (Smith pers. comm.). Furthermore, Stips possesses four visible coxite lobes, which enables direct comparisons with the sepidiinoid terminalia. However, the ovipositors of many other Pimellinae are strongly modified, which at this point disrupts clarity of any potential phylogenetic conclusions within Sepidiini. Therefore, representatives of other investigated Pimeliinae tribes were not included in the cladistic analysis.
The data matrix was created in Mesquite v.3.61 (Maddison and Maddison, 2019) and later exported to TNT v.1.5 (Goloboff andCatalano, 2016) andMrBayes v.3.2.7 (Ronquist et al., 2012) for phylogenetic analyses. Maximum parsimony analyses were implemented with implicit enumeration, with equal weight assigned to all characters (states treated as unordered), and conducted locally. Branch support was determined using Bremer supports (Bremer, 1994) in TNT. The performance of the characters in the parsimony analysis was verified by calculating the consistency (ci) and retention indices (ri) in Mesquite, whereas the values of the consistency (CI) and retention (RI) indices for the whole matrix were obtained in TNT by using the stats.run script. Bayesian analyses were run through the CIPRES portal (Miller et al., 2010) using the Mkv + G model (Lewis, 2001) and flat priors. Two searches were performed with four chains each for 20 million generations with a 0.5 burn-in. Trees were sampled every 1000 generations. Branch support for Bayesian analyses was inferred based on the posterior probability (PP) distribution of tree topologies. Posterior probabilities were mapped onto the strict consensus cladogram from the parsimony analyses. Resulting trees were studied in Winclada 1.61 (Nixon, 2002) and FigTree 1.4.4 (Rambaut, 2009).
Combined analysis of female terminalia traits and molecular data. The majority of the molecular dataset was assembled from previous studies Kami nski et al., 2021). Seven additional species were sequenced to increase  (Bouchard et al., 2011), were used as outgroups.
For newly sequenced taxa, DNA was extracted from ethanolpreserved specimens (95% EtOH) using DNeasy Blood & Tissue Kits (Qiagen, Germantown, MD, USA) following the manufacturer's protocols. Extractions were performed on soft tissue from the head or thorax; no cuticle was ground during the extraction process. Voucher specimens are deposited in the PERC. Polymerase chain reactions (PCRs) were performed using ExTaq (TaKaRa, Moutain View, CA, USA) with thermocycler protocols and primers described in Kanda et al. (2015). PCR cleanup, quantification and sequencing were performed by the University of Arizona's Genetics Core Facility. Cleaned PCR products were sequenced on an Applied Biosystems 3730XL DNA Analyser (Foster City, CA, USA). Assembly of chromatograms was performed as described in Kami nski et al. (2021).
Combined (DNA + morphological data), as well as molecularonly, Bayesian analyses were performed in MrBayes v.3.2.7 on XSEDE via CIPRES. Data partitions and models of sequence evolution were assessed with Partitionfinder v.2.0 (Lanfear et al., 2017) implemented on CIPRES. Only models for "mrbayes" were considered. The search was conducted using greedy searches and the Bayesian information criteria (BIC). For the morphological partition the Mkv model (Lewis, 2001) with gamma-distributed rate variation was implemented. For each, combined and molecular-only, analysis, two independent runs with four chains were performed. Analyses were run for 40 million generations, and parameters were sampled every 1000 generations. A burnin fraction of 25% was used, and convergence was determined by the standard deviation (SD) of the split frequencies; runs were considered to have converged at <0.01. Nodes with PP > 0.95 were considered strongly supported, with PP = 0.90-0.94 moderately supported and with PP = 0.70-0.89 weakly supported.
The performance of the characters in the combined analysis was verified by calculating the consistency (ci) and retention indices (ri) in Mesquite. Furthermore, character optimization ( Fig. 7) has been carried out with the maximum parsimony method implemented in the same software.
The ovipositor, with primary focus on coxites, of the studied EtOH-preserved female specimen was manually segmented without application of 10% potassium hydroxide. During this process the proctiger plate was detached. Subsequently the dissected ovipositor was dried with hexamethyldisilazane (HMDS) (Bray et al., 1993), and placed in a PCR tube (0.2 ml) for microCT scanning.
MicroCT analyses were performed at MIZ PAS using a SkyScan 1172 system (SkyScan, 2008). During the scanning processes, the Xray source was set to a voltage of 40 kV and a current of 250 mA. As a result of the elongate shape of the analysed ovipositor, the specimen was scanned using four oversized scans (Iwan et al., 2015;Ra s et al., 2018); image pixel size equals 1.66 lm. Reconstruction was performed using NRecon v.1.6.4.7. Primary segmentation was conducted in CTAn v.1.18.4.0+. Specimen anatomy was later investigated in Blender v. 2.91.0.The resulting 3D model is available online (Ra s, 2021).
