Current Biology, Volume 24 Supplemental Information Specialized Myrmecophily at the Ecological Dawn of Modern Ants Joseph Parker and David A. Grimaldi Systematic Palaeontology Diagnosis. Clavigerite pselaphines distinguished from all other Clavigeritae by Family Staphylinidae Latreille, 1802 possession of distinct, unfused tergites IV–VI; Subfamily Pselaphinae Latreille, 1802 further distinguished by possession of 8 antennomeres, maxillary palpi emerging well Supertribe Clavigeritae Leach, 1815 outside buccal cavity, presence of paired Revised diagnosis (modified from hook-like trichomes on paratergites IV and V Chandler [S1]). Head with 3-8 antennomeres, with smaller trichome on VI, and partially terminal antennomere with setose cavity in overlapping sternites indicating abdominal truncate apex (cavity absent in Colilodion, flexibility. Articerodes, Kurbatoviella, Disarthricerus and Protoclaviger gen. nov. possibly others); lacking ocular mandibular carinae; mouthparts small, barely visible; Type species: Protoclaviger trichodens sp. maxillary palpi small, often with only one nov., here designated. segment. Pronotum lacking paranotal carinae. Diagnosis. Protoclaviger, with its single Abdomen with visible tergites IV–VI fused into species P. trichodens, is presently the only a tergal plate (tergites unfused in known genus of Protoclavigerini. At this time, Protoclaviger). Paratergites IV (paratergites diagnoses of the new genus and species thus IV–VI in Protoclaviger) bearing tufts of match the tribal diagnosis above. specialized setae (trichomes). Legs with Description. Body length ~1.6 mm (Fig 1A, trochanters of middle and hind legs elongate, B), body form somewhat flattened dorsal junction of mesofemur and dorsoventrally. mesotrochanter distant from mesocoxa; third tarsomeres longer than length of basal two Head: Length: ~0.3 mm. Width across eyes: tarsomeres combined, first and second ~0.2 mm. Form triangular in lateral view, tarsomeres subequal in length (second expanding apically to flat, steep frontal margin tarsomeres elongate in Colilodion), with single (Fig 1C). Vertex convex, raised and narrowed tarsal claws. from eyes to clypeus to form short, prominent frontal rostrum. Vertexal fovea absent or not Comments. Creation of the new tribe apparent (possibly obscured by trapped film Protoclavigerini brings the number of of air). Lateral margins of head in dorsal view Clavigeritae tribes to four and necessitates a tapering weakly behind eyes to occipital revised supertribal diagnosis. Morphological constriction. Occipital carina absent. Neck characters of the basal abdomen (the basal largely hidden dorsally by pronotum. Gular sulcus and basal fovea) [S1] are omitted since area relatively flat, gular sulcus or fovea their form varies substantially across extant absent or not apparent. Neck ventrally Clavigeritae, and these characters are not exposed and broadening from occipital visible in Protoclaviger. Symmetric male constriction to apical margin of prosternum. genitalia [S1] is removed from the supertribal Eyes prominent (Fig 1C), positioned along diagnosis because Pseudacerus [S2] and lateral margins of head, crescent-shaped with some undescribed New Caledonian species ocular canthi formed from genal projection; (P. Hlavac, personal communication) have ventral half of eye crescent extending into since been discovered to possess asymmetric gular area. Region anterior to eyes excavate median lobes. to form frontal rostrum. Antenna received under roof of frontal rostrum (Fig 1D, 3A). Tribe Protoclavigerini trib. nov. Antenna 8-segmented, antennomere 1 relatively large and exposed, easily visible Type genus: Protoclaviger Parker & Grimaldi (Fig 1D, 3A), antennal club formed by here designated. enlarged antennomere 8 (Fig 1D, 3A). Apex sternite. Sternite III short, 0.4x length of IV but of antennomere 8 abruptly truncate (Fig 1D) extending well past metacoxae and spanning with apical face forming a recessed cavity (Fig entire width of abdomen. Sternite IV longest, 1E). Mandibular apex relatively blunt and 1.2x V; V and VI subequal in length. Sternite weakly directed adorally, protruding slightly VII shorter, 0.6x V; VIII smallest, 0.5x VII. outside of buccal cavity (Fig 3A). Maxillary Legs: Procoxae contiguous, mesocoxae palpus projecting from side of buccal cavity; moderately separated, metacoxae separated palpus evident as a single small, curved by one-third metaventral width. Trochanters of spatulate segment with truncate apex (Fig 3A). all legs elongate (“macrosceline-type”), with Setiferous brushes of maxillary galea/lacinia femur distant from coxa. Femora and tibiae protruding from buccal cavity (Fig 1C). flattened, and broadened apically. Femora Thorax: Pronotum length ~0.3 mm, width with apical excavation for retracted tibiae. ~0.3 mm at widest point. 1.3x wider than head. Tarsi with tarsomeres 1 and 2 very short (I Lateral margins smoothly rounded, difficult to see within surrounding tibia apex); broadening to just over half pronotum length tarsomere 3 much longer (Fig 1H). Single before narrowing slightly to base (Fig 1B). tarsal claws. Basal margin of pronotum angled convexly in Elytra: Elytral length ~0.5 mm; width at outline, received between sloping elytral widest point ~0.3 mm, strongly convex in bases. Depressions at base evident, perhaps lateral view. Elytral bases sloping shallowly corresponding to median antebasal fovea and from humeri to midline. Lateral margins lateral antebasal foveae. Mesoventral- smooth and rounded, broadening from base metaventral area strongly convex, with to three-quarters elytral length before metaventrite sloping to apical margin in lateral narrowing slightly to apex; apical elytral view (Fig 1A). Metaventral apex produced margins sinuate (Fig 1G). Elytra without between metacoxae into broad shelf that evidence of any fovea or striae. covers base of sternite III (visible ventrite 1). Ventral thoracic foveae unobservable. Abdomen: Abdomen length ~0.6 mm, width Protoclaviger trichodens sp. nov. at base ~0.6 mm slightly narrower than elytra Holotype Material. Sex unknown (putative and with dorsal surface strongly flattened. male). Data label: INDIA: Gujarat Tadkeshwar Three tergites (IV–VI) clearly evident, with lignite mine. Cambay Form. (Paleo-Eocene) distinct boundaries between them (Fig 1G). 21°21.400ʼN, 73°4.532ʼE Jan 11-16, 2012 Broad paratergites present on all three Grimaldi/Nascimbene/Singh/Barden/Tribull/ segments. Thick, hook-shaped trichomes Luzzi/Rana No. Tad-490. Specimen in AMNH. emerging from paratergites IV and V (Fig 1F, G), with smaller cluster of shorter setae Description. Body length 1.56 mm (Fig 1A, B, present on paratergite VI (Fig 1F). Tergites IV S1B). Colour uniform blackish brown, most and V subequal in length, VI 1.4x longer. dorsal surfaces coated with short, sparse, Apical edges of tergites IV and V moderately aciculate pubescence. approximately straight to shallowly concave in Head: Length: 0.28 mm, width across eyes: outline; edge of VII broadly and convexly 0.2 mm. Vertex with rugose sculpture (Fig 1C), rounded (Fig 1G). Tergite VII barely visible in covered with moderately dense, short erect dorsal view, covered by apical margin of VI. setae orientated posteriorly. Median gular Abdomen ventrally with six visible sternites region with several short erect setae (III–VIII). Sternites seemingly articulating (Fig orientated slightly anteriorly. Eye (Fig 1C) with 1A), with extensive intersegmental membrane approximately 23 facets, ommatidia large, indicated by apical edge of each sternite with golden sheen. Antenna (Fig 1D) 0.52 mm overlapping the base of the proceeding long. Antennomere 1 subquadrate; 2 Gujarat state, western India, Tadkeshwar (N cylindrical, narrower than 1, 1.5x wider than 21° 21.400ʼ, E 073° 4.532ʼ), which cuts long; 3–7 conical with ventral edges straight through extensive sequences of the Cambay and dorsal edges curved (Fig 1D), 3–6 Shale Formation, a 75-1500 m-thick layer of subequal in length and width, 7 larger. Dorsal dense, glauconitic clay with seams of lignite, faces of antennomeres 3–6 each with one or which was deposited in an intracratonic two prominent bristles pointing apically (Fig graben that trends NNW-SSE, called the 1D). Dorsal and ventral regions of Cambay Basin. The lowermost basin fill units antennomere 7 apex each with cluster of 3–4 are the Paleocene-Lower Eocene-aged thick, blunt setae; antennomere 8 largest, Vagadkhol Formation (and its equivalent, the equal in length to 6+7 combined; concave Olpad Formation), which directly overlies the apex of antennomere 8 densely covered in Deccan Traps. These are overlain by the Late short, thick setae and several longer setae Paleocene-Middle Eocene-aged Cambay (Fig 1D, E). Formation, which can be subdivided into the Older Cambay Shale and the Younger Thorax: Pronotum length 0.26 mm, width Cambay Shale. In the Vastan mine is a 20- 0.28 mm at widest point. Venter of thoracic 145 m thick unit of interbedded lignite beds, segments lacking pubescence. shales, carbonates, and clay-marls, which is Abdomen: Abdomen length 0.57 mm, width lithologically equivalent to the Older Cambay at base 0.56 mm. Trichomes on paratergites Shale. A mid- to early-Ypresian age (ca. 50- IV and V and trichome-like cluster on VI, all 52 Ma) of the Older Cambay is indicated by formed from shining, golden brown hair-like fossil shark teeth and by the index setae (Fig 1F). Medial tergite regions with foraminiferan Nummulites b. burdigalensis, short, sparse setae, some appearing large found between the upper and lower lignite and flattened, possibly squamous-type. seams at Vastan. In the Cambay Shale Sternites IV–VI with short setation on system, sediment input was probably basolateral regions; IV and V also with medial seasonal with an equatorial climate, and the regions sparsely setiferous. bulk of the sediment derived from chemical Elytra: Elytron length 0.51 mm; width at weathering of the thick Deccan Traps. Nature widest point 0.3 mm, Uniformly covered with of the sediments suggests deposition in a low- sparse, short setae. energy, nearshore/coastal setting similar to extensive 1600-km long chenier systems Legs: Setae on tibiae and femora short, along the coast of northeastern South sparse and aciculate; longer, thicker setae America. Dense mangrove forests fringed the around apex and on dorsal face of metatibia. coast of the Eocene muddy shelf that formed Middle of ventral face of metafemora bearing the Cambay shale, with large dipterocarp distinct bract of 2–3 short, very thick bristles, forests further inland where the amber was possibly indicating the specimen is male. formed. Horizon and locality. The amber in which Protoclaviger is preserved was collected from outcrops exposed within large lignite mines in

