AMERICANt MUSEUM NovJitates PUBLISHED BY THE AMERICAN MUSEUM OF NATKTIUA-1ISTORY CENTRAL PARK WEST AT 79TH STREET, , N.Y. 10024 Number 3208, 43 pp., 24 figures, 1 tables October 23, 1997

New and Rediscovered Primitive (: Formicidae) in from , and Their Phylogenetic Relationships*

DAVID GRIMALDI,1 DONAT AGOSTI,2 AND JAMES M. CARPENTER'

ABSTRACT All Cretaceous records of ants are reviewed, worker are external and even some internal fea- and evidence originally given for their placement tures of the metapleural gland-a definitive for- is evaluated. Cretaceous Formicoidea preserved in micid synapomorphy-which are described in de- rocks lack preservation of critical synapomor- tail for the first time. This specimen is designated phies like the metapleural gland, which are pre- as a neotype to replace the disintegrated holotype served in specimens in amber. For this reason, the of this . One complete male specimen is Dlussky, from the Cretaceous of Rus- tentatively assigned to , which sia, are placed here as Formicoidea , would be the first known male of the . One not in the Formicidae proper. The oldest amber complete and one partial male, both from the with ants is from New Jersey. same piece of amber, are a new, plesiomorphic Seven complete and partial specimens of For- species of Baikuris Dlussky (Baikuris casei, n. micidae, recently discovered in (92 Ma) sp.), the genus previously known only from upper amber from central New Jersey, are described and Cretaceous (Santonian) amber of Taymyr, north- discussed. The specimens consist of four males ern Siberia. Two workers represent a new genus and three workers. One complete and well-pre- and species of Cretaceous Formicidae, Browni- served worker is Sphecomyrma freyi Wilson and mecia clavata n. gen., n. sp., based on distinc- Brown 1967, known previously from two workers tively clubbed antennae, proportions of the anten- in a piece of Turonian amber from Cliffwood nal articles, and thin mandibles that lack teeth and Beach, New Jersey. Well preserved in the new extensively cross, and a ponerine-like constriction

* Paleontology of , Part VI. ' Curator, Department of Entomology, American Museum of Natural History. 2 Research Scientist, Department of Entomology, American Museum of Natural History.

Copyright C American Museum of Natural History 1997 ISSN 0003-0082 / Price $5.60 2 AMERICAN MUSEUM NOVITATES NO. 3208 between abdominal segments M and IV (gaster crepencies with the cladograms of Baroni Urbani segments I and II). Brownimecia is more closely et al. (1992) are discussed. Including the related to extant Formicidae than are Sphecomyr- Lutz (compressed remains minae, the genus being the earliest known mem- from the of Germany) in the cladistic ber of the extant . The fourth analysis, as done by Baroni Urbani et al., imparts male belongs to a third genus, undescribed. These too many missing values for meaningful analysis. are the oldest definitive ants. Sphecomyrma, not discussed by Baroni Urbani et Composition, monophyly, and systematic po- al., retains its basal position in the Formicidae in sition of the subfamily are dis- the present analysis. Discovery of new and exclu- cussed. Modifications were made to the data of sively primitive ants in upper Cretaceous Baroni Urbani et al. (1992) and reanalyses were indicates an origin of the ants probably in the low- made of the basal relationships of the Formicidae ermost Cretaceous, but no older, contrary to a re- including Sphecomyrma and Brownimecia. Dis- cent molecular hypothesis.

INTRODUCTION The ants can arguably be said to be the son and Brown, in Cretaceous amber from ecologically and numerically dominant fam- New Jersey. The piece of amber that con- ily of -perhaps of all organisms-on tained two workers was discovered in 1966 earth. As Holldobler and Wilson (1990) dis- by an amateur collector, Edmund Frey, who cussed, ants are so abundant that approxi- found it in strata of the Raritan-Magothy For- mately 3 million exist in a hectare of Ama- mations exposed in bluffs at Cliffwood zonian forest soil, and they represent some Beach, New Jersey. It was studied and re- 10-15% of the entire biomass in ter- ported by Wilson et al. (1967a, 1967b). De- restrial ecosystems. On this basis alone, the spite such a sensational find, little attention origins of the ants would be important for was paid to New Jersey amber other than by understanding the of modem ter- another fossil collector, Gerard R. Case. Case restrial biological communities. collected Cretaceous amber from at least a The outstanding hallmark of the ants is that dozen localities between 1962 and 1986, all all species are eusocial, which is what probably of it found in rather trace quantities on the caused their tremendous success. The wingless surface of lignitic exposures. Many of his workers can easily penetrate minute niches. samples were given to Princeton University Their array of glands and secretions enables (Dept. of Geology) where they were exam- sophisticated chemical communication, such ined for inclusions. (The other samples were that they can rapidly recruit workers for se- given to the AMNH [American Museum of questering food; defend the colony, especially Natural History] in 1992.) The only piece with soldiers and stinging workers; and even with inclusions was found in 1966-the prey. has actually subdue large same year as the Sphecomyrma discovery- 12 in the Hymenoptera: once in evolved times in the Such Clay Pits in Parlin, New Jersey. the ants, and all other times in the It contained a ceratopogonid and two chiron- (social/paper wasps) and the s.s. revealed that (). It was the ants that became predomi- omid midges. Case's collections nant, though, because they are thought to have amber in New Jersey was actually wide- been the earliest, perhaps the first, eusocial in- spread in appropriate Cretaceous exposures, sect predators on the ground (Holldobler and and that other insects existed in the amber Wilson, 1990). Clearly, knowing the earliest besides Sphecomyrmna. These collections re- history of the ants can allow much better un- mained largely unnoticed for more than 20 derstanding of the unparalleled success of this years, and even the ceratopogonid wasn't de- remarkable group. It is a history that began in scribed until 1986, as Culicoides casei. the late . In 1976 Gerard Case brought Robert Lan- genheim (University of Illinois, Champagne- SPHECOMYRMA AND NEW JERSEY AMBER Urbana) to several of the amber localities in The first Mesozoic ant to be discovered Sayreville and adjacent towns in New Jersey, and described was Sphecomyrma freyi Wil- where they collected samples. Langenheim 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 3 had worked on the stratigraphy of late Oli- deep deposits of the South Amboy Fire Clay gocene amber from Mexico and Cretaceous (Tlronian). The peats are the stranded re- amber from Alaska. No results were reported mains of deltaic deposits, probably formed from those samples from New Jersey. From by coastal cedar swamps. Inclusions of leafy 1986-1989 the senior author resumed the shoots and fibrous bark indicate that the bo- collection of amber from various Cretaceous tanical origin of the amber is not Araucan- exposures in New Jersey, some of them vis- aceae, as originally proposed based on chem- ited 20 years earlier by Case. The samples istry (Langenheim, 1969; Grimaldi et al., collected by Grimaldi were used in a strati- 1989), but in the . A detailed graphic and chemical study of the amber account of the stratigraphy, taphonomy, and (Grimaldi et al., 1989), and only one piece paleoecology of the amber sites is provided was found to contain an inclusion (a elsewhere (Grimaldi, 1997). partial midge). By 1986, some sites had been Approximately 70% of the amber is vir- developed and were no longer accessible. tually opaque, due to a thick suspension of The bluffs at Cliffwood Beach- locality particulate debris and bubbles. The transpar- for Sphecomyrma freyi-for example, were ent pieces are mostly very brittle and fracture covered in 1974-75 with boulders and a skin easily. To trim and polish a window in this of cyclone steel fencing onto which concrete amber with an orientation crucial for obser- was poured. All collecting done in New Jer- vation, a technique of vacuum-embedding sey up until 1989 uncovered small amounts needed to be developed. The technique is de- of amber, and the likelihood of discovering scribed in more detail elsewhere (Silverstein pieces with insects seemed so remote that and Nascimbene, 1997), but involves embed- prospecting was then abandoned. ding the amber in a synthetic casting resin of In 1992 Gerard Case brought to Grimaldi's low viscosity and refractive index (see also attention two remarkable new deposits from appendix). While the resin is still fluid, the central New Jersey rich in fossiliferous Cre- preparation is subjected to a reduced pressure taceous amber. They lay between some of the of approximately 4 psi in a bell jar attached sites prospected earlier that were only a few to a vacuum pump. This removes air from miles apart. Exact locations and maps are the fine cracks and allows resin to seep in. available through the senior author, but they Filling cracks not only cements the fragile correspond most closely to site number five piece, but allows one to see through cracks on the map in Grimald et al. (1989), approx- where the air would otherwise cause a crack imately 7 km NW of Cliffwood Beach. Since to be like a small mirror. November 1992, excavations by dedicated Work on the manuscript was as follows: AMNH volunteers (see below) have amassed Grimaldi did the final trims and polishing of several hundred pounds of amber and hun- specimens for observation, did sketches and dreds of inclusions. Among these inclusions final drawings, photographs, and wrote the were the additional specimens treated here. manuscript, including descriptions. Agosti The new material allows us to more carefully also examined morphology, checked the consider the phylogenetic position of the sketches and descriptions, and worked with sphecomyrmines. Although amber has been Carpenter on the cladistics. known from New Jersey and other places in the Atlantic Coastal Plain for at least 150 ACKNOWLEDGMENTS years (reviewed in Grimaldi et al., 1989), it was the 1966 discovery of Sphecomyrrna that It is a pleasure to acknowledge the hard brought serious attention to the substance. work of Keith Luzzi, Gerard Case, and Paul Nascimbene, the collectors of the specimens MATERIALS AND METHODS reported here, who generously donated their specimens to the AMNH. A wonderful spec- Specimens were excavated with hundreds imen-the holotype of Brownimecia clava- of other inclusions in approximately 80 kg of ta-was collected by Yale Goldman, who raw amber. The amber occurred in veins of provided it for purchase. In addition, Henry compacted, lignitic peat, just above the very Silverstein developed the vacuum-embed- AMERICAN MUSEUM NOVITATES NO. 3208

ding technique for the amber laboratory, and Sphecomyrminae), which was a classification Paul Nascimbene did many embeddings and adopted yet again by Dlussky (1996). As we further refined the technique. Vladimir discuss below, until more completely pre- Ovtsharenko kindly provided some Russian served material becomes available, it is most translations; and Stefan Cover loaned the ho- prudent to regard the Armaniidae as Formi- lotype of Sphecomyrma freyi from the MCZ, coidea incertae sedis. which tragically disintegrated during the em- bedding process (see appendix). Funding of THE MESozoIc FORMICOIDS the research was made possible by a grant from Henry G. Walter, trustee of the AMNH Sphecomyrma freyi Wilson and Brown, and by generous donations from Henry and 1967. Described on the basis of two well- Meryl Silverstein. The original manuscript preserved workers in a piece of amber from was reviewed by Barry Bolton, Bill Brown, Cliffwood Beach, New Jersey. It was origi- E. 0. Wilson, and especially detailed and nally thought (Wilson et al., 1967a, 1967b) helpful comments were provided by Phil that this amber was Cenomanian (ca. 100 Ward. To all we owe great thanks. Ma), then late Santonian, ca. 80 Ma (Wilson, 1985; HIlldobler and Wilson, 1990) (this THE SPHECOMYRMINAE AND was a date quoted by Dlussky, 1975). Dlus- MESOZOIC "FORMICOIDEA" sky (1983: p. 63 of Paleontol. J. translation) later mentioned that "in the past few years A synopsis of all taxa described as ants they [the New Jersey amber deposits] have and close relatives from the Mesozoic, in the been redated as Santonian or even early in which they were reported, is given ," which is erroneous. Like the below. Dlussky (1975, 1983, 1987) described material we are reporting, the type material 16 species and 11 genera from the Creta- of S. freyi is probably Turonian (90-94 Ma). ceous of Russia and Kazakhstan, based on 30 In the original paper describing the species compression and inclusions in amber. (Wilson et al., 1967b), they mentioned the The Armaniidae was described for piece of amber as being in the private col- several new "formicoid" forms: Archaeo- lection of Edmund Frey, of Mountainside, pone, Armania, Armaniella, Poneropterus, New Jersey. It was donated to the Museum and Pseudarmania. Dlussky (1975) original- of Comparative Zoology, Harvard, soon after ly kept the Sphecomyrminae as a subfamily publication of the paper, residing in the fossil of the Formicidae, but later (Dlussky, 1983) insect collection. The specimen disintegrated elevated Sphecomyrminae to family, main- during the vacuum-embedding process, for taining that the primitive condition of two which a neotype is below. :eeth on the mandible and an with a designated, ;hort scape and long, flexible funiculus ex- Cretomyrma arnoldii Dlussky, 1975: 116. Aluded them from the Formicidae. Wilson Known from a partial worker in a piece of (1987), using a morphometric analysis, amber from Yantardakh, Taymyr Peninsula, showed that Dlussky's genera are possibly al- northem Siberia (Paleontological Institute lometric variations of the gynes and workers Nauka [PIN] number 3130/113). It is largely for only two taxa. He placed Dlussky's gen- headless (only imprints of some mouthparts era into two genera: Sphecomyrma and Cre- and the second segment of the antenna re- tomyrma. In lieu of reexamining original ma- main), but details of petiole, gaster, and legs terial, as E. 0. Wilson did for some taxa, we are well described. This deposit is dated as will not address the identity of Armaniidae Santonian. Cretomyrma was placed by Dlus- here, but our results do bear on the identities sky (1975; 1987) in the Sphecomyrminae/- and relationships of the genera that Dlussky idae. placed in the Sphecomyrminae/-idae: Baiku- ris, Cretomyrma, Paleomyrmex (later re- Cretomyrma unicornis Dlussky, 1975: named Dlusskyidris), and Sphecomyrma. 116. A worker specimen even more partial Bolton (1994) placed the Armaniidae as a than C. arnoldii, with just a portion of the subfamily of the Formicidae (along with the , petiole, gaster, and the entire left 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 5 middle leg preserved. From the same deposit posit as previous taxa). Like Archaeopone, as C. arnoldii, PIN 3311/363. Dlussky originally placed this unique speci- Dlusskyidris zherichini (Dlussky), 1975: men (PIN 2383/145) and a similar one (de- 118. Dlusskyidris Bolton, 1994, is a replace- scribed as Dolichomyrma? latipes Dlussky, ment name for Paleomyrmex Dlussky (pre- 1975 [PIN 2383/144]) in the Ponerinae, then occupied by Paleomyrmex Heer, 1865). in the Armaniidae. Carpenter (1992) men- Known from a virtually complete male (ho- tioned this taxon also to be of doubtful as- lotype, PIN 3311/364) and two incomplete signment as a formicid. The type and only males (PIN 3311/365 and 3311/366) in Si- known specimen of D. longiceps has pro- berian amber. Placed by Dlussky (1975; portions of the head, alitrunk, and gaster sim- in ilar to that of a worker ant, and a possible 1987) the Sphecomyrminae/-idae. petiole is rendered in his figure of it, but no Petropone petiolata Dlussky, 1975: 119. other "diagnostic" features are obvious. Compression fossil of an apparent wingless specimen, possibly a worker, from the Cre- Annania robusta Dlussky, 1983. Based on taceous (Turonian) of southern Kazakhastan a large, winged female compressed in rocks [PIN 2783/158]. Dlussky actually placed this from the Al'skaya Formation (Cenomanian) (and two other genera, immediately below) of Ten'ki District, Magadan region, Russia. in the Ponerinae, but later (Dlussky, 1983: p. Holotype (PIN 3901/155) is in lateral view 64 Paleontol. J. translation) mentioned that showing a distinct, large petiole; paratype the "systematic position of these genera [Pe- (PIN 3901/160), also winged, is preserved tropone and Cretopone] must be changed dorsally with a petiole barely preserved. .... [but] remains unclear, because their im- Pseudarmania rasnitsyni Dlussky, 1983. pressions are incompletely preserved." Car- Known only from a single winged, well-pre- penter (1992) mentioned this taxon to be served female from the same deposit as Ar- only doubtfully assigned to the Formicidae. mania (PIN 3901/156). Head, alitrunk, and Cretopone magna Dlussky, 1975: 119. gaster, some legs, are well preserved, and Compression of a partial and badly distorted only basal portions of the wings. insect from the same deposit as Petropone. Dlussky (1983) removed it, and Petropone, Pseudarmania aberrans Dlussky, 1983: a from the Ponerinae to a genus incertae sedis for P. rasnitsyni, except that the unique spec in the Formicoidea. In our view, based on imen (PIN 3901/152) is missing the head an, Dlussky's (1975) illustration, it is impossible anterior portion of the alitrunk. to assign the specimen to any aculeate fam- Armaniella curiosa Dlussky, 1983. De- ily. scribed on basis of a single, incomplete fe- Archaeopone kzylzharica Dlussky, 1975: male (PIN 3901/158) from the Al'skaya For- 121. Based on a single, incomplete compres- mation of Ten'ki, Magadan. Portions of the sion fossil from the Cretaceous of Kazakha- head, antenna, wing, legs, gaster, and the pet- stan (same deposit as previous two taxa) iole are preserved. (PIN 2383/263). Originally placed in the Po- nerinae (Dlussky, 1975), it was later placed Archaeopone taylori Dlussky, 1983. Based in the Armaniidae (Dlussky, 1983). Carpen- on two specimens from the Al'skaya For- ter (1992) gave this taxon a doubtful assign- mation of Magadan. The holotype (PIN ment to the Formicidae, with which we must 3901/154) is an apparent male with the head, concur. As illustrated (Dlussky, 1975, p. 120, antenna, alitrunk, anterior half of gaster, and fig. 132), the specimen is headless and vir- portions of wing preserved. tually legless. The alitrunk and gaster are also shows but without Poneropterus sphecoides Dlussky, 1983: preserved; a petiole Complete apparent male specimen (PIN an obvious node. 3901/157) from the Al'skaya Formation of Dolichomyrma longiceps Dlussky, 1975: Magadan. Virtually entire body preserved, in 121. Cretaceous of Kazakhastan (same de- dorsal aspect. 6 AMERICAN MUSEUM NOVITATES NO. 3208

