The first fossil rove beetle from the middle Eocene Kishenehn Formation (North America) provides evidence for ancient Eocene relicts within the hyperdiverse Staphylinini (Coleoptera: Staphylinidae: Staphylininae)

A new rove beetle (Coleoptera: Staphylinidae) is described from the middle Eocene Kishenehn Formation in Montana, USA. †Tympanophorus greenwalti Chatzimanolis, Brunke & Schillhammer sp. nov. is the oldest known definitive member of the subtribe Anisolinina (Staphylininae: Staphylinini) and the entire Staphylinini propria clade, which contains the majority of the tribe's over 5500 described extant species. In order to provide robust justification for the systematic placement of the Kishenehn fossil, all genera of the Tympanophorus lineage are reviewed and redefined. A key to these genera is provided for the first time. Paratympanophorus Lecoq becomes a junior synonym of Trigonopalpus Cameron, with the following new combinations: Trigonopalpus africanus (Lecoq), Tr. peyrierasi (Lecoq), Tr. pubescens (Lecoq), Tr. punctatus (Lecoq) and Tr. steineri (Lecoq). Tympanophorus schenklingi Bernhauer is moved to Trigonopalpus (comb. nov.) and Ty. clavicornis (Lecoq) is moved to Barygnathus (comb. nov.), and thus, the genus Tympanophorus no longer occurs in the Afrotropical region. Tympanophorus is shown to be at least as old as the middle Eocene, and its disjunct New and Old World lineages are hypothesized to have been separated in the early Eocene. http://zoobank.org/urn:lsid:zoobank.org:pub:CA1993B8-1251-45C3-877E-C4604F78E781


Introduction
The last decade has seen a renaissance in fossil rove beetle systematics, with dozens of new species of Staphylinidae coming to light (for papers summarizing most of these fossils see: Chatzimanolis & Engel 2011Chatzimanolis et al. 2012;Solodovnikov et al. 2013;Peris et al. 2014). One of the major barriers to the detailed integration of palaeontological taxa with extant rove beetle lineages is that the majority of the supra-specific taxa are poorly defined and are in need of revision. Recently, there has been renewed interest in reconstructing phylogenetic relationships within the tribe Staphylinini of Staphylininae (Chatzimanolis et al. 2010;Brunke & Solodovnikov 2013;Brunke et al. 2016), a hyperdiverse group of more than 5500 relatively large and predatory beetle species. While a consensus topology is emerging for Staphylinini and its higher classification is partially revised (e.g. Chatzimanolis et al. 2010;Brunke et al. 2016), the fossil record of this group is in need of revision as the majority of taxa were either described in large polyphyletic genera or incorrectly placed systematically at the genus, subtribe or even subfamily level (Brunke, pers. obs.). One exception is the diverse Early Cretaceous (Yixian Formation, China) assemblage of Staphylinini described and placed in a phylogenetic context by Solodovnikov et al. (2013). However, with the evidence currently available, none of these taxa could be confidently placed within extant lineages and they are probably best treated as early diverging stem groups.
Resolving the position and demonstrating monophyly of the Staphylinini subtribe Anisolinina (sensu Schillhammer 2004) has been particularly problematical, though a recent analysis of molecular data suggests that it may be a natural group (Brunke et al. 2016). However, morphological character evidence for this group as a whole remains elusive (Brunke & Solodovnikov 2013), making the direct assignment of fossils to Anisolinina difficult. At present, Anisolinina is a group of 14 poorly known genera (Schillhammer & Brunke, in prep.) and just over 200 species, which is unknown from the fossil record (A. Newton, unpublished catalogue). The exclusion of Algon Sharp and Philothalpus Kraatz from the subtribe is supported by morphological (Schillhammer 2004) and molecular evidence (Chatzimanolis 2014), though their sister groups remain unknown. The systematic position of Rientis Sharp is also unknown but is thought to be near Algon (Schillhammer 2006).