Supplementary Discussion inside the nest (Fig 2B; Supplemental Video S2). Analogous trichome structures have Systematic and functional morphology of arisen in some other obligately Protoclaviger myrmecophilous beetle taxa, including It is evident on first sight that Protoclaviger chlamydopsine histerids, paussine carabids, represents a transitional form between the cremastocheiline scarabaeids, lomechusine morphologically derived myrmecophilous staphylinids, as well as in scattered other Clavigeritae and largely free-living ancestral Pselaphinae genera (e.g. Attapsenius, Pselaphinae. In addition to possessing the Epicaris, Desimia, Janusculus, Baceysus, same overall habitus as most Clavigeritae, Batrisiotes, Dendrolasiophilus, Gadgarra, Protoclaviger exhibits key synapomorphies Songius, Millaa). When the biology is that support its phylogenetic affinity to this unknown, trichomes are a reliable indicator supertribe. Several such characters serve that a given species is myrmecophilous. They adaptive functions inside ant colonies, and are the “badges” of the socially integrated thus testify to Protoclavigerʼs likely myrmecophile [S8]. myrmecophilous biology. Importantly however, Protoclaviger bears dense bundles of certain autapomorphies of Recent yellow setae forming hooked-shaped Clavigeritae are missing in Protoclaviger, trichomes in exactly the same position as while others are present at an intermediate Recent Clavigeritae—on paratergites IV (Fig state of evolutionary development. In what 1F). Additionally, Protoclaviger has trichomes follows, important functional and diagnostic on paratergites V, and smaller trichome-like characters of Recent Clavigeritae are tufts on paratergites VI (Fig 1F). We propose compared to the states found in Protoclaviger. that trichomes may have initially developed as Reference is made to Figures S1 that shows serially homologous structures on abdominal confocal reconstructions of a typical segments IV-VI, before the derived clavigerite, Diartiger fossulatus, from Japan. modification of the abdomen in crown-group 1. Abdominal trichomes. Recent Clavigeritae restricted trichome development Clavigeritae possess trichomes—dense, to paratergites IV alone. Trichomes on the elaborately sculpted brushes of yellow hair- elytral margins and within tergite IV, seen in like setae—at the base of the abdomen. some genera of Recent Clavigeritae, would Trichomes typically emerge from paratergites then represent derived additions to the IV (flanking the first visible tergite) (Fig S1D, symplesiomorphic trichomes emerging from E). In some genera, additional trichomes also the paratergites. emerge from within tergite IV itself, or from 2. Abdominal “tergal plate”. The first three the posterior margin of the elytron. Trichomes visible tergites (IV–VI) of all Recent act as wicks, conducting substances from Clavigeritae are fused into a composite tergal large “Wasmann glands” embedded at the plate (Fig S1D). Fusion of the tergites may base of the abdomen [SS3-5]. The strengthen the abdomen when the beetle is substances themselves are unidentified, but being carried or licked by host ants, and may host ants find them attractive and are also facilitate the spreading of exudates from commonly observed licking the trichomes the Wasmann glands and other glandular (Supplemental Video S3). Trichome exudate units that decorate the abdomen [S5]. is thought to appease the ant, and induce it to Remarkably, in Protoclaviger, the tergites are feed the beetle via stomodeal trophallaxis [S6]. distinct, with true segment boundaries plainly Trichomes are also positioned at a part of the evident (Fig 1G). A further difference lies in beetleʼs body that is fashioned into a the relative lengths of the dorsal abdominal “grasping notch” [S7], used by worker ants to segments. Unlike the tergites, the paratergites pick the beetles up and carry them around in Recent Clavigeritae are still demarcated by boundaries, and thus betray the relative Like trichomes, consolidation or compaction lengths of the segments to which they belong of antennomeres is common among (Fig S1D). The paratergites reveal that tergite myrmecophilous beetles [S10-12], and IV typically exceeds the length of V, which is presumably strengthens the antenna, subequal in length to or longer than VI. In protecting against damage or loss of contrast, Protoclaviger has tergite IV and V segments when the beetle is gripped by ant subequal in length, while tergite VI is the mandibles. Evolutionary loss of the pedicels longest (Fig 1G). Hence, the overt also increases the surface area for the restructuring of the abdomen of Recent development of exocrine glands [S6], and Clavigeritae is not seen in Protoclaviger with may facilitate the spreading of exocrine its largely primitive tergite morphology. secretions across an uninterrupted However, Protoclavigerʼs broad, relatively flat integument. Protoclaviger possesses 8 tergites may anticipate the very wide and antennomeres—a degree of antennomere often flattened form of the tergal plate of consolidation intermediate between other Recent Clavigeritae. Whether Protoclaviger Pselaphinae and Recent Clavigeritae. possesses the deep basal excavation of 5. Reduced antennomere pedicels. An tergite IV seen in Recent Clavigeritae [S1] additional means of antennomere cannot be ascertained at this time, due to the consolidation is achieved by elytra covering the abdomen base, as well as reducing/concealing the thin antennomere damage to the right hand side of the abdomen pedicels that form the junctions between any (Fig 1G). remaining, still-distinct antennomeres (Fig 3. Abdominal intersegmental membranes. S1A, C). This gives the antenna the distinctive Although the abdominal segments of Recent appearance of a stacked set of inverted cones. Clavigeritae are still distinct ventrally, their As in Recent Clavigeritae, the antennomere intersegmental membranes are extremely pedicels are indistinct in Protoclaviger, and short [S9]. Neighbouring sternites appear to the conical form of the antennomeres is be fully contiguous with each other (Fig S1F), identical (Fig 1D). and in lateral view, the abdomen has a 7. Truncate antennal apex with setose smooth, continuous ventral profile, cavity. The tip of the antenna of Clavigeritae uninterrupted by obvious segment boundaries. is characteristically truncate and excavate, Reduction of membrane creates a ventrally with the apical cavity filled with short, thick rigid abdomen, matching the absolute setae (Fig S1C and inset). The apex is immobility of the fused, dorsal tergites IV–VI. covered with glandular units that ants lick and Protoclaviger, like non-clavigerite Pselaphinae, may encourage trophallaxis from ant to beetle has extensive intersegmental membranes, so [S6]. In Clavigeritae, the setose cavity is the sternites appear to be partially articulating known to be absent in a few genera, but these (Fig 1A). This flexibility of the ventral still have the apex relatively truncate and abdomen matches Protoclavigerʼs still-distinct covered with short, thick setae. Protoclaviger and possibly articulating tergites IV–VI (Fig possesses the stereotypical apically truncate, 1G). setae-filled excavation of the vast majority of 4. Antennomere consolidation. Most Recent Clavigeritae (Fig 1D, E). We regard pselaphines possess 11 antennomeres, the absence of the cavity in some Recent whereas Clavigeritae have between three and genera as a derived loss. six through loss of a varying subset of 8. Antennal scape. Recent Clavigeritae antennomere pedicels and seamless fusion of typically have the derived condition of very the segments (Fig S1A, C; note that Diartiger short scapes (antennomeres 1), which are fossulatus has four antennomeres but the hidden when the head is viewed from above minute scape is not visible in these images). or below, and do not extend past the shelf-like possess plesiomorphic 4-segmented overhangs of the frontal rostrum (Fig S1A, B; maxillary palpi, implying that