Sphecomyrma canadensis Wilson, 1985. from an incomplete male in a piece of amber Known from a well-preserved worker (ho- from the same locality as B. mandibularis lotype) and poorly preserved one in amber (PIN 3730/8). from near Medicine Hat, Alberta, Canada, numbers 330 (holotype) and 205 (paratype) Cariridris bipetiolata Brandao and Mar- in Biosystematics Research Institute, Agri- tins-Neto, 1989: Based on an apterous spec- culture Canada, Ottawa. These specimens imen in Lower Cretaceous (Santana Forma- were from collections made by McAlpine tion, Aptian, ca. 110 Ma) limestone from and Martin (1966), who also collected Cre- Ceara, Brazil, and interpreted as a worker. taceous amber from Cedar Lake, Manitoba. This specimen has a definite, narrow and It is generally believed that the amber from elongate petiole, and a scape that appears Medicine Hat and the redeposited amber slightly longer than in Sphecomyrma or Cre- from Cedar Lake is the same age as amber tomyrma workers; but the bases of the anten- from Grassy Lake, Alberta, dated in situ (ac- nae, on the other hand, are obscured. Ex- cording to Pike, 1995) as from the upper part amination by DG of an excellent color slide of the Judith River Group of the Foremost of the specimen indicates that the proposed Formation, ca. 78-79 Ma. existence of a postpetiole in the specimen (a Pike (1995) reported three additional ants, constriction just anterior to true abdominal identified as "Sphecomyrna sp.," in the am- segment IV) may be a preservational artifact. ber from Grassy Lake. These are in the Royal Prior and continuing studies of Santana For- Tyrrell Museum, Drumheller, Alberta. The mation insects (e.g., Grimaldi, 1990) docu- specimens were examined by E. 0. Wilson, ment excellent and detailed preservation of two of them being Sphecomynna [canaden- external cuticles, the study of which requires sis?], the other being a poorly preserved proper physical cleaning and removal of ma- worker with mandibles and antenna more trix with acetic acid digestion. There are sub- highly derived than in Sphecomyrma (Pike, stantial areas of the specimen still uncleaned, personal commun. to DG, Jan. 1997). with limestone matrix still covering the pro- notum, bases of the legs, all of the mid left Cretacoformica explicata Jell and Duncan, leg, and margins of the alitrunk and petiole. 1986. Based on (an) alate specimen(s) from What is interpreted as petiole and gastric the Lower Cretaceous of Koonwarra, Victo- constriction could be an appearance due to ria, Australia, in the Aptian (ca. 115 Ma) the overburden of matrix in these areas. As Koonwarra Fossil Bed Formation. The spec- we discuss later, placement of Cariridris as imen is not well preserved. For example, the possibly in the Myrmiciinae (Brandao et al., petiole, if present, is obscured. Naumann 1990) is consistent with the basal phyloge- (1993) reexamined the unique specimen (no. netic position of this subfamily. However, NMVP 102501A,B in Museum of Victoria, since mandibular dentition, calcars, and pres- Melbourne), and presented a detailed, cor- ence of a metapleural gland cannot be ob- rected description. He could not place it to a served, placement of Cariridris in the family family of , but discussed features Formicidae cannot be confirmed with present that were both consistent and inconsistent material. with the Diapriidae, as proposed by Darling and Sharkey (1990). The genus was cited as Lastly, Poinar (1992: 203) published an Cretaceoformica [sic] in Poinar (1992). unclear photograph of an aculeate in Leba- nese amber (Neocomian) from the Acra Col- Baikuris mandibularis Dlussky, 1987. lection, as "The earliest known ant?" That Known from three males in a piece of amber specimen is preserved very well and was ex- from Lake Taymyr, Baykura-Neru Bay, Tay- amined in 1990 by the senior author. Al- myr Peninsula of Siberia, Russia (amber though an aculeate, it is definitely not For- piece is PIN 3730/5 in the Paleontological micoidea. Institute, Moscow). This amber is dated as Santonian. Using a criterion of just the presence of a petiole-a feature easily fossilized even as a Baikuris mirabilis Dlussky, 1987. Known compression-there is certainty in the iden- 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 7 tity of the following genera being formi- sey amber described below. We mostly agree coids: Armania, Amaniella, Archaeopone with the classification of Wilson (1987), in taylori, Cariridris, Dolichomyrma, Poner- that Sphecomyrminae are the most primitive opterus, and Pseudarmania. Wilson (1987) and oldest known subfamily of the Formici- reexamined Dlussky's holotypes of Annania dae (see cladistic results, below). However, robusta and Pseudarmania rasnitsyni, and we are unprepared for the formal synonymies found that a trochantellus was actually ab- of Dlussky's genera made by Wilson, without sent, as is the case in all living Formicidae. the basis of definitive synapomorphies. Dlussky used its apparent primitive presence The following is a proposed summary to exclude the Armaniidae from the formic- classification of Cretaceous aculeate fossils ids. It is an intriguing character, since it oc- that have been described as convincing ants curs as a ground plan of the , the or close ant relatives: (with ) to the Formici- dae (Brothers, 1975; Brothers and Carpenter, APOCRITA, incertae sedis 1993; Rasnitsyn, 1988). The character, how- Cretacoformica ever, seems as if it would be exceedingly dif- , incertae sedis ficult to resolve in a compression fossil, since Cretopone it is a segment largely fused to the base of Petropone the femur, demarcated merely by a fine su- Superfamily FORMICOIDEA ture. A trochantellus definitely occurs in at Family ARMANIIDAE (monophyly least the New Jersey amber specimens, mak- uncertain) ing it plausible that the Armaniidae also pos- Arrnania sessed it. In Dlussky's diagnosis of the Ar- Armaniella maniidae he mentioned the broad attachment Archaeopone of the petiole to the gaster (a plesiomorphic Dolichomyrma feature), which Wilson also critiqued. If the Poneropterus large gasters of the winged females described Pseudarmania by Dlussky indicate that they are indeed Family uncertain queens, as argued by Wilson, the most par- Cariridris simonious argument is that the broadly at- Family FORMICIDAE (true ants) tached petiole is a sexually dimorphic fea- Subfamily Sphecomyrminae ture, as in living Formicidae. Lack of pre- ?Baikuris (d d only) served details like dentition of the mandible Cretomyrma and presence of a metapleural gland pre- Dlusskyidris cludes a more detailed placement of these Sphecomyrma taxa, at least with the present material. ?Haidomyrmex (: Incomplete and/or poor preservation of Tertiary?) Archaeopone kzylzharica, Cretacoformica, Subfamily Ponerinae Cretopone, and Petropone (as described by Brownimecia Dlussky) makes their identities even as for- micoids questionable in our view. This FAMILY FORMICIDAE leaves the four genera described in Creta- SUBFAMILY SPHECOMYRMINAE ceous amber from Siberia, Canada, and New Jersey: Sphecomyrma, Cretomyrma, Baiku- INCLUDED GENERA: Baikuris, Cretomyrna, ris, and Dlusskyidris. The latter two genera Dlusskyidris, and Sphecomyrna. are known only from males, and metapleural All features diagnostic for the subfamily glands cannot be observed. The first two as given by Wilson et al. (1967b) and Dlus- genera are known from workers where the sky (1975, 1983, 1987) are symplesiomor- metapleural gland can be observed, making phic at the level of the family: worker with these the oldest true ants, family Formicidae short mandibles with only two teeth; anten- (see below, for discussion of ). The nae with short scape and long, flexible fu- oldest definitive Formicidae is still Spheco- niculus; low, rounded petiole; presence of myrma, along with the new taxa in New Jer- metapleural gland; an unconstricted gaster; AMERICAN MUSEUM NOVITATES NO. 3208 extrusible sting; mid and hind tibiae with proportions and social existence in aculeates. double spurs. Using 37 genera in various families, he de- The only synapomorphies we could find vised several indices of the antennae, which for the group are: presence of a short pedicel, for purposes here are best represented by his which has a length slightly longer than its IK1 index (length of scape/total length of an- width; a second funicular segment (first fla- tenna). This index and the other two were gellar segment) twice or more the length of divided into three categories: one represented other segments (an exception would be Sphe- by the parasitoid aculeates (IK1 = 10-25%), comyrma canadensis, see below); and males the second by the solitary provisioning without a distal, evanescent segment of wing wasps and bees (IK1 = 17-35%), and the vein CuA (CuAl). All three features are third by the social bees, ants, and a few sol- unique within the Formicidae. However, the itary bees (IK1 = 30-40%). In lieu of seeing short pedicel also appears in some Vespidae, the complete data, it is reasonable to assume so it is possibly plesiomorphic. these are arbitrary groupings. Statistics on Dlussky removed sphecomyrmines from ANOVA or confidence limits were not given, the Formicidae, as mentioned above, based and the great overlap in the ranges would on the structure of the antenna and the man- suggest that the groupings are not significant. dibles. Specifically, the antenna has a first We do not want to belabor the antennal segment too short for the antenna to be suf- index argument, because the manipulation of ficiently elbowed-a condition necessary, the antennae in any sphecomyrmine will nev- Dlussky maintained, for eusociality, since it er be observed, only inferred. For Dlussky allows brood and food manipulation. If not (1983) to claim that "[sphecomyrmines] social, Dlussky concluded, then Sphecomyr- could not have transported their brood or en- ma and relatives cannot be ants. Although he tered into trophallaxis with their larvae," is did not cite studies that address how ants ma- thus speculation at this point. As Wilson nipulate objects with their antennae, Dlussky (1987) pointed out, the eusocial wasps have presumably meant that the apices of the an- antennal proportions similar to those of the tennae must not be too distant from the man- sphecomyrmines, and this obviously hasn't dibles, for optimal manipulation of food and prevented them from feeding their larvae. In tactile communication ("antennation"). In- fact, an extremely short antennal scape is deed, antennation is a behavior conunonly present in various leptanilloidine ants, which used for such social activities as trophallaxis, are eusocial like other ants. Morphology of although (as Wilson [1987] pointed out), sphecomyrmines is not at all inconsistent trophallaxis may not occur among workers with a social existence, but perhaps the only of many primitive ants such as Nothomyr- way to confirm would be the dis- mecia, Amblyoponini, Apomyrminae and covery of part of a fossilized colony. Por- even some advanced species like Pogono- tions of fossilized colonies occur, rarely, in myrnex badius (reviewed in Holldobler and younger, Tertiary ambers, such as with some Wilson, 1990). Workers of some ants merely dolichoderines like Azteca, and another (un- transport larvae to cached prey and do not described) genus in amber from the engage in trophallaxis. Dominican Republic (Waggensberg et al., Evidence in favor of Dlussky's argument, 1996). Wilson's (1987) morphometric evi- however, is the fact that the proportions of dence on the relative size of the gaster is the scape and funiculus appear to be grossly strongly supportive of castes in the Creta- maintained throughout the ants. For example, ceous formicoids. there is a convergent development in ants Dlussky probably underestimated the sig- with extremely long, gaff-shaped mandibles nificance of the metapleural gland, a feature and a corresponding lengthening of the scape agreed even by Dlussky to be in sphecomyr- (e.g., Acanthognathus, , Myr- mines. The metapleural gland, which pro- mecia, and Myrnomoteras). This suggests a duces large quantities of microcidal and fun- functional constraint in proportions of the an- gistatic phenylacetic acid, is unique to ants. tennae and mandibles. Dlussky attempted to It presumably acts as a disinfectant, and al- quantify the relationship between antennal lows nesting in soil/leaf litter without infec- 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 9 tion of the brood and colony (Holldobler and Genus Cretomyrma Dlussky Wilson, 1990), which explains its loss in tree-nesting genera (Oecophylla, Polyrachis, DIAGNOSIS (extracted and translated from Dendromyrmex, and most Camponotus) and Dlussky, 1975, 1983): Known only from some socially parasitic ants. Despite claims workers. Propodeum: low; with single, tooth- by Dlussky (1987) that sphecomyrmines like process; spiracles circular; differs from were solitary or semicommunal, the meta- Sphecomyrma on the basis of a petiole that pleural gland is almost certainly evidence of is lower and rounder, with anterior and pos- a social existence. terior connections that are less constricted. Brandao et al. (1990) stated that the po- Legs elongate, with calcars on mid and hind sition of sphecomyrmines as ancestral/prim- tibiae. Mandibles with "additional tooth on itive to all Formicidae " . .. was challenged top." by Dlussky [sic] 1987, with whom we con- One character mentioned by Dlussky, the cur." They then concluded that Cariridris is size of the sting, is questionable, since the the oldest formicid on the basis of its differ- extent that it is extruded will affect estimates entiated scape and the presence of a petiole. of its size. Baroni-Urbani (1988) (and as cited in Poinar, TYPE SPECIES: C. arnoldii Dlussky. 1992) also offered a conclusion similar to that of Dlussky, that Sphecomyrma could not Genus Dlusskyidris Bolton be an ant because of the short basal antennal Paleomyrmex Dlussky, 1975: 118 (preoccupied article, and "the apparent absence of a meta- by Paleomyrmex Heer, 1865: 91). pleural gland." This is one reason why Bar- Dlusskyidris Bolton, 1994: 187 (replacement oni-Urbani et al. (1992) did not include name). Sphecomyrma in their phylogenetic analysis DIAGNOSIS (extracted and translated from of ant tribes (personal commun. CBU to DG Dlussky, 1975): Known only on basis of and DA). Our new material definitively male. Mid and hind tibia with calcars. Wing proves the existence of the gland. Also, Ba- venation complete and primitive, with closed roni Urbani and de Andrade (1994: p48) cells lr + 2r, 3r, 2rm and lmcu [Dlussky's commented that "only one probably true ant terminology]. Subgenital plate triangular, is known from the lower Cretaceous [Car- reaching valvulae. Last tergite of abdomen iridris] ... other presumed Cretaceous For- triangular, with well-developed cerci. Para- micidae are currently regarded as non-for- meres narrow, pointed, and slightly curved micid Aculeates." As discussed above, crude medially. preservation of Cariridris in limestone pre- Dlussky mentioned that, on the basis of vents observation of critical details. hardly being able to associate males and The other reason given by Baroni Urbani workers in ants, it is "quite possible that Pa- et al. (1992) for not including sphecomyrmi- leomyrnex are males of Sphecomyrna or nes in their ant cladogram was that "many of Cretomyrma." the characters ... cannot be assessed ..." TYPE SPECIES: D. zherikhini (Dlussky) Oddly, they included the Formiciinae, an ex- tinct subfamily described from compressed Genus Sphecomyrma Wilson and Brown remains in Eocene oil shales of Messel, Ger- many (Lutz, 1988), the preservation of which DIAGNOSIS: Scape short, 0.57-0.68 mm; is hardly as fine as that in amber. Indeed, only funiculus long and filiform, approximately 20% of the characters in their matrix can be four times the length of scape; pro/mesotho- scored for the Formiciinae, but slightly over racic suture complete and well formed; pet- 40% can be scored for Sphecomyrma workers iole with distinct, domed node widely sepa- (67% if characters of the apparent male are rated from propodeum and gaster by deep included, but which were unavailable to Bar- constrictions; cuticle without sculpturing, oni Urbani et al. [1992]). This probably ex- just superficially microscopic relief, with plains why the Formiciinae was the basalmost scattered and sparse setae. lineage in their analysis, which we discuss in Significant features not mentioned in the the cladistic section of this paper. original reports (Wilson et al., 1967a,b), 10 AMERICAN MUSEUM NOVITATES NO. 32081 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER

Fig. 1. Photomicrographs of Formicidae in Turonian amber from New Jersey. [Left page] a. Sphe- comyrma freyi, AMNH NJ-1 12. b. Male, Sphecomyrma? AMNH NJ-242. c. Male, Baikurus casei (ho- lotype), AMNH NJ-9Obb. [Above] d. Male, undetermined genus, AMNH NJ-107. e-f: Brownimecia clavata. e. Holotype. f. Paratype, AMNH NJ-231. 12 AMERICAN MUSEUM NOVITATES NO. 3208 which we report here based on AMNH NJ-1 12, are: helcium apparently lacking; tro- chantellus lacking; and a pair of long, ex- tremely fine setae present in the middle of the clypeus. TYPE SPECIES: S. freyi Wilson and Brown.

Sphecomyrma freyi Wilson and Brown Figures la, 2, 3 SPECIMEN (NEOTYPE): AMNH NJ-i 12, from the "Sunrise Landing" site, East Brunswick, New Jersey. Collected by Gerard Case, James Leggett, and Paul Borodin, 1994. The ant is a complete and beautifully preserved wingless female in a transparent, deep amber-colored piece that originally was half filled with debris. After embedding, the debris half was sliced off; now the piece is 18 X 14 X 5 mm, and contains some debris, bubbles, as well as one mite, six podurid Collembola, stellate trichomes, and portion of a spider. A piece of debris is lying in front of the head, obscuring a portion of the face. Deposited in the Museum of Comparative Zoology, Harvard University, in replacement for the original type specimen (see appen- dix). DESCRIPTION: Measurements of AMNH NJ- 112: total length 4.1 mm; head width 1.08; depth 0.28; scape length 0.57; fu- niculus length 2.02; alitrunk length 1.48; gas- Fig. 2. Newly discovered worker and neotype ter length 1.51. These measurements and oth- of Sphecomyrma freyi (AMNH NJ-1 12, now in er features agree in virtually all respects with the MCZ). Above. Detail of head, frontal view (at the detailed descriptions and illustrations in a slight ventrolateral angle). Total length of funic- Wilson et al. (1967a,b), to which we can add ulus and relative size of flagellomeres are distort- some observations on the metapleural gland ed, since the antenna is curled back and somewhat (MG). The MG was distinctly observed un- underneath the head. The left side of the head is der 60-lOOX better obscured by a large piece of debris. Note that magnifications, although short space between scape and pedicel is a mem- results were obtained using a Leitz stereo- branous area, not an antennomere. Below. Detail scope at 144X and different angles of fiber of pretarsal claws. optic illumination. Best results were obtained by examining the propodeum at 160X and 400X using a Zeiss compound microscope are mistaken about the presence of the meta- and reflected fiber optic illumination. This pleural gland, it would be possible to con- method of examination absolutely requires a sider Sphecomynna as a wasp rather than an flat amber surface parallel to and very close ant." to the surface of the structure, as in the prep- The metapleural gland in AMNH NJ- 112 aration of AMNH NJ-1 12. In the original re- distinctly shows a wide oval orifice, virtually port by Wilson et al. (1967b: p. 11), there round, with a shallow lip extended antero- was some equivocation as to the presence of ventrad to it for a length of approximately 6 the gland, which perhaps explained why it to 7 diameters of the orifice. Distance of the was not described in any detail: " . . . if we orifice from the posterior margin of the pro- 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER

Posterior end -MGO

Fig. 3. Sphecomyrma freyi worker, AMNH NJ-1 12. Photomicrograph and illustrated rendering (to the same scale) of the alitrunk and anterior portion of the gaster, showing details of the metapleural gland opening [MGO] (arrows) and petiole. The cuticle of the bulla appears thinner than surrounding cuticle and, as shown in the photo, this area is actually collapsed into a slight depression. The bulla was rendered in the drawing as it might have appeared: slightly bulging. Subcuticular detail beneath the MGO is also shown, which is probably the atrium of the gland, with remnants of the glandular tissue to one corner. Scale = 0.1 mm. AMERICAN MUSEUM NOVITATES NO. 3208

Fig. 4. Male, AMNH NJ-242, probably Sphecomyrma. Scale = 0.5 mm. podeum is approximately twice the diameter (which is about as long as the second anten- of the orifice. The cuticle of the area anterior nal segment; in freyi the third segment is to the orifice is quite transparent and shows slightly more than twice the length of seg- little sclerotization, suggesting it to be con- ment two). siderably thinner than surrounding cuticle of the alitrunk. What was probably a slightly Sphecomyrma? sp. raised area (the bulla) is now a shallow con- Figures lb, 4, 12a cavity, perhaps compressed because of the SPECIMEN: Male, AMNH NJ-242, from thinner cuticle in this area. Consequently, an White Oaks site, Sayreville, New Jersey, col- irregularly shaped chamber is visible beneath lected by Keith Luzzi, 1995. A superbly pre- the cuticle, probably the atrium for the se- served, complete, and well-displayed speci- cretions of the metapleural gland. In the an- men in a piece of amber that was originally terodorsal corner of the lightly sclerotized cylindrical and 5 mm diameter, 10 mm long, subcuticular chamber is a whitish mass with with the ant occupying almost the entire di- four small lobes: perhaps the gland itself, ameter. The amber is slightly turbid with shrunken from its normal size. minute particles and bubbles, and contains a few fine cracks. The piece was vacuum-em- Sphecomyrma canadensis Wilson bedded and sliced lengthwise to remove the The only other sphecomyrmine known debris-filled half and improve visibility of from can be distinguished the wings and venation. from S. freyi (according to Wilson [1985]) DESCRIPTION: Total length 3.13 mm. HEAD: based on the workers with a more robust With three large ocelli on vertex, but not on trunk and a shorter third antennal segment tubercle(s); pair of fine, stiff hairs on frons, 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 15 vertex, and postocciput, with finer hairs in- which is based mostly on workers. Without terspersed. Face with two pairs of longer, fine males taken directly from a colony, assign- setae and several smaller ones. Eye reniform, ment of them to a particular species or even with middle of hind margin emarginate, in- genus is rarely done. Nonetheless, some fea- dented to a depth of about 4 facets. Eye tures allow us to tentatively associate this large, its depth 0.6x depth of head; width of male with Sphecomyrma workers: similar eye 0.72X depth of eye. Supra-antennal su- size; similarity in structure of the petiole, ture present, extended to level of anterior which has a pair of fine, long setae on the ocellus. No microtrichia on head. Antennal apex of the node (albeit widespread in many proportions: scape short, approximately same ants); lack of a trochantellus; and the pro- length as flagellomere I (funicular segment portions of the basal segments of the anten- II); pedicel very short, ca. 0.3 x length of fla- na. This male is certainly not Baikurus or the gellomere I. Flagellomere I 1.1-1 .2X length same genus as AMNH NJ-23 1. of longest distal funicular segments. Antenna with 11 flagellomeres, densely covered with GENUS Baikuris Dlussky fine microtrichia, total length of antenna 2.6 DIAGNOSIS: According to Dlussky mm. Mouthparts largely covered with bub- (1987): Only males known, having bles, with two palpal segments protruding. kidney-shaped; mandibles narrow, margins ALITRUNK: 1.25 mm long, length of petiole virtually parallel and without teeth; base of 0.35 mm. Petiole covered with mricrotrichia, outer surface with oval area and ridge having pair of longer setae on apex of nodus, (opening of mandibular gland?); "lower and shorter setae posteriad. Depth of petiole jaw" palps long, 6-segmented; labral palps under node 1.5>X depth of where it is at- evidently 3-segmented; scutum with deep tached to trunk or gaster. Trunk largely cov- parapsidal grooves; forewing with cell lr + ered with fine microtrichia, nota with scat- 2r (Dlussky terminology) partly divided by tered, larger setae. Legs distal to and includ- rudimentary crossvein; mid and hind legs ing femora with microtrichia in even, longi- with small trochantellus; tibial spurs 2, 2, tudinal rows. Mid and hind tibia each with a 2; claws with preapical tooth; petiole dis- pair of setose, apical spurs; one on the hind tinct, with small node; indistinct constric- tibia is pectinate. Propodeal spiracle slitlike, tion between first and second segment of with small lobe covering part of orifice. abdomen [sic?] [gaster]; genitalia small, re- WINGS: Forewing length 2.25 mm. Hind- tracted into abdominal segments; cerci well wing 0.7x length of forewing, shapes typi- developed. cally formicid, with hindwing having more acute apex and angulate posteroapical mar- Baikurus casei, new species gin. Distinctive features of forewing are: cos- Figures lc, 5a-d, 12b tal vein thickened beyond pterostigma, ta- pering to nearly wing apex; vein R + Sc with DIAGNOSIS: Known only from male spec- small spur just before it meets with ptero- imens, and distinguished from other males of stigma; discoidal cell narrow, with distinctive genus on basis of eyes not being reniform veins connecting it with R + Sc and M + and there being no complete or partial cross- CuA. Hindwing with 7 fine, long distal ham- vein between cell lr + 2r. Otherwise, simi- uli, no anterior hamuli; basal part of vein M larities with Baikurus are striking: mandibles straight, distal spur virtually lost; vein Cu have no or virtually no teeth; labial palps are fairly long, nearly reaching wing margin. long, 6 segmented; alitrunk with deep pa- GASTER: Length 1.22 mm; segments 1 and rapsidal grooves; small trochantellus present 2 equal in length, comprising approximately on mid and hind legs; petiole elongate, as in 2/3 of gaster. Terminalia largely obscured by Baikurus mirabilis; and genitalia very simi- layer of froth, with two pairs of setose lobes lar. protruding (cerci?). Anterior pair (parameres) DESCRIPTION: Males large, 7.5 mm body twice as broad as posterior ones. length, forewing length 5.0 mm. DIscUSSION: Genera of ants are not rou- HEAD: With three large ocelli, hind pair on tinely surveyed for males, the systematics of low tubercles; head bare of setae. Eyes 16 AMERICAN MUSEUM NOVITATES NO. 3208