Based on morphological characters of the mesoventrite, and maxillary and labial palpi, Schillhammer (2004) defined two groups of Anisolinina: the Tympanophorus and Anisolinus lineages. While the genera of the latter are diagnosable, the rarely collected genera of the Tympanophorus lineage are poorly defined and have never been critically re-evaluated. Nearly all genera of Anisolinina are restricted to the Afrotropical and Oriental regions (i.e. the Old World subtropics and tropics). Misantlius and Tympanophorus are exceptional in that they exhibit a disjunct subtropical to tropical distribution between the Old and New World: Oriental-Neotropical, or Oriental C east Palearctic (Japan and Korea) ¡ Nearctic C Neotropical, respectively. The two Afrotropical species of Tympanophorus have not been re-examined since their descriptions and may belong to other genera. Although currently disjunct, these distributions in Anisolinina and other lineages of Staphylinini were hypothesized by Brunke & Solodovnikov (2013) to be formerly widespread in the northern hemisphere during the Palaeogene/early Eocene, becoming relictual after the cooling and polarization of global climate in the late Eocene/early Oligocene. However, evidence for an ancient Eocene age of these genera, from either fossils or divergence estimates, is currently lacking.
Evidence for boreotropical relictualism may be discovered amongst the diverse fossil insect faunas of subtropical to tropical Eocene North America (e.g. Archibald et al. 2011). We here report and describe the first fossil staphylinid from the middle Eocene Kishenehn Formation (c. 46 Ma, Constenius et al. 1989) in Montana, United States, that possesses morphological characters particular to some members of the Tympanophorus lineage of Anisolinina (Fig. 1). As this would currently be the only described fossil of Anisolinina, an accurate and taxonomically precise placement is critical for estimating divergence times within and testing biogeographical hypotheses about the Staphylinini super-radiation. Therefore, the morphological limits of several genera in the Tympanophorus lineage are revised based on the study of a broad taxon sample, and serve as robust justification for the taxonomic assignment of this fossil.

Specimen examination and documentation
The fossil was examined submerged in 100% ethanol under an Olympus Ò SZX-12 stereomicroscope and measurements were made using an ocular micrometer. Photographs of the fossil were taken using a Cognisys Ò StackShot TM 3X macro rail and controller, a Canon Ò EOS40D and a Canon Ò MP-E 65 mm macro lens, and automontaged using Helicon Focus Pro TM 6. The age, taphonomy and biotic diversity of the middle Eocene Kishenehn Formation have been summarized by Constenius et al. (1989) and Greenwalt et al. (2015). Specimens of extant taxa were examined using a Nikon Ò SMZ 745T microscope. Photomontage was accomplished using a motorized Nikon Ò SMZ25 microscope and NIS Elements BR v. 4.5. Photographs were processed in Adobe Photoshop Ò CS6.

Systematic palaeontology
Family Staphylinidae Latreille, 1802 Subfamily Staphylininae Latreille, 1802 Tribe Staphylinini Latreille, 1802 Remarks. The Kishenehn fossil is clearly a staphylinid beetle based on its habitus, the presence of the exposed abdomen and the apically truncate and shortened elytra. All subfamilies of Staphylinidae can be excluded except Aleocharinae, Paederinae and Staphylininae based on the overall body shape. The exposed antennal insertions exclude the Paederinae. Although the habitus and extremely transverse antennomeres of the fossil resemble those of the diverse tribe Lomechusini of Aleocharinae (e.g. those genera near Zyras), the insertions of the antennae occur at the level of the anterior margin of the eyes ( Fig. 2A, C) rather than distinctly posteriad of this level (Fig. 2B). In Aleocharinae, the mandibles in repose are nearly entirely covered by the labrum such that only the lateral portions are normally visible from above (Fig. 2B). The anterior outline of the head capsule of the fossil is not abruptly narrowed and produced anteriad as in the Aleocharinae (Fig. 2B). The fossil beetle is therefore most consistent morphologically with some Staphylininae. The staphylinine tribes Diochini, Xantholinini, Maorothiini and Othiini can be excluded based on the antennal insertions of the fossil ( Fig. 2A), which are each closer to the eye than to each other (Brunke et al. 2011). The fossil does not resemble the characteristically cylindrical members of the tribe Platyprosopini. The tribes Arrowinini and Thayeralini have a robust and parallel-sided body shape similar to Platyprosopini and lack accessory basal lines on the abdominal tergites. Thus, they can also be excluded, leaving the morphologically diverse Staphylinini.
Subtribe Anisolinina Hayashi, 1993 Remarks. Within the remaining tribe Staphylinini, the presence of accessory basal lines on the abdominal tergites (Fig. 2D, E) occurs in the subtribes Staphylinina (e.g. genera related to Eucibdelus and Ocypus) and Anisolinina (all genera), and the incertae sedis genera Barypalpus, Rientis and Philothalpus. The proportions of the body segments in the fossil are unusual for Staphylinini but among the above taxa, the broad abdomen relative to the narrow forebody is restricted to the Tympanophorus lineage of Anisolinina (sensu Schillhammer 2004), specifically to species of the poorly defined genera Tympanophorus and Pammegus (Fig. 3A, B). The position of these lines laterally rather than medially is also unique to the same lineage, though both states occur. In addition to the above differences, members of the Anisolinus lineage of Ansiolinina possess long, rod-like palpi and a more elongate body shape ( Fig. 3F) (Schillhammer 2004) compared to the Kishenehn fossil. However, the genera of the Tympanophorus lineage are poorly defined and have never been critically evaluated. A revision of these generic concepts is provided below so that the Kishenehn fossil may be placed more confidently to a specific genus.
Generic revision of the Tympanophorus lineage. In this lineage, Schillhammer (2004) included the genera Amelinus Bernhauer, Barygnathus Bernhauer, Bombylodes Fauvel, Diatrechus Bernhauer, Pammegus Fauvel, Tolmerinus Bernhauer, Turgiditarsus Schillhammer and Tympanophorus. All members of this lineage differ from those of the Anisolinus lineage in the ridge of mesoventrite, which extends close to its lateral margins, the broad and short mandibles and the basally placed setae on the medial margin of second labial palpomere (Schillhammer 2004). The genus Turgiditarsus should be excluded from Anisolinina entirely (Schillhammer & Brunke, in prep), lacks accessory basal lines and does not resemble the fossil specimen. We have examined the obscure Staphylinini genera Paratympanophorus Lecoq and Trigonopalpus Cameron and conclude that the former is a junior synonym of the latter, and that they are almost morphologically indistinguishable from Tympanophorus (see below for details). The morphologically diverse and speciose genera Diatrechus Bernhauer and Tolmerinus Bernhauer differ from all other genera of the Tympanophorus lineage in the fusiform apical labial palpomeres. Based on the description of Paradiatrechus by Cameron (1944), this genus is similar to Diatrechus and Tolmerinus. These three genera need revision, as at present they cannot be separated by us, do not contain taxa with a distinctly narrow forebody relative to a broad abdomen, and are not redefined herein. The remaining genera are redefined based on a sample as representative as possible, especially given that some taxa are known from very few specimens.
Diagnosis. Species of Amelinus (Fig. 3C) can be recognized by the following combination of character states: labrum small and transverse; apical maxillary palpomere elongate, distinctly longer than previous segment and narrowed to pointed apex, not distinctly narrower than previous segment at widest point, not narrowed at base (Fig. 4A); apical labial palpomere distinctly dilated toward apex and flattened laterally, apical face narrow (similar to Fig. 4E); pronotum with median impunctate line (Fig. 3C); spines on hind tibia thickened (Fig. 3C); most species without secondary sexual structures on male sternite VII (see below). All species where male characters are known bear a short, compact aedeagus, with a short, broad paramere bearing many peg setae; this type of aedeagus also occurs in all species of Barygnathus and Diatrechus known to us.
Remarks. Currently, Amelinus consists of 19 species known from tropical mainland Africa and New Guinea. 'Algon' africanus Bernhauer and several similar undescribed species will key out as Amelinus here. As these species possess several character states not occurring in other Amelinus (reduced punctation on pronotal disc, a lateral group of setose punctures on pronotum and secondary sexual structures on male sternite VII), a new genus could be erected for them. However, this would require a more detailed morphological study that is outside of the scope of this paper. The single species from New Guinea, A. punctus Last, does not belong in this genus (or possibly Anisolinina at all) based on the drawing of the aedeagus in Last (1989) but further taxonomic assignment is impossible at this point.