This investigation of coxities musculature is, to the authors' knowledge, the first within Coleoptera (Doyen, 1966;Lawrence et al., 2011;Medvedev, 2001). In the absence of previous nomenclature for particular muscle groups, new terminology is introduced within the present paper (see the Results).

Results
In order to clarify the characters used in the following phylogenetic analyses, a description of the sepidiinoid female terminalia is given here. The ovipositor is the most distal structure of the female reproductive system. It can be fully extended from the body and is composed of two sets of paired sclerites (Fig. 1b): the distal, more strongly sclerotized sclerites are referred to as the coxites (Fig. 1b, c), and the proximal, more membranous sclerites the paraprocts (Fig. 1b, p). Each paraproct also bears a longitudinal sclerotization known as the baculus of the paraproct (Fig. 1b, bp). It should be noted that Doyen (1966) referred to the paraprocts and coxites as valvifer 1 and 2, respectively. However, in subsequent papers he used the term valvifer to refer only to the proximal plate of the coxite and used the term paraproct to refer to the proximal pair of sclerites of the ovipositor (Tschinkel and Doyen, 1980), which is the terminology followed here. In Tenebrionidae, the coxites are usually composed of four visible lobes in ventral view, named c1 to c4 from proximal to distal. However, it is hypothesized that the basal plate of Sepidiini is the product of the fusion of the c1 and c2 lobes . Therefore, here the coxite plates are referred to as c1 + c2 (basal plate), c3, and c4 (apical plate) (Fig. 1b). The apical plate bears latero-ventral sensory setae (ss), whereas the gonostyli are absent. The base of each c1 + c2 plate is strongly sclerotized, forming a transverse or oblique baculus of the coxite (Fig. 1b, bc) that directly articulates to the apex of the corresponding baculus of the paraproct (Fig. 1b). Each paraproct is proximally attached to abdominal segment 8 by a flexible membrane (Fig. 1c). In Sepidiini, each paraproct ( Fig. 1b) is composed of inner (ip) and outer (op) plates separated by the longitudinal baculus (bp). In some other groups of darkling beetles, for example Blaptinae Leach, the inner plate can be largely reduced or absent (Fig. 1e). The anus (an) is located dorsal to the ovipositor, and both these structures are shielded dorsally by the proctiger (Fig. 1b, pr). The proctiger is equipped with a pair of longitudinal lateral baculi (bpr) along the basal two thirds of its length (Fig. 1b). Within Sepidiini the apical margin of the proctiger is variable in shape ( Fig. 5q-s). Ovipositor, anus, and proctiger, when retracted within the body cavity, are apically surrounded by abdominal segment 8 (Fig. 1a, s8). Sternite VIII usually forms a single plate ( Fig. 5t-w), but sometimes it is formed by two lateral plates (hemisternites) medially joined together by membrane (e.g. Fig. 2f, l). Proximal to the sternite VIII (or hemisternites), a second pair of lateral sclerites form the spiculum ventrale (Fig. 1a, sv). In the majority of the studied Pimeliinae species, including Sepidiini, the spicula ventralia are proximally fused and thus can be divided into arms (distal) and base (proximal). The presence of an unfused base is exclusively reported in the case of studied Cryptochilini, Idisiini, and Pimeliini ( Fig. 3d, f, l). Medially, the ovipositor possesses the vulva, which is proximally linked to a sac-like vagina (Fig. 1f, vg). The vagina of Sepidiini is devoid of sclerites or spikes. During the present investigation no eggs or larvae were found inside of the vaginas of the studied specimens. Although some Afrotropical lineages of darkling beetles were proven to be ovoviviparous (Fig. 1d), this mode of reproduction is unlikely for Sepidiini mainly as a consequence of the elongation of the ovipositor. The presence of paired oviducts was found in Sepidiini ( Fig. 1f, ov1, ov2). However, both oviducts are merged into a single canal before entering the vagina distally (a) Fig. 1f, ov). The spermatheca (sp) is present near the proximal end of the vagina. In Sepidiini the spermatheca is composed of several smaller tubes (multibranched spermatheca; Fig. 1g, sb), whereas in Adelostomini it splits over into two smaller ducts ( Fig. 1g, h). Laterally, the spermatheca connects with the spermathecal accessory gland (Fig. 1g, h, ag). Together, vagina, oviducts, spermatheca and spermathecal accessory gland form the genital tubes. Table 1 provides logical definitions for each of the structures here discussed, along with their term equivalency in the ontology for the Anatomy of the Insect SkeletoMuscular System (AISM; Gir on et al., 2021).

