A preliminary phylogeny and review of the genus Tasmanitachoides, with descriptions of two new species (Coleoptera, Carabidae, Bembidarenini)

Abstract The genus Tasmanitachoides Erwin, a genus of very small carabid beetle endemic to Australia, is reviewed. Although uncommon in collections, they can be abundant and diverse on banks of fine gravel or coarse sand next to bodies of fresh water; samples from southeastern Australia suggest numerous undescribed species. An initial phylogenetic hypothesis for the genus is presented, including 19 of the 32 known species. The inferred phylogeny, based upon one mitochondrial and four nuclear genes, shows the kingi group to be sister to remaining Tasmanitachoides, with the wattsensis group and T. lutus (Darlington) also being phylogenetically isolated. Two new species are described: T. baehrisp. nov., from the Australian Capital Territory, is a member of the kingi group; T. erwinisp. nov., from Tasmania, is a member of the wattsensis group. Identification tools for described and some undescribed species are presented, including photographs of all known species.


Introduction
The genus Tasmanitachoides Erwin, 1972 (Fig. 1) comprises very small carabid beetles found on fine gravel and coarse sand shores of bodies of fresh water throughout Australia. Tasmanitachoides was proposed to include six species described between 1895 and 1962 by Thomas G. Sloane (1895Sloane ( , 1896Sloane ( , 1921, Thomas Blackburn (1901), and Philip J. Darlington (1962), plus two additional species described by Erwin (1972). Martin Baehr (1990) revised the genus, adding five species, followed by a later series of papers (Baehr 2001(Baehr , 2008a(Baehr , b, 2009(Baehr , 2010(Baehr , 2013 in which he brought the total of known species to 25. This enigmatic group has migrated through the classification of carabids. In describing the first known species, which he called Tachys murrumbidgensis, Sloane (1895) noted "I am in some doubt as to the position of this species… It seems a thoroughly isolated species"; Blackburn (1901), in contrast, considered the two species he described to belong to Bembidion. Darlington (1962) noted that "Although they are certainly Tachys rather than Bembidion by current classification, the species of this group are anomalous (primitive ?) in some ways and should be specially considered by students of bembidiine phylogeny". Erwin (1972) recognized that they were not Tachys, but instead considered them "an early off-shoot of the tachyine lineage which gave rise to the Anillina." However, DNA sequences and morphological data (Grebennikov 2008;Maddison and Ober 2011;Maddison et al. 2019) convincingly indicate that they are not closely related to either Bembidion or tachyines, but instead are members of a Southern Hemisphere clade including the South American genera Bembidarenas, Argentinatachoides, and Andinodontis; this clade of four genera is now known as the isolated tribe Bembidarenini (Maddison et al. 2019).
During the last twenty years, two threads wove together to yield the discovery, documented here, that Tasmanitachoides are likely notably more diverse than reported in the 13 papers describing the 25 known species. NP began collecting Tasmanitachoides in 1998, intrigued by their described diversity given the relatively few specimens that had been studied by Martin Baehr. Baehr (1990) had documented 16 species based upon an examination of only 157 specimens; nine of the species were known from only 1-6 specimens each. These limited numbers suggest either a rarity in nature, a cryptic habitat, a lack of collecting effort, or a combination of these factors. Both this paucity of studied specimens and the difficulty NP encountered in identifying his specimens hinted at a potential for undocumented diversity. Separately, the first DNA sequence data of Tasmanitachoides fitzroyi Darlington acquired by DRM in 2000 showed that it was not a tachyine, and that it held an isolated position within the supertribe Trechitae. A desire to confirm and extend the DNA results led to a collaboration between us, with NP collecting a small series of specimens of Tasmanitachoides from the Murrumbidgee River at Uriarra Crossing in 2002. The diversity of species revealed by the DNA sequence data and associated morphological study suggested that in that small sample there was at least one undescribed species, and possibly a second. We posited that this genus held hidden diversity, perhaps unnoticed because the beetles' very small size made both their capture and their examination difficult. A plan was formed to more seriously re-examine the diversity of Tasmanitachoides using both DNA sequence data and morphological structures, but it was not until 2019, when DRM visited NP in Australia, that the project gained momentum.
After two decades of pondering Tasmanitachoides, our first encounter together with living Tasmanitachoides was memorable. DRM had collected numerous species of the three South American genera of Bembidarenini on trips to Chile, Argentina, and Ecuador between 2006 and 2011, but had only seen preserved Tasmanitachoides. As neontological systematists we often are embedded in rooms full of dead carcasses of the organisms we study, and as beautiful as they may be, the experience of observing pinned specimens is very different than seeing alive, in nature, the biodiversity we seek to document and discover. In the early evening of 8 January 2019, we approached Uriarra Crossing of the Murrumbidgee River, not knowing whether we would find these relatively rarely collected organisms. At our first footsteps on the fine gravel banks of the river specimens of Tasmanitachoides emerged from the substrate, and in short order we had collected three species and more than 110 individuals.
These beetles are not rare in their preferred habitat (Fig. 2). We found them to be common at the ten sites we visited (two on the Murrumbidgee River in the Australian Capital Territory, three in Victoria, and five in Tasmania). Almost all collecting was during daylight hours. At some localities Tasmanitachoides were so abundant that the limiting factor was not finding specimens but capturing them; with dozens of specimens per square meter, many would escape as we were collecting other specimens. In total, we found more than 1100 specimens at those ten sites, representing 15 species, at least four of which are undescribed. It became evident that the group was poorly collected. For example, in Tasmania, we found T. leai (Sloane) to be abundant at three of our five sites, but it had yet to be reported from the island in the literature; although we found more than 200 specimens of T. hobarti (Blackburn) at four of our sites in Tasmania, these apparently were the first specimens collected since the type series at least 118 years earlier.
Tasmanitachoides are found on the shores of larger rivers ( Fig. 2A; see also the 360° view at https://goo.gl/maps/gCpfBCHxueUCr3Kf9), or smaller, more shaded upland creeks (Fig. 2B), or smaller rivers (Fig. 2C), or lake shores (Fig. 2D), with different species appearing to prefer different elevations, levels of shade, water flow regimes, and water body sizes. At Angle Crossing on the Murrumbidgee River in the Australian Capital Territory ( Fig. 2A), we found eight species, but at the smaller and more shaded Flat Rock Creek in eastern Victoria (Fig. 2B) we found a different fauna, with three other species. In Tasmania, Lake St. Clair had only Tasmanitachoides hobarti on its banks, whereas the mouth of Machinery Creek into the River Forth (Fig. 2C) had four other species, but no T. hobarti. There are also differences in species distributions among microhabitats at a single site. For example, at the Machinery Creek / River Forth site (Fig. 2C), T. kingi Darlington was found in drier areas higher up on the bank, approximately 1.5-2 m from the water, whereas T. leai was found primarily lower and closer to the water. In these water-shore habitats, Tasmanitachoides are concentrated in those regions with no or minimal vegetation, within 3 meters of the shoreline, with at least some moisture a centimeter or two below the surface. Most critically, though, they are found where the substrate is composed of moderately well-sorted fine gravel and coarse sand, with particles mostly approximately 1-4 mm (Fig. 3), with at most small amounts of finer sand, silt, or clay mixed in. This substrate is extremely porous, such that water splashed upon it quickly drains through. The beetles emerge when water is poured on the surface; they then run up the bank. Microhabitats with clean, fine gravel can be widespread at a site, or quite localized. For example, along the South Esk River at Avoca we found Tasmanitachoides in only one small patch a few meters long (Fig. 4A, B). This patch was composed of clean, fine gravel, and had many Tasmanitachoides (we collected more than 60). In other rivers and creeks, they were extensively distributed along much of the gravel shoreline (e.g., Figs 2, 4C, D).  ing, 35.5803°S, 149.1109°E, 600 m. This is the substrate from an area in which T. murrumbidgensis and T. sp. "Tambo R" were abundant, with some specimens of T. maior and T. rufescens. Scale bar: 10 mm. This is the first in a series of planned papers about diversity within Tasmanitachoides. We infer an initial phylogeny of the genus based upon DNA sequence data, document some aspects of the diversity we found, describe two new species, and provide an improved identification key as well as images of the species. We plan a more complete revision of the genus after more focused collecting throughout Australia, and a more detailed phylogenetic analysis including more species.

