/Of3

CHAPTER2. EPHEMEROPTERA

W. P. MCCAFFERTY1

ABSTRACT I Fossils of 26 alate and 88 larval were logical data. Among the , the newly studied from the Santana Formation, Lower discovered extinct genus Colocrus is shown to be Cretaceous, Ceara Crato, Brazil. New descriptions cladistically more derived than Chromarcyinae, are as follows: larva of Siphgondwanus occiden­ with forewing venation intermediate between the talis, new genus and species (); alate plesiomorphic venation ofChromarcyinae and the forms of Siphlonuridae (?) spp. 1, 2, and 3; larva highly specialized venation ofOligoneuriinae. The and alate form of Colocrus indivicum, new genus larva of Colocrus retains a plesiomorphic dorsal and species (Oligoneuriidae: Colocrurinae, new first gill similar to that of Chromarcyinae. subfamily); alate forms of Australiphemera reve­ Although in general the mayflies studied are lata, new genus and species, and Microphemera characteristically similar to modern schistonote neotropica, new genus and species (); forms, a majority of the lineages· represented did alate form of Pristiplocia rupestris, new genus and not survive to the present in the Neotropics, either species (); alate form of Epheme­ becoming entirely extinct or displaced biogeo­ roidea sp. 1; alate form of sp.; graphically. Both Oligoneuriidae and Ephemero­ larva ofLeptophlebiidae (?) sp. 1, and alate forms idea apparently radiated into their major lineages of {?) spp. 2 and 3; larvae of in­ by Lower Cretaceous time. Pannotes and , certae sedis spp. 1 and 2. Protoligoneuria limai however, remain unknown from the Mesozoic. Demoulin is redescribed from a very large series Finds of Siphlonuridae, Ephemeridae, and possi­ oflarval specimens and is clearly a member of the bly in Brazil indicate previous Hexagenitidae rather than the Oligoneuriidae, widespread distributions for these. families {extant where it was previously classified. Neotropical ephemerids being of more recent The study material provides the first fossils of north-temperate origin). The discovery of Hex­ Oligoneuriidae (as here restricted) and Euthyplo­ agenitidae in the Southern Hemisphere indicates ciidae, and possibly Potamanthidae. Oligoneuri­ a widespread Pangaean distribution for this extinct idae along with (new status), Iso- Mesozoic group. The presence of Oiigoneuriidae . nychiidae, and are recognized as and Euthyplociidae in West Gondwana suggests the monophyletic superfamily Heptagenioidea. The that continental vicariance in the Southern Hemi­ higher classification is based on cladistic relation­ sphere accounts for their present Pantropical dis­ ships offamilial lineages. The relationships of the tributions. Mayflies from the fossil site apparently extinct families Hexagenitidae and Epeoromimi­ include forms from both lentic and lotic aquatic dae are reevaluated in light of the new paleonto- habitats.

INTRODUCTION

The unearthing oflarval and alate Ephem­ As a result of this find, questions regarding eroptera from the Santana Formation in the affinities of Lower Cretaceous mayflies, Ceara Crato, Brazil, is a major discovery, with either an essentially modem fauna or a providing the opportunity for a critical con­ more ancient one, can be resolved with more tribution to our knowledge of history. certainty. In addition, questions about bio­ Not only are fossil ephemeropteran remains geographic elements present iq West Gond­ rare from the Lower Cretaceous, but prior to wana that, because of the conriection of cer­ this they have been very poorly known from tain continents, may accou'nt for some South America, with only brief accounts of present-day disjunct world disfubutions, can a few larvae from the Lower Cretaceous of also begin to be resolved. Limited hypotheses Brazil (Costa Lima, 1950; Brito, 1987) and ofcausal mayfly biogeography in the past have the Eocene of Argentina (Rossi de Garcia, been based almost entirely on phylogenetics 1983). and other inferences from exi\mt taxa (Ed­ munds, 1972, 1975;McCafferttetal., 1990). Many of these hypotheses, which have yet to 1 Professor, Department of Entomology, Purdue Uni· be tested because of the paucitylofa Southern versity, West Lafayette, IN 47907. Hemisphere fossil record, are now testable. 20 !990 McCAFFERTY: EPHEMEROPTERA 21

Ephemeropteran paleontology has been re­ tant taxa , Ephetnerellidae, Co­ viewed by Tshernova (1970, 1980). Ed­ loburiscinae, Leptophlebiidad (Mesonetinae), munds ( 1972) discussed fragmentary data on , and Siphlonuridae. Of these fossil relationships, and Landa and Soldan extant groups, only Siphlorturidae in the ( 19 8 5) accounted for most extinct higher taxa broadest sense (e.g., Tshem9va, 1967) ap­ in their classificatory synopsis. Sinitshenkova pears to be unquestionably represented. Ju­ (1984) gave a brief account of the paleoecol­ rassic fossils assigned to other\extant families ogy of mayflies, and Hubbard (1987) has pro­ require review. The extinct Jurassic families vided a useful catalog ofnearly all fossils that are all quite similar to the Si~hlonuridae. have been referred to Ephemeroptera. For the A Cretaceous mayfly fauna has only re­ most part, extinct families have not con­ cently been documented from some fossils vincingly been placed within a phyletic from Palearctic Laurasia (Tsl:lemova, 1971; scheme of extant families, and differences of Tshemova and Sinitshenkov8., 1974; Sinit­ opinion remain as to the superfamilial clas­ shenkova, 1976, 1986), Australia (Jell and sification of several extinct families. In ad­ Duncan, 1986), and one froni Algeria (Sin­ dition, the familial classification of many fos­ itshenkova, 1975). These include the Lower sils requires reevaluation. Cretaceous genera Mesoneta 1 (Leptophlebi­ from the Paleozoic that have been idae ?) and Epeoromimus (Ep6oromimidae), considered to be mayflies represent an an­ both of which had also occurred in the Ju­ cient extinct fauna known mainly from rassic, as well as Hexameropsis and Mon­ northern Pangaea. However, Triplosoba pul­ gologenites (Hexagenitidae), Pr(Jameletus and chella is the only Carboniferous that Australurus (Siphlonuridae), Promirara has consistently been considered in the (Arneletopsidae ?), and Duici"tanna (family Ephemeroptera, and placement of a number incertae). The new data represented herein of Carboniferous insects in the Ephemerop­ greatly expand our knowledge of Cretaceous tera, for example, by Kukalova-Peck (1985) mayfl ies.. I, (Protoephemeroptera, sensu Hubbard, 1987) Sinitshenkova (1984) mentioned, without requires a very broad definition of the order. detail, some undescribed Bntzilian fossils In my opinion, basic ephemeropteran wing from the Lower Cretaceous. Th,ose are prob­ venation may be highly plesiomorphic and ably referable to Protoligoneuria limai (Hex­ thus several widely diverging early insect lin­ agenