Phylogenetic position of genera Acrostilicus Hubbard and Pachystilicus Casey (Staphylinidae, Paederinae) and their redescription

. Acrostilicus Hubbard, 1896 and Pachystilicus Casey, 1905 are North American genera known from only one and two species, respectively, and have never been a subject of a modern revision. In fact, Acrostilicus was not even properly described as its author provided only a sketchy diagnosis of the genus and species. Here, we provide a redescription of the genus Acrostilicus and its species and illustrate the habitus and male genital features. For the first time, we also redescribe Pachystilicus and its two species, and provide their differential diagnoses. Additionally, we tested the phylogenetic position of both genera. They were scored into a morphological matrix supplemented with molecular data and the analyses were run using Bayesian inference and maximum likelihood methods. A total of 119 morphological characters and 4859 bp of nuclear (28S, TP, Wg, CADA, CADC, ArgK) and mitochondrial (COI) sequences were analysed for 46 taxa. The results confirmed that both Acrostilicus and Pachystilicus are members of the subtribe Stilicina, but at the same time challenged the monophyly of the subtribe in its current composition. Additionally, we provided further evidence for non-monophyly of the subtribe Medonina and discussed the biology of Acrostilicus and Pachystilicus .


Introduction
Paederinae Fleming, 1821 is one of the most species-rich subfamilies of Staphylinidae Latreille, 1802, with around 7600 described species in more than 225 genera (Newton 2018;). As such a large but still poorly known group, they face the problem of the taxonomic impediment. Although new species are described rather regularly, Paederinae are rarely the subject of modern taxonomic revisions. The vast majority of the known genera and species remains unrevised, and a large proportion of publications did not include photographs or drawings of habitus and genitalia with their descriptions, nor did they attempt to establish the phylogenetic placement of these taxa. Examples of taxa that were originally described more than a hundred years ago and have never been the subject of phylogenetic study are the two exclusively Nearctic genera Acrostilicus Hubbard, 1896 andPachystilicus Casey, 1905. Acrostilicus was only very recently redescribed based on newly collected material (Brunke & Schnepp 2021) but Pachystilicus has never been studied using a modern taxonomic approach.
Acrostilicus are slender beetles with rather unusually long legs and were originally found in burrows made by the gopher tortoise, Gopherus polyphemus (Daudin, 1801), in Florida by Hubbard (1896). The description of the new genus and species, Acrostilicus hospes Hubbard, 1896, consisted of a single character stating that the labrum has only one tooth, and it was compared to Stilicopsis Sachse, 1852 (subtribe Stilicopsina Casey, 1905). Although Hubbard (1896), establishing the new genus and species name, provided a note that he would like to give a description for it, he never fulfilled his plans. Blackwelder (1952) mentioned the species in a catalogue of generic names and included the genus in his key to paederine genera (Blackwelder 1939), where he keyed it to a group of genera that are currently in the subtribe Stilicina Casey, 1905, not Stilicopsina. It was placed in a couplet with Stiliderus Motschulsky, 1858, and the presence of a single median tooth on the labrum was the separating character (more than one tooth in Stiliderus). The first more detailed morphological description for the genus and its habitus, together with characters in a key, were provided in Moore & Legner's (1979) illustrated guide to the genera of Staphylinidae of America north of Mexico excluding Aleocharinae Fleming, 1821, but Moore and Legner did not provide a species description or a formal diagnoses for the genus or its single species. However, they stated that the genus is distinguished from others by the narrow neck and a single tooth on the labrum and it was again keyed with the same group of genera in Stilicina. Newton et al. (2001) included Acrostilicus in the key to the Nearctic genera of Stilicina and commented on the poor taxonomic status of the genus, indicating the need for its re-evaluation. In addition to the single tooth on the labrum, Newton et al. (2001) also included one additional character to distinguish the genus, i.e., punctures of the head not very dense. Brunke & Buffam (2018) listed the species in their review of the Nearctic rove beetles specialised on the burrows and nests of vertebrates and mentioned its uncertain status. Recently, Brunke & Schnepp (2021) redescribed Acrostilicus based on the study of the type material and two recently collected specimens. They confirmed the placement of the genus in Stilicina based on shared morphological characters and tentatively considered the genus as valid pending more detailed phylogenetic study.
as Stilicus quadriceps by Lucas (1920). Further, the genus was also included in Blackwelder's key (1939), again in a couplet with Stilicus. However, the characters separating them were different from those suggested by Casey (1905) and were as follows: head emarginate at the base, labrum with median teeth separated by twice their average width, notch rounded, pronotum punctured very differently from the head. Moore & Legner (1979) used the last character to separate Pachystilicus from Rugilus in their key and provided a habitus illustration and updated description for this genus. They also noted that both species were found with ants, which was the first such mention for P. quadriceps as LeConte (1880) in his original description wrote that all Stilicus species that he collected were found on the seashore under seaweed, cast up by the waves. This information was not confirmed later by Newton et al. (2001) who stated that the biology of Pachystilicus quadriceps was unknown, and only P. hanhami was confirmed to be found in ant nests. Newton et al. (2001) again keyed the genus together with Rugilus and used the same characters as Blackwelder (1939), adding Casey's (1905) "robust habitus" character. Frania (1986), in his study on several Stilicina genera, suggested that Pachystilicus is closely related to or congeneric with Rugilus. No further studies were undertaken on this genus or any of its species.
The uncertain taxonomic status of both genera, partly resulting from an unclear separation from the diverse genus Rugilus, motivated us to redescribe each species of Pachystilicus and supplement the recent redescription of Acrostilicus, as well as conducting a phylogenetic analysis to confirm their assignment to the subtribe Stilicina. In this study, we provide the habitus photographs and illustrations of the male genitalia of all three species, as well as the results of a total-evidence phylogenetic analysis, where both genera were included in the morphological partition of the data matrix.