Comparative analysis of female terminalia within Pimeliinae
Sepidiinoid beetles differ from the remaining investigated Pimeliinae tribes by having (Table 2) Table 2) and sclerotization of coxite plate c4 seem to be common among the investigated pimeliinoid tribes (Figs 2-4), although these features can be highly variable at the tribal levels (e.g. Adelostomini; Fig. 5a-e).
The presence of well-developed gonostyli was observed only in Akidini, Elenophorini, Nyctoporini and Vacronini (Table 2). Coxites divided into four lobes seem to be relatively rare among pimeliinoid beetles and were reported for selected Sepidiini, Adelostomini, Akidini, Elenophorini, Erodiini, Nyctoporini and Vacronini (e.g. Figs 2n, 3m). Likewise, fully fused lobes also are infrequent ( Table 2). The coxite plate c2 is membranous and folded under the coxite plate c1 in Adesmiini, Evaniosomini, Nycteliini, Tentyriini, Trilobocarinim and Zophosini ( Fig. 2g-j). It can be uncovered by manual manipulation, unlike in other investigated species. Elongation of the base of the spiculum ventrale is present in the majority of the tribes ( Table 2). The base is commonly fused with the exception of studied Cryptochilini, Idisiini and Pimeliini ( Fig. 3d, f, l). In Edrotini, Epitragini and Thinobatini the spiculum ventrale is missing the arms (Fig. 4d, i), whereas in Nycteliini the arms are relatively short when compared to the remaining tribes (Fig. 3q).

Cladistic analysis of female terminalia features within Sepidiini
Morphological matrix. A matrix of 23 characters was constructed for 47 OTUs representing all five currently recognized subtribes of Sepidiini. The complete dataset consists of 13 characters related to the ovipositor, five for the proctiger, four for the spiculum ventrale, and a single character for the genital tubes. Characters 1, 8, 13, 14, 15 and 20 are parsimony-uninformative within the tribe, but constitute the morphological definition of Sepidiini when compared with other lineages of Pimeliinae. The character states, values of ci/ri indexes for morphology-only and combined analyses are provided below, whereas the coded matrix is available at MorphoBank (http://morphobank.org/permalink/? P4101).
A well-developed coxite plate c3 was exclusively observed within Molurina.
State 1 is used here as a reliable supporting diagnostic character for a clade containing Argenticrinis, Bombocnodulus and Psammodes sens.n. 8. Coxites: base of basal plate (c1 + c2) in lateral view: (0) straight, transverse; (1) oblique, with section corresponding to inner plate of paraproct more distally projected than section corresponding to outer plate of paraproct (Fig. 5k).
This character directly corresponds to character 56 in Doyen (1994). All of the analysed representatives of Sepidiini possess oblique bases of the basal plate.
The presence of sclerotization is synapomorphic for Sepidiina.
All of the studied Sepidiini species possess multibranched spermathecae. This observation is convergent with data presented by Doyen (1994) for Sepidiini.
State 1 is shared by all studied Sepidiina species and Cyrtoderes sp. of Trachynotina.
The observed diversity of the shape of the apical margin of the proctiger is unexpectedly high. To the best of the authors' knowledge, this is the first time that this character has been considered as a phylogenetic and diagnostic feature within darkling beetles (Doyen, 1994;Doyen and Tschinkel, 1982;Tschinkel and Doyen, 1980). State 3 is characteristic for all studied Trachynotina species; however, it is also present in Oxura setosa (Oxurina). State 4 is shared between Molurina, Argenticrinis and Bombocnodulus (both classified within Hypomelina).
State 2 is shared by the examined Ocnodes and Tarsocnodes species.
State 1 is shared by the studied Sepidiina species, and Cyrtoderes and Somaticus of Trachynotina.
All of the analysed Sepidiini species have elongate bases of spiculum. Within Pimeliinae, and many other tenebrionids, this feature is apparently linked with the elongation of the ovipositor (Medvedev, 2001).
State 1 is shared by several representatives of Trachynotina and Vieta speculifera (Sepidiina), with the exception of included Ethmus species. This finding contradicts the diagnostic character proposed by Doyen (1994), who concluded that in Sepidiini the spiculum arms are reflexed (i.e. with tips pointing basally, as in Somaticus, Fig. 6).