a

c 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 17 large, oval (not reniform or emarginate), 1996) (details of the piece are presented depth ca. 0.6x depth of head. No supra-an- elsewhere [Grimaldi, 1997]). A large sliver tennal suture seen. Antennal proportions: from the entire piece, measuring 30 X 30 scape short, 0.5X length of flagellomere 1 x 8 mm, was vacuum-embedded, and the (antennal segment 3); pedicel very small, portion with the ant isolated by cutting out O.45X length of scape; antennal segment 3 a pielike slice. One surface of the slice with slightly longer than immediately distal one, the ant is irregular and coarsely ground, re- segments increasingly shorter distad, nar- quiring glycerine with a glass coverslip on rowly tapered at apex. Antenna with 11 fla- top for viewing the inclusion. Observing gellomeres. Labial palps long, 5 segments the wing venation requires viewing from visible; maxillary palps shorter, 4 segments both sides. visible. Mandibles barely toothed; clypeus ETYMOLOGY: Patronym in recognition of well developed. the collecting efforts of Gerard R. Case. ALITRUNK: Pleura somewhat obscured; OTHER SPECIMENS: NJ-90aa. A very par- legs long and slender, forelegs with one api- tial male in a thin, chipped, clear yellow cal tibial spur; mid and hind tibia with two piece 10 x 6 X 2 mm, which also contains apical tibial spurs; mid and hind coxa (best the remains of a very large insect unidenti- seen in NJ-9Oaa) with slight suture at prox- fiable to order. The dorsal portion of the imal end of femur (the trochantellus). Peti- ant's alitrunk is lost at the surface of the ole long and slender: in NJ-90bb length is amber, the head is completely gone, and 2.3X the width, in NJ-9Oaa length is only only one-third of the wing is preserved. 1.7 x width, difference possibly due to an- However, the entire fore and mid legs are gles of observation. WINGS: Forewing with well preserved and the hind legs up to the vein C not reaching pterostigma, vein R ta- femur. The petiole and base of the gaster are pering out midway between pterostigma and well preserved, and have proportions very apex of Rs. Vein M + lm-cu with slight similar to that of specimen NJ-9Obb. That, bend in middle, but no extra crossvein in- tersecting submarginal cell- 1. Discal cell plus the fact that both specimens came from and subdiscal cell- 1 trapezoidal. Hindwing the same large piece, makes it almost certain with apical half of posterior margin acute they are the same species, perhaps even and very straight; 10 minute apical hamuli, males from a swarm. no proximal hamuli. DIsCUSSION: Wilson et al. (1967b) dis- GASTER: First segment 1.2-1.3X length of cussed the similarities between Sphecomyr- petiole, with slight constriction between gas- ma and amblyoponines, the latter being the ter segments 1 and 2. Genitalia best seen in most primitive or among the most primitive ventral view of NJ-9Obb, with two pairs of "poneroids," by virtue of the petiole being lobes apparent: anterior pair slightly clavate, broadly attached to the gaster (much more posterior pair with apices tapered. so than in Sphecomyrma or Cretomyrma). HOLOTYPE: AMNH NJ-90bb, a large, Interestingly, they mentioned that "one completely preserved male in a large, ex- character that has never received any par tensively fractured, clear yellow piece of ticular attention is the form of the ambly amber. Piece AMNH NJ-90 is one of the oponine male mandibles ...'" which ar( most remarkable pieces of New Jersey am- "narrow . .. sometimes bidentate, an( ber discovered thus far, containing 40 sometimes tapering to a single acute poin whole organism inclusions, including the ... [and] close tightly against the . . . clyp oldest fossil mushrooms (Hibbett et al., eal margin." This is the situation in Baik

Fig. 5. Baikurus casei, n. sp. a-c. Holotype (AMNH NJ-90bb). a. Left habitus, with a slightly ventral view, also showing details of genitalia. b. Oblique dorsal view of head. c. Oblique dorsal view of gaster and petiole. d. Incomplete specimen (AMNH NJ-9Oaa), showing petiole, legs, and portions of alitrunk and gaster. Scales = 0.5 mm. 18 AMERICAN MUSEUM NOVITATES NO. 3208 uris, which also has a slight constriction from B. casei and other Cretaceous ant between gaster segments 1 and 2. wings by the open discal cell- 1 (lacking the apical cross vein) and subdiscal cell-1 not GENUS? being trapezoidal but with the anterior vein Figures Id, 6, 13c curved down toward posterior vein. DISCUSSION: Although this male is distinct DESCRIPTION: Male, AMNH NJ-107: A from the Baikuris and probably Sphecomyr- virtually complete specimen in a small, ma males also in the deposit, we prefer not clear yellow chip of amber, 7 x 9 x 3 mm, to formally name it. Ant taxonomy presently missing distal portions of forelegs, anten- relies on diagnoses using worker characters, nae distal to first two segments. which is a criterion that should apply to ex- HEAD: Eyes large, oval, posterior margin tinct taxa until modem taxa are better sur- not emarginate. Position of specimen in the veyed for males. Genera named for isolated amber prevents measurements on proportions males and workers are very likely to lead to of the eye and head. Mouthparts largely ob- a confusing taxonomy. We describe a new scured by forthy coating, although two pairs species of Baikuris above, based on males, palps protrude, each with four visible seg- because this genus is presently diagnosed ments. Only three basal segments of right an- only by males. tenna remain, the pedicel being very short Unfortunately, critical features of the an- (ca. 0.3X length of scape). If portion of the tenna are lost that would confirm this ant right antenna distal to pedicel is a complete as a sphecomyrmine, which appears very first flagellomere, it is relatively short (ca. likely. It primitively possesses the trochan- 2.2X length of pedicel). tellus (Sphecomyrma lacks a trochantellus), ALITRUNK: Mesonotum with spare, short, and wing venation-although with a few erect hairs, cuticle otherwise devoid of mi- deviant features-is very similar to that of crotrichia. Scutellum with shallow trans- other male sphecomyrmines. The apparent verse trough on anterior margin, containing absence of a distal spur on vein CuA would row of 5 pits; pair of widely separated, fine also place this ant in the Sphecomyrminae. setae on dorsal surface. Mesothoracic spi- WING VENATION: Perhaps not surprisingly, racle recessed in deep, large pit. Propo- there is little comparative work on the wing deum large; spiracle centrally placed and venation in the living Formicidae. Ant sys- aperture not slitlike (no flap covers the ap- tematists have concentrated their collecting, erture), with pair of fine setae dorsally. and thus the systematic characters, on the ap- Posterior face of propodeum with very terous workers. One of the few studies on lightly rugose oval area, bordered by 8 comparative wing venation was done by erect fine setae. Petiole relatively short; Brown and Nutting (1949); on which many conically shaped; narrow and articulating of our comments on Cretaceous taxa are with propodeum, with a central ring of 8 based. fine setae. Where gaster articulates with In general, the venation of aculeates is petiole, short collar is formed on anterior rather conservative, particularly so in the margin of first gaster segment. LEGS: Fore- Formicidae. The discal/discoidal cell of the leg with apical tibial spur and proximal tar- forewing, for example is virtually constant, sal segment with basal "kink" and comb very rarely lost in ants. The second sub- of 8-10 fine setae that oppose spur (anten- marginal cell ("discal cell 2") is apo- nal cleaner). Mid and hind tibia each with morphically absent in some ants; the cell is pair of apical spurs, one spur half the present in many primitive ants and in the length of other. Mid and hind femora with Cretaceous taxa. But, valuable insight in short trochantellus at proximal end, faintly sphecomyrmine monophyly, for example, sutured from femur. Tarsal claws with small was provided by wing venation: loss of the preapical tooth. evanescent distal section of vein Cu-A WINGS: Shape of submarginal cell most (also termed Cu-Al). Conversely, wing ve- like that of Baikurus casei, with slight bend nation reinforces the primitive nature of in vein M + lm-cu. Easily distinguished sphecomyrmines. The presence of vein lr, 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 19

Fig. 6. Male, AMNH NJ-107: genus unknown. a. Detail of pretarsal claw. b. Detail of spurs/calcar at apex of tibia on fore and hind legs. c. Alitrunk, petiole, and anterior portion of gaster, dorsolateral view. Scale = 0.1 mm. 20 AMERICAN MUSEUM NOVITATES NO. 3208 for example, which bisects the submarginal gellomere 2.3. For AMNH NJ-667: pedicel cell, is seen in primitive, extant Myrmeci- 1.4: 1.8: 1.2:: 1:1:1:1.2: 1.2: 1.7 inae and some Ponerinae, as well as in the : apical flagellomere 3.2. Shortest funicular Cretaceous Baikuris mandibularis. Loss of segments are the third through sixth; sec- this vein occurs early in ant cladogenesis. ond funicular segment slightly swollen in Likewise, the absence of vein Mf2 (the dis- middle. Scape lost in AMNH NJ-23 1; in tal wall of the second submarginal cell) is AMNH NJ-667 scape is 0.4X length of fu- apomorphic, but the vein is present in Baik- niculus, 1.7X height of eye. Mouthparts in- uris, Dlusskyidris, and the apparent male tact only in NJ-667. Clypeus shallow and Sphecomyrma reported here. very broad, nearly width of head. Mandi- bles distinctive: apical half of left mandible Brownimecia, new genus protruding beyond clypeus (but best seen in lateral view of specimen), right mandible DIAGNOSIS: Distinguished from other Cre- adpressed against anterior margin of clyp- taceous worker ants by antenna (seen in both eus, but oral surface visible in anterodorsal specimens) and the mandibles (seen in view: mandibles are narrow, toothless, AMNH NJ-667). Antenna distinctly clubbed, scimitarshaped, crossing approximately at apical flagellomere twice the width of basal midlength, oral surface with about 30 short ones and pedicel. Mandibles long, thin, scim- and spiculelike setae. Basal portion of gena itar-shaped, strongly cruciate, without teeth with a lateral tooth. Palps long and thin: or crenulations, but with oral surface bearing labial palp with 4 segments (basal one min- ca. 30 short, spiculelike setae. Gaster with ute), maxillary palp with 2 visible seg- slight but definite constriction between first ments (probably more present). Fimbricate and second segment (abdominal segments III glossa present (observable only in lateral and IV). view). ETYMOLOGY: For the renowned system- ALITRUNK: Dorsally constricted at level of atist on ants, the late William L. Brown, Jr., middle of mesonotum; pronotum and meso- whom the senior author had the pleasure of notum with sparse microtrichia, mesonotum 'cnowing while a graduate student at Cornell with flat, oval-shaped dorsal surface. Metatho- Jniversity. Bill died suddenly on March 31, racic spiracle raised on small cone-shaped pro- 1997, several days after reviewing this manu- tuberance. Articulation between propodeum script. We are grateful he was able to learn and mesonotum largely fused (suture barely of the discoveries and see the patronym in visible). Propodeum with dome smooth and his honor. hairless, in lateral view helmet-shaped. Pro- TYPE SPECIES: B. clavata, new species. podeal spiracle situated in middle. Metapleural orifice Brownimecia clavata, new species gland obscured by froth of fine bubbles, Figures le, f; 7-11 but transverse ridge to MGO is visible (this is also seen in a scanning electron micrograph of DESCRIPTION: Body largely covered with the cavity of AMTvINH NJ-23 1, see below). Area microtrichia, sparsest on notum, absent on surrounding the MGO not conspicuously propodeum. raised into a bulla. Legs densely covered with HEAD: Large, width (excluding eyes) microtrichia, except coxae. Calcars on mid 1.6X greatest width of notum. Eye rela- and hind tibiae. Pretarsal claw with preapical tively small, length 0.3X length of head, tooth. Petiole with dome of node not hemi- virtually round, with about 100 ommatidia. spherical but slightly pointed, microtrichose, Ocelli not observed. Postocciput with re- with pair of fine setae near summit. Petiolar cessed, trapezoidal area bearing no micro- spiracle at basal third of petiole. Petiole nar- trichia. Antenna distinctively clubbed, with rowly attached to propodeum, broadly at- apical segment 2.OX width of basal flagel- tached to gaster. lar segments plus pedicel. Relative lengths GASTER: Helcium well developed, with of funicular segments: for AMNH NJ-23 1: slight crenulations on gastral edge (best pedicel 2.1 : flagellomere-I 1.6: 1.0: 1.1.: seen in SEMs). Abdominal segment 3 (gas- 1.3: 1.3: 1.1: 1.3 : 1.3: 1.6 :apical fla- ter segment 1) microsetose, 1.5 X length of 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 21 petiole, with slight constriction between vacuum embedded, trimmed close and paral- abdominal segments 3 and 4. Abdomen lost lel to the lateral and dorsal surfaces of the at surface of amber in AMNH NJ-23 1. In body, which exposed windows into the gold- AMNH NJ-667, tip of abdomen is obscured coated amber. Surprisingly, translucency of by a coating of froth, through which a sting the coated surface was retained. is discernible. ETYMOLOGY: In reference to the distinc- tively clubbed antennae. Measurements (in mm) DIscusSION: Brownimecia is not referable holotype paratype to the Sphecomyrminae, by virtue of its more derived proportions of antennal segments, the Body length 3.43 Head width pedicel being relatively longer and the first Scape length 0.52 flagellomere relatively shorter than in all the Funiculus length 1.02 other sphecomyrmine workers in Canadian, Alitrunk length 1.32 Siberian, and New Jersey ambers. The scape Gaster length 1.29 is much longer in proportion to the funiculus (1:2) than in the worker sphecomyrmines TYPES: Holotype, AMNH NJ-667, a su- (e.g., 1:4 in Sphecomyrma freyi). The man- perbly preserved, complete specimen in a dibles are distinctively autapomorphic, and clear yellow piece of amber. It was vacuum the clavate funiculus is found in most living embedded for better observation of certain ants, probably a groundplan feature of the liv- details. Collected by Yale Goldman at White ing ants. The slight gastric constriction indi- Oaks site. cates Brownimecia to be a primitive ponerine. Paratype, partial remains of a worker, Ward (1994) discussed many characters of AMNH NJ-231 in a tiny, oval, clear yellow primitive ponerines, and amblyoponines in piece, only 2 X 3 X 6 mm. We are grateful particular, which are useful in assessing the to Alan Graffam of Geological Enterprises phylogenetic position of Brownimecia. One (Ardmore, OK), who sent it to us, having ob- feature that he did not mention, however, was tained it from a private collector whose iden- genal spurs, which occur in some ambly- tity was not disclosed. Almost undoubtedly oponines like and Mystrium. this piece was collected at the White Oaks Other features that Brownimecia shares with site, since this is where virtually all New Jer- certain amblyoponines are: mandibles nar- sey amber collectors obtained their material. row, long, articulation points very widely Someone had obviously tumbled the piece separated, and mandibles extensively cross- since the surface is rather polished, and the ing (again, much like Amblyopone and Mys- ant is merely an exposed and clean cast on trium); eyes small; and the mesonotum short the flat surface of the piece. The front legs are and flattened. In virtually all other respects, virtually all gone, the mid and hind legs lost though, Brownimecia is extremely primitive distally from the middle of the tibiae. A cast compared to extant Amblyoponini or Poner- of the dorsal part of the head remains (in- inae: the genus lacks tergosternal fusion of cluding the ommatidia), but the ventral half abdominal segments III and IV, anterior mar- of the head is lost. Funiculus of left antenna gin of the clypeus is simple (without peglike is intact, but scape is largely lost (the right setae), and the eyes are situated near the mid- antenna is entirely lost). The entire alitrunk is dle of the sides of the head (not slightly pos- largely intact within the amber. Exposure of terior to the middle). Unfortunately, some the specimen is fortuitous, since it exposed characters discussed by Ward (1994) could casts of some critical areas and details, partic- not be observed in either specimen of ularly the petiole, which allowed viewing Brownimecia: metacoxal cavities, metapleu- with an SEM. The specimen was first gold ral gland orifice, abdominal sternum and ter- coated for SEM and examined at low voltage gum II, stridulatory file on abdominal (5 kV) using a Zeiss DSM-l scanning EM. IV, and details of the sting. Details of the exposed cuticle were observa- Very recently, Dlussky (1996) described a ble, the most significant being the propodeum, bizarre ant in the Sphecomyrminae. His re- petiole, and helcium. The specimen was then port is of exceptional interest because of the 22 AMERICAN MUSEUM NOVITATES NO. 3208