Examination and deposition of taxa
Specimens of the redescribed taxa were studied using a Nikon SMZ1500 and an Olympus SZX9 stereo microscopes. All measurements are given in millimetres and were made with an ocular micrometer mounted on a stereoscopic microscope.
Abbreviations for measurements are as follows: EL = elytra length EW = elytra width HL = head length HW = head width PL = pronotum length PW = pronotum width A Nikon DS-Fi1 camera was used for taking pictures of habitus, and photographs were stacked using Zerene Stacker software (ver. 1.04, Zerene Systems LLC, Richland, Washington State, USA, 2009). The pictures were further edited in Adobe Photoshop CS6, while illustrations were made in Adobe Illustrator CS6 (ver. 16.0.0, Adobe Systems Incorporated, San Jose, California, USA, 2007) based on a photograph and observations. Specimens were relaxed in warm water and dissected. Morphological terminology follows Bogri et al. (2020). Data from the specimen labels are cited verbatim (text between double quotes, ""), with original spelling retained. A slash '/' separates labels for each specimen. Additional morphological observations on North American Stilicina were made during the visits of the first and last authors to the respective collections, without dissecting the specimens. Whenever possible, the type specimens were studied for comparison. If no actual specimen was present in the collection, the original descriptions were used (Fall 1901;Casey 1905;Fall & Cockerell 1907). Information on the status and repositories of all specimens used for such comparison is given in Supp. file 1.
Material from the following collections was used in the current study and its currently deposited there: Most of the specimens used for phylogenetic analysis were scored as in Żyła et al. (2021). Those that were newly scored into the matrix were relaxed in warm water and dissected. Information on the repositories of all taxa used for phylogenetic analysis is given in Supp. file 2.

Taxon sampling and outgroup for phylogenetic analyses
Representatives of all currently recognised tribes of Paederinae were included in the analysis. Since the main goal of our research was testing the phylogenetic position of Acrostilicus and Pachystilicus, we increased the number of representatives of the subtribe Stilicina, where they are traditionally classified. We also scored more representatives of the subtribe Medonina Casey, 1905 as it is potentially closely related to the genera under study (Żyła et al. 2021). In total, 46 taxa were included in the final combined dataset. Both molecular and morphological data were available for most representatives, except for 10 taxa for which only morphological characters were included in the matrix, including four Stilicina (Acrostilicus hospes Hubbard, 1896, Megastilicus formicarius Casey, 1889, Pachystilicus hanhami (Wickham, 1898), Panscopaeus lithocharoides (Sharp, 1889)); two representatives of Medonina (Deroderus Sharp, 1886, Ecitocleptis Borgmeier, 1949; and four genera classified as Lathrobiini Laporte, 1835 incertae sedis (Micrillus Raffray, 1873, Mimophites Fauvel, 1904, Scymbalium Erichson, 1839, and Synecitonides Reichensperger, 1936. We chose representatives of the subfamily Staphylininae Latreille, 1802 as the closest related outgroup and Tachyporinae MacLeay, 1825 as a more distantly related outgroup.