Materials and methods
Members of Tasmanitachoides were examined from or deposited in the collections listed below. Each collection's listing begins with the code used in the text.

ANIC
Australian

Collecting methods
Specimens were collected with the aid of an aspirator after splashing water on fine gravel or coarse sand, and waiting for the beetles to appear on the surface. Specimens for morphological studies were killed and preserved in Acer sawdust to which ethyl acetate was added. Specimens for DNA sequencing were collected into 95% or 100% ethanol.

Morphological methods
General methods of specimen preparation for morphological work, and terms used, follow Maddison (1993;. Genitalia were prepared, after dissection from the body, by treatment in 10% KOH at 65 °C for 10 minutes followed by multi-hour baths of distilled water, 5% glacial acetic acid, distilled water, and then ethanol. Male genitalia were then mounted in Euparal on a small circular coverslip attached to archival-quality heavyweight watercolor paper, and, once dried, pinned beneath the specimen. Photographs of entire beetles, elytra, and heads were taken with a Leica M165C dissecting scope and a Sony NEX-7 camera, and of male genitalia with a Leica DM5500B compound microscope and DMC425C camera. Microsculpture photographs were taken with a DMC425C camera attached to a DM5500B compound scope equipped with an X-Cite 110LED light source, which provides co-axial illumination, and a 20× epi-illumination objective lens. For all photographs of specimens or body parts, a stack of images from different focal positions was merged using the PMax procedure in Zerene Systems' Zerene Stacker; the final images thus potentially have some artefacts caused by the merging algorithm. Measurements were made using Leica Application Suite v4.9 from images acquired using these either a Leica Z6 Apo lens and DMC4500 camera or a Leica DM5500B compound microscope and DMC425C camera. We follow Baehr (1990) in measuring the body length of specimens from the anterior edge of the labrum to the tip of the longest elytron.