palpomere note that the basal-most segment visible in reduction in some arhytodines may have these images is antennomere 2). In contrast, arisen convergently, possibly in response to Protoclaviger has enlarged scapes, protruding similar myrmecophilous lifestyles. out from under the frontal rostrum, akin to The mandibles of Recent Clavigeritae outgroup Pselaphinae (Fig 1D, Fig 3A). are only weakly toothed compared to other 9. Reduced mouthparts. Pselaphinae Pselaphinae [S14], consistent with their possess 4-segmented maxillary palpi with a derived use in novel feeding modes such as 5th apical “pseudosegment”—an piercing ant eggs and larval cuticles, scraping autapomorphy of the subfamily [S9]. The palpi larval exudates, and feeding via trophallaxis are often large and elaborate, and in at least [S8, 16]. The mandibles are also largely some species are used to trap moving prey recessed inside the buccal cavity, with only items [S13]. Clavigeritae have miniscule their tips protruding outside. Protoclaviger maxillary palpi, reduced to a single main also appears to have somewhat weakly palpomere [S14] which does not extend toothed mandibles (Fig 3A), and the outside of the confines of the buccal cavity mandibles do not strongly protrude outside (Fig S1B). Diminution of the palpi occurs in the buccal cavity, approaching the state seen some other myrmecophilous or presumed in Recent Clavigeritae. myrmecophilous Pselaphinae (e.g. 11. Tarsal morphology. Pselaphinae have 3- Attapsenius [S8] and some genera of segmented tarsi, and in most groups, Arhytodini) and may facilitate novel modes of tarsomere 2 exceeds the length of tarsomere feeding associated with this lifestyle. However, 1, which is usually extremely short. In it could also be due to evolutionary Clavigeritae, tarsomeres 1 and 2 are both degeneration, as enlarged palpi are no longer very short, and tarsomere 3 comprises most required for dealing with large, moving prey of the length of the tarsus (Fig S1I). A single objects as the mode of feeding has shifted to tarsal claw is present. Protoclaviger has tarsi trophallaxis and consuming immobile ant identical to those of Recent Clavigeritae (Fig eggs and larvae. The maxillary palpi of 1H)—a clear synapomorphy. One genus of Protoclaviger are comprised of a single main Clavigeritae, Colilodion, has tarsomere 2 segment (Fig 3A), which is bent midway, in elongate, like most other Pselaphinae, and similar fashion to genera of Recent this has led to the notion that Colilodion Clavigeritae (Fig 3B, S1B). The palpi are very represents an evolutionary intermediate small, but nonetheless still larger than those between Clavigeritae and outgroup of Recent Clavigeritae, and they extend some Pselaphinae [S17-19]. Such a view is difficult way outside of the buccal cavity (Fig 3A). We to reconcile with Protoclaviger, which interpret Protoclavigerʼs maxillary palpi as possesses a larger number of primitive transitional between those of ancestral character states than does Colilodion, and in Pselaphinae and Recent Clavigeritae. our cladistic analysis forms the basal-most Similarly vestigial maxillary palpi—reduced to lineage within Clavigeritae (Fig 3C). If a single, bent segment that extends outside of Colilodion is indeed a true member of the buccal cavity—occur in some genera of Clavigeritae (a placement which has been Arhytodini (the form seen in Protoclaviger is questioned [S18]), its tarsal morphology is a most closely approximated in Sabarhytus derived character, and the product of a [S15]). This tribe, together with Pselaphini, reversal to the ancestral condition seen are closely allied to Clavigeritae (Fig 3A, 4A). elsewhere in Pselaphinae. Evaluation of However, many other genera of and Colilodionʼs phylogenetic position within Arhytodini (and all genera of Pselaphini) Pselaphinae awaits collection of DNA-grade specimens of this extremely scarce South pubescence seen in Arhytodini and Pselaphini, East Asian genus. the putative sister tribes of Clavigeritae 12. Additional characters for systematic 13. Autapomorphies of Protoclaviger. placement of Protoclaviger. Like Recent Protoclaviger presents a mostly transitional Clavigeritae, Protoclaviger possesses morphology with few evident autapomorphies. elongate “macrosceline”-type trochanters on However, several structures may represent all legs, where the coxa and femur are distant unique, derived characters. The thick bristles from each other, and the joint between on the dorsal apex of antennomeres III–VI are trochanter and femur is almost perpendicular distinctive (Fig 1D), although possibly only in to their longitudinal axes (Fig S1G, H) [S1]. their configuration rather than type, since thick Elongate trochanters also occur in bristles adorn the antennae of most Pselaphitae, so this character is not a Clavigeritae. The crescent-shaped eyes, synapomorphy linking Protoclaviger to bisected by deep ocular canthi (Fig 1C) are Clavigeritae. However, their presence in also notable but not unique in Clavigeritae Protoclaviger is nonetheless consistent with (even more exaggerated “split” eyes are seen our phylogenetic placement of the new genus. in Colilodion [S19]). The bract of thick, short The convex, ventrally protruding meso- setae on the metafemur— potentially a sex- metaventrite of Protoclaviger (Fig 1A) is also linked character—is unusual, and perhaps characteristic of the form seen in many autapomorphic. In Clavigeritae, it is typically Clavigeritae. It should also be noted that the male mesoleg that bears similar structures, Protoclaviger has few foveae—internally usually in the form of tubercles or spines on striated pits that decorate the body of the tibia, femur or trochanter. Additionally, the Pselaphinae in a stereotypical pattern, and dorsally convex (“hollowed out”) form of the which have important systematic value [S1]. apex of the spatulate maxillary palpi (Fig 3A) Foveae may serve a structural purpose [S1], is, to our knowledge, highly atypical, and a and the trend is for them to become further possible autapomorphy. Finally, the evolutionarily lost in myrmecophilous taxa (e.g. apparent lack of vertexal foveae—a character [S1, 20-22]). Clavigeritae have a highly present in some Recent Clavigeritae—may reduced number of foveae. Preservation also represent a derived loss in Protoclaviger. precludes detailed examination of Given these unique traits, Protoclavigerʼs foveation pattern, but there is Protoclaviger may not be directly ancestral to little evidence of foveae in exposed positions: Recent Clavigeritae, but rather a highly vertexal foveae (the dorsal tentorial pits), plesiomorphic member of the stem-group that elytral foveae, and ventral thoracic foveae diverged early, inheriting a body plan midway such as the lateral mesocoxal and lateral during the major remodelling that produced metaventral foveae, are not apparent. Lateral the crown-group form. Such a notion fits with and medial antebasal foveae or depressions our molecular dating analysis, which indicates on the pronotum—which some Recent that crown-group Clavigeritae may predate Clavigeritae possess—may be present, but Protoclaviger by some 10 Ma (Fig 4A; see these inferred structures could be artefacts following discussion). caused by fossilization. Interestingly, the medial regions of the abdominal tergites appear to have blunt, flattened rectangular- shaped pubescence (Fig 1G). This kind of pubescence is wholly absent among genera of Recent Clavigeritae. It is very similar— perhaps homologous—to the squamous Phylogenetics and diversification of Pselaphini and Arhytodini are distinct among Clavigeritae Pselaphitae in possessing single tarsal claws, like Clavigeritae. Notably, both Pselaphini and The apparent rareness of the vast majority of Arhytodini also exhibit flattened “squamous”- Clavigeritae in nature has been a major type pubescence on various body regions. barrier to conducting any kind of phylogenetic This kind of pubescence is absent among analysis of this beetle taxon. We accumulated genera of Recent Clavigeritae, but sufficient freshly-collected and dried museum Protoclaviger appears to have patches of material to allow the assembly of a 5-gene squamous pubescence on the medial regions data set for almost 25% of recognised genera, of tergites IV–VI, as seen in some genera of enabling us to perform the first-ever Arhytodini. This relationship between phylogenetic analysis of Clavigeritae Clavigeritae and Arhytodini is congruent with (molecular or otherwise). Our analysis results from a taxonomically comprehensive includes 2 of the 3 tribes (Clavigerini and phylogenetic analysis of the entire Tiracerini; only the vanishingly rare SE Asian Pselaphinae (~240 ingroup taxa from all monobasic Colilodionini—6 described species supertribes and 37 of 39 tribes), to be known from fewer than 20 specimens—was published elsewhere (J. Parker, in unavailable, but we discuss the implications of preparation). this taxonʼs absence). Within Clavigerini, we obtained representatives of 8 of the 10 Within Clavigeritae, three tribes are subtribes (only Lunillina and Disarthricerina, presently recognised [S17]: Tiracerini, were unavailable, and the absence of these Colilodionini and Clavigerini. Our analysis obscure and probably derived subtribes does supports the reciprocal monophyly of not likely influence our conclusions). We thus Tiracerini and Clavigerini (Fig 