Morphological characters
We constructed the morphological matrix in Mesquite ver. 3.5 (Maddison & Maddison 2018) using 119 characters, which were primarily derived from the matrices of Bogri et al. (2020) and Żyła et al. (2021). Unknown character states were coded using '?', while inapplicable states were marked as '-'. The list of characters is provided in Supp. file 3. The nexus file containing the character matrix is available as Supp. file 4 and in MorphoBank (project no 4067) under this permalink: http://morphobank.org/permalink/?P4067.

GenBank data
To construct our molecular matrix, we used seven gene fragments: the nuclear protein-encoding genes carbamoylphosphate synthetase (CADA and CADC), topoisomerase I (TP), arginine kinase (ArgK), and wingless (Wg), the mitochondrial protein-encoding cytochrome c oxidase I (COI), and the nuclear ribosomal 28S. The Genbank accession numbers of all sequences are given in Supp. file 2. All sequences were already used in Żyła et al. (2021); thus, the amplification, sequencing, sequence editing and assembly protocols are described there.

Sequence alignment
Sequences were newly aligned in Geneious ver. 9.1.7 (Biomatters Ltd, Auckland, New Zealand, 2005) using the MAFFT plugin ver. 1.3.6, based on MAFFT (Katoh et al. 2002). 28S was aligned using the E-INS-i algorithm of MAFFT, and ambiguously aligned regions were identified and removed with the server version of Gblocks (Talavera & Castresana 2007). We allowed gap positions within the final blocks and less strict flanking positions but did not allow many contiguous non-conserved positions. The resulting 28S alignment was 866 bp and had very few scattered and, usually, single-nucleotide gaps. Individual gene alignments were concatenated with the 'concatenate' function of Geneious. The concatenated sequence alignment is provided in Supp. file 5 in fasta format.

Data matrix and partitioning
Our combined matrix of molecular (4982 bp) and morphological (119 characters) data for the total number of taxa under study (46) was analysed using Bayesian inference (BI) and maximum likelihood (ML). For the molecular data matrix, the alignment was initially partitioned by gene and, for proteinencoding genes, by codon position. The optimal partitioning scheme and the corresponding models of nucleotide evolution were determined by PartitionFinder ver. 2.1.1 (Lanfear et al. 2016) using the Bayesian Information Criterion running on CIPRES Science Gateway ver. 3.3 (Miller et al. 2010). Models for MrBayes and IQ-TREE were considered, branch lengths were unlinked, and the search was set to the 'greedy' algorithm (Lanfear et al. 2012). The morphological data in the combined matrix were analysed as a single, separate partition using the maximum likelihood model for discrete morphological character data, under the assumption that only characters that varied among taxa were included (Mkv) (Lewis 2001).

Phylogenetic analysis
Bayesian analysis was performed using MrBayes ver. 3.2.6 (Ronquist et al. 2012) running on CIPRES. The analysis used four chains (one cold and three heated) and two runs of 30 million generations with default prior settings, except for the temperature, which was set to 'temp = 0.08' for better mixing. The analysis was conducted with a gamma distribution for the morphological partition, and autapomorphic characters were included. The third COI codon positions were excluded as in previous studies (e.g., Żyła et al. 2021). A script for the combined analysis in MrBayes is given in Supp. file 6 and is also available at https://github.com/DagmaraZyla/Acrostilicus_Pachystilicus.