Molecular methods
Genes studied, and abbreviations used in this paper, are:
DNA was extracted using a Qiagen DNeasy Blood and Tissue Kit. Gene fragments were amplified using the Polymerase Chain Reaction on an Eppendorf Mastercycler ProS Thermal Cycler, using TaKaRa Ex Taq and the basic protocols recommended by the manufacturers. Primers and details of the cycling reactions used are given in Maddison et al. (2019), with the addition that some of the 18S sequences were amplified with primers 18S5 (GACAACCTGGTTGATCCTGCCAGT) and 18Sb5 (TAAC-CGCAACAACTTTAAT) (Shull et al. 2001) using two rounds of cycling, with the first round beginning with an annealing temperature of 51 °C, which was sequentially reduced by 0.5 °C for each of 10 cycles, and the second round of 27 cycles using an annealing temperature of 46 °C; all cycles used an extension time of 60 seconds. The ampli-fied products were then cleaned, quantified, and sequenced at the University of Arizona's Genomic and Technology Core Facility using a 3730 XL Applied Biosystems automatic sequencer. Assembly of multiple chromatograms for each gene fragment and initial base calls were made with Phred (Green and Ewing 2002) and Phrap (Green 1999) as orchestrated by Mesquite's Chromaseq package (Maddison and Maddison 2020a, c) with subsequent modifications by Chromaseq and manual inspection. Multiple peaks at a single position in multiple reads were coded using IUPAC ambiguity codes.
We sampled DNA from 54 specimens of 19 species of Tasmanitachoides, as well as specimens of seven outgroup species, which belonged to other genera of Bembidarenini (Table 1). Of the 214 sequences examined, 153 were newly acquired, with 61 being from previous publications (Maddison and Ober 2011;Maddison 2012;Kanda et al. 2015;Maddison, et al. 2019). Of the 19 species of Tasmanitachoides sampled, we consider seven to belong to undescribed species (T. erwini sp. nov., T. baehri sp. nov., T. sp. "Lerderderg R", T. sp. "Angel Crossing #1", T. sp. "Angel Crossing #2", T. sp. "River Forth", and T. sp. "Tambo R"), the first two of which are described in this paper. Locality information for the Tasmanitachodes whose DNA was sequenced is provided in Table 2. For Tasmanitachoides fitzroyi, the terminal taxon used in the analyses is a chimera of two different specimens from the same locality, with 28S from specimen DNA0762 and the remaining genes from DNA1575. Sequences of the two holotypes listed in Table 1 are "genseq-1", of paratypes "genseq-2", and the remainder are all "genseq-4" (Chakrabarty et al. 2013).
Alignment was not difficult for any of the protein-coding genes. There were no insertion or deletions (indels) evident in the sampled CAD4, CAD2, wg, or COI sequences. Alignments of 28S and 18S were conducted in MAFFT version 7.130b (Katoh and Standley 2013), using the L-INS-i search option and otherwise default parameter values.
Sites in 28S were chosen to be excluded from consideration using the modified GBLOCKS analysis (Talavera and Castresana 2007) present in Mesquite with the following options: minimum fraction of identical residues for a conserved position = 0.2, minimum fraction of identical residues for a highly-conserved position = 0.4, counting fraction within only those taxa that have non-gaps at that position, maximum number of contiguous non-conserved positions = 4, minimum length of a block = 4, and allowed fraction of gaps within a position = 0.5. No sites were excluded for 18S.
Maximum likelihood (ML) analysis was conducted for each gene individually using IQ-TREE version 1.6.12 (Nguyen et al. 2015), as orchestrated by Mesquite's Zephyr package (Maddison and Maddison 2020b, c). The ModelFinder feature (Kalyaanamoorthy et al. 2017) within IQ-TREE was used to find the optimal character evolution models. The MFP model option was used for 28S and 18S, and the TESTMERGE option for the protein-coding genes. The TESTMERGE option sought the optimal partition of sites, beginning with codon positions in different parts. Fifty searches were conducted for the ML tree for each matrix.
In addition, analyses of a matrix formed by concatenation of all six gene fragments were conducted, with the TESTMERGE option also being used, beginning with each  codon position for each gene as a separate part (thus, the analysis began allowing for up to 11 parts: three for each of the three protein-coding genes, as well as one for 28S and one for 18S). Fifty searches were conducted for the ML tree for each matrix; for bootstrap analyses, 500 replicates were performed. In addition, an equivalent ML search was conducted for a matrix formed by the concatenation of all gene fragments except COI.

Data resources
Sequences have been deposited in GenBank with accession numbers MW291161 through MW291313. Aligned data for each gene as well as files containing inferred trees for each gene are available in Suppl. material 1, and have been deposited in the Dryad Digital Repository, https://doi.org/10.5061/dryad.8931zcrpv.   Table 1) not depicted.