Morphology-based analysis. The heuristic search with characters equally weighted yielded 12 most parsimonious trees of 45 steps, with a CI of 0.644 and a RI of 0.908. Topological differences between the maximum parsimony (MP) trees concerned the shallow nodes (mostly the clade including Sepidiina + Trachynotina). The strict consensus tree (CI = 0.652, RI = 0.909) with Bremer support values is presented in Fig. 6. Branch support for the majority of recovered nodes was low.
Only Sepidiina was recovered as monophyletic in the MP analysis. Trachynotina was recovered paraphyletic with the majority of representatives clustered within a single clade additionally containing Sepidiina, whereas the genus Ethmus was recovered separately in a clade containing representatives of Hypomelina, Molurina and Oxurina. However, Ethmus did not cluster within any other major lineage in that clade (Fig. 6). The close affinity of Sepidiina and Trachynotina (excluding Ethmus) is supported by the specific structure of the ovipositor and proctiger; i.e. apical margin of the basal plate of the coxites (c1 + c2) with elongate transverse sclerotization (character state 11:1), basal margin of the same plate oblique (6:1) and proctiger with sclerotized apical margin (19:1).
Representatives of Oxurina were recovered in three different clades. The first contains Oxura (the type genus of the subtribe), Decoriplus, Synhimba and Pterostichula kung (Fig. 6). This grouping is supported by a unique structure of coxite plate c3--base with short transverse sclerotization (10:1). The second clade with Oxurina representatives contains Palpomodes and Namibomodes (Fig. 6), and is differentiated by a thin/ reduced apical plate of coxites (c4; 12:0). The last Oxurina representative, Pterostichula solitudo, was recovered in a clade containing the majority of sampled Hypomelina, i.e. Brinckia, Iugidorsum, Sulcipectus and Trachynotidus; this grouping was supported by a synapomorphic placement of the sensory field, which is situated in the middle of plate c4 (9:1).
Bayesian analyses were less conclusive (Appendix S2). Although the taxonomic composition of the major clades was largely congruent with the results of the MP analysis (Molurina + Argenticrinis-Bombocnodulus, Hypomelina + Pterostichula solitudo), the relationships between them remained unresolved. The branch support values were variable (Fig. 6).

Combined analysis of female terminalia traits and molecular data
As in the morphological analysis, Sepidiina was recovered as monophyletic in the combined dataset (Fig. 7), and placed as sister to Trachynotina, with strong support for the relationship (Sepidiina: PP = 0.98, Sepidiina + Trachynotina: 1.00). Additionally, Trachynotina and Hypomelina were both recovered as monophyletic (Trachynotina: 1.00, Hypomelina: 1.00), but it should be noted that owing to a lack of molecular data Ethmus and Pterostichula solitudo were not included in this analysis. Oxurina was again recovered as paraphyletic, with the clade containing Namibomodes and Palpomodes sister to the remaining Sepidiini, excluding other Oxurina, and the rest of the Oxurina taxa recovered as a clade sister to all other Sepidiini (including Namibomodes + Palpomodes). Support for Namibomodes + Palpomodes was strong (1.00), however, support was low for the placement of this clade in the tree (0.54). Clustering and placement of the remaining Oxurina taxa was strongly supported (1.00, 1.00). Argenticrinis was recovered as sister to Psammodes longicornis (1.00) within the clade containing all Molurina taxa.
The relationship between Argenticrinis and P. longicornis was supported by the structure of the coxites, specifically, the elongated plate c3 and the strongly sclerotized baculus of the basal plate with a noticeable inner branching (synapomorphies 4:1, 7:1). The included Molurina taxa + Argenticrinis share a well-developed plate c3 of the coxites and the apical margin of the proctiger is rounded and strongly sclerotized (5:1, 17:1). The clade containing Somaticus (Trachynotina) was supported by the presence of an elongate transverse sclerotization of the apical margin of the basal coxite plate, and the apical edge of the proctiger, which has a sclerotized margin (11:1, 19:1). In the Vieta clade (Sepidiina), the apical area of the proctiger possessed clearly visible longitudinal fibres (16:1). Hypomelina was supported by two synapomorphies: the sensory field of the coxites is situated in the middle of plate c4 and the apical edge of the proctiger is pointed (9:1, 17:2). Finally, the clade containing members of Oxurina, excluding Namibomodes and Palpomodes, share a short transverse sclerotization on the base of plate c3 of the coxites (10:1).
The phylogeny generated from the molecular dataset alone (Fig. 7b) recovered the same topology as the combined dataset, but resolved the polytomies found in the combined tree, which means that even though morphological features can support certain clades and represent synapomorphies, they might dilute the resolution of molecular data.

Ovipositor musculature
MicroCT analysis revealed the existence of five different pairs of muscles within the coxites of the studied specimen. Four were located within the basal plate c1 + c2 (Fig. 8), namely: the dorsal longitudinal muscle (dlm), which extends from the antero-dorsal margin to the posterior margin of plate c1 + c2 (this muscle is probably homologous with M25 in Medvedev, 2001); the ventral longitudinal muscle (vlm) that connects the antero-ventral margin and the postero-ventral margins of plate c1 + c2 (probably homologous with M23 in Medvedev, 2001); the proximal dorso-ventral muscle (pdvm) connecting the bases of the dorsal and ventral sections of c1 + c2; and the distal dorso-ventral muscle (ddvm), located on the dorso-medial part of of c1 + c2. The dorso-ventral muscle is external to the longitudinal muscles. The last of the recovered muscles is the oblique muscle of c4 (omc4), which extends between the dorso-proximal and the ventro-distal margins of the base of c4 (Fig. 8).

Terminology for the morphology of the female terminalia
The terminology to refer to the female terminalia in this paper partly overlaps with the general terminology applied to Insecta, as depicted in the ontology for the Anatomy of the Insect SkeletoMuscular system (AISM; see Table 1). However, a Coleoptera-specific ontology probably would be more appropriate. For some terms in the AISM, even though their definitions are generally compatible with Coleoptera, their labels do not match commonly used beetle anatomy terms. As the beetle-specific ontology currently is being developed (Gir on pers. comm.), the present study should be considered as a reliable starting point for development of logical definitions linked to the female terminalia, especially as it highlights important nomenclatural inconsistencies between the most commonly referenced papers for Tenebrionidae (i.e. Doyen, 1966;Tschinkel and Doyen, 1980).

Terminalia of Pimeliinae
This study demonstrates the usefulness of the morphology of female terminalia for reconstructing phylogenetic relations among darkling beetles. Although this observation has been made already by previous contributors who investigated higher-level relationships within the family (Doyen and Tschinkel, 1982;Iwan and Kami nski, 2016;Kergoat et al., 2014;Watt, 1974), up to now, specific data on pimelinoid beetles have remained scarce (see Watt, 1992). The structures of the investigated ovipositors, spermathecae, proctigers and spicula ventralia proved to be extremely diverse not only in tok-tok beetles, but also in other groups of pimeliinoid beetles. For example, within Adelostomini two major lineages differentiated by the structure of the coxites were distinguished. The first consists of Lepidochora and Stips and is characterized by elongate apical coxite lobes c4 (Fig. 2b), whereas in the other, which groups the remaining Adelostomini, the lobes c4 are not extending beyond the outline of coxites (Fig. 2a, e). This cursory observation highlights the potential informativeness of the morphology of the female terminalia within this group, and should be used as a reference by future revisionary workers. Up to now, the ovipositor and genital tubes of Adelostomini have remained almost fully uninvestigated (however, see Doyen, 1994). The material analysed here reveals that to some extent the morphology of the female terminalia is stable among the studied tribes. This enables the presentation of new and unambiguous diagnostic features for these taxa. This is well-illustrated in three morphologically diverse tribes Adesmiini, Asidini and Nycteliini. The ovipositors of the representatives of Adesmiini, regardless of the high distinctiveness of the beetle forms (Adesmia, Epiphysa, Onymacris and Stenocara), are characterized by the enlarged basal plates of coxites c1 + c2, the membranous and folded coxite c3 lobes (Fig. 2g-j), and the presence of three spermathecal ducts. Asidini, regardless of their geographical origin (Europe, South Africa, North and South America), are easily recognizable from most Pimeliinae by the strong sclerotization of the coxites, which also are mostly fully fused into a single plate. Furthermore, the inner plate of the paraproct characteristically covers a large portion of the basal plate of the coxites (Fig. 2p). Finally, the representatives of Nycteliini are well-characterized by having short arms of the spiculum ventrale and rhomboidal coxite plate c1 (Fig. 3p-s).