Fig. 7. Brownimecia clavata, n. gen., n. sp., paratype (AMNH NJ-23 1). Above, left lateral view. Right, dorsal view. Bubbles are emanating from spiracles and the metapleural gland orifice, obscuring the shapes of the apertures. Scale = 0.5 mm. 23 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER

I /

Fig. 7. (Continued) 24 AMERICAN MUSEUM NOVITATES NO. 3208

Figs. 8-10. Scanning electron micrographs of Brownimecia clavata, paratype AMNH NJ-231. 8: (above) Entire specimen, as exposed on surface of amber (compare fig. 7). Scale = 200 purm.

structure of the ant, and its preservation in rounded amber from northern Burma, which Burmese amber. Moreover, it had been sug- was mined for at least a millennium until the gested to us that perhaps Haidomyrmex Dlus- early quarter of the 20th century (Grimaldi, sky and Brownimecia were closely related, 1996). The only museum with a significant on the basis of the bizarre, sickle-shaped, collection of the material having inclusions toothless mandibles (they also both possess is the Natural History Museum, London. a narrow, elongate pronotal "neck," and Cockerell (1922), who originally studied the small eyes). NHM material, stated that the material was First, a great deal of mystique has sur- no younger than mid-Eocene, but, unfortu- 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 25

viding the ant is accurately placed (see be- low). Composition of the Burmese amber, based on visual inspection by D. Grimaldi and pyrolysis gas chromatography by Alex- ander Shedrinsky (unpubl.), suggests it to be Tertiary. Granted, there are no means pres- ently known that perfectly correlate amber's color, hardness, brittleness, or any other physical and chemical features of "matura- tion," with age. Nonetheless, all Cretaceous ambers are brittle and often highly fractured; Burmese amber is soft enough that it had been used for carvings, as early as 1000 A.D. and before Baltic amber became so available (Grimaldi, 1996). The general features of Burmese amber suggest a or early Eocene age. Secondly, the ant itself, Haidomyrmex, is startling: it possesses small eyes, huge genae, and huge, L-shaped mandibles. Depth and width of the mandible are approximately equal to those of the ant's cranium; the outer margin of the mandible has a squared angle, the inner margin being curved and toothless. These mandibles may have functioned in a manner analogous to that of other, living ants with "trap jaws," where long, fine hairs trig- ger the mandibles to suddenly snap closed. Haidomyrmex has a pair of such hairs on the margin of the clypeus. We doubt that this new genus is in th Sphecomyrminae, for the lack of ocelli at structure of the head, and a scape that is d tinctly shorter than any definitive, Cretacea sphecomyrmine. Its lack of a trochantell (cited by Dlussky) does not differ from ti situation in Sphecomyrma (reported here i the new worker of S. freyi). Clearly, Haido Figs. 9, 10. 9: Detail of propodeum, showing and Brownimecia are unrelated channels leading to the metapleural gland open- myrmex ing. 10: Petiole, showing crenulate helcium. based at the very least on proportions of the a constric- Scales = 50 ,um. antennae, apparent lack of gastral tion in Haidomyrmex, position of spiracles, and pilosity of the alitrunk. Haidomyrmex nately, the stratigraphy of Burmese amber may be a sphecomyrmine, but until more de- has never been reliably determined. Recent- finitive worker synapomorphies are discov- ly, Rasnitsyn (1996) surveyed 1200 arthro- ered for the subfamily, its exact position will pod inclusions in the NHM Burmese amber remain obscure. collection, and concluded, at least on the ba- sis of the Hymenoptera (including the ants described by Dlussky), that the amber was most likely upper Cretaceous in origin. A It is virtually self-evident that the fullest sphecomyrmine in Burmese amber would systematic value of fossils can be revealed suggest a Cretaceous age of this amber, pro- by an understanding of phylogenetic posi- 26 AMERICAN MUSEUM NOVITATES NO. 3208

Fig. 11. Brownimecia clavata, holotype: lateral habitus. Scale line: 0.5 mm (for habitus). Detail of left mandible (slightly more magnified) from anterodorsal view, showing spiculate oral surface. tion. Species diversity and rich morphology METHODS: Cladistic analysis (Hennig, of modem forms have dictated the bases for 1966) was implemented using the programs insect phylogenies, including the Hymenop- Hennig86 (Farris, 1988), Nona (Goloboff, tera. Indeed, the lack of many characters in 1996a), Phast (Goloboff, 1996b), Dada (Nix- fossil forms compromises the resolution of on, 1995a) and Clados (Nixon, 1995b). Dada phylogenetic hypotheses (Wheeler, 1992), was used for data matrix editing, and as a and the placement of fossils accordingly de- shell for submission of character matrices to pends on preserved characters whose phylo- Hennig86, Nona, and Clados. Hennig86 was genetic values are known. used for cladogram construction, and succes- The most comprehensive and widely cited sive weighting with the rescaled consistency study on ant phylogeny is by Baroni Urbani index (Farris, 1989). Typically, the multiple et al. (1992), which provides a cladistic tree search command was applied with framework for our discussion of the positions branch swapping (mhennig*), followed by of Sphecomyrma and Brownimecia, the old- branch breaking (bb*). This procedure was est definitive worker ants. First, we discuss supplemented by use of the "autospin" fea- some character recoding for the matrix of ture of Dada, which randomly shuffles order Baroni Urbani et al. (1992), reanalyze the of taxa in the matrix and resubmits the ma- slightly revised data, then discuss implica- trix to Hennig86. Autospin was run 100 tions of the revised (albeit, still unstable) times with mhennig* to try to find separate cladograms for placing the two Cretaceous stands of multiple equally parsimonious genera. trees, with the results all subjected to bb*. 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 27

a

Sphecomyrma ? AMNH NJ-242

Baikuris casei AMNH NJ-9Obb Fig. 12. Fore and hind wing venation of male sphecomyrmines. a. AMNH NJ-242 (Sphecomyrma?). b. Baikurus casei, holotype, AMNH NJ-9Obb.

Because Hennig86 does not support analysis 1996), Nona was used as a filter for strictly of polymorphic data, such characters are re- supported trees, as described by Carpenter coded as missing by Dada before submission (1996), using the "unique" command after to Hennig86. Results of Hennig86 analyses reading in Hennig86 trees. Nona was also were checked with Nona, which does support used for cladogram construction, by applying polymorphisms. Because Hennig86 also out- the multiple tree search command via ran- puts semistrictly supported trees (with reso- dom addition sequence, with tree-bisection- lutions supported by just one among the pos- reconnection branch swapping (mult* with sible optimizations; see Nixon and Carpenter, 20-50 random additions), followed by ad- AMERICAN MUSEUM NOVITATES NO. 3208

Baik.rismandibularis

Baikuris mandibularis

b

Formicium simillimum Fig. 13. Wing venation of male sphecomymines in upper Cretaceous amber from Taymyr, Siberia (a,d), New Jersey (c), and the Eocene Formiciinae (Messel, Germany; redrawn from Lutz, 1988). a. Baikuris mandibularis Dlussky (redrawn from Dlussky, 1987 [fig. 2]). b. simillimum. c. AMNH NJ-107, genus unknown. d. Dlusskyidris zherichini (Dlussky) (redrawn from Dlussky, 1975 [fig. 1301). ditional branch swapping (command max*). modifications to the original data matrix by Results were filtered for collapsibility as de- Baroni Urbani et al. (1992) (see appendix, scribed by Nixon and Carpenter (1996), us- table 1). ing the "ksv" and "best" commands on tree files. Clados was used for tree visualization 1. Character 30 of the matrix was recoded and manipulation, checking for strict support, for Leptanilloidinae (1 instead of 0), a cor- calculation of consensus trees, and printing. rection suggested by Phil Ward (personal Bremer support values (Bremer, 1988) were commun.). The results from analysis of the calculated using Phast, by invoking the corrected matrix are shown, which are essen- "bsupport" command and searching on sub- tially the same as those reported by Baroni optimal trees up to six steps longer. Urbani et al. (1992). The following analyses were based on 2. The corrected data matrix was expand- 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 29

AMNH NJ- 107

d

Dlusskyidris zherikhini Fig. 13. (Continued) ed by the addition of , a genus 3. The following characters used in the that complicates the proposed phylogeny of matrix of Baroni Urbani et al (1992) were Baroni Urbani et al, as this new genus does recoded, due to unnecessary splitting of mul- not fall within their interpretation of the Po- tistate characters: 11 and 12 (new character nerinae (Ward, 1994). 9): metapleural gland absent (0), present 30 AMERICAN MUSEUM NOVITATES NO. 3208

TABLE 1 Characters 1-62 Multistate characters are treated as nonadditive, as noted. A question mark denotes an unknown state, while a dash denotes an inapplicable trait. An asterisk denotes a polymorphism showing all applicable states; a dollar sign denotes a subset polymorphism.