Phylogeny
The ML tree for all six gene fragments combined is shown in Fig. 5. Many of the clades are well supported, as measured by bootstrap values (Fig. 6). The ML trees for individual genes are shown in Figs 7-9. Based upon these analyses, the kingi group of Tasmanitachoides appears to be a clade that is sister to the remaining species. This result is supported by bootstrap values of 100%, and by ML trees for all gene fragments except for CAD4 (Figs 7-9). Within the remaining species, the two members sampled of the wattsensis group, T. erwini and T. sp. "Lerderderg R", are sisters, as supported strongly by the combined analysis and by individual gene trees for 28S, COI, wg, and CAD4. This pair appears to  be sister of all Tasmanitachoides except the kingi group; this is strongly supported by the concatenated analysis, but in individual genes only by 28S, wg, and CAD2. The morphologically distinctive T. lutus is isolated, with no near relatives. All remaining Tasmanitachoides (all but the kingi and wattsensis groups, as well as T. lutus) form a strongly supported clade; that clade is present in every gene tree except that of 18S (Figs 7-9). Within these remaining Tasmanitachoides there are some consistent results T wilsoni ACT  from gene to gene, in particular the close relationship between T. fitzroyi and T. maior, between T. cf. gerdi and T. sp. "Angle Crossing #1" (the two species of the katherinei group that were sampled), and between T. rufescens and T. sp. "River Forth". With one exception, for all species for which multiple specimens were sampled, the sequences of a species form a clade in the gene trees separate from specimens of other species. This is evident in the tree for 28S (Fig. 7), and for most of the COI tree (Fig.  8). The one exception is T. murrumbidgensis, which shows two distinctive clades in COI that are not each other's sisters; in fact, one of those COI clades is in a clade with T. sp. "Tambo R" and T. bicolor (Fig. 8). Representatives of these two clades of T. murrumbidgensis were found together at both Angle Crossing (Murrumbidgee River) and along the Tambo River at Bruthen. The sequences in these two clades consistently differ at 13 of the 658 sites, for a divergence of approximately 2%. This is a very large difference in mitochondrial haplotypes within a species relative to divergences within other carabid species (e.g., Maddison 2008;Maddison and Cooper 2014;Maddison and Anderson 2016;Maddison and Sproul 2020). It is possible that there might be two species within what we call T. murrumbidgensis, but we can detect no morphological differences and there are no differences in other genes. Another possibility is that for one of these clades we have sequenced a nuclear copy (a numt, Thalmann et al. 2004), with the other clade representing the true mitochondrial gene, or they could represent the effects of Wolbachia infections (Smith et al. 2012), but we have no independent evidence supporting this. Whatever the nature of the sequences, the diversity within COI causes T. murrumbidgensis to appear as two separate clades within the multi-gene analyses (Fig. 5, main tree); these two separate clades are not evident if COI is excluded (Fig. 5, inset).

Morphological variation
The known species of Tasmanitachoides vary in shape, form, and color (Figs 10-14). The elytral striation shows notable species-specific variation (Figs 15, 16), as do the structure of the clypeus and extent and structure of the frontal furrows of the head (Fig. 17), microsculpture (Fig. 18), and male genitalia (Figs 19, 20). We provide more details about this variation below in the Taxonomic treatment.

Taxonomic treatment
Diagnoses and descriptions of the genus are provided by Erwin (1972) and Baehr (2013). We are aware of 32 species of Tasmanitachoides: 25 previously described, two described in this paper, and five additional species whose descriptions await future research. Based upon the phylogenetic results and morphological variation, most species can be tentatively arrayed into six species groups, as follows: The placement of species into groups may change once more species are better known, including those we have not sampled for DNA.