Furthermore, the analysed data revealed similarities of female terminalia among some seemingly unrelated tribes. The spicula ventralia of studied Cryptochilini (Sub-Saharan Africa), Idisiini (East Asia) and Pimeliini (Palaearctic) are characterized by unfused bases (Fig. 3d, f, l). However, it has to be mentioned that not all Pimeliini share this feature (e.g. Podhomala Solier; see Chigray, 2019). Nevertheless, outside the above-mentioned tribes this feature has been reported only for Morica Dejean of Akidini (Doyen, 1994;Doyen and Tschinkel, 1982;Tschinkel and Doyen, 1980;Watt, 1992). Until now, Cryptochilini, Idisiini and Pimeliini were not considered to be closely related (Endr€ ody-Younga, 1989;Medvedev, 1973;Watt, 1992). However, different authors, while discussing unrelated features, unknowingly provided support for this relationship. Namely, over the course of history, externally open procoxal cavities were reported for Idisia (of Idisiini), Platyope (Pimeliini), Cryptochile and Pachynotelus (of Cryptochilini) (Doyen, 1994;Watt, 1992). This feature previously has been linked with the close association of the pro-and pterothoraces in these species and treated as a secondary adaptation in contrast to the primarily opened cavities of Zolodininae Watt, which were interpreted as plesiomorphic among Tenebrionidae (Watt, 1992). Furthermore, data analysed here show that studied Pimeliini and Cryptochilini are further linked by the strongly rotated apical lobes of coxites, similarly structured coxite plates and baculi of coxite plate c1 + c2 (Fig. 3a, b, e, j, k). The ovipositor of Idisia is highly reduced and thus cannot be used in this comparison ( Fig. 5e; Watt, 1992). Nevertheless, the similar structure of the spicula ventralia is interpreted here as evidence of a close phylogenetic relationship between Cryptochilini, Idisiini and Pimeliini. This affiliation, and basal placement of those tribes within Pimeliinae, also is strongly supported by genomic data (Smith et al., in prep.). By contrast, the observed diversity of female terminalia of pimeliinoid beetles sometimes leads to problems with homology recognition. The greatest challenge concerns the ovipositor, especially in cases where the four coxite lobes are not fully visible (e.g .  Figs 2-4). Some previous authors have evaded this problem by developing new nomenclature systems for describing more challenging terminalia (e.g. P erez Vera, 2014). However, this approach prevents the incorporation of such data into analyses concerning a wider spectrum of taxa. Examination of the proportions of the plates of closely related tribes and the position of key reference points, such as sensory fields or baculi, might be helpful when assessing the homology of structures in problematic groups (e.g. Banaszkiewicz, 2006;Kami nski et al., 2019). It is hypothesized here that the musculature of the ovipositor can also be used as a reliable navigation point. Although preliminary in this aspect, this study reveals the specific connections of particular muscles and different chitinous elements of the coxities (Fig. 8). The nomenclature introduced herein, as well as the threedimensional models, provides a basis for future studies. To our knowledge, the data presented here constitute the first description of the muscular system of coxities (Doyen, 1966;Eggs et al., 2018;Ernst et al., 2013;Medvedev, 2001;Scudder, 1961).

Terminalia of Sepidiini
The results of this study refute the diagnostic characters proposed by Doyen (1994) for Sepidiini. Although the oblique baculus of coxites, elongated paraprocts and strongly sclerotized apical lobes were present in all of the studied tok-tok beetle species, these features have also been reported for many other unrelated pimeliinoid taxa (Table 2). In addition, the morphology of the arms of the spiculum ventrale is too variable to be used as a reliable diagnostic character at the tribal level (Figs 5t-w, 6). Instead, the analysed data, including previous contributions (Doyen, 1994), suggest that Sepidiini is the only pimeliinoid tribe in which representatives possess laterally flattened ovipositors and elongate proctigers (Fig. 1b). Furthermore, the diversity of the forms of the latter structure is surprisingly high when compared to other tribes of Pimeliinae. For example, no other tribe of Pimeliinae is currently known to possess medially indented proctigers (Figs 2-4;Doyen, 1994). Although the evolutionary drivers leading to this increased diversity remain unknown, it can be hypothesiszed that, at least in Molurina, these modifications enable the proctiger to enfold, and thus protect, the elongate ovipositor (Fig. 5g, h).
The analysed data reveal that the ovipositors of toktok beetles are ventrally directed (Fig. 5a-m). Although this feature was also reported for Erodiini and for Stips and Lepidochora of Adelostomini, it seems to be quite rare among Pimeliinae (Table 1). In the majority of the investigated tribes, the apical lobes of the coxites are rotated inwards. This is wellvisualized by the ventrally located sensory fields (Fig. 5e). Such arrangement impedes the recognition of homologous structures in different pimelinoid beetles, as the area homologous to c3 often is rotated (see Kami nski et al., 2021).