Phylogenetic analyses
PartitionFinder found the following five partitions: 1) 28S; 2) TP2, CADC2, CADA2, COI2, ArgK2, COI1, ArgK1, Wg2, Wg1, TP1, CADC1, CADA1; 3) Wg3, ArgK3, TP3; 4) CADC3, CADA3 and 5) COI3. In the case of MrBayes, GTR+I+G was found to be the best supported model for the first four partitions. For the last one containing the third codon positions of COI, the HKY+G model was selected as the best supported. For IQ-TREE, GTR+I+G was also found to be the best supported model for the first two partitions. For the third and fourth partitions, TVM+I+G and TRN+I+G were selected, respectively. For the fifth one, the HKY+G model was also found as the best supported. However, this partition was excluded in both analyses as it has been shown that it can potentially bias phylogenetic analyses (e.g., Swofford et al. 1996;Lin & Danforth 2004). Our independent Markov chains converged on the same stationary distribution as visualized in Tracer ver. 1.7.1, and both combined and individual traces were inspected. The effective sample size (ESS) values were greater than 200 for all parameters indicating good mixing of the chains. The tree topology presented in Fig. 1 is the 50% majority-rule consensus tree of BI analysis with support values of both BI and ML.
Both BI and ML analyses showed almost identical topologies ( In the BI tree, this clade was recovered as sister (P = 0.93) to the well supported 'true' Lathrobiina (PP = 1, UFB = 100), which was comprised of four taxa, and altogether were recovered as sister (PP = 1) to the well-supported 'Medonina and allied taxa' clade (PP = 1, UFB = 100). In this part of the tree, the ML analysis resulted in a different topology, and the clade was resolved as sister directly to the 'Medonina and allied taxa' clade but with no support.

Diagnosis
As in species redescription.

Diagnosis
The species and genus can be differentiated from all other Stilicina based on the combination of the following characters: the presence of a single median tooth on the labrum, mandibles symmetrical, posterior margin of head rounded, and legs elongated. Habitus. Slender, elongated, with long legs. Integument shining, with coarse, umbilicate sculpture. Head and pronotum differently punctured.
abdoMen. Finely sparsely punctate, widest at tergite V. Tergites III-VI shallowly impressed at base, with fringe of setae on posterior margin. Tergites III-VII with pair of paratergites on each side. Tergite VIII with posterior margin rounded. Sternite III without keel between coxae. Sternite VII with straight apical margin. Female: posterior margin of sternite VIII straight. Male: sternite VIII with moderately shallow and broad median emargination of posterior margin; sides of emargination on sternum VIII rounded (Fig. 2B). Aedeagus with parameres reduced and fused to median lobe; ventral process slightly shorter than uneverted internal sac; in parameral view with apex of ventral process rounded, sides sinuate; dorsal plate rather large and relatively weakly sclerotised (Fig. 2C-D).

Distribution
The species has only been recorded from Florida (USA) (Brunke & Schnepp 2021

Diagnosis
The genus can be recognised based on the following combination of features: body robust (usually slender in Rugilus), covered with fine, dense, golden setae (absent in Acrostilicus, Eustilicus Sharp, 1886 and Megastilicus Casey, 1889); antennal tomentose pubescence starting from antennomere 4; paired teeth on labrum (single in Acrostilicus); the base of head broadly emarginate, punctation of head clearly denser and coarser than that of pronotum, head without dense ground sculpture (often present in Rugilus); absence of scutellar ridges of mesoscutellum (one or two present in other North American Stilicina).

Redescription
Habitus. Medium sized, robust, covered with fine golden setae, integument densely punctured. Head more densely punctured than pronotum, without short and stout bristles.
Colouration. Dark brown; head darker, legs and antennae reddish brown.
tHorax. Pronotum wider than long, rhomboid, with anterior angles obtuse, narrower in anterior and posterior part, widest in middle, disk finely densely punctured, midline less punctured. Long black seta on each side of pronotum, in apical portion of widest part; shorter black seta in posterior portion of widest part. Superior marginal line deflexed, not meeting with inferior line. Basisternum of prosternum without macrosetae or microsculpture, but surface wrinkled, longitudinal carina present. Furcasternum of prosternum longer than ½ of basisternum length, reaching farther than tip of postcoxal process, triangular, acute, with sharp longitudinal carina and transversal carina. Hypomeron not delimited from pronotal disc by carina. Furcasternum of mesosternum with longitudinal carina, short, reaching less than ⅓ of distance between coxae, rectangular. Elytra quadrate, longer and wider than pronotum, with row of setae on edge of posterior margin, without epipleural ridge; humeral angle indistinct, rounded; surface glossy, interstices without microsculpture. Scutellum without ridges, integument reticulate, anterior margin rounded. Hind wing fully developed, with MP3 vein present. Legs, as rest of body, covered with shiny gold setae. Trochantins large, quadrate. Middle coxa contiguous, ridge below coxal rests present. Tibiae without spines or long bristles on outer edge. Protibia with two fully developed, longitudinally placed, comb-like rows of setae, and three associated macrosetae. Tarsi 5-segmented, stout, with one pair of empodial setae on each tarsus, equal to or slightly shorter than claws. Protarsus narrowly dilated, protarsomere 1 slightly longer than protarsomere 5, protarsomeres 2-4 decreasing in length, protarsomeres 1-4 narrower or equal to meso-and metatarsomeres 1-4, with dense pale adhesive setae on ventral side, protarsomere 4 not bilobed. Mesotarsus with mesotarsomere 1 longer than mesotarsomere 2, mesotarsomere 4 similar to preceding one. Metatarsus with metatarsomere 1 longer than metatarsomere 2, metatarsomere 4 similar to preceding one, metatarsomere 5 equal to metatarsomere 1, longer than metatarsomere 4, but shorter than metatarsomeres 2-4 combined.
abdoMen. Finely pubescent, with fine and dense punctation, wider than elytra, widest at tergite V. Both ventral and dorsal sides with golden setae mixed with numerous longer black ones. Tergites III-VI impressed at base. Tergites III-VII with pair of paratergites on each side. Tergite VIII with posterior margin rounded. Sternite III without keel between coxae. Female: posterior margin of sternum VIII straight. Male: sternum VIII with moderately deep and broad median emargination of posterior margin; sides of emargination rounded (Figs 3B, 4C).
aedeagus. With parameres reduced and fused to median lobe; ventral process stout, apically broadly truncate, slightly longer than uneverted internal sac; in parameral view ventral process bent and with apex acute (
abdoMen. As in generic description; sternite VII with apical margin straight.