4A, S2), but the achieved a broad and taxonomically unbiased systematic position of Colilodionini could not sample from across the supertribe, of be resolved at this time. Our morphological sufficient density to infer the groupʼs analysis (Fig 3C) places Colilodionini basal to approximate pattern of diversification. The (Tiracerini, Clavigerini)—a relationship we variable and heavily modified morphology of hope to test molecularly when DNA-grade Clavigeritae has hindered attempts to specimens of Colilodion become available. confidently resolve the supertribeʼs Clavigerini, the largest tribe, currently relationship to other Pselaphinae; furthermore, contains 10 subtribes of questionable validity within Clavigeritae itself, extraordinary [S17]. Some are monobasic, while others morphological variation has precluded bring together genera based on characters creation of a stable system of internal that are likely homoplasious, such as relationships. Our analysis is the first to antennomere number and trichome position explicitly study the relationship between [S23, 24]. In our analysis, basal relationships Clavigeritae and other Pselaphinae, as well within Clavigerini are relatively weakly as the first to shed light on some key internal supported, consistent with the tribe radiating relationships. quickly. However, consistent with the inadequacy of the subtribal classification We find that Clavigeritae are system, the 10 included genera of the largest monophyletic (Fig 4A, S2), and form a well- subtribe, Clavigerodina, are distributed across supported relationship with the tribes the tree in a way that is irreconcilable with the Pselaphini and Arhytodini of the supertribe monophyly of this subtribe. Similarly, Pselaphitae. This scenario which has been Besuchetʼs expanded concept of the subtribe alluded to by previous authors [S9, 17]. All Clavigerina [S25], based on the form of the tribes of Pselaphitae share with Clavigeritae basal excavation of the tergal plate and the elongate (“macrosceline”) trochanters, incorporating several genera including most clearly seen in the mesolegs, but Claviger, Adranes, Diartiger and Triartiger, is richness to that of Madagascar, albeit only partially supported. Our results confirm a morphologically more conservative. relationship among Diartiger, Triartiger and Our dating analysis using BEAST Adranes, but do not support this clade illuminates the temporal sequence of grouping with Claviger. This raises the clavigerite diversification. We infer a date for possibility that eyelessness evolved the clavigerite stem of 73 Ma, followed by a independently in the Palearctic Claviger and short, ~11 Ma period of stem evolution, and a Nearctic Adranes. Monophyly of other crown-group age estimate for Clavigeritae of subtribes including Apoderigerina and ~62 Ma. This crown-group age predates Mastigerina is also in conflict with our analysis. Protoclaviger (dated to 52 Ma), indicating that A detailed study of morphology in Clavigerini, Protoclaviger may belong to a stem lineage guided by molecular data, is badly needed to that diverged between 73–62 Ma, persisting create a new higher-level classification for this alongside crown-group Clavigeritae until at tribe. least the early Eocene. Since our analysis did In contrast to the lack of support for not include Colilodion, a crown-group genus subtribes, our analysis reveals some that may be sister to the (Tiracerini, unexpected relationships within Clavigerini. Clavigerini) clade (Fig 3C), the actual period Notably, the Madagascan genera we included of stem-group evolution may be shorter still. form a well-supported monophyletic group. Our topology places Indomalayan and Madagascar is home to a remarkably rich Australasian genera basal within Clavigeritae, clavigerite fauna: 30 described genera are which together with Protoclaviger in Indian known to occur there, 29 of which are amber, indicates that this region may be the endemic [S26]. However, estimating their groupʼs evolutionary center of origin. We infer relationships has historically been impeded by that, from this region, Clavigeritae reached dramatic morphological variation, leading Madagascar and began radiating 43–37 Ma, some authors to erect new subtribes and and reached the New World on at least two even new tribes to accommodate particularly independent occasions, once to form Fustiger enigmatic genera [S23, 24]. Our results are (37–29 Ma), and slightly later to form Adranes consistent with Clavigeritae having undergone (35–12 Ma). The exclusively Nearctic range of an unprecedented radiation within the the latter genus, and its affinity to the East confines of the Madagascan landmass, but Asian Diartiger and Triartiger, suggest a further taxon sampling from the Islandʼs possible Beringian route into North America. clavigerite fauna, as well from the continental One future avenue of study will be to correlate Afrotropics, will be needed to fully test this the global spread of Clavigeritae with the hypothesis. Other relationships elsewhere in historical dispersal of their host ants. Clavigerini are also notable. The genus Our inferences about the temporal Fustiger is polymorphic and diagnosed by dynamics of diversification are based on a extremely weak criteria: a New World BEAST analysis with relatively non-stringent geographic occurrence, and possession of fossil calibrations (Analysis #1; see Materials three antennomeres (many Old World genera and Methods). To explore the effects of our of Clavigeritae have three antennomeres). fossil constraints, we performed an otherwise Surprisingly, our five included representatives identical BEAST run but with tighter priors of Fustiger from the USA, Costa Rica, Peru (Analysis #2), which led to comparable results and Brazil do form a clade, indicating that for the radiation of Clavigeritae (Fig S4A), with Fustiger may in fact be monophyletic. It is only slightly younger ages for stem- and possible that Fustiger represents a largely crown-group Clavigeritae (Fig S4G). In Neotropical radiation of equivalent species contrast, testing whether our priors alone (Analysis #3; Fig S4B) or the molecular data without fossil calibrations (Analysis #4; Fig the Paleocene/early Eocene until approaching S4C) were constraining the outcome of our the present day (Fig 4A, B). analysis [S23, 24, 27, 28] led to vast differences in age estimates (Fig S4B, C, G). Hence, our inferences concerning the origin The myrmecophile fossil record and diversification of Clavigeritae are not The general scarcity of myrmecophiles driven by either overly-strong priors or relative to other arthropods means they are molecular data, but are instead the outcome mostly known from only the richest fossil of the priors and molecular data working deposits. Our survey of described and synergistically. Notably, performing additional undescribed fossil myrmecophiles indicates analyses depicted in Figure S4H, using an that Protoclaviger, in Ypressian-age Cambay alternative placement of fossil A (Arhytodini), amber (~52 Ma) is demonstrably the oldest- or without fossils A and E (Protoclaviger), known myrmecophile. The next oldest yields age estimates for stem- and crown- myrmecophiles occur in Lutetian Baltic amber, group Clavigeritae that are slightly older than ~8–10 Ma younger than Cambay amber. those obtained in our focal analysis (Fig S4D, Pselaphinae of the tribes Tmesiphorini, E, G). Hence, we conclude with confidence Ctenistini and Batrisini have been described that our inferred age for the origin of obligate [S29], all of which contain obligately myrmecophily in Clavigeritae is, if anything, myrmecophilous extant species, although conservative, and the lifestyle may be slightly none as highly socially-integrated as older still. Finally, running the analysis without Clavigeritae. Paussine carabids of the tribe constraining the topology yielded a tree that Paussini—another large, Recent group of differed only in several weakly-supported obligate myrmecophiles—are also known regions within the Clavigeritae clade, and from Baltic amber [S30] (also found as produced a near-identical lineages through compression fossils in the contemporaneous time plot (Fig S4F). Hence, our inferences Eckfelder Maar of Germany [S31]). The concerning the diversification of Clavigeritae presence of these putative myrmecophiles is are not contingent on our constraint tree, and consistent with the high ant diversity of Baltic are robust to any weakly-supported internal amber [S32] and increasing ecological relationships inside Clavigeritae. presence of ants by the middle to late Eocene Lastly, we do not think that increased [S33, 34]. Although not socially-integrated taxon sampling will fundamentally alter our myrmecophiles, Baltic amber also contains an conclusions. Our taxon sampling is spread in adult wasp of the family Eucharitidae [S35], an unbiased fashion across the higher modern members of which have larvae that classification of Clavigeritae, and not are endo- and ectoparasitoids of ants, and a concentrated on one group in particular. More putatively parasitic mesostigmatid mite [S66]. importantly, we see already from our analysis, The late Eocene (Priabonian) using ~25% of genera from the supertribe, Bembridge Marls fossils (Isle of Wight, UK) that Clavigeritae begin radiating soon after include a butterfly, Lithopsyche antiqua Butler, they first arise; hence, there