The present study shows no direct indications as to which Pimeliinae tribes have a close relationship with the tok-tok beetles. On account of the inflexed spiculum arms of some of the species and the above-mentioned ventral orientation of the ovipositor, the close affiliation between tok-tok beetles and Adelostomini might be considered (Fig. 2a-f). Nevertheless, both tribes can be separated easily by the differently structured spermathecae (multibranched in Sepidiini, with two ducts in Adelostomini) and the orientation of the baculi of the basal coxite lobe c1 (oblique in Sepidiini, straight in Adelostomini), and no external characters are currently known to link these tribes together (Doyen, 1994). The morphology-based phylogenetic analyses conducted by Doyen (1994) recovered Sepidiini close to Akidini and Ceratanisini Gebien. Judging solely on the morphology of the female terminalia, this relationship seems unlikely as no common features were found to link these three tribes (Figs 2-4;Doyen, 1994).
Within Sepidiini, the morphology of the female terminalia proved to be extremely informative for reconstructing phylogenetic relationships (Fig. 6) and produced topologies that were largely congruent with molecular data (Fig. 7). Several phylogenetically informative features were identified on the apical parts of the ovipositor and proctiger. These structures are often exposed in nondissected specimens, making the female terminalia an easily accessible resource for identification purposes. Our analyses of female terminalia data reveal the need for further research into the taxonomy of Sepidiini, as several previously unrecognized evolutionary lineages were identified. One of the most unexpected findings concerns the recovery of a wellsupported clade containing the type species of the genus Psammodes (P. longicornis), and the following two genera of Hypomelina (Figs 6, 7a)--Argenticrinis and Bombocnodulus. In order to restore the monophyly of Molurina, Argenticrinis and Bombocnodulus are transferred into this subtribe. However, the position of P. longicornis creates a cascade of taxonomic and nomenclatural problems.
With about 150 species, Psammodes is the most diverse genus of tok-tok beetles (Kami nski et al., 2019). Although the monophyly of this genus already has been questioned by previous authors Kami nski et al., 2021), this study is the first to revise the morphology of the type species. When taking into consideration the results of the combined (morphology + DNA) analyses, representatives of Psammodes were recovered in three separate clades (Fig. 7). One containing P. longicornis (hereinafter referred to as Psammodes sens.n.), a second clade containing P. longipes + P. gerstaeckeri, and a third clade containing species with the convex forms traditionally attributed to Psammodes (Fig. 7). An examination of Molurina species, including many type specimens, resulted in recognition of the morphological features delimiting these lineages (for details see Taxonomy section). This information was used to divide the species previously attributed to Psammodes into these three lineages. Only three species were included in Psammodes sens.n., 12 were included in the P. longipes + P. gerstaeckeri clade, and >140 with the most third lineage. Owing to the inclusion of Pimelia scabra Fabricius, the second group is hereinafter referred to as Piesomera Solier (Kami nski et al., 2019). Although, the third lineage contains Psammodes caffra F ahraeus, 1870, which is the type species of Parmularia Koch, this generic name cannot be used as it is a junior homonym of Parmularia Macgillivray (Bryozoa: Cheilostomida). Therefore, a new replacement name--Mariazofia Kami nski nom.n.--is hereby introduced. Piesomera and Mariazofia possess similarly structured female terminalia (3:1, 12:1), however, the latter genus can be easily distinguished by the presence of confined abdominal setal patches in males (Fig. 9a-e). This feature groups Mariazofia together with the genus Moluris--which also is reflected in the molecular analyses (Fig. 7). All new combinations and related nomenclatural acts are presented in the Taxonomy section.
The subtribe Oxurina was recovered as polyphyletic in all phylogenetic analyses. Specifically, a clade composed of Palpomodes + Namibomodes was recovered outside the core Oxurina clade (Figs 6, 7). This result was not unexpected, as the morphological distinctiveness of Palpomodes and Namibiodes from other Sepidiini was already postulated by Koch (1952Koch ( , 1958. To accommodate these findings, a new subtribe Palpomodina Kami nski & Gearner subtr.n. is erected. Detailed diagnosis of this taxon is provided in the Taxonomy section. Furthermore, the morphological analyses recovered Pterostichula solitudo within the main Hypomelina clade. Based on female terminalia (especially proctiger morphology) this species clearly groups with Brinkia, Iugidorsum, Sulcipectus and Trachynotidus. However, at the moment it is not clear if it should be accommodated in a separate monotypic genus. Detailed investigation of other Pterostichula species should be conducted as, besides the data presented here, no information is available on the morphology of female terminalia within this genus. As a result of these uncertainties, no taxonomic decisions are introduced.