Identification of species of Tasmanitachoides
Species of Tasmanitachoides are currently very difficult to identify using morphological characteristics, in part because they are small, and as the known external differences between many species are subtle. In addition, although the internal sac of the male aedeagus has a complex pattern of sclerites, and thus could be a very valuable source of characters for identification, genitalic variation is not well documented or understood. One difficulty with comparing male genitalia is that the internal sac sclerites are oriented in a plane that is nearly edge-on in the standard left lateral view. This causes them to appear very differently as a function of slight differences in the orientation of the genitalia (compare, for example, Fig. 20B to Fig. 20C), causing interpretation of sclerites in the standard left lateral view troublesome. In contrast, a ventral view (e.g., Fig. 19) shows patterns much less sensitive to slight differences in angle. As previously published images of male genitalia (e.g., Erwin 1972;Baehr 1990) are of the standard left lateral view, the genitalia of all species will need to be re-examined to provide a more robust understanding of variation.
The key we present below is only an incremental improvement on Baehr's (2010;2013). We began with his key and modified it to include the two new species we describe, as well as some (but not all) of the undescribed species of which we are aware. We have not included T. sp. "Lerderderg R", T. sp. "Angle Crossing #1", and T. sp. "River Forth". We have included T. balli, although with some doubt, as the only known specimen is now missing its head.
Based upon our examination of specimens of all known species, we have removed some of the inconsistencies in the key, simplified it, and changed its structure somewhat. However, we view this as a provisional key. Although we have previously seen specimens of all known species, for some of them (e.g., T. glabellus, T. comes, T. gerdi) we modified the key without those specimens at hand, and depended upon our notes and photographs of the primary types, as well as Martin Baehr's papers. In addition, the variation within many species is not yet known, as there is limited material available (ten of the described species are known from fewer than five specimens). For example, Baehr's key uses size to separate T. maior from other species, noting that the only specimen he knew was 2.9 mm in length; however, based upon our somewhat larger sample (we have measured seven specimens) the holotype is at the upper end of the size range, with some specimens as small as 2.44 mm in length, overlapping in length with related species. We suspect that the sizes given in the key for many species will need to be modified once more material is examined. The same will likely be true for color, as we have seen more variation in color in our large samples of some species than Martin Baehr had seen in his smaller samples. In addition, the geographic distributions mentioned in the key should be viewed with suspicion, as the ranges of species are very poorly known.
In Baehr's keys, the couplet which divides Tasmanitachoides into the largest two groups is that which focuses on whether or not the clypeus is "distinctly impressed anteriorly". We find this character difficult to ascertain, with many specimens appearing ambiguous, in part because of the more or less continuous variation in this trait across Tasmanitachoides species. For this reason we have replaced this couplet with one that focuses instead on a related trait, the presence or absence of tubercles on the anterior lateral corners of the clypeus, with associated modifications to other regions of the clypeus; this latter character is easier to judge.
Provisional key to all described and a few undescribed species of Tasmanitachoides   1 Only the sutural stria distinct, others completely effaced or almost so (Fig. 10F)  Body larger and wider (Fig. 14A, B) Pronotum narrow, much narrower than the elytra at the shoulders (Fig. 10E), and approximately the same width as head; elytral intervals 2-5 convex, with striae evident as broad, shallow, impunctate grooves between them (Figs 15D,  18A). On head a groove extends from the anterior supraorbital puncture anteriad and mediad to approximately halfway toward the frontal furrow ( Pronotum closer to the width of the elytra at the shoulders ( Fig. 10A-D), and more evidently wider than the head; intervals 3 and 4 not convex, striae 2 either absent, or, if present, composed of a narrow striation rather than a broad groove. If there is a groove extending from the anterior supraorbital puncture, it is very short (Fig. 17A). Hind angle of pronotum obtuse to acute ..........12 12 Body short and convex (Fig. 10D); elytra considerably less than 1.5 × longer than wide; pronotum wide, base (at hind angles) as wide as apex, hind angle greater than 90°, laterally not projected. Body orange or orange-brown. Frontal furrows short (Fig. 17D, E) Frontal furrows attaining but the anterior third of the eyes, ended abruptly (Fig. 17D); pronotum wider, ratio width/length 1.33, barely sinuate in front of hind angles, with wider base compared with apex, ratio width of apex/ width of base 1.10 (Fig. 11D) Color paler, head and pronotum pale reddish, elytra pale yellow (Fig. 11B); elytra longer, ratio length/width > 1.81, striae 2 and 3 as deeply impressed as striae 1 and 4. Northwestern QLD & northeastern NT ...T. elongatulus Baehr 19 Pronotum constricted posteriad such that the hind margin is notably narrower than width at widest point, with lateral margin distinctly sinuate (Fig. 13D); hind angles rectangular or acute. Elytra relatively flat. Body color uniformly reddish or dark reddish. Body size large, 2.6-2. First elytral stria straighter, less abruptly sinuate ( Fig. 16E-G, J), with or without distinct punctures in the anterior half. Fifth elytral interval distinctly impressed in the anterior fifth to third of the elytra, abruptly less distinct behind that point ( Fig. 16E-G, J). Body relatively flat (Fig. 13A-C), except for T. bicolor which is slightly convex (Fig. 13E). With distinct microsculpture on the elytra, and thus the surface is duller. Body color either almost uniformly reddish or dark reddish, or piceous with disk of each elytron contrastingly lighter, or body uniformly piceous (if body uniformly piceous, then length < 2.0 mm). Body size generally smaller, 1. First elytral stria abruptly sinuate, very close to the suture in the anterior fifth or fourth, at which point it abruptly bends away from the suture (Fig. 16A-D); with distinct punctures in the anterior half. First interval is at its widest at approximately the one-third point, as wide or wider than the second interval, and from that point back it gradually narrows. Fifth elytral interval not abruptly shallower within the first third of the elytra, well-impressed for at least the first half ( Fig. 16A-C), except for T. leai (which is convex, dark, and shiny, Fig. 12D). Body convex (except for T. erwini, which is very dark, Fig. 12C). Microsculpture on elytra more effaced, and thus the surface is shinier. Body color uniformly dark piceous to black, or piceous with elytra slightly (not contrastingly) lighter (Fig. 12). Body size in general larger, 2.1-2.6 mm ........... 24 21 Eyes large, more protruded, temples almost wanting ( Fig. 17G Other material examined. We have seen an addition specimen labeled "Paddy's River, 1 mi. S. of Cotter Dam, ACT, 17.iv.1969. S. Misko" (ANIC;currently in ZSM).
Derivation of specific epithet. We are honored to name this species after the late Martin Baehr, who discovered and documented many of the carabid species of Australia, and who described 14 of the known species of Tasmanitachoides.