is no long branch a questionable member of Lycaenidae [S36], leading up to their radiation that addition of many Recent members of which have socially further taxa might “break up”. It is possible parasitic caterpillars. The earliest definitive that addition of further taxa might benignly record of Lycaenidae is Aquisextana irenaei alter their inferred rate of diversification, but Théobald, from the late Oligocene/Early this would not negate our hypothesis for when Miocene (Chattian–Aquitanian) of Aix-en- the group arose, and our finding that they Provence (France) [S36]. This deposit also experienced a radiation from approximately provides the first record of an adult Microdon [S37], a hover fly genus (Syrphidae) with myrmecophilous larvae. Mexican amber of pselaphine in Arhytodini [S46] was erroneous, approximately the same age, as well as and a re-evaluation of its position will be slightly younger Domincan amber from the published shortly. early Miocene (Burdigalian; ~20 Ma), contain Finally, we note that Protoclaviger many putatively myrmecophilous groups, represents the only definitive fossil of including Paussini [S38, 39] and haeteriine Clavigeritae thus far discovered. A specimen histerid beetles [S40], modern members of of Articerus Dalman has been documented in which include many guests of Neotropical Copal, but is most probably a Recent species army ants. Staphylinids of the subfamily [S47]. Oxytelinae [S41], and Pselaphinae of the modern, putatively myrmecophilous genera Rhytus, Caccoplectus and Pselaphomorphus (specimens in the AMNH collection) have also been recovered from these ambers. The diversity of myrmecophiles in Dominican amber in particular, and their similarity to Recent forms, suggests a relatively complete collection of modern colony guests had arisen by at least ~20 Ma, consistent with the high percentage of ants in this tropical palaeoenvironment. We reject the claim of Martins-Neto [S42] of Myrmecophilinae (Orthoptera) in the Cretaceous Crato formation of Brazil. Myrmecophilinae are a small group of specialised, myrmecophilous “ant crickets”. Like many Crato compressions, Araripemyrmecophilops gracilis Martins-Neto appears to be a cockroach. We mention also that a member of the pselaphine tribe Arhytodini has been recovered from Cambay amber (“Fossil A” in our dating analysis; Fig S3A). Some Arhytodini have small, recessed mouthparts suggesting a mode of trophallactic feeding similar to Clavigeritae [S43, 44]. Hence, these beetles are potentially myrmecophilous, although only a single report exists of their collection from ant colonies [S45]. Nevertheless, this arhytodine inclusion may thus represent the joint oldest-known myrmecophile, along with Protoclaviger. However, unlike Protoclaviger, the arhytodine fossil—which will be published elsewhere— has no unambiguous adaptive characters for myrmecophily, and unlike all Recent Arhytodini, has enlarged and morphologically elaborate mouthparts. Note that the recent placement of a mid-Cretaceous amber

Experimental Procedures combined to produce montage images using Helicon Focus. Fossil collection and preparation Filming of living Clavigeritae interacting The holotype and unique specimen of with host ants Protoclaviger was discovered amidst To our knowledge, interactions between approximately 20 kgs of raw Cambay amber, Clavigeritae and their host ants have not been which has yielded approximately 550 filmed before. We collected live Fustiger sp. arthropod inclusions, including diverse from a natural population near Austin, Texas, myrmecine, dolichoderine, and formicine ants. together with workers of its Nylanderia sp. The raw piece was trimmed using a 4” host ant. A beetle and some workers were diameter diamond-edged, water-fed trim saw, placed into a plastic bottle cap with damp into a prism approximately 20 mm x 10 mm x tissue paper flooring, and filmed using a USB 8 mm. It was then embedded in EpoTek digital microscope. Over the course of two 301B synthetic resin, to protect the amber and days, we were able to capture several typical inclusion for final trimming. The piece was behaviours. then trimmed much smaller, to maximize full dorsal and lateral views of the beetle and to Morphological placement of Protoclaviger minimize the amount of amber matrix To resolve Protoclavigerʼs relationship to between the beetle and amber surface. Clavigeritae, we scored 33 morphological Finally, it was carefully ground and polished characters from Protoclaviger and members using a series of wet emory papers of of the three tribes of Recent Clavigeritae— decreasing grit sizes (400, 600, 800, 1200, Clavigerini, Colilodionini and Tiracerini. We 2400). also scored putative sister taxa from the tribes Confocal microscopy Pselaphini and Arhytodini [S17], and the “tyrine group” of tribes from Pselaphitae For confocal imaging, a specimen of Diartiger (equivalent to all Pselaphitae except fossulatus was incubated in DNA extraction Arhytodini and Pselaphini, which unpublished buffer to dissolve soft tissues (recipe in [S48]), molecular work has revealed form a strongly washed in ethanol and then disarticulated. supported monophyletic group; J. Parker, in Temporary slide preparations of body parts preparation). In addition, we included further were made using Vectashield (Vector Labs) outgroup Pselaphinae belonging to the four mounting medium. A Leica SP5 confocal other supertribes: Faronitae, Euplectitae, microscope with a 488 nm laser was used to Goniaceritae and Batrisitae. Although obtain image stacks of structures with a 1 mm monophyly of Euplectitae and Goniaceritae is step size between sections (typically 100-250 doubtful [S9], our sole aim was to assess the sections were needed, depending on the phylogenetic position of Protoclaviger, and for structure). Image stacks were collapsed in this purpose these supertribes represent units LAS AF to make maximum projections. of sufficient character state consistency. A Montage photography character matrix was developed in Mesquite [S49] using outgroup character states For habitus images of Pselaphinae (including corresponding to those seen in Clavigeritae), specimens were incubated in Protopselaphinae and allied taxa of the DNA extraction buffer to relax and clean them. pselaphine lineage of omaliine group Appendages were spread and beetles were staphylinid subfamilies, following Newton and mounted on temporary points for photography Thayer [S9] and further discussion by with a Visionary Digital photomicrographic Chandler [S1]. Parsimony analyses were apparatus with Infinity optics and a Canon conducted using TNT [S50] (mult = tbr replic 60D camera. Photograph stacks were 10000 hold 1000 ratchet drift;) with all characters treated as unordered multistate. approximate, mounted on raised projection Clade support was estimated by bootstrap between eyes (1). [S51] (resample boot replic 10000;). 6. Antennal bases. Antennal bases largely Character states indicated for a given exposed, visible in dorsal view (0). Antennae higher taxon are those we could confidently inserted under shelf-like projections (1). conclude are ancestral within that taxon; (NT_17) when there was any doubt, we coded taxa 7. Antennal scape length. Scape prominent, with multiple states. Because of our focus on apical portion visible in dorsal view (0). Scape the placement of Protoclaviger, certain short, concealed in dorsal view (1). isolated examples of independently derived, (CB_2/IL_2) homoplasious character state changes within higher taxa were disregarded, but we note 8. Antennomere number. Antenna with these instances in the character descriptions more than six antennomeres (0). Antenna below. Characters used by Newton and with at most six antennomeres (1). Thayer [S9] in their supertribal phylogeny of (CB_2/IL_2) Pselaphinae are denoted “NT” followed by the 9. Antennal apex. Apex of terminal character number as listed in that paper. antennomere rounded or acuminate (0). Those used by Claude Besuchet [S17] or Ivan Antenna apically truncate (1). (CB_2/IL_2) Lobl [S18] in their discussions of Clavigeritae relationships are denoted CB and IL, 10. Antennal apical excavation. Apex of respectively. terminal antennomere lacking cavity (0). Antenna apically with setose apical cavity (1). 1. Hind body width. Narrow, body (CB_2/IL_2). The apical cavity is missing in a approximately parallel sided (0). Broad, with few genera of Clavigeritae. both elytra and abdomen distinctly wider than head and prothorax (1). 11. Mandibles recessed in buccal cavity. Not recessed, exposed to at least half 2. Hind body convexity. Body form linear in mandible length (0). Recessed, at most only profile, with pterothorax, elytra and abdomen apical teeth visible externally beyond buccal relatively flattened, shallow in dorsoventral cavity (1). (CB_3/IL_3) axis (0). Body form convex, with pterothorax, elytra and abdomen cylindrical or globular, 12. Mandibular prostheca. Prostheca absent long in dorsoventral axis with compact (0). Prostheca present (1). (NT_34) abdomen (1). 