Distribution
The species has previously been recorded from the following US states: California, Massachusetts, and Missouri (Newton 2018). We newly report this species from the Canadian province of British Columbia. Pachystilicus hanhami (Wickham, 1898) Fig. 4 Stilicus hanhami Wickham, 1898: 220 (original description, notes).

Diagnosis
Differs from P. quadriceps by shorter and slightly sparser setae covering body, longitudinal carina of basisternum less elevated. Additionally, the median emargination of posterior margin of male sternite VIII is shallower and narrower than in P. quadriceps, with the depth of emargination ca 1/5 of sternite length.
abdoMen. As in generic description; sternite III with short medial longitudinal keel.

Distribution
The species has previously been recorded from the Canadian provinces of Alberta, British Columbia, Manitoba, New Brunswick, and Saskatchewan and the US state of Oregon (Newton 2018). New records include the following US states: Massachusetts, New Hampshire, and Oklahoma.

Position of Acrostilicus and Pachystilicus in the phylogeny
Our total-evidence dataset combining molecular and morphological data allowed us to demonstrate the phylogenetic position of Acrostilicus and Pachystilicus for the first time and confirm their classification as subtribe Stilicina within the tribe Lathrobiini. The characters supporting their placement in Stilicina are as follows: small, acicular maxillary palpomere 4; expanded labrum, almost or completely covering mandibles when closed; gular sutures at least partially fused; narrow neck, equal to or less than 1/5 of head width; prosternal apophysis present as clear invagination; and metatibia with apical ctenidium on one side only. Acrostilicus was resolved as sister to Rugilus, and this relationship is supported by several morphological characters: postoccipital suture not crossing gular sutures and indistinct at the base of gula; lack of ridge below coxal rests on mesosternum, lack of MP3 vein on the hind wing, and lack of a row of setae on the edge of elytral posterior margin. The position of Pachystilicus was unresolved within a group consisting of Acrostilicus + Rugilus and Megastilicus. The clade consisting of these four genera was supported by the lack of a marginal ridge on the elytral epipleuron. All four genera were resolved as sister to Eustilicus + Stilicoderus and together form the 'core' of subtribe Stilicina.