Diagnosis and description. Very small, length 1.59-1.63 mm (n = 4). A pale species, body mostly orange, with the front half of the elytra and head a darker reddish orange. Antennae pale testaceous, with antennomeres 5-11 slightly infuscated. Head  with moderately long but shallow frontal furrows, reaching approximately the center of the eye, and at least to the anterior supraorbital seta (Fig. 17B); with a groove extending from anterior supraorbital puncture anteriad and mediad to approximately halfway toward the frontal furrow (Fig. 17B). Pronotum convex, narrow, only slightly wider than head (Fig. 10E). Hind angle of pronotum obtuse. Elytra more parallel-sided than T. wilsoni. Striae 2 and 3 shallow, broad, impunctate grooves (Figs 15D, 18A); nearby intervals convex. Stria 5 deeply engraved in anterior half of elytron; stria 5 reaching or nearly reaching the second discal seta (ed5; Fig. 15D). Striae 6 and 7 effaced. Discal setae ed6 apparently in stria 2. Microsculpture without engraved lines; where present on the dorsal surface, the microsculpture is formed as low papillae without defined boundaries (Fig. 18A). Pronotum and head very shiny, virtually without microsculpture. Aedeagus (Figs 19A,20A) with internal sac sclerites compact, and sinuate, very similar to those of T. wilsoni (Fig. 19B).
Comparison with related species. Likely to be confused only with similarly small and compact T. wilsoni, from which it can be distinguished by the narrower pronotum with less rounded lateral margins, and narrower, less rounded elytra. In addition, T. wilsoni has much shorter frontal furrows, which do not reach the anterior supraorbital seta (Fig. 17A); T. wilsoni also lacks the notable groove extending forward from the anterior supraorbital seta. The elytral striae in T. wilsoni are less evident than in T. baehri: T. baehri has an evident (if shallow and broad) stria 3 between the two anterior discal setae (Figs 15D, 18A), whereas in T. wilsoni it is either absent or extremely faint and shallow (Fig.  15C); stria 5 in T. wilsoni is much shorter, only reaching to around half-way in between the two anterior discal setae (Fig. 15C), as opposed to reaching or nearly reaching the second discal seta (ed5) as it does in T. baehri (Fig. 15D) T. baehri and T. wilsoni look very much like small members of the tribe Tachyini (e.g., Elaphropus, Tachyura). The two Tasmanitachoides can be distinguished by the presence of four setae on the clypeus, as opposed to the two setae present in tachyines.
Geographic distribution. Only known from the Australian Capital Territory (Fig. 21), but very likely occurring in similar habitats in NSW. Habitat. Collected from pockets of gravelly cobble at the edge of still water of the Murrumbidgee River. The collection locality was amongst riverbank sheoaks (Allocasuarina) and relatively protected. Specimens were recovered by splashing the gravel bank after removal of cobbles. The species was collected with T. murrumbidgensis, T. rufescens, and a single specimen of T. leai.
Phylogenetic relationships. This species belongs to the kingi species group and appears to be sister to T. wilsoni among the sampled species (Figs 5-9).
Notes. This species was called "Tasmanitachoides cf. rufescens" in Maddison et al. (2019).   . In addition to these, we have seen three additional specimens, all in the MCZ, which we have designated as paratypes. Two are a labeled "L. StClaire-Queenstown Jan. '57 Tas Darlingtons" "Tachys hobarti (Sl.) det Darl. '61"; according to Darlington (1962:117) these two specimens are from the crossing of the King River by the Queenstown road, which at the time (before the Crotty Dam) would have been approximately 42.074°S 145.652°E. The third is labeled "Mersey R Vy. Mar. '57 Tas Darlingtons" "Tachys hobarti (Sl.) det Darl. '61". According to the map in Darlington (1960), this locality is at approximately 41.532°S, 146.426°E. These specimens formed Darlington's concept of Tasmanitachoides hobarti. They also are specimens studied and figured by Erwin (1972)  Derivation of specific epithet. We are honored to name this species after the late Terry Lee Erwin, for his many contributions to carabidology and systematics in general, and to our knowledge of Tasmanitachoides and other bembidarenines in particular.
Diagnosis and description. Length 2.25-2.75 mm (n = 7); most specimens less than 2.6 mm. One of the darker species of Tasmanitachoides ( Fig. 1): body piceous to black; appendages piceous, including basal antennomeres, with the exception of the tarsi, which are slightly paler. Body relatively flat and parallel-sided; elytra narrowing posteriorly, and thus more pointed than other species. Head without tubercles at anterior corners of clypeus, and without concave region in anterior half. Frontal furrows (Fig. 17G) more or less straight, reaching backward to approximately the center of the eye, parallel or slightly diverging posteriorly; bottom of furrows rugose. Pronotum relatively narrow (Fig. 12C), slightly sinuate laterally in front of the right or slightly acute hind angle. First stria abruptly sinuate, very close to the suture in the anterior fifth or fourth, at which point it abruptly bends away from the suture. Striae 3 and 4 very weak, almost absent in some specimens; the striae 3 and 4 are joined at the anterior discal seta (ed3; Fig. 16B), and in most specimens are merged in front of that point. Stria 5 distinctly engraved throughout the entire anterior half; in posterior half it gradually weakens toward the rear. Stria 6 consisting of a few isolated punctures; stria 7 absent. Discal setae ed6 in stria 3. Microsculpture weak, sculpticells weakly engraved, and thus the surface is shiny; sculpticells isodiametric on head and pronotum, slightly longitudinally stretched on elytra (Fig. 18B). Aedeagus (Figs 19C,20B,C) with internal sac sclerites elongated and relatively straight, very similar to those of T. sp. "Lerderderg R" (Fig.19D). Ventral surface of the aedeagus quite straight (Fig. 20B, C).
Comparison with related species. As with other members of the wattsensis group, this species has a relatively unmodified clypeus, without anterior lateral tubercles, and with the third and fourth elytral striae nearly effaced. Its darker color (including the entirely piceous antenna) and flatness distinguish it from other members of the group. It is the only known species of the group from Tasmania. From the other two large and dark Tasmanitachoides from Tasmania, T. hobarti and T. leai, T. erwini is distinguished by having a darker antennomere 1 and flatter body. From T. hobarti it is further distinguished by the much weaker striae 3 and 4; from T. leai by the longer stria 5.
Habitat. At the type locality, members of this species were found during daylight hours in fine gravel on the banks of Mineral Creek at its mouth into the River Forth (Fig. 2C); specimens were found after splashing the gravel with water. The banks had no visible vegetation. Present in the same habitat were Tasmanitachoides leai, T. kingi, and T. sp. "River Forth".
Phylogenetic relationships. This species belongs to the wattsensis species group, and appears to be the sister to T. sp. "Lerderderg R" among the sampled species (Figs 5-9). Notes. This is the species illustrated by Erwin (1972) as T. hobarti. This is evident both by the localities of the specimens he examined (as the localities match Darlington's), and because of the figures themselves, including the features of the genitalia, which match those of this species rather than T. hobarti. The male genitalia of "Tasmanitachoides hobarti" figured in Baehr (1990: Fig. 12) is of this species as well. Some specimens from the type series (collected 14 January 2019) are teneral.