13. Maxillary palpus apical (fifth) 3. Fovea. Foveae (striated pit-like pseudosegment. Apical pseudosegment invaginations) absent on body (0). Fovea absent on palpomere IV (0). Digitiform present on at least some body regions (1). pseudosegment present on palpomere IV (1). (NT_37). We follow Newton and Thayer 4. Squamous pubescence. Flattened, (1995) in treating the unsclerotized, seta-like squamous-type pubescence entirely absent structure at the apex of the single-segmented on body (0). Squamous pubescence present clavigerite maxillary palpus as homologous to on some body regions (1). the apical 5th pseudosegment of other 5. Frontal rostrum. Absent or only weakly Pselaphinae. evident, with front of head broad between 14. Maxillary palpus size. Small to large, but widely separated antennal bases that are not extending outside buccal cavity (1). Scarcely mounted on a raised projection between eyes visible, largely recessed inside buccal cavity (0). Pronounced, with narrowing of head to (2). (CB_4/IL_4) clypeus and antennal bases closely 15. Maxillary palpus segment number Caccoplectus (Arhytodini). (apparent). Maxillary palpus 4-segmented (NT_83/CB_7/IL_7) (not counting apical pseudosegment) (0). 24. Tarsal claw number. Two equally-sized Palpi apparently 1-segmented (1). tarsal claws (0). Two unequally-sized tarsal (CB_4/IL_4). Palpomere reduction in some claws (1). A single tarsal claw present (2). myrmecophilous groups such as Ctenistini (NT_81) and Attapseniini of the tyrine lineage is evidently independently derived and ignored 25. Composite abdominal tergal plate. for the purposes of this analysis. Plate absent, segment boundaries of abdominal tergites IV-VI (visible segments 1- 16. Pronotal antebasal sulcus or 3) distinct (0). Plate present, tergites IV-VI impression. Pronotum with transverse fused, their boundaries not apparent (1). impression or well developed antebasal (CB_8/IL_8). Fusion of tergites has occurred sulcus present (0). Prontoum without trace of independently in the SE Asian Plagiophorus transverse impression (1). (NT_53). See (Cyathigerini) and partial fusion is seen in discussion in Chandler (2001) p 30 for a some Neotropical genera of Gonicarini, and discussion of the polarity of this character. these are ignored in this analysis. 17. Elytral sutural striae. Elytron lacking 26. Basal abdominal excavation. Abdominal stria along suture (0). Sutural stria present (1). tergite IV (visible tergite 1) without deep (NT_70) excavation at base (0). Tergite IV basally 18. Protrochanter length. Protrochanter excavate, with deep impression (1). length short, dorsal face of proximal femur (CB_8/IL_8). The basal excavation of nearly touching proccoxa (0). Protrochanter Clavigeritae is independently approached in much longer, profemur and procoxa widely some Tyrini (e.g. Ceophyllus) and Attapseniini separated (1). of the tyrine lineage. 19. Mesotrochanter length. Mesotrochanter 27. Basal abdominal trichomes. Trichomes length short, dorsal face of proximal femur absent from base of abdomen (0). nearly touching mesocoxa (0). (CB_11/IL_11). Trichomes present (1). Mesotrochanter much longer, mesofemur and Trichomes are present in this body region mesocoxa widely separated (1). sporadically outside Clavigeritae, in (NT_84/CB_6/IL_6) Attapseniini, Ctenistini, Trichonychini and Batrisini. 20. Metacoxal projection. Metacoxa projecting at articulation with metatrochanter 28. Abdominal lateral margins. Interfaces (0). Metacoxae not projecting (1). between segments IV-IV angulate in dorsal view, making the lateral abdominal margins 21. Metatrochanter length. Metatrochanter appear discontinuous (0). Interfaces between length short, dorsal face of proximal femur segments IV-IV not angulate, near-seamless nearly touching metacoxa (0). Metatrochanter continuity between edges of paratergites much longer, metafemur and metacoxa widely producing lateral margins of abdomen that are separated (1). smoothly rounded to almost straight (1). 22. Metacoxal separation. Metacoxae 29. Relative length of abdominal tergite IV. contiguous (0). Metacoxae separated (1). Segment IV subequal in length to segment V 23. Tarsomere 2 length. Second tarsomere (0). Segment IV at least 1.3 times as long as clearly longer than tarsomere 1 (0). than segment V (1). In Clavigeritae, the state Tarsomeres 1 and 2 subequal in length (1). can be assessed using the still-segmented Tarsomere 2 approaches the length of 1 in paratergites flanking the composite tergal plate. 30. Sternite III and IV setal concentration. Disarthricerina were not included). Within Basal abdominal impression (apex of sternite Clavigerini we added increased sampling from III/ base of sternite IV) without pronounced the two larger subtribes Clavigerodina and accumulation of setae (0). Basal impression Clavigerina, and the diverse Madagascan of ventral abdomen setiferous (1). (NT_89) fauna. We could not obtain specimens of the rare, monobasic tribe Colilodionini, but as we 31. Sternite VIII defensive gland. Medial, explain in “Phylogenetics and Diversification unpaired omaliine-type defensive gland of Clavigeritae”, absence of this tribe does not, present (0). Sternal gland absent (1). (NT_92) we think, significantly affect our conclusions. 32. Ventral abdominal intersegmental Hence, little of any consequence from the pre- membranes. Intersegmental membranes existing taxonomy of Clavigeritae was left out long, equal to or greater than 1/10 sternite of our analysis. Our data set thus included length (0). Intersegmental membranes short, appropriate unbiased taxonomic, less than 1/10th sternite length (1). morphological and biogeographic coverage to 33. Ventral abdominal intersegmental enable us to gauge the groupʼs temporal membrane pattern. Membrane with brick- pattern of diversification. We were also able to wall pattern of minute sclerites (0). Brick-wall provide some insights into relationships sclerite pattern absent (1). (NT_96) between genera, and test the validity of several Clavigerini subtribes. The resulting morphological character matrix for TNT is shown in Table S1. DNA extraction and sequencing. Ethanol- preserved specimens were vacuum dried, and pinned specimens were removed from card points. Specimens were incubated whole, without destruction in an SDS/Proteinase-K- Molecular phylogenetic analysis and based extraction buffer (see [S48] for recipe) diversification of Clavigeritae for 2 days at 55°C. Following digestion,

specimens were removed, and supernatant Taxon sampling. Taxa are listed in Table S2. was phenol-chloroform extracted following the We included representatives of the six protocol detailed in [S21]. Purified DNA was supertribes of Pselaphinae, incorporating taxa resuspended in Tris-EDTA. Clontech that enabled us to use all Pselaphinae fossils Advantage 2 polymerase was used to amplify at our disposal for molecular dating of nodes gene fragments and an annealing outside of Clavigeritae. Within Clavigeritae, temperature of 51°C was typically used in all we aimed for as much taxonomic coverage as PCR reactions, with only the extension time possible to provide a reliable and unbiased being varied depending on target amplicon estimate of Recent diversity from across the length. Bands were excised from gels and breadth of the supertribe. This was crucial for ligated into TOPO-XL-PCR vector (Life dating the supertribeʼs origin, and for making Technologies), and transformed into DH5a assessments about its pattern of cells. Colonies were miniprepped and test diversification. Given that the majority of digested with EcoRI. Plasmids containing genera are extremely rare and known from correct inserts were batch-sequenced with T7 one or a few specimens, we had to resort to and M13R primers using Macrogen Corp. (NY, pinned museum specimens for some critical USA). The following primer combinations groups. Ultimately, we were able to include 31 were used (asterisks indicate primers species (~10% of the total) representing 26 designed for this study): genera (~25% of the total), including the monobasic Tiracerini and 8/10 subtribes of by 28s rRNA (~650-800 bp): (28sDD 5ʼ- far the largest tribe, Clavigerini (only the GGGACCCGTCTTGAAACAC / 28sFF 5ʼ- likely-derived subtribes Lunillina and CACACTCC TTAGCGGAT) 18s rRNA (~1900-2140 bp): (18s5ʼshort* 5ʼ- consensus tree was rooted with Faronitae, the CAACCTGGTTGATCCTGCC / 18s3ʼl 5ʼ- basal-most lineage of Pselaphinae based on CACCTACGG AAACCTTGTTACGAC or for morphological [S9] and molecular problematic taxa, the 3ʼ end of the locus was phylogenetic analyses [S21]. Faronitaeʼs targeted using 18sa2.0 5ʼ- basal position is also unequivocally supported ATGGTTGCAAAGCTGAAAC) by a forthcoming multilocus phylogeny of the entire subfamily Pselaphinae (Parker, in prep). 