Generic status of Acrostilicus and Pachystilicus
As both genera are easy to diagnose and recognise, we confirm that they should be treated as separate genera in the subtribe Stilicina and not synonymised with any other genus. Acrostilicus is the only genus within the Nearctic representatives of the subtribe with a single median tooth on the labrum, and when combined with its slender habitus and prolonged legs, make it easy to identify. As for Pachystilicus, which has been suggested to be a synonym of Rugilus, the following characters separate the two genera in our morphological matrix: the presence of the tomentose pubescence on the antennomere 4 (absent in Rugilus), postoccipital suture that crosses gular sutures and continues through the base of gula (in Rugilus, the suture does not cross gular sutures and is indistinct at the base of gula), head longer than pronotum (head shorter or as long as pronotum in some Rugilus), presence of longitudinal median carina on basisternum (absent in some Rugilus), presence of ridge below mesosternal coxal rests (absent in Rugilus), absence of scutellar ridges of scutellum (one ridge present in Rugilus), presence of ridge below mesosternal coxal rests (absent in Rugilus), presence of the row of setae on the edge of the posterior margin of elytra (absent in Rugilus), and presence of MP3 vein on the hind wing (absent in Rugilus). Additional differences are the body shape (stout in Pachystilicus, usually rather slender in Rugilus), separation of labral teeth by twice their average width and rounded labral medial notch (although according to Frania (1986) present also in some Rugilus species), distinctive emargination of the head in Pachystilicus (in a majority of specimens), and differently punctured head and pronotum (very fine and dense in Pachystilicus, while coarse and elongate in many Rugilus). It is, however, possible that among the mega-diverse genus Rugilus with 265 described species (Newton 2018), there might be species that possess some of the abovementioned characters. Nevertheless, the combination of features given in the Pachystilicus diagnosis is unique for the genus. Among North American Stilicina, Pachystilicus can be easily distinguished from the genera Acrostilicus, Eustilicus and Megastilicus by its rectangular head with straight temples, emargination at the head base, and dense, gold setation. All three genera have a (sub)orbicular head with temples rounded and without the emargination. Additionally, Megastilicus has its body covered with characteristic short, black, stout setae and the dorsal surface of the head, pronotum, and elytra are granulate (Żyła & Koszela 2021). The same combination of characters above distinguishes Pachystilicus from the following North American Rugilus species: R. angularis (Erichson, 1840), R. angustatus (Geoffroy, 1785), R. apicalis (Casey, 1905), R. biarmatus (LeConte, 1880), R. dentatus Say, 1831, R. lacustrinus (Casey, 1905), R. occiduus (Fall, 1901), R. orbiculatus (Paykull, 1789) (head more quadrate but converging in the posterior part), and R. rudis (LeConte, 1863). From Rugilus species with a more rectangular head, i.e., R. ceylanensis (Kraatz, 1859), R. oblitus (Fall, 1907), R. opaculus (LeConte, 1880), and R. rufipes Germar, 1836, Pachystilicus differs in the overall stout habitus, presence of head emargination, dense gold setation, and differently punctured head and pronotum.
As opposed to other Stilicina, both Acrostilicus and Pachystilicus do not have any scutellar ridges.

Non-monophyly of Stilicina and Medonina
Another genus present in the analysis and currently assigned to the subtribe, Panscopaeus Sharp, 1889, was not resolved as part of the 'core' Stilicina clade and renders Stilicina non-monophyletic.
Panscopaeus is a small genus with seven described species distributed in the Indomalayan region (Assing 2011). It was originally described from a single species Scopaeus lithocharoides Sharp, 1874 and later considered as a subgenus of Achenomorphus Motschulsky, 1858 (Blackwelder 1952). Herman (2003) revalidated its status as a distinct genus and later it was placed in the subtribe Medonina (Smetana 2004). Assing (2011), in his revision of Panscopaeus, moved it to Stilicina, mostly based on the morphology of the mouthparts, some other external characters, and the male sexual characters. Characters that morphologically distinguish Panscopaeus from the rest of Stilicina are: smaller labrum, not fully covering mandibles from above (but it occurs at least in some Rugilus species as well), separated gular sutures, and the presence of both anterior and posterior ridges on mesoscutellum. In our BI tree, it was separated from the rest of Stilicina by a clade consisting of the three Medonina genera Thinocharis Kraatz, 1859, Deroderus, and Ecitocleptis, and highly modified ant inquilines Mimophites and Synecitonides (subtribe incertae sedis in Lathrobiini in both cases). As all these genera, except for Thinocharis, were included in the morphological dataset only, we treat the result with caution and do not make any taxonomic changes. Moreover, in our ML tree, Panscopaeus was resolved in a position even more distant from Stilicina. A possible explanation is that the current limit of Stilicina is too narrow and more genera are, in fact, part of this subtribe. Alternatively, it is also likely that we found the sister clade to a more restricted Stilicina. Future research, preferably also including molecular data when DNA grade samples become available, should investigate whether our results are due to sampling bias and/or lack of molecular data in the analysed dataset. This is the first time the medonine genera Deroderus and Ecitocleptis have been included in broader phylogenetic studies, and the non-monophyly of Medonina is confirmed once more (e.g., Żyła et al. 2019, 2021). The sister-group relationship of Deroderus and Eustilicus, suggested by Frania (1986), was not confirmed in our studies. Myrmecophilous Mimophites and Synecitonides have also never been included in a phylogenetic analysis, but it was suggested before that at least Mimophites could belong to Stilicina (Seevers 1965). Both genera were resolved together in a clade, which might be a result of their morphological modifications that could have evolved convergently. A test of this potential relationship would be possible with a molecular-based phylogenetic study.