Diagnosis. Species of Barygnathus (Fig. 3D, E) can be recognized by the following combination of character states: labrum expanded, each lobe less than 2 times wider than long and not widely separated medially (Fig. 4E); apical maxillary palpomere distinctly longer than previous segment, varying from weakly dilated fusiform (Fig. 4E) to slightly expanded at apex, about as broad as previous segment at widest point (Figs 4B); apical labial palpomere distinctly dilated toward the apex and flattened laterally, apical face narrow (Fig. 4E); spines on hind tibia thin (Fig. 3D, E). All known extant species bear a short, compact aedeagus, with a short, broad paramere bearing many peg setae; this type of aedeagus also occurs in all species of Amelinus and Diatrechus known to us.
Remarks. Prior to this study, the genus Barygnathus consisted of two highly similar species with a characteristic habitus due to elongate apical antennomeres, legs and pronotum: B. opacus Bernhauer described from Sri Lanka and later reported from China (Fujian, Sichuan) by Schillhammer (2004); and B. sasajii (Hayashi) from Borneo (Sabah, Malaysia). The morphological concept of Barygnathus is expanded here to include the Madagascan B. clavicornis (Lecoq) comb. nov. (see below), which possesses a less elongate pronotum, extremely transverse antennae and shorter legs. A new genus could be erected for this species, but without any pre-existing phylogenetic analyses for the Tympanophorus group we prefer to include it in Barygnathus. Based on the available material, Barygnathus occurs across southern China, Borneo, Java, Thailand, Sri Lanka and Madagascar. Due to the rarity of its species (fewer than a dozen specimens are known), Barygnathus may occur, albeit undetected, in continental Africa.
Barygnathus clavicornis (Lecoq, 2012) comb. nov. (Fig. 3E) 2012 Remarks. Despite the strongly transverse antennomeres typically associated with most species of the genus Tympanophorus, Tympanophorus clavicornis Lecoq possesses all generic characters of the genus Barygnathus (see diagnosis of both genera) and is therefore moved to this genus.
Species examined. The redescription by Schillhammer (2002) of the single known specimen was used in this study.
Diagnosis. This genus can be recognized easily by the triangular head with temples almost straight and converging strongly posteriad of eyes, and the areas of dense golden and silvery pubescence on the pronotum, elytra and abdominal tergites. Male characters are unknown.
Remarks. The generic concept of Bombylodes (Schillhammer 2002) remains unchanged. This apomorphy-rich and monotypic genus known only from Sumatra might be phylogenetically nested within Pammegus, to which it is similar.
Species examined. Twelve species. All described species were available for study.
Diagnosis. Species of Pammegus (Fig. 3B) can be recognized by the following combination of character states: labrum weakly transverse, lobes not broadly separated; apical maxillary palpomere elongate fusiform, about as twice as long as previous segment and about as broad at widest point (Fig. 4C); apical labial palpomere distinctly dilated, not flattened laterally and with broad apical face (similar to Fig. 4F); spines on hind tibia thin (Fig. 3B); abdomen broad (Fig. 3B). The paramere of Pammegus is also uniquely shaped among Anisolinina: in parameral view with lateral expansion at midlength and long slender apical portion.
Remarks. Pammegus, with 12 described species, is restricted to the Oriental region as far as is known (Schillhammer 2002 . 4F); apical maxillary palpomere distinctly constricted apically and entirely setose, at most slightly longer than and always narrower than previous segment, with acuminate apical portion (similar to Figs 2C, 4D); apical labial palpomere distinctly dilated, not flattened laterally and with broad apical face (similar to Fig. 4F); spines on hind tibia variable; ventral surface of head with pair of gular setae approximate and placed in middle of gula (Fig. 4G); prosternum rounded in lateral view, not projected acutely ventrad.