16s rRNA (~425-535 bp): (16saR 5ʼ- CGCCTGTTTATCAAAAACAT / 16sb 5ʼ- Molecular dating. The rooted and fully CTCCGGT TTGAA CTCAGATCA or 16sb_3* resolved maximum clade credibility tree from 5ʼ-TTAATCCAACATCGAGGTCG) the MrBayes analysis was created by combining log files in TreeAnnotator [S57]. cytochrome oxidase I (820 bp): (TL2-N- Branch lengths were transformed to create an 3014PAT 5ʼ-TCCAATGCACTAA ultrametric tree conforming to dating priors in TCTGCCATATTA / C1-J-2183JERRY 5ʼ- TreeEdit [S58]. The constraint tree was CAACATTTATTTTGATTTTTTGG or imported into BEAST [S57] for fossil Jerry2nd* 5ʼ-GATTTTTTG calibration using separate Bayesian lognormal GWCAYCCWGAAG) relaxed clocks [S59] for the mitochondrial and wingless (~450-510 bp): (wg550f 5ʼ- nuclear gene partitions. For our focal analysis ATGCGTCAGGARTGYAARTGYCAYGGYA (Analysis #1), we employed seven fossils and TGTC / wgABRZ 5ʼ- one ancient biogeographic event. For the CACTTNACYTCRCARCACCARTG with fossil calibrations, fossil ages were used as secondary PCR if necessary wg578f 5ʼ- offsets that defined hard minimum bounds on TGACNGTGAARACYTGCTGGATG / wgABR lognormal distributions. Because we could not 5ʼ-ACYTCGCAGCACCARTGGAA) justify a priori how stringent the soft maximum In our final matrix, >90% of taxa were bounds of our lognormal distributions should sequenced for all five loci (Table S2). be [S27], we employed relatively weak soft maxima to shape their tails (using a high Phylogenetic analysis. Sequences were standard deviation) but performed an aligned in MUSCLE [S52] with default additional analysis (Analysis #2) with tighter parameters, and visually inspected in Se-Al soft maxima to determine how this affected with minimal editing. Model selection in the results. jModeltest2 [S53] yielded the GTR+I+G model for the Wg, COI and 18s rRNA partitions, Analysis #1 used the dating priors listed GTR+G for 16s rRNA, and SYM+G for 28s below, with their hard minimum ages, the type rRNA. Sequences were concatenated using of distribution followed by the mean (in real Sequence Matrix, producing a 51-taxon matrix space) and standard deviation (if lognormal) comprised of 5944 characters. Bayesian in parentheses: inference, partitioned by locus, was performed A) Arhytodini (52 Ma [S60], lognormal, 10.0, using MrBayes 3.2.2 [S54], available through 1.0). An undescribed Arhtyodine in Cambay the Cipres Science Gateway [S55]. Two amber (AMNH TAD-23; Fig S3A). Although MCMC runs of one cold and three hot chains the two genera of Arhytodini included in our (temperature parameter 0.2) were run until analysis (Rhytus and Caccoplectus) are the standard deviation of split frequencies closely allied and emerge as a monophyletic dropped below 0.01, which was reached after sister group to Pselaphini (Fig 4A, S2), the 5.1 million generations. Tracer [S56] was monophyly of Arhytodini as a whole is used to judge adequate sampling from the questionable [S1], and the tribe may posterior distribution. The first 25% of trees paraphyletic with respect to Pselaphini. The were discarded as burn-in. The resulting phylogenetic position of some genera of Arhytodini, including the undescribed Cambay zoogeographic distributions), indicates the fossil, is presently unresolved, raising the higher Pselaphinae likely originated on possibility that it could have diverged earlier Pangaea, hence at the latest by the end of the with respect to the four genera that comprise Jurassic. We used a conservative date of 150 the (Arhytodini, Pselaphini) clade in our tree. Ma and an exponential distribution with mean Accordingly, rather than placing fossil A at the = 20.0 to incorporate uncertainty about when, ancestral node joining (Arhytodini, Pselaphini), before this minimum age, the higher we opted for a conservative placement at the Pselaphinae arose. Because we applied this ancestral node of ((Arhytodini, Pselaphini) date so close to the root, we did not specify a Clavigeritae) (fossil placement configurations root height in our analysis. are shown in Fig S4H; see also the alternative Four independent BEAST runs of 200 placement of Fossil A in Figure S2). In million generations were performed using a Analysis #1, Fossil A is therefore redundant speciation birth-death process tree prior [S62]. with Protoclaviger (Fossil E). However, we We enforced the tree topology during the also performed an additional analysis analysis [S63], but also ran an identical (Analysis #5), in which Fossil A was placed at analysis where topology was not enforced the less conservative position of the node (Analysis #7). Convergence of runs was joining (Arhytodini, Pselaphini) (Fig S4H). judged based on high ESS values in Tracer B) Bythinini (99 Ma [S61], lognormal, 10.0, [S56]. Tree files were combined in 1.0). Undescribed Bythinini in Burmese amber LogCombiner with the first 10% of trees (AMNH Bu-248 and AMNH B-023; Fig S3B). discarded as burn-in, yielding the time- calibrated maximum clade credibility tree C) Bythinoplectini: Bythinoplectina (52 Ma shown in Figure 4A. The tree was imported [S60], lognormal, 10.0, 1.0). Undescribed into R (http://www.R-project.org) to produce Bythinoplectina in Cambay amber (AMNH the Lineages-Through-Time plot in Figure 4B TAD-130; Fig S3C). using the Ape package [S64]. Data for the D) Caccoplectus (20 Ma, lognormal, 5, 0.5). relative abundance of ants in different fossil Caccoplectus in Dominican amber [S44]; also deposits, used in Figure 4B, are from [S33] AMNH DR8-429, Fig S3D). and [S65]. E) Clavigeritae (52 Ma [S60], lognormal, 10.0, Six additional BEAST analyses were 1.0). Protoclaviger in Cambay amber (AMNH performed to explore how our priors TAD 490; Fig S3E). influenced our results. These differ from F) Ctenistini (52 Ma [S60], lognormal, 10.0, Analysis #1 as follows: 1.0). Undescribed Ctenistini in Cambay amber Analysis #2. Tighter maximum bounds on (AMNH TAD-491; Fig S3F). fossil calibration densities. Standard G) Faronus (44 Ma, lognormal, 10.0, 1.0). deviations on all fossil calibrations (A-G) were Faronus in Baltic amber [S29] (and AMNH lowered to 0.4 to constrain the variance of the Ba-JWJ 334; Fig S3G). soft maximum bounds of our dating priors. H) “Higher Pselaphinae” (150 Ma, exponential, Analysis #3. Molecular data excluded. 20.0). The higher Pselaphinae are comprised Using overly-strict dating priors can override of all Pselaphinae except the basal-most information from the molecular data, and supertribe, Faronitae. The presence of non- strongly constrain estimation of the posterior faronite pselaphines in mid-Cretaceous [S27, 28]. An analysis run without any Laurasian Burmese amber, combined with the molecular data will sample from the prior evidently Gondwanan origin of many higher alone, enabling one to gauge whether the pselaphine tribes (inferred from their modern priors used are dictating the outcome. Analysis #4. All fossil calibrations excluded. To ascertain how are our fossil dating priors were influencing the outcome of our analysis, all our fossil calibrations (A-G) were removed, and only calibration H (higher Pselaphinae) was included. Analysis #5. Alternative placement of fossil A. The impact of placing Fossil A at the less conservative, more recent ancestral node of (Arhytodini, Pselaphini) (scheme depicted in Fig S4H) was explored. Analysis #6. Fossils A+E excluded. To assess how removal of Protoclaviger (Fossil E) and redundant Fossil A altered our inferences concerning dating of Clavigeritae, both E and A calibration points were excluded from the analysis (Fig S4H). Analysis #7. Tree topology unconstrained. We determined the extent to which using a rigid constraint tree affected our analysis by running an otherwise identical analysis in which the topology was simultaneously estimated in BEAST.

Supplemental References phylogenetic analysis and review of the Omaliine Group of Staphylinidae

including Pselaphidae (Coleoptera). In S1. Chandler, D. S. (2001). Biology, Biology, Phylogeny, and Classification Morphology and Systematics of the of Coleoptera: Papers Celebrating the Ant-Like Litter Beetles of Australia 80th Birthday of Roy A. Crowson., J. (Associated Publishers). Pakaluk and S. A. Slipinski, eds. S2. Hlavac, P. (2011). Contribution to the (Muzeum i Instytut Zoologii PAN, knowledge of the tribe Mastigerina Warszawa), pp. 221–320. (Coleoptera: Staphylinidae: S10. Jeannel, R. (1936). Monographie des Pselaphinae, Clavigeritae), with a Catopidae (Insectes Coléopteres). description of a new genus from Mémoires du Muséum National Borneo. Zootaxa 3070, 51–59. d'Histoire Naturelle N.S. 1, 1–438. S3. Krüger, E. (1910). Beiträge zur S11. Geiselhardt, S. F., Peschke, K., and anatomie und biologie des Claviger Nagel, P. (2007). A review of testaceus Preyssl. Zeitschrift für myrmecophily in ant nest beetles wissenschaftliche Zoologie 95, 327– (Coleoptera: Carabidae: Paussinae): 381. linking early observations with recent 4. Hill, W., Akre, R., and Huber, J. (1976). findings. Naturwissenschaften 94, 871– Structure of some epidermal glands in 894. the myrmecophilous beetle Adranes S12. Jałoszyński, P. (2013). A new species taylori (Coleoptera: Pselaphidae). of the putatively myrmecophilous genus Journal of The Kansas Entomological Plaumanniola Costa Lima, with notes Society 49, 367–384. on the systematic position of S5. Cammaerts, R. (1974). Le systeme Plaumanniolini (Coleoptera: glandulaire tegumentaire du coleoptere Staphylinidae: Scydmaeninae). myrmecophile Claviger testaceus Zootaxa 3670, 317. Preyssler, 1790 (Pselaphidae). Z. S13. Schomann, A., Afflerbach, K., and Betz, Morph. Tiere 77, 187–219. O. (2008). Predatory behaviour of S6. Cammaerts, R. (1992). Stimuli inducing some Central European pselaphine the regurgitation of the workers of beetles (Coleoptera: Staphylinidae: Lasius flavus (Formicidae) upon the Pselaphinae) with descriptions of myrmecophilous beetle Claviger relevant morphological features of their testaceus (Pselaphidae). Behav heads. European Journal of Process 28, 81–96. Entomology 105, 889–907. S7. Leschen, R. A. B. (1991). Behavioral S14. Ohishi, H. (2000). Notes on the observations on the myrmecophile mouthparts of Pselaphinae Fustiger knausii (Coleoptera: (Coleoptera: Staphylinidae). Insecta Pselaphidae: Clavigerinae) with a Miyatakeana, a special publication from discussion of grasping notches in the Entomological Laboratory, Osaka myrmecophiles. Entomological News Museum of Natural History, 37–46. 