Biology of Acrostilicus and Pachystilicus
One of the specialised habitats where Paederinae rove beetles may occur are the nests and burrows of various vertebrates. The genus Acrostilicus is an example of such a nidicolous rove beetle, which lives in the burrow of the gopher tortoise in Florida. The gopher tortoise is a burrowing reptile living in the dry and sandy longleaf pine ecosystem of the southeastern USA (Jones & Dorr 2004). Tortoise dung, which accumulates at the blind end of the burrow, provides a resource for many invertebrates, including beetles (Jackson & Milstrey 1989). So far, three species of Staphylinidae were reported from there, i.e., Acrostilicus hospes, Philonthus gopheri Hubbard, 1894, andPhilonthus testudo Smetana, 1995 (the latter two belong to Staphylininae) (Brunke & Buffam 2018). All three nidicolous species are rather pale in colouration, which is consistent with the other invertebrates of this assemblage (Brunke & Buffam 2018). Another morphological modification of Acrostilicus may include elongated legs and an overall slender body. Brunke & Buffam (2018) hypothesised that all three Staphylinidae species could be members of a specialised dung community. It is possible that the distribution of Acrostilicus hospes is much wider and overlaps with the occurrence range of its tortoise host and/or that more species of Acrostilicus are awaiting discovery since only a few samples of rove beetles have been collected from these burrows.
Paederinae is one of a few subfamilies within Staphylinidae where myrmecophily has evolved in several different lineages. Many of them possess significant morphological modifications, e.g., in the body shape and appendages or developed special glands. None of these are obvious in Pachystilicus, which might mean that the beetles are not integrated into the host nest and live in their vicinity. However, many records of beetles from ants' nests were isolated events and most likely not indicative of myrmecophily (Parker 2016). This could be the case for Pachystilicus hanhami, where a direct interaction with ants has never been observed, but the labels indicate that specimens were collected both in the ant nest and outside, although the former situation occured more often. None of the known specimens of P. quadriceps have labels that mention co-collection with ants.

Comparison with previous phylogenetic results
The tree topologies obtained as a result of BI (Fig. 1) and ML (Supp. file 7) analyses were consistent with the previous result of Żyła et al. (2021), the dataset of which served as a base for the current study. The most important differences concerned the relations of the Paederine tribes and the position of the Pseudolathra + Cylindroxystina clade. In Żyła et al. (2021), the tribe Pinophilini was resolved as sister to Lathrobiini + Paederini, but in our analyses only to Lathrobiini. In the second case, the Pseudolathra + Cylindroxystina clade was resolved as sister to 'Medonina and allied taxa' in Żyła et al. (2021), while we here obtained such a result only in our ML analysis. A different topology was recovered in the BI analysis, where this clade was resolved as sister to the 'true' Lathrobiina and they together formed the sister group to the 'Medonina and allied taxa' clade. The above differences are attributed to sparser taxon sampling in the present study for non-Lathrobiini. Within the 'Medonina and allied taxa' clade, topologies differed between BI and ML but Medonina, the largest subtribe within this lineage, was always recovered as non-monophyletic, consistent with other studies (Schomann & Solodovnikov 2017;Żyła et al. 2019Bogri et al. 2020). that greatly improved the manuscript. This project has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 797823 (postdoctoral fellowship of DŻ) and from the Polish National Science Centre, grant number 2019/35/B/NZ8/03431 (to DŻ). DŻ was additionally funded by the Ernst Mayr Travel Grant in Animal Systematics through the Harvard Museum of Comparative Zoology to visit the MCZ collection. We also thank the CIPRES Scientific Gateway, which provided access to computational resources.