Tasmanitachoides angulicollis Baehr and T. hendrichi Baehr
We have examined the holotype of T. hendrichi and a paratype of T. angulicollis, and found them to be extremely similar; it is possible that they are synonyms.

Tasmanitachoides comes Baehr and T. gerdi Baehr
There is only one known specimen of T. comes and only one of T. gerdi (Baehr 2010). In Martin Baehr's collection in ZSM, one of those specimens (Fig. 11C) exactly matches the description of T. comes, including in the pattern of punctures of the head (compare Fig. 11C to Baehr 2010: fig. 4). However, the locality label on the pin with that specimen matches that listed in Baehr (2010) as the type locality of T. gerdi (Mt. Elliot, QLD), and the pin bears a label declaring it to be the holotype of T. gerdi. In contrast, the specimen matching the description and figures of T. gerdi (Fig. 11D) bears the locality label (Little Panton R., WA) and holotype label of T. comes. The simplest explanation is that the labels were accidentally switched at some point. In resolving whatever accidents of history yielded the contradiction between description and labels, the published description, including figures, take precedence, and as there is no doubt to which specimens Baehr was referring in his 2010 description, the specimen in our Fig. 11C should be considered the holotype of T. comes, and the specimen in our Fig. 11D should be considered the holotype of T. gerdi. The specimens we are calling T. cf. gerdi may be T. gerdi, but we await more detailed study of the holotype, and better understanding of the distribution of Tasmanitachoides species, before we can be more definitive.

Tasmanitachoides hobarti (Blackburn) and T. glabellus Baehr
We have examined a photograph of the type (or syntype -see Baehr 1990) of Bembidium hobarti Blackburn in the NHMUK (courtesy of Beulah Garner), and from that photograph it is clear that the type is conspecific with the large, convex species we are treating here as T. hobarti. However, specimens collected by Philip Darlington in Tasmania, and illustrated by Erwin (1972), are T. erwini, not T. hobarti, and Baehr's (1990) concept of T. hobarti included T. erwini, as noted above. It is not clear if Baehr's concept of T. hobarti included true T. hobarti. Before our fieldwork in 2019, the only specimens of true T. hobarti in museums of which we are aware are members of the type series (NHMUK), and the only specimens of T. erwini those in the MCZ. Based upon a search by DRM in 2019, Baehr's collection (ZSM) includes neither T. hobarti nor T. erwini, although there is a specimen of T. kingi from the Meander River, Tasmania, identified by Baehr in 2011 as T. hobarti. Baehr thus apparently had a mixed concept of "Tasmanitachoides hobarti" which may or may not have included true T. hobarti.
Baehr's mixed concept of T. hobarti may be relevant to understand the history of T. glabellus. We have examined high-quality photographs of the paratype of T. glabellus (in ZSM, courtesy of Michael Balke), and it looks extremely similar, if not identical, to true T. hobarti. We could see no evident differences. As it seems very unlikely that a species would be known from only Tasmania and one mountain top in North Queensland, even given how poorly Tasmanitachoides is collected, it seems more likely that these are distinct species or that the label data for the two T. glabellus specimens is in error. We leave it to further field work and closer examination of the types to resolve the status of T. glabellus.