Remarks. The genus Trigonopalpus had remained more or less in obscurity since its description by Cameron (1951) based on a single species from Uganda. An examination of the only known syntype in the NHMUK revealed that it was morphologically similar to Tympanophorus and the Afrotropical genus Paratympanophorus recently described by Lecoq (2002). The type species of Trigonopalpus was found to possess a setose apical maxillary palpomere and medially placed gular setae, both features of Paratympanophorus given by Lecoq (2002) to differentiate it from Tympanophorus. As we cannot find any morphological evidence to maintain separate generic concepts, we herein synonymize Paratympanophorus with Trigonopalpus. Of the diagnostic character states given by Lecoq (2002), only the approximate gular setae were found to be truly diagnostic. A pubescent apical maxillary palpomere was also observed in two undescribed species of Tympanophorus from New Guinea and Australia that do not bear medially placed gular setae. In the four species of Trigonopalpus available for study, the prosternum was observed to be rounded rather than projected ventrad as in all studied species of Tympanophorus and Pammegus. It is possible that Trigonopalpus is phylogenetically nested inside Tympanophorus, as all known unique character states of the latter genus are shared by the former genus. Currently, Trigonopalpus consists of seven species distributed in the Afrotropical region, mostly in Madagascar.
Trigonopalpus schenklingi (Bernhauer, 1912) comb. nov. Diagnosis. Species of Tympanophorus can be recognized by the following combination of character states: lobes of labrum distinctly transverse more than twice as long as wide, or strongly reduced, lobes distinctly separated by Ushaped emargination (Fig. 4F); apical maxillary palpomere distinctly constricted apically, rarely setose, with acuminate apical portion, at most slightly longer than and always narrower than previous segment (Figs 2C, 4D, F); apical labial palpomere distinctly dilated, not flattened laterally and with broad apical face (Fig. 4F); accessory basal lines placed laterally and well developed, reaching at least half the length of each tergite (Fig. 2E); ventral surface of head with pair of gular setae distant, placed in each anterior corner (Fig. 4H); prosternum projected acutely ventrad. Aedeagus usually asymmetrical.

Remarks.
A large variation in size (6.7 mm in an undescribed Caribbean species, to nearly 16 mm in Ty. canaliculatus), head shape (entirely rounded and convergent to strongly angular hind angles) and body shapes (dilated toward apex to parallel-sided) was observed between species of Tympanophorus despite consistent shapes of the maxillary and labial palpi, the labrum and the strong development of the accessory basal lines. In many species, the antennae were removed from the margin of the frons such that they were positioned about halfway between the margin and the eye, a character state also occurring in Pammegus. We also observed a distinct tendency toward more transverse antennal segments and a more triangular apical maxillary segment in the smallest species. The 12 described species of Tympanophorus are distributed in the Oriental, Palearctic (Korea and Japan), Nearctic and Neotropical Regions. The African species originally described in this genus belong to Barygnathus and Trigonopalpus. We herein report Tympanophorus and the entire subtribe Anisolinina from the Caribbean subregion for the first time based on one male specimen of an undescribed species from Jamaica (CNC). Diagnosis. Among most species of Tympanophorus, yTy. greenwalti can be distinguished by the small size alone. From the few Tympanophorus smaller than 9 mm in body length (all undescribed), it can be easily distinguished by the extremely transverse antennomeres 5-9 (more than 3 times as wide as long).
Description. Total length at least 6.0 mm, probably about 6.5 mm (body strongly contracted); body colouration dark brown to black except antennae, mouthparts, legs and tip of segment VIII reddish brown. Body elongate, with head and pronotum both distinctly narrower than elytra and abdomen. Head ovoid, hind angles entirely rounded. Head 0.9 mm long and 0.9 wide (at middle). Eyes moderately large, positioned anteriorly and slightly longer than temples. Mandibles prominent, broad, curved, sharp distally; as long as half the length of head. Maxillary palpi short, labial palpi not preserved; last maxillary palpomere triangular and distinctly constricted apically, glabrous, slightly longer and narrower at base than previous segment, apical segment of right palpus apparently broken at about midlength, palpomeres 1 and 2 not clearly preserved. Labrum with lobes transverse, more than twice as long as wide and broadly separated. Antennal inserts situated about halfway between frontal margin and eye. Antennomeres 1-3 longer than wide, 4 subquadrate, 5-9 extremely transverse, antennomeres 10 and 11 not preserved. Antennomere 1 longer than antennomere 2; antennomere 3 slightly longer than 2; antennomere 4 wider than 3 but narrower than 5; antennomere 4 slight longer than 5; antennomeres 5-9 subequal in length and width. Head constricted, neck only partially visible but about one-third of posterior head width. Head and pronotum with dense small punctures. Pronotum widest medially, anterolateral corners curved ventrad; lateral sides slightly concave after middle, pronotum constricted posteriorly. Pronotum slightly longer than wide; 1.2 mm long and 1.1 wide (at widest part); on lateral sides with long seta near anterior 2/3. Elytra wider than long; 1.2 mm long and 1.7 mm wide (at widest part). Meso-and metatibia slightly curved; metatibia slender, not conspicuously spinose; mesofemur and mesotibia setose; mesotibia with coarse apical spines; metatarsi slender; apical metatarsomere long, longer than previous two tarsomeres. Abdomen broad, compact, setose; with prominent paratergites; abdominal tergites VI and V (III obscured) with laterally positioned and well-developed accessory basal lines that extend beyond half the length of tergite.