This study is the first to include members of Sepidiina within a phylogenetic framework Kami nski et al., 2021). This subtribe was recovered as being closely related to Trachynotina (Figs 6, 7). This relation is well-supported by structures of the female terminalia (synapomorphies 11:1, 19:1; homoplasy 6:1). The recovery of a close affinity between Sepidiina and Trachynotina is interesting, as representatives of both of these subtribes strongly differ in terms of external morphology. In the morphology-based analyses, Trachynotina was rendered polyphyletic, as the genus Ethmus was recovered close to Hypomelina, Molurina and Oxurina (Fig. 6). However, no direct external characters are currently known to link Ethmus with the representatives of the abovementioned subtribes (Koch, 1958). Therefore, the recovered position of this genus is believed to be biased by the relatively apomorphic structure of its female terminalia. Furthermore, due to the unavailability of EtOH-preserved specimens, the position of Ethmus could not be verified by molecular data (Fig. 7b). As a result, no taxonomic decisions are made in the context of this genus.
Although the usefulness of female terminalia morphology for alpha-taxonomy was not directly tested in this paper, the analysed data suggest that at least in some of the groups the structure of the ovipositor seems to be variable between closely related species.
Particularly good examples of this phenomenon are the taxonomically challenging genera Mariazofia of Molurina and Vieta of Sepidiina. In both cases, the coxite plate c4 seems to be the most variable structure. Future alpha-taxonomic and revisionary works also should consider this finding.

Taxonomy
This section summarizes the taxonomic decisions proposed in this paper. A revised key to the subtribes of Sepidiini is available at MorphoBank (http:// morphobank.org/permalink/?P4101).
Diagnosis. This subtribe can be unambiguously distinguished from the remaining subtribes of Sepidiini by a unique structure of the proctiger (17:4)apical margin indented, with mid-section even with lateral sections (Fig. 5g). Furthermore, Molurina is the only lineage among the tribe whose representatives possess well-developed 3rd lobes of the coxites (5:1). An alternative definition of Molurina, based solely on external features, was recently provided by Gearner et al. (2021).
Diagnosis. Piesomera can be distinguished from all other Molurina by the following combination of characters (Fig. 9d, h): basal pronotal margination complete, prosternal process deflated, epipleuron with a distinct groove in median part, male setal patch large, covering several ventrites, and elytral surface covered with microtubercles.
From Argenticrinis, Psammodes can be separated by not having pronotal and elytral surfaces covered with very dense, long, silver-coloured setae (Louw, 1979), and by having elevated edges of the pronotal disc (Fig. 9l). From Bombocnodulus (Fig. 9m), Psammodes differs by having rounded prosternal apophysis (in lateral view)--flat and sunk between procoxae, not fol-other subtribes of Sepidiini (Fig. 10). From Hypomelina, the newly designated subtribe can be distinguished by having the sensory fields of the coxites situated at the base of lobe 4 (9:0)--situated in the middle of lobe 4 in Hypomelina; and rounded apical margin of proctiger (17:1)--pointed in Hypomelina. The structure of the proctiger also differentiates Palpomodina from Molurina--the apical margin is indented, with the mid-section even with lateral sections in Molurina (Fig. 5g). Furthermore, both those subtribes can be distinguished by the structure of spiculum ventrale--thin in Palpomodina, thick in Molurina (Fig. 5d, m). The new subtribe differs from Oxurina by not having a short transverse sclerotization on the base of the coxite lobe 3 (10:0), whereas it can be separated from Sepidiina and Trachynotina by the lack of the sclerotized apical margin of the proctiger (19:0), absence of sclerotization of the apical margin of basal plate (11:0) and Y-shaped spiculum ventrale (21:0). Furthermore, from Sepidiina it differs by the absence of pronotal appendages and by having enlarged mesotrochantin--punctiform or absent in Sepidiina. From Trachynotina it also can be differentiated by the lack of helotatic eyes. Additionally, from Hypomelina, Molurina and Trachynotina the newly designated subtribe can be distinguished by having deep and complete episternal suture of metasternum (Fig. 10c, d); this is fine to obsolescent or abbreviated in the other subtribes (Koch, 1955(Koch, , 1958. Notes. Tools for species and generic identification are available in Koch (1952Koch ( , 1958Koch ( , 1962 and Louw (1979).