Tasmanitachoides leai (Sloane) and T. hackeri Baehr
In his description of T. hackeri, Baehr (2008) notes that this species has "Stria 5 near base deeply sulcate, abruptly ended behind basal third", and indeed, the paratype from the type locality that we have examined (ZSM) has this trait. This is the character by which he separates this species from, for example, T. leai in the dichotomous key he presents. However, in this regard T. hackeri exactly matches all specimens of T. leai we have examined, including the lectotype (ANIC), as T. leai also has a short, abruptly ending stria 5, against Baehr's (2008) and later keys. We can find no significant differences between the paratype of T. hackeri and T. leai, and it is likely that they are synonyms. However, we do not formally synonymize them now, awaiting study of additional specimens from NSW and QLD.

Tasmanitachoides maior Baehr
The only specimen Baehr (1990) had seen of T. maior was a female 2.9 mm in length. We have measured seven additional specimens, and found that Baehr's female is at the upper end of the range; the specimens we measured range from 2.44 to 2.89 mm in length.

Tasmanitachoides wattsensis (Blackburn) and relatives
The specimens we have in hand of the T. wattsensis group from Victoria and New South Wales show a great deal of variation, hinting at a complex of closely related species. The specimens from the Lerderderg River (T. sp. "Lerderderg R") are distinctly broader than the remainder, and appear to be a separate species. This is the species that was called "Tasmanitachoides cf. leai" in Maddison et al. (2019). East and north of Melbourne are other forms, including true T. wattsensis (of which the specimen shown in Fig. 12A may be a member). Additional research will be necessary to understand the diversity in this complex.

Concluding remarks
George Eugene Ball died on 12 January 2019, as the two authors of this current paper were travelling on the ferry from Melbourne, Victoria to Devonport, Tasmania, in the midst of the field work that produced the bulk of the specimens on which this paper was based. In less than two years since that day, the world has lost most of the remaining senior figures in carabid systematics, and in the process a tremendous amount of knowledge about carabid beetles that had never been written down. George's death was followed by that of Martin Baehr, who knew the Australian carabid fauna better than anyone. We lost Augusto Vigna Taglianti and Ross Taylor Bell later in 2019. In the early spring of 2020 we lost Terry Lee Erwin, and in early October, Shun-Ichi Uéno.
To have lost six of our grand masters in less than two years is stunning. Our naming a species after Terry and one after Martin are but small gestures to help us honor and remember these two carabidologists, and all the others, like George, Augusto, Ross, and Shun-Ichi, who have devoted their lives to uncover the hidden diversity in the small organisms with which we share our planet.
journal ZooKeys, and who contributed to carabid systematics in particular, including naming the genus Tasmanitachoides.
Specimens were collected under permit FA19258 issued by the Tasmanian Department of Primary Industries, Parks, Water and Environment and permit 10008992 issued by the Victorian Department of Environment, Land, Water, and Planning. As with most biodiversity projects, this work could not have been completed without the people who are stewards of insect collections at museums around the world. We thank Cate Lemann (ANIC) for all of her help during our visits, and for the photographs of types she took at our request. We also thank Adam Ślipiński (ANIC) for his hospitality. We are very grateful to Michael Balke (ZSM) for his generosity during DRM's visit to ZSM in 2017 and 2019. He and Ditta Amran Balke went above and beyond the call of duty to photograph specimens of Tasmanitachoides for us. We are also very thankful for Martin Baehr opening his house and collection to DRM in 2017. We thank Beulah Garner (NHMUK) for photographing the type specimens of Bembidium wattsense Blackburn and Bembidium hobarti Blackburn for us, and Crystal Maier (MCZ) for loaning to us Philip Darlington's Tasmanitachoides specimens. For help with obtaining DNA sequence data by conducting some of the PCRs, we thank Lili S. Adams, Caitlin E. Hudecek, Tiana S.L. Week, and Rhea K. Sellitto. We thank Arnaud Faille for providing protocol details for amplifying the 18S5-18Sb5 piece of the 18S gene. For their helpful reviews of the manuscript, we thank David H. Kavanaugh, Vasily Grebennikov, and James K. Liebherr. This project was supported by the Harold E. and Leona M. Rice Endowment Fund at Oregon State University.