5 10 15 20 25 30 35 40 45 50 55 60

Vespidae 000000000000000000000oo--oooooo00 00000000000000000000000000 0-0 Bradynobaenidae 0000000$00000101000000$--000000000???0000000000000*0000000??-? Aenictinae 10010100121100100011001000100100010100110001010101011010000010 Aenictogiton ???????0??1?????????????????????????????0??111011100101000???? 100000010100010100000oo--ooooooooo0o0oooo0000011000010010000oo Anomalomyrmini 10-0110002100111011100200000000000000010010????????????? Apomyrmini 11010100011000110101000--000000000000010000000????????????0000 Cerapachyinae 100*0000*21*001000*100$100101000010000*0000*010111100010000000 10000*01011001*10000000--0000000011010*000000*01100001*0100010 10010100121100100001001000110000010100110011010101001010000010 Ecitoninae 10010*001211001000*10010001001000100001100000101001000110000*0 10-00001*11001110000000--0000001111000100000010110000**01000*0 Leptanillini 1011010002100111011100200000000000000011001*011111?00010111011 Leptanilloides 10-1010002100?100011?01000100110000000*???????????????????????...... Myrmeciini 10000011020010010010002100000000000001000100000000100100100000 10100*0$111001*11*10*0$0100000000*0000*001000*0110100**0*00010 Nothomyrmeciinae 10-00001020010010000010--000000000000100000000000000?110100000 Ponerinae 10-*0**$*1*0**1100*1**210100*0000*0000*00*0*0*0*11100**0*000*0 100000010210*0010010101010000000000000000100000110100100100110 Formiciinae ???????????????????????????????000?????010?000??0???????00???? Prionomyrmecini 10?00001??????????1????????????00???????0??00000?????????????? Adetomyrma 10010101010001110000000??010000000???0???????????????????????? Sphecomyrma 10?0000102?00??1?0000?0???0000000??????????0000?1?00010000???? 10000001010001100001002001000000000???1000?000001111000000???? 10000001010001110000012100000000000???1000?000000011000000???? Amblyopone 100000010110001100000020010000000000??1000?000011001000000???? Brownimecia 10?0000?11??0????0010?100?00?000?0???????????????????????????? CHARACTER LIST 1. Head: hypognathous (0); prognathous (1). 2. Labrum: without rows of peglike teeth (0); with 2 rows of peglike teeth (1). 3. Labium: with pair of cylindrical pegs (0); without (1). 4. Clypeus: normally developed (0); extremely reduced (1). 5. Clypeus: rounded (0); with flat medial surface (10. 6. Worker eyes: present (0); absent (I). 7. Malar area in dorsal view: visible (0); reduced (1). 8. Antennal socket: exposed (0); not covered (1); covered by torulus (2) [nonadditive]. 9. Worker promesonotal suture: mobile (0); fused (1). 10. Metapleural gland: absent (0); not covered (1); covered by cuticular flange (2) [nonadditive]. l I. Metacoxal cavities: open or sutured (0); closed (1). 12. Metatibial gland: absent (0); present (1). 13. Basitarsal sulcus on metatarsi: absent (0); present (1). 14. Petiole: without tergosternal fusion (0); with (1). 15. Worker helcium: unfused (0); fused (1). 16. Helcium sternite: protruding ventrally (0); not (1). 17. Helcium sternite: overlapped laterally by pretergite (0); not (1). 18. Worker spiracle of III: situated posteriorly (0); situated close to anterior face of tergite (1). 19. Worker diameter of segment III: subequal to segment IV (0); smaller (1). 20. Worker segment III: tergum and sternum not fused (0); with complete fusion (10). 21. Worker dorsal stridulatory organ: absent (0); present between segments III & IV (1). 22. Worker ventral stridulatory organ: absent (0); present between segments III & IV (1). 23. Segment IV: without presclerites (0); with presternite & pretergite, unfused (1); fused (2) [nonadditive]. 24. Segment IV presclerite length: <.5 IV (0); >.5 IV (1). 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 31 without a dorsally covering flange (1), cov- The following characters were deleted: 9, ered with a flange (2); 25 and 26 (new char- the "short scape" is difficult to define, and acter 22): abdominal segment IV without furthermore present in Leptanilloidinae. presclerites (0), with presclerites and preter- Dlussky (1983) and Baroni Urbani (1989) in- gites unfused (1), with presclerites and pre- troduced this character essentially as a sur- tergites fused (2). Baroni Urbani et al. (1992) rogate for social behavior (also see our pre- scored only binary variables in their data ma- vious discussion, p. 8). 51, basal hamuli have trix, but in doing so, they potentially lost in- not been seen in Amblyopone spp. and in formation (Pogue and Mickevich, 1990). Nothomyrnecia. 67 and 68, we consider the Multistate characters are readily analyzed as empirical basis too meager to include these additive, if states are in a morphocline or two behavioral characters in the analyses. similarity is nested, or as nonadditive. The Polymorphic characters were resolved for two new variables were treated as nonaddi- the ponerines by splitting the ponerines into tive. monomorphic subtaxa (admittedly, still a slim

25. Presternite of IV: subequal to pretergite (0); shorter (1). 26. Segment IV tergum & sternum: not fused (0); completely fused (1). 27. Spiracles of V-VII: not visible (0); visible (1). 28. Pygidium: not bidentate (0); bidentate (1). 29. Pygidial spines: absent (0); present (1). 30. Pygidial reduction: absent (0); narrow U-shaped sclerite (1). 31. Worker pygidium: not covered by tergum VI (0); covered (10. 32. Acidopore: absent (0); present (1). 33. Sting and lancets: articulated (0); disarticulated (1). 34. Furcula: well developed (0); reduced to absent (1). 35. Proventriculus: flaccid (0); sclerotized (1). 36. Dufour gland epithelium: not crenellate (0); crenellate (1). 37. Pavan's gland: absent (0); present (1). 38. Sting bulb gland: absent (0); present (1). 39. Worker-gyne dimorphism: limited to wings & alar sclerites (0); pronounced (1). 40. Gyne: not dichthadiiform (0); dichthadiiform (1). 41. Wing venation: not crowded (0); crowded (1). 42. Gyne segment III: poorly differentiated (0); forming postpetiole (1). 43. Bursa copulatrix: not exposed (0); exposed (1). 44. Male antennal sockets: clypeus visible (0); at anterior margin of head (1). 45. Male head incision: absent (0); present posteriorly (1). 46. Male antennal scape: shorter than next 2 articles (0); longer (1). 47. Male forewing pterostigma: present (0); absent (1). 48. Male hindwing jugal lobe: present (0); absent (1). 49. Male propodeal spiracle: slit-shaped (0); round to elliptical (1). 50. Male segment III: no tergosternal fusion (0); present (1). 51. Male segment IV: without presclerites (0); with differentiated presclerites (1). 52. Male tergite VII: sclerotized (0); desclerotized (1). 53. Male stemnite VIII: without anterior apodemes (0); with long anterior apodemes (1). 54. Male cerci: present (0); absent (1). 55. Male subgenital plate: not biaculeate (0); biaculeate (1). 56. Lamina annularis: thin (0); enlarged (1). 57. Male genitalia: retractile (0); not retractile (1). 58. Male genitalia size: normal (0); larger than rest of gaster (1). 59. Larval hemolymph feeding organ: absent (0); present on abdominal segment III (1). 60. Larval mandibles: inner masticatory margin (0); outer (1). 61. Trophothylax: absent (0); present (1). 62. Pupa: with cocoon (0); naked (1). 32 AMERICAN MUSEUM NOVITATES NO. 3208

Vespidae Bradyn obaenidae KFormiciinae L I Aneuretinae Dolichoderinae Formicinae Myrmicinae Pseudomyrmecinae Myrmeciini Nothomyrmeciinae LPrionomyrmecini

I- Apomyrmini

PnnPrin P I Anomalomyrmini Vespidae Leptanillini Bradynobaenidae Cerapachyinae I Formiciinae 1 1-2-2 Formicinae Aneuretinae Leptanilloides L -Dol ichode rinaeL 1- Myrmicinae 1 Pseudomyrmecinae Ecitoninae 1- Myrmeciini E Not homyrmec iinae E Prionomyrmecini -Aenictinae -1- Apomyrmini - Ponerinae Aenictogiton -i- Anomalomyrmilni L Leptanillini 1.j Leptanilloides Dorylinae -Cerapachyinae E1 Ecitoninae 1--1 Aenictinae - Aenictogiton Dorylinae 14 Fig. 14. Consensus cladogram based on successive weighting of 54 trees with length of 537, using data revised from Baroni Urbani et al. (1992). Fig. 15. Bremer support values for cladogram in fig. 14. sampling), which is preferable to the ambiguity polymorphism was resolved by extrapolation, introduced by polymorphic scorings (Nixon based on the groundplans established by and Davis, 1991; Nixon, 1996). For that rea- Brothers and Carpenter (1993). son, Paraponera, Platythyrea, and Amblyopo- ne were scored and included. For the out- 4. The New Jersey amber fossils Spheco- groups, Vespidae and Bradynobaenidae, some myrma and Brownimecia were added to the 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 33

-Vespidae -Bradynobaenidae Formiciinae - Nothomyrmeciinae Myrmeciini Prionomyrmecini i --i p~~~~~~~~~~~~~~~

Vespidae Bradynobaenidae Anomalomyrmini 1-E_Formiciinae.1-.-1-r- Formicinae Leptanillini L Aneuretinae oMyrmicinae - Leptanilloides Pseudomyrmecinae 1 Myrmeciini Nothomyrmeciinae Cerapachy Prionomyrmecini L-1LAdetomyrma 11. A-pomyrmini Ecitoninae 1 Ponerinae I1T- Anomalomyrmini Aenicti Leptanillini .1, Cerapachyinae ELeptanilloides A jiton E1IL Ecitoninae _I Aenictinae Dorylin Aenictogiton 17 Dorylinae

Fig. 16. Consensus cladogram with the Malagasy genus Adetomyrma added. Fig. 17. Bremer support values for cladogram in fig. 16. matrix, scoring characters for the worker of of high levels of missing data is to weaken Sphecomyrma and the apparent male. the application of the parsimony criterion (Nixon, 1996), and this is seen in the present 5. Finally, all taxa with more than 70% study, where the taxon Formiciinae (an Eo- missing values were excluded, i.e., Formici- cene compression fossil) in particular intro- inae (81%), Prionomyrmecinae (75%), and duces instability into the results, with no dis- Aenictogitoninae (71%). The overall effect cernible benefit. 34 AMERICAN MUSEUM NOVITATES NO. 3208

-Vespidae Bradynobaenidae Formiciinae Paraponera Amblyopone Ponerinae Platythyrea Formicinae Aneuretinae Dolichoderinae Myrmicinae Pseudomyrmecinae Nothomyrmeciinae Myrmeciini Prionomyrmecini

Vespidae Adetomyrma Formiciinae -Apomyrmini Bradynobaenidae -11- Nothomyrmeciinae Anomalomyrmini I1E Myrmeciini Prionomyrmecini 1- Myrmicinae Leptanillini - Pseudomyrmecinae 1, Formicinae Leptanilloides 1, Aneuretinae . Dolichoderinae I Cerapachyinae -Amblyopone Paraponera 1-,- Ponerinae Ecitoninae Platythyrea L3.Adetomyrma Aenictogiton 2pomyrminiA 1AELnomalomyrmirai -Aenictinae 1 LeptanillLini -~2-ELeptaLePt anilloides Dorylinae 1 IT Cerapachyinae Ecitoninae 2- Aenictogiton l-, Aenictinae 19 Dorylinae 18 Fig. 18. Cladogram using revised characters and the ponerines divided into three smaller taxa. Strict consensus of successively weighted trees. Fig. 19. Bremer support values for cladogram in fig. 18.

CLADISTIC RESULTS cladograms, only eight are strictly supported when filtered by Nona (four for the original, 1. The reanalysis of Baroni Urbani et al.'s uncorrected matrix). Successive weighting revised data resulted in 144 cladograms of results in 54 cladograms of weighted length length 110; the original result was 72 clado- 537; their consensus tree (fig. 14) is slightly grams of that length. However, of these 144 less resolved than the consensus presented by 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 35

Baroni Urbani et al. (1992: fig. 4), with Lep- retention index 0.69. Only two of these are tanilloides and Cerapachyinae now part of a strictly supported, and the results are stable to trifurcation. successive weighting (weighted length 468). Baroni Urbani et al. (1992) reported var- The consensus tree of the two strictly support- ious statistical tests of "structure" in their ed trees is poorly resolved basally, but exam- matrix. They cited the PTP test (Faith and ination of the underlying trees shows that this Cranston, 1991) as indicating "significant" is caused by alternative placements of one tax- phylogenetic structure in the data, but the on: Formiciinae is either the sister group of problems and validity of that test have been Bradynobaenidae, or of Paraponera! Clearly, presented (Kallersjo et al., 1992; Carpenter, that taxon is contributing little useful infor- 1997). Baroni Urbani et al. also reported mation. Aside from that (artifactual) instability, results of bootstrapping (Felsenstein, a notable difference is the grouping within the 1985), another technique that has been cri- sensu Bolton (cf. the position tiqued (Carpenter, 1992); but, criticism of Cerapachyinae). Bremer supports remain aside, the bootstrapping result (Baroni Ur- low, although now two have supports bani et al., 1992: fig. 7) was notable for of two steps (fig. 19). showing no "significant" nodes. Indeed, 4. The addition of the two fossil taxa, their bootstrapped tree, a fully resolved ma- Brownimecia and Sphecomyrma, leads to a jority-rule consensus, contained a great ma- report of 120 cladograms by Hennig86, with jority of nodes with replicate frequencies of length of 124, consistency index of 0.51, and less than 50%! Although commonly pre- retention index 0.68. Just 16 of these are sented, this is certainly a misapplication: strictly supported, and successive weighting groups with frequencies less than 50% may results in 12 cladograms (weighting length be contradicted more often than they ap- 470). The consensus of these 12 cladograms pear (see examples in Farris et al., 1996). (fig. 20) is again poorly resolved basally, but The Bremer support was calculated for the again that results primarily from instability data matrix, and the results (fig. 15) con- brought about by inclusion of the Formici- firm low support for all groupings-only inae. The position of the sphecomyrmines is one clade has a Bremer support of as much thus ambiguous, but Brownimecia turns out as two steps, all other nodes being sup- to be part of a distant ponerine clade. The ported by one step, or with no support (un- Bremer supports did not change overall (fig. resolved in that figure). 21). 2. The addition of the data for the living 5. The deletion of taxa with more than Malagasy genus Adetomyra produced a 70% missing values caused significant rather different tree (fig. 16). Hennig86 re- changes in topology, as predicted in simula- ported 264 cladograms of length 114, con- tions by Wheeler (1992). Hennig86 reports sistency index 0.59, and retention index of 356 cladograms of length 122, consistency 0.71. Only 40 of these trees are strictly sup- index 0.50, and retention index 0.67. Just 82 ported, and successive weighting resulted of these cladograms are strictly supported, in six cladograms of weighted length 532. and successive weighting results in 36 clado- The consensus of these six cladograms (fig. grams (weighted length 449). The consensus 16) shows major rearrangements of basal of these 36 cladograms (fig. 22) is much bet- nodes, and that the enigmatic, apparent po- ter resolved. The basal formicid node is a nerine genus Adetomyra is not the sister quadritomy with Sphecomyrna, Myrmeciini group of the Ponerinae, as already indicat- + Nothomyrmeciinae, Myrmicinae + Pseu- ed by Ward (1994). Bremer supports (fig. domyrmecinae, and the remaining taxa. 17) remain low. Brownimecia is part of a larger ponerine 3. Recoding of the abovementioned char- clade, apparently a very basal member (not acters, and the addition of three more ponerine reflected in the cladistic analysis); and, there taxa produced yet another topology (fig. 18). are two changes in the army ant clade (Apo- Hennig86 reports four cladograms (with autos- myrmini to Dorylinae). The Bremer supports pin) of length 121, consistency index 0.52, and (fig. 23) are low. 36 AMERICAN MUSEUM NOVITATES NO. 3208