Etymology. The species is named in honour of Dr Dale Greenwalt who provided the specimen to us and has been instrumental in uncovering Kishenehn fossils.
Remarks. According to the new diagnoses and limits of the genera given below under 'Tympanophorus lineage', the fossil specimen can be placed within Tympanophorus or Trigonopalpus based on the posteriorly shifted antennal insertions, strongly developed and laterally positioned accessory basal lines, short, triangular and glabrous apical maxillary palpomere (best visible on left palpus) and strongly transverse labral lobes. The combination of the antennal insertions and accessory basal lines are unique to Tympanophorus/Trigonopalpus and are sufficient to make a confident identification, even though the diagnostic mouthpart characters are less than optimally preserved in the fossil. Although the position of the gular setae is the only definitive diagnostic difference between Tympanophorus and Trigonopalpus, the apical maxillary palpomere of the fossil specimen is certainly not entirely setose, excluding the latter. The fossil bears transverse subapical antennomeres, consistent with Tympanphorus but far more transverse than any other known species. It is possible either that these segments have become compressed during fossilization or that the unusual antennal morphology reflects an association with social insects as in other members of host-reliant Staphylinini (e.g. Quedius (Velleius) Leach, Haematodes Laporte, Agacerus Fauvel). In general, the biology of all Tympanophorus is unknown, though they have been collected in forests in decaying wood and leaf litter (Brunke, pers. obs.).
Other species of Staphylinidae are present in the Kishenehn Formation (Greenwalt et al. 2016;Chatzimanolis, unpublished data) from multiple subfamilies (e.g. Oxytelinae, Omaliinae, Paederinae), though their generic assignment is uncertain. The reason for this is mostly due to the state of the rove beetle taxonomy (limited or not up-todate genus-level revisions of extant taxa that render the placement of a fossil uncertain) rather than the preservation of the fossils. A revision of the generic concepts of the Tympanophorus lineage revealed that all African species of Tympanophorus belonged to other genera, and that the revised distribution of this genus is disjunct in a similar way to other Staphylinini that were hypothesized to be previously widespread across the northern hemisphere during the early Eocene (see Introduction). The discovery of yTympanophorus greenwalti provides evidence that Tympanophorus is at least as old as the middle Eocene, making it possible that the divergence between disjunct New and Old World lineages dates to the early Eocene. The genus Tympanophorus occurs in the Recent fauna of North America and is represented by two species: eastern Tympanophorus puncticollis (Erichson) and south-western Ty. concolor Sharp, which extends into Central America. The small body size, strongly transverse antennomeres, elongate pronotum and posteriorly inserted antennae of yTy. greenwalti suggest that it is related to undescribed species of Tympanophorus from Jamaica, Costa Rica and Panama (not illustrated here) rather than to Ty. puncticollis LeConte or Ty. concolor, which are much larger, and have a broad and circular pronotum, regular antennal insertions and much less transverse antennomeres. The Eocene was a period of globally warmer temperatures and radically different faunal distributions compared to the present (Archibald et al. 2011), and the presence of a tropical group of Tympanophorus in the middle Eocene of North America is congruent with the discovery of small tarsier-like primates (Ni et al. 2010) and tropical ant groups (LaPolla et al. 2015) from the Kishenehn Formation.