102, 215–222. S15. Lobl, I. (2000). Pachacuti chandleri sp. S8. Park, O. (1942). A Study in Neotropical n. and Sabarhytus kinabalu gen. et sp. Pselaphidae (Evanston and Chicago: n. with comments on Arhytodini and Northwestern University). Pselaphini (Coleoptera, Staphylinidae, S9. Newton, A. F., and Thayer, M. K. Pselaphinae). Biological Journal of the (1995). Protopselaphinae new Linnean Society 55, 143–149. subfamily for Protopselaphus new S16. Park, O. (1932). The myrmecocoles of genus from Malaysia, with a Lasius umbratus mixtus aphidicola Walsh. Ann Entomol Soc Am 25, 77– que dʼun genre inédit. Revue de 88. Zoologie et de Botanique Africaines 80, S17. Besuchet, C. (1991). Révolution chez 415–424. les Clavigerinae (Coleoptera, S25. Besuchet, C. (2008). Synonymies et Pselaphidae). Rev. Suisse Zool. 98, combinaisons nouvelles, revalidations 499–515. et description de taxa nouveaux de S18. Lobl, I. (1994). The Systematic Position Pselaphinae (Coleoptera: of Colilodionini with Description of a Staphylinidae). Mitteilungen der New Species (Coleoptera, Schweizerischen Entomologischen Pselaphidae). Revue Suisse de Gesellschaft 81, 61–82. Zoologie 101, 289–297. S26. Hlaváč, P., Baňař, P., and Parker, J. S19. Nomura, S., and Sugaya, H. (2007). A (2013) The Pselaphinae of New Species of the Genus Colilodion Madagascar. II. Redescription of the Besuchet (Coleoptera: Staphylinidae: genus Semiclaviger Wasmann, 1893 Pselaphinae) from Vietnam. Annales (Coleoptera: Staphylinidae: de la Société Entomologique de France Pselaphinae: Clavigeritae) and (n.s.) 43, 333–339. synonymy of the subtribe Radamina S20. Zhao, M.-J., Yin, Z.-W., and Li, L.-Z. Jeannel, 1954. Zootaxa 3736, 265. (2010). Contributions to the knowledge S27. Ho, S. Y. W., and Phillips, M. J. (2009). of the myrmecophilous pselaphines Accounting for Calibration Uncertainty (Coleoptera, Staphylinidae, in Phylogenetic Estimation of Pselaphinae) from China. IV. The Evolutionary Divergence Times. Syst. second species of the genus Songius Biol. 58, 367–380. (Coleoptera, Staphylinidae, S28. Heled, J., and Drummond, A. J. (2012). Pselaphinae), with description of its Calibrated tree priors for relaxed probable mature larva. Sociobiology 56, phylogenetics and divergence time 77–89. estimation. Syst. Biol. 61, 138–149. S21. Parker, J., and Maruyama, M. (2013). S29. Schaufuss, L. W. (1890). System- Jubogaster towai, a new Neotropical schema der Pselaphiden, ein Blick in genus and species of Trogastrini die Vorzeit, in die Gegenwart und In die (Coleoptera: Staphylinidae: Zukunft. Tijdschrift voor Entomologie Pselaphinae) exhibiting myrmecophily 33, 101–162. and extreme body enlargement. S30. Wasmann, E. (1929). Die Paussiden Zootaxa 3630, 369–378. des baltischen Bernsteins und die S22. Zi-Wei Yin, L.-Z. L. (2013). Stammesgeschichte der Paussiden. In Pengzhongiella daicongchaoi gen. et Bernstein-Forschungen (Amber sp. n., a remarkable myrmecophile studies), K. Andrée, ed., pp. 1–110. (Staphylinidae, Pselaphinae, Batrisitae) S31. Wappler, T. (2003). Systematik, from the Gaoligong Mountains. Phylogenie, Taphonomie und Zookeys 326, 17–26. Paläoökologie der Insekten aus dem S23. Jeannel, R. (1954). Les Pselaphides de Mittel-Eozän des Eckfelder Maares, Madagascar. Memoires de l'Institut Vulkaneifel. Clausthaler Scientifique de Madagascar 4, 139– Geowissenschaften 2, 1–241. 344. S32. Wheeler, W. M. (1914). The ants of the S24. Celis, M. J. (1969). Contribution á Baltic amber. Schriften der l'etude des Clavigerinae de lʼAfrique Physikalisch-okonotnischen (Coleoptera Pselaphidae) 3. Gesellschaft Königsberg 55, 1-142. Démembremant des Fustigerini et S33. Grimaldi, D., and Agosti, D. (2000). A création de deux tribus nouvelles ainsi formicine in New Jersey Cretaceous amber (Hymenoptera : Formicidae) and Pselaphidae). The Coleopterists early evolution of the ants. Proc. Natl. Bulletin 29, 301–316. Acad. Sci. U.S.A. 97, 13678–13683. S44. Chandler, D. S., and Wolda, H. (1986). S34. Wilson, E. O., and Hölldobler, B. (2005). Seasonality and diversity of The rise of the ants: A phylogenetic Caccoplectus, with a review of the and ecological explanation. genus and description of a new genus, Proceedings of the National Academy Caccoplectinus (Coleoptera: of Sciences 102, 7411–7414. Pselaphidae). Zoologische Jahrbiicher. S35. Heraty, J. M., and Darling, D. C. (2009). Abteilung flir Systematik, Okologie und Fossil Eucharitidae and Perilampidae Geographie 113, 469–524. (Hymenoptera: Chalcidoidea) from S45. Bruch, C. (1918). Nuevos huéspedes Baltic Amber. Zootaxa 2306, 1–16. de hormigas procedentes de Córdoba. S36. Sohn, J. C., Labandeira, C., Davis, D. Physis 4, 186–195. R., and Mitter, C. (2012). An annotated S46. Peris, D., Chatzimanolis, S., and catalog of fossil and subfossil Delclòs, X. (2014). Diversity of rove Lepidoptera (Insecta: Holometabola) of beetles (Coleoptera: Staphylinidae) in the world. Zootaxa 3286, 1–132. Early Cretaceous Spanish amber. S37. Hull, F. M. (1945). A revisional study of Cretaceous Research 48, 85–95. the fossil Syrphidae. Bulletin of The S47. Newton, A. F., and Chandler, D. S. Museum of Comparative Zoology 95, (1989). World catalog of the genera of 251–355. Pselaphidae (Coleoptera). Fieldiana S38. Nagel, P. (1997). New fossil paussids Zoology 53, 1–110. from Dominican amber with notes on S48. Gilbert, M. T. P., Moore, W., Melchior, the phylogenetic systematics of the L., and Worobey, M. (2007). DNA paussine complex (Coleoptera: Extraction from Dry Museum Beetles Carabidae). System Entomol 22, 345– without Conferring External 362. Morphological Damage. PLoS ONE 2, S39. Kraemer, M. M. S. (2006). The first e272. fossil paussine (Coleoptera: Carabidae) S49. Maddison, W. P., and Maddison, D. R. from Mexican amber. Paläontol Z 80, (2011). Mesquite: a modular system for 107–111. evolutionary analysis. Version 2.75. S40. Tishechkin, A. K. (2005). Phylogenetic mesquiteproject.org. Revision of the Genus Mesynodites S50. Goloboff, P. A., Farris, J. S., and Nixon, Reichardt (Coleoptera: Histeridae: K. C. (2008). TNT, a free program for Hetaeriinae). Dissertation. (Louisiana phylogenetic analysis - Goloboff - 2008 State University). - Cladistics - Wiley Online Library. S41. Engel, M. S., and Chatzimanolis, S. Cladistics 24, 774–786. (2009). An oxyteline rove beetle in S51. Felsenstein, J. (1985). Confidence- dominican amber with possible African Limits on Phylogenies - an Approach affinities (Coleoptera: Staphylinidae: Using the Bootstrap. Evolution 39, Oxytelinae). Annals of Carnegie 783–791. Museum 77, 425–429. 52. Edgar, R. C. (2004). MUSCLE: multiple S42. Martins-Neto, R. G. (1991). Sistemática sequence alignment with high accuracy dos Ensifera (Insecta, Orthopteroida) and high throughput. Nucleic Acids Res. da formação Santana, Cretáceo Inferior 32, 1792–1797. do Nordeste do Brasil. Acta Geologica 53. Darriba, D., Taboada, G. L., Doallo, R., Leopoldensia 32, 5–160. and Posada, D. (2012). jModelTest 2: S43. Chandler, D. S. (1975). A Revision of more models, new heuristics and the Genus Caccoplectus (Coleoptera: parallel computing. Nat. Methods 9, 772. S63. Moreau, C. S., and Bell, C. D. (2013). 54. Ronquist, F., Teslenko, M., van der Testing the museum versus cradle Mark, P., Ayres, D. L., Darling, A., tropical biological diversity hypothesis: Höhna, S., Larget, B., Liu, L., Suchard, phylogeny, diversification, and M. A., and Huelsenbeck, J. P. (2012). ancestral biogeographic range MrBayes 3.2: efficient Bayesian evolution of the ants. Evolution 67, phylogenetic inference and model 2240–2257. choice across a large model space. S64. Paradis, E., Claude, J., and Strimmer, Syst. Biol. 61, 539–542. K. (2004). APE: Analyses of S55. Miller, M. A., Pfeiffer, W., and Schwartz, Phylogenetics and Evolution in R T. (2010). Creating the CIPRES language. Bioinformatics 20, 289–290. Science Gateway for Inference of S65. LaPolla, J. S., and Dlussky, G. M. Large Phylogenetic Trees. In, M. A. (2013). Ants and the Fossil Record. Miller, W. Pfeiffer, and T. Schwartz, eds. Annu. Rev. Entomol. 58, 609–630. (New Orleans), pp. 1–8. S66. Dunlop, J. A., Kontschán, J., Walter, D. S56. Rambaut, A., Suchard, M. A., Xie, D., E., and Perrichot, V. (2014). An ant- and Drummond, A. J. (2013). Tracer associated mesostigmatid mite in Baltic v1.5. Available from amber. Biol. Lett. 10, 20140531 http://beast.bio.ed.ac.uk/Tracer. S57. Drummond, A. J., Suchard, M. A., Xie, D., and Rambaut, A. (2012). Bayesian Phylogenetics with BEAUti and the BEAST 1.7. Mol. Biol. Evol. 29, 1969– 1973. S58. Rambaut, A., and Charleston, M. (2002). TreeEdit. Phylogenetic Tree Editor. tree.bio.ed.ac.uk. S59. Drummond, A. J., Ho, S. Y. W., Phillips, M. J., and Rambaut, A. (2006). Relaxed phylogenetics and dating with confidence. PLoS Biol. 4, e88. S60. Rust, J., Singh, H., Rana, R. S., McCann, T., Singh, L., Anderson, K., Sarkar, N., Nascimbene, P. C., Stebner, F., Thomas, J. C., et al. (2010). Biogeographic and evolutionary implications of a diverse paleobiota in amber from the early Eocene of India. Proc. Natl. Acad. Sci. U.S.A. 107, 18360–18365. S61. Shi, G., Grimaldi, D. A., Harlow, G. E., Wang, J., Wang, J., Yang, M., Lei, W., Li, Q., and Li, X. (2012). Age constraint on Burmese amber based on U-Pb dating of zircons. Cretaceous Research 37, 155–163. S62. Gernhard, T. (2008). The conditioned reconstructed process. Journal of Theoretical Biology 253, 769–778.