Vespidae Bradynobaenidae -Formiciinae Sphecomyrma Paraponera Amblyopone Ponerinae Platythyrea Brownimecia Formicinae Aneuretinae Dolichoderinae Myrmicinae Pseudomyrmecinae Nothomyrmeciinae Myrmeciini Prionomyrmecini -Adetomyrma Apomyrmini Anomalomyrmini Vespidae Leptanillini U Bradynobaenidae e - Formiciinae Leptanilloides 1-pNothomnyrrne c i inae Myrmec:iini Cerapachyinae El-r Prionormyrmecini - Sphecomyrma ' Ecitoninae 1 Myrmicinae 7- PseudomyrmEecinlae Aenictogiton -I-E-Formicinae-1, Aneuret Dolich oderinae Aenictinae Amblyopone - Paraponera Dorylinae 7 Ponerinae - Platythyrea - Brownimecia -1 Adetomyrma [lL Apomyrmi.ni 1[-J Anomalomyrcmini __ELLeptanill ini 2 Leptanilloides 2 _I-l Cerapachyinae n1~~~~~1 Ecitoninae z1 2L~~~2Aenictogiton 1, Aenictinae Doryl inae Fig. 20. Consensus of 12 successively weighted cladograms, with the New Jersey amber Cretaceous fossils Sphecomyrma and Brownimecia incorporated. Fig. 21. Bremer support values for cladogram in fig. 20. 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 37

fespidae -Bradynobaenidae rSphecomyrma

icinae

-Amblyopone _ --Paraponera

I- Ponerinae Platythyrea -Brownimecia r-Adetomvrma Vespidae

Bradynobaenidae Apomyrmini 1-- Sphecomyrma , Mrmeciini. Anomalomyrmini Nothomyrmeciinae - Paraponera Leptanillini Brownimecia - Ponerinae Leptanilloides - Platythyrea - Amblyopone C 1 Myrmicinae Cerapachyinae Pseudomyrmecinae 1 Formicinae Ecitoninae E1.F Aneuretinae Dolichoderinae Aenictinae Adetomyrma Apomyrmini Dorylinae I1- Anomalomyrmini - Leptanillini 1-- Leptanilloides - Cerapachyinae - Ecitoninae 23i Aenictinae 22 -/2-c Dorylinae

Fig. 22. Preferred cladogram: consensus of 36 successively weighted cladograms, deleting three taxa with >70% missing values (Formiciinae [Eocene compression fossil], and the living Prion- omyrmecinae and Aenictogitoninae). Fig. 23. Bremer support values for cladogram in fig. 22.

DISCUSSION: CLADISTICS vealed in the studies by Bolton (1994). Ad- etomyrma, however, replaces the Ponerinae Our preferred result (fig. 22) differs in as their sister group in figure 22, confirming some fundamental ways from that of Baroni Ward's (1994) comment that Adetomyra, if Urbani et al. (their figs. 9-15). The grouping considered a ponerine, would render the sub- of army ants is stable through both analyses, family polyphyletic. Even more fundamental reflecting the numerous synapomorphies re- is the finding that the clade ((Aneuretinae + 38 AMERICAN MUSEUM NOVITATES NO. 3208

Dolichoderinae) + Formicinae) is now the sphecomyrmines are ants. The reanalyses of sister group to the Ponerinae s.l. + (Adeto- earlier published data and the inclusion of mynna + army ants), and not to (Myrmeci- fossil data (indeed, data on fossilized inclu- inae, Nothomyrmeciinae, Pseudomyrmeci- sions!), for which more than 35% of the nae, Myrmicinae). Also, the basal node of characters could be coded, confirm the basal Formicidae is now unresolved, whereas it position of the sphecomymines. The discov- was resolved in Baroni Urbani et al.'s (1992) ery in particular of Brownimecia, a new and trees. In the hand-derived cladogram by primitive Cretaceous ponerine, and the pres- Ward (1994), the most significant difference ence of four taxa from the same deposit, in- is the switch of the Myrmicinae from being dicate that the radiation of ants 90-94 Ma one of the most basal clades, to perhaps be- was more developed than previously be- ing allied with a portion of the Ponerinae. lieved. Stratigraphic distribution of fossil The analyses presented here raise caution Formicoidea is thus: the oldest known forms concerning the robustness of the proposed are from ca. 110 Ma, and the oldest definitive phylogeny of the ants, caused by the addition Formicidae are known from approximately and deletion of taxa as well as the reinter- 20 Ma later; only primitive taxa occur pretation of some of the character states, throughout the first 50 Ma-or pre-Tertia- which have great effects on the topology of ry-history of fossil ants. the trees. First, many missing values, es- A reasonable estimation would place the pecially in three taxa (Aenictogiton, Priono- origin of the ants in the lowermost Creta- myrmex, and the fossil Formicium) create a ceous, but almost certainly no earlier. This highly ambiguous matrix-with 16% of the conclusion is also consistent with the close cells missing. Second, many taxa have a high relationship of the Formicidae to the Vespi- degree of polymorphism (6% of cells). These dae + Scoliidae (Brothers and Carpenter, two factors result in an ambiguous matrix 1993), and the phylogenetic position of Cre- (more than 22% of the cells), which is ac- taceous Vespidae (Carpenter and Rasnitsyn, cordingly problematic (Nixon, 1996). The 1990). Vespid fossils belonging only to the addition or deletion of taxa always resulted two basalmost clades of the family occur in in different topologies, in our view another sediments from the Barremian (ca. 118 Ma) clear sign of insufficient data. to Turonian of Russia. Vespids may extend As important as phylogenetic analyses of to the earliest Cretaceous. Recently, Crozier Formicidae are for evolutionary studies, the et al. (1997) estimated a controversial age of inderstanding of ant phylogeny still requires the Formicidae as being in the lower Juras- ,nous attention. The fine preservation of sic. This hypothesis was based on estimated )ssils in amber allows retention of many divergence times of mitochondrial cyto- nore characters than in compression fossils, chrome b sequences compared among extant which is made abundantly clear in how much species of , calibrating nucleotide less ambiguous is the placement of the New substitution rates using Cariridris, and an as- Jersey amber ants as compared to the Eocene sumption of clocklike rates. Previously we compression fossil taxon. Placement of even discussed the ambiguous nature of Cariri- the finest fossils, though, depends on a stable dris, but Crozier et al. regard it as an "un- phylogeny, itself subject to the discovery of doubted ant" and a myrmeciine in particular. fossils with unpredicted combinations of Our reanalyses have confirmed a basal phy- characters. Ultimately, all robust phylogenies logenetic position of the myrmeciines, mak- depend on the discovery of new characters ing the existence of a lower Cretaceous one and close scrutiny of the homologies of like Cariridris plausible. While Cariridris, known characters. like other insects from the Crato Member of the Santana Formation, is assignable to the CONCLUSION: THE AGE OF ANTS Aptian (Maisey, 1990), an absolute age of 124.5 Ma (cited by Crozier et al., 1997) is The discovery of further sphecomyrmine too old: 110 Ma is the recognized age. Use ants in amber and the detailed structure of of an ambiguous fossil to calibrate nucleotide their metapleural gland clearly show that change among extant species of Myrmecia 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 39 no doubt has led to an estimate of ant origins date 140 Ma, or the earliest Cretaceous. Lat- that is contrary to the whole fossil record of est Cretaceous and earliest Tertiary records the Hymenoptera. Despite the vicissitudes of of ants are paultry to nonexistent, but all ev- insect fossilization, given the consistent pres- idence thus far indicates that by the early ervation of Symphyta and basal Apocrita in Tertiary there is consistent preservation of the , and the appearance of most liv- living subfamilies and even some extant gen- ing and extinct families of Aculeata in the era (Dlussky, 1988, 1996; W. M. Wheeler, lower to mid-Cretaceous (Rasnitsyn, 1988), 1915; Wilson, 1985). The radiations that a lower Jurassic origin of the ants is highly spawned approximately 11,000 living spe- improbable. The earliest Hymenoptera, in cies of ants occurred only 40-50 million fact, are xyelid sawflies from the , years ago, or one-third to one-half the ap- and the earliest aculeates are the extinct fam- proximate duration of ants. Why the ants did ily Bethylonymidae from the upper Jurassic not become abundant, diverse, and dominant of Russia. until the Tertiary is a matter for future con- With little doubt, ant origins do not pre- sideration.

REFERENCES Baroni Urbani, C. leate Hymenoptera, with special refer- 1988. Phylogeny and behavioral evolution in ence to Mutillidae. Univ. Kansas Sci. ants. Paper presented at International Bull. 50: 483-648. Conference, Evolution and Ecology of Brothers, D. J., and J. M. Carpenter Social Behavior, Florence, 19-24 1993. Phylogeny of Aculeata: March, 1988. and Vespoidea (Hymenoptera). J. Hym. 1989. Phylogeny and behavioral evolution in Res. 2: 227-304. ants, with a discussion of the role of Brown, W. L., Jr., and W. L. Nutting behavior in evolutionary process. Ethol. 1949. Wing venation and the phylogeny of Ecol. Evol. 1: 137-168. the Formicidae. Trans. Am. Entomol. Baroni Urbani, C., and M. de Andrade Soc. 75: 113-134. 1994. First description of fossil Dacetini ants Carpenter, F M. with a critical analysis of the current 1992. Superclass Hexapoda, vol. 4 (Part R, classification of the (amber collec- Arthropoda 4), Treatise on Invertebrate tion Stuttgart: Hymenoptera, Formici- Paleontology. Lawrence, KS: Geol. dae. VI. Dacetini). Stuttg. Beitr. Nat. Soc. Am./Univ. Kansas. (B) 198: 65 pp. Carpenter, J. M. Baroni Urbani, C., B. Bolton, and P S. Ward 1992. Random cladistics. Cladistics 8: 147- 1992. The internal phylogeny of ants (Hy- 153. menoptera: Formicidae). Syst. Ento- mol. 17: 301-329. 1996. Phylogeny and biogeography of Pol- Bolton, B. istes. In S. Thrillazzi and M. J. West- 1994. Identification guide to the ant genera of Eberhard (eds.), Natural history and the world. Cambridge, MA: Harvard evolution of paper Wasps, pp. 18-57. Univ. Press. Oxford: Oxford Univ. Press. Brandao, C. R. E, R. G. Martins-Neto, and M. A. 1997. PTP is meaningless, T-PTP is contra- Vulcano dictory: A reply to Trueman. Cladistics, 1990. The earliest known fossil ant (first in press. southern hemisphere mesozoic [sic] Carpenter, J. M., and A. P. Rasnitsyn record) (Hymenoptera: Formicidae: 1990. Mesozoic Vespidae. Psyche 97: 1-20. ). Psyche 96: 195-208. Cockerell, T. D. A. Bremer, K. 1922. Fossils in Burmese amber. Nature 109: 1988. The limits of amino-acid sequence data 713-714. in angiosperm phylogenetic reconstruc- Crozier, R., L. S. Jermiin, and M. Chiotis tion. Evolution 42: 795-803. 1997. Molecular evidence for a Jurassic ori- Brothers, D. J. gin of ants. Naturwissenschaften 84: 1975. Phylogeny and classification of the acu- 22-23. AMERICAN MUSEUM NOVITATES NO. 3208

Darling, D. C., and M. Sharkey ceous (Turonian) of New Jersey. MS, 1990. Hymentopera. In D. Grimaldi (ed.), In- Am. Mus. Novitates/Bull. sects from the Santana Formation, Grimaldi, D., C. Beck, and J. J. Boon Lower Cretaceous, of Brazil. Bull. Am. 1989. Occurrence, chemical characteristics, Mus. Nat. Hist. 195: 123-153. and paleontology of the fossil resins Dlussky, G. M. from New Jersey. Am. Mus. Novitates 1975. Formicidae. In A. P. Rasnitysn, Hyme- 2948: 28 pp. noptera Apocrita of the Mesozoic. Hennig, W. Trans. Paleontol. Inst. Acad. Sci. USSR 1966. Phylogenetic systematics. Urbana: 147: 115-121 [In Russian]. Univ. Illinois Press. 1983. A new family of Upper Cretaceous Hy- Hibbett, D. M., D. Grimaldi, and M. Donoghue menoptera-an "intermediate link" be- 1996. Cretaceous mushrooms in amber. Na- tween the ants and the scolioids. Pa- ture 377: 487. leontol. J. 3: 65-78. H6lldobler, B., and E. 0. Wilson 1987. New Formicoidea (Hymenoptera) of 1990. The Ants. Cambridge, MA: Harvard the Upper Cretaceous. Ibid. 1987: 146- Univ. Press. 150. Jell, P. A., and P M. Duncan 1988. Ants from (Paleocene?) Sakhalin am- 1986. Invertebrates, mainly insects, from the ber. Ibid. 1988 (1): 50-61. freshwater, Lower Cretaceous, Koon- 1996. Ants (Hymenoptera: Formicidae) from warra fossil bed (Korumberra Group), Burmese amber. Ibid. 30: 449-454. South Gippsland, Victoria. In P. A. Jell Faith, D. P., and P. S. Cranston and J. Roberts (eds.), Plants and inver- 1991. Could a cladogram this short have aris- tebrates from the Lower Cretaceous en by chance alone? On permutation Koonwarra fossil bed, South Gipps- tests for cladistic structure. Cladistics 7: 1-28. land, Victoria, Sydney: Assoc. Austral- Farris, J. S. asian Paleontologists. 1988. Hennig86, version 1.5. Program and Kdllersj6, M., J. S. Farris, A. G. Kluge, and C. documentation. Port Jefferson Station, Bult New York. 1992. Skewness and permutation. Cladistics 1989. The retention index and the rescaled 8: 275-287. consistency index. Cladistics 5: 417- Langenheim, J. H. 419. 1969. Amber: a botanical inquiry. Science Farris, J. S., V. Albert, M. Killersjo, D. Lipscomb, 163: 1157-1169. and A. G. Kluge Lutz, H. 1996. Parsimony jacknifing outperforms 1988. Eine neue Unterfamilie der Formicidae neighbor-joining. Cladistics 12: 99- (Insecta: Hymenoptera) aus dem mittel- 124. eozanen Olschiefer der "Grube Mes- Felsenstein, J. sel" bei Darmstadt (Deutschland, 1985. Confidence limits on phylogenies: An S-Hessen). Senckenb. Lethaea 67: 177- approach using the bootstrap. Evolution 218. 39: 783-791. Maisey, J. G. Goloboff, P. A. 1990. Stratigraphy and depositional environ- 1996a. NONA, version 1.5. Program and doc- ment of the Crato Member (Santana umentation. Tucumdn, Argentina. Formation, Lower Cretaceous of N.E. 1996b. PHast, version 1. Program and docu- Brazil). In D. Grimaldi (ed.), Insects mentation. Tucumdn, Argentina. from the Santana Formation, Lower Grimaldi, D. (ed.) Cretaceous, of Brazil. Bull. Am. Mus. 1990. Insects from the Santana Formation, Nat. Hist. 195: 15-19. Lower Cretaceous, of Brazil. Bull. Am. McAlpine, J. F, and J. E. H. Martin Mus. Nat. Hist. 195: 191 pp. 1966. Canadian amber: a paleontological 1993. The care and study of fossiliferous am- treasure chest. Can. Entomol. 101: ber. Curator 36: 31-49. 819-838. 1996. Amber: window to the Past. New York: Naumann, I. D. Abrams/Am. Mus. Nat. Hist. 1993. The supposed Cretaceous ant Cretaco- 1997. Stratigraphy, taphonomy, and paleo- explicata Jell and Duncan (Hy- ecology of a remarkable deposit of fos- menoptera). J. Aust. Entomol. Soc. 32: siliferous amber from the mid Creta- 355-356. 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER

Nixon, K. C. eny, pp. 205-215. New York: Columbia 1995a. Dada, ver. 1.000. Clados ver. 1.49. Pro- Univ. Press. gram and documentation. Ithaca NY: L. Wheeler, W. M. H. Bailey Hortorium, Cornell Univ. 1915. The ants of the Baltic amber. Schr. Phys. 1995b. Clados ver. 1.49. Program and docu- -Okon Ges. Konigsberg 55: 1-142. mentation. Ithaca NY: L. H. Bailey Wilson, E. 0. Hortorium, Cornell Univ. 1985. Ants from the Cretaceous and Eocene 1996. Paleobotany in cladistics and cladistics amber of North America. Psyche 92: in paleobotany: enlightenment and un- 205-216. certainty. Rev. Palaeobot. Palynol. 90: 1987. The earliest known ants: an analysis of 361-373. the Cretaceous species and an inference Nixon, K. C., and J. M. Carpenter concering their social organization. Pa- 1996. On consensus, collapsibility and clade leobiology 13: 44-53. contention. Cladistics 12 (in press). Wilson, E. O., F M. Carpenter, and W. L. Brown, Nixon, K. C., and J. I. Davis Jr. 1991. Polymorphic taxa, missing values and 1967a. The first Mesozoic ants. Science 157: cladistic analysis. Cladistics 7: 233-241. 1038-1040. Pike, E. M. 1967b. The first Mesozoic ants, with the de- 1995. Amber taphonomy and the Grassy Lake, scription of a new subfamily. Psyche 74: Alberta, amber fauna. Ph.D. diss., Univ. 1-19. Calgary, Alberta. 264 pp. Pogue, M. G. and M. F Mickevich APPENDIX 1990. Character definitions and character state Vacuum Embedding and Disintegration delineation: The b6te noire of phyloge- of the Type of Sphecomyrma freyi netic inference. Cladistics 6: 319-361. Poinar, G. O., Jr. The holotype of Sphecomyrma freyi occurrec 1992. Life in amber. Palo Alto, CA: Stanford in a deep red piece of amber collected in 196( Univ. Press. from Cliffwood Beach, New Jersey. The piece Rasnitysn, A. P. originally contained two workers, but it cracked, 1988. An outline of the evolution of the hy- separating the two workers (Wilson et al., menopterous insects (Order Vespida). 1967a,b). The piece containing the holotype (pho- Orient. Insects 22: 115-145. tograph in Holldobler and Wilson, 1990) was 1996. Burmese amber at the Natural History roughly cubical, approximately 22 X 18 X 17 Museum [London]. Inclusion (Wrostek) mm, very brittle, with some fine, superficial 23: 18-22. cracks; larger cracks permeated deeply into the [Krakow] piece. It was stored in a wooden cabinet in an Schlee, D., and H. G. Dietrich uncovered drawer of the fossil insect collection in 1970. InsektenfUihrender Bernstein aus der Un- the Museum of Comparative Zoology for 30 terkreide des Lebanon. N. Jahrb. Geol. years, padded with cotton in a small plastic box Palaontol. Monatsh. 1970: 40-50. with a snap clasp. Over the years there had been Silverstein, H., and P. Nascimbene deterioration of the piece; it become darker and 1997. Techniques for screening and preparing more fractured, which is typical of even softer, inclusions in brittle ambers from the less brittle amber like Baltic material (Grimaldi, Cretaceous period. MS. Am. Mus. Nov- 1993). Cretaceous material is even more suscep- itates. tible to oxidative damage. To slow or prevent fur- Waggensberg, J., C. R. F Brandao, and C. Baroni ther deterioration, Canada Balsam had been ap- Urbani plied to one or more faces of the piece (F M. 1996. Le mystere de la chambre jaune. Re- Carpenter, personal commun. to DG, 1988), al- cherche 288: 54-59. though when and how often is unknown. Ward, P S. The ant itself was very close (ca. 0.3 mm) to a 1994. Adetomyrma, an enigmatic new ant ge- flat, prepared surface. This surface had numerous, nus from Madagascar (Hymenoptera: fine, parallel scratches, the result of grinding/pol- Formicidae), and its implications for ant ishing by Carpenter or others. On top of this sur- phylogeny. Syst. Entomol. 19: 159-175. face was a layer of clear, hardened material with Wheeler, W. a slightly irregular surface, 0.1-0.2 mm thick, that 1992. , sampling, and molecular prevented detailed observation of the holotype. A phylogenetics. In M. Novacek and Q. D. cotton swab moistened with 70% ethanol gently Wheeler (eds.), Extinction and phylog- rubbed on the surface resulted in no visible effect. 42 AMERICAN MUSEUM NOVITATES NO. 3208

Balsam would become tacky under such treat- eliminating mirrorlike fractures filled with air). ment. The coating(s) applied just over the ant no After 10 minutes under vacuum, the rubber cups doubt were made to reduce surface distortions for are removed and the specimen is allowed to hard- photography and close examinations. It later be- ened at room temperature and pressure, generally came apparent that this material was almost cer- taking several hours. The hardened block can then tainly dried mineral oil, or perhaps a synthetic be trimmed with a water-fed diamond saw and slide mounting medium like Euparol (or both), ground with water-fed emory wheels of 320 and based on the manner in which this substance be- 600 grits. haved during curing in a synthetic embedding res- Shortly after the MCZ piece was removed from in, and some other observations. the vacuum, a foam of bubbles suddenly appeared Mineral oil has traditionally been used for ex- on the coated surface over the ant, signaling a tremely clear observation of inclusions in amber reaction within the epoxide no doubt initiated by (e.g., Grimaldi, 1993). F. M. Carpenter used min- the heat of polymerization. Within several min- eral oil, in fact, to store some pieces of Baltic and utes the epoxide had become thickened enough to Canadian amber in the MCZ, as a method to pre- prevent removing the amber without damage. Nu- vent oxidative decomposition, but he also embed- merous previous embeddings done in this manner, ded some pieces in Canada Balsam (Grimaldi, using pieces of New Jersey amber and epoxide 1993; also see comments below). Lastly, the cot- volume several times this size, never yielded vis- ton stored in the box with the type specimen was ible effects to the amber, let alone a froth of bub- stained yellowish, similar to mineral oil residues bles. The heat alone from the polymerization re- on cotton from specimens in the AMNH collec- action would not cause such an effect on the am- tion. If the layer was even partially dried mineral ber (amber, in fact, requires temperatures in ex- oil, numerous coatings must have been applied cess of 350°F for any softening to occur). The over the years, perhaps by various investigators coating over the ant had caused the intense bub- inspecting the specimen. bling, since uncoated surfaces were largely unaf- To prevent further deterioration of the MCZ fected. To our deep disappointment, we found that specimen, a process of vacuum embedding in a the coating had penetrated several millimeters into synthetic resin was used, refined from that of the amber. Schlee and Dietrich (1970), who developed it for After the epoxy was completely hardened, it preserving and studying very brittle Lebanese am- the from was trimmed and polished. No view of the ant ber. The technique not only seals piece could be found, the surface where it now was be- atmospheric oxygen, it is an excellent physical protection from accidental shattering, and is the ing opaque with a suspension of yellowish bub- bles. The block was cut through the middle of the reliable way to trim into a piece of cracked only in an amber for accurate views of inclusions without amber piece, distant from the ant, attempt to view the ant from the other, interior side. (In further fracturing or splitting the piece. The tech- nique has been used successfully and consistently cutting through the center, other, minute insects have been which we considered on over 800 pieces of New Jersey and Lebanese may obliterated, amber at the AMNH. A detailed, illustrated de- worth sacrificing for the sake of the ant. One un- scription will be presented elsewhere (Silverstein, usual psychodid midge deep in the piece, though, and Nascimbene, 1997), but with basic techniques was revealed this way and remains very well pre- mentioned here. served and displayed, a silent and enduring wit- The embedding medium is an epoxide (Buehler, ness). Cutting through the interior of the amber Inc.) that is hardened using a polymerization/cat- exposed highly fractured and brittle amber too alyzing agent. Heat is generated during the poly- deep for the epoxide to have permeated, even un- merization; the larger the volume of medium, the der vacuum. Both halves were vacuum embedded hotter the reaction. Epoxide and catalyzing agent again, with perfect results (the coated surface of are mixed in a separate container, then poured into the amber was already sealed). Since the exposed shallow cylindrical rubber molds (Buehler, Inc). inner surface of the amber was cemented by the The amber piece is then gently slipped beneath second embedding, grinding of the interior sur- the surface of the liquid epoxy, bubbles removed face toward the outer surface with the ant could from the epoxide, and the preparation placed un- now be done safely. Reaching 2 mm within the der a bell jar that is sealed and brought to ap- presumed surface of the ant revealed no specimen. proximately 4 psi of pressure (atmospheric pres- It was only after a cross section was ground away sure at sea level being 14 psi). The reduced pres- from one edge that it became apparent that the sure removes air from fine cracks in the amber, bubbling involved the 1-2 mm of surface amber allowing epoxide to seep into the cracks and bind containing the ant. During the bubbling of the fractures together (it also improves visibility, by coating + amber mixture, the ant had become dis- 1997 GRIMALDI ET AL.: PRIMITIVE ANTS IN CRETACEOUS AMBER 43

articulated and the parts entirely obscured, with the specimen effectively disintegrated. What was so stunning was the dramatic and destructive effect of the coating on the amber it- self, a substance popularly considered inert. The effect must have been greatly enhanced by oil that penetrated numerous fine fractures in the piece, and by presumably 30 years of contact with and replenishment of oil. If any of these microscopic and oil-laden fractures extended to the ant, its body cavity would probably have contained the coating too. Oddly, Baltic amber stored for de- cades in vials of mineral oil at the MCZ showed little surface degradation (D. Grimaldi, personal obs., 1995). A small collection of Arkansas am- epoxide- ber, however, given to the AMNH and stored for 20 years in mineral oil at the Univ. Illinois by Ellis MacLeod, was soft and malleable even to

iIIK the core of pieces 20 mm thick, and totally de- graded. Mineral oil obviously has dramatically different effects on different ambers. Even though Baltic and Arkansas ambers are similar in age, they have completely different botanical origin and chemistry, which must account for the differ- ent reactivity. New Jersey amber is no different. The fact that this amber visibly reacts with organ- ic solvents like acetone, whereas most other am- bers (including Cretaceous amber from Lebanon and Canada) do not, indicates a particularly frag- ile and reactive composition. This preparation is a tragic lesson in the study -=- and stewardship of valuable amber fossils. First, the use of mineral oil-almost certainly the coat- ing that caused the disintegration-must be strin- gently avoided. Despite advocacy of the use of mineral oil for viewing inclusions (e.g., Grimaldi, 1993; Poinar, 1992), this oil clearly chemically reacts with some ambers much more than others. Sugar syrup or glycerine, or other water-based substance, should be used. Hopefully it is not too late for some specimens. Many pieces of amber in the Acra collection of Lebanese amber, and even some of the New Jersey amber ants reported in this study, had been coated with mineral oil at various times for better viewing of inclusions. o o Thin films of the oil must still reside on the sur- face and in the fine cracks. Secondly, failures like this must temper future curatorial decisions. Given a particularly valuable amber fossil that will gradually disintegrate with Fig. 24. Schematic interpretation in cross sec- age, but which has been treated in unknown ways, tion of damaged incurred to the amber piece con- keen insight must be excercised as to the appro- taining the type of Sphecomyrma freyi during em- priate method of conservation. bedding process. Top, prior to embedding. Mid- dle, formation of reaction bubbles. Bottom, end David Grimaldi, Curator, AMNH result. Uncoated surfaces were largely unaffected. Paul Nascimbene, Curatorial Assistant Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates and Bulletin published during the last five years are available at World Wide Web site http://nim,idi.amnh.org. Or address mail orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, N.Y. 10024. TEL: (212) 769-5545. FAX: (212) 769-5009. E-MAIL: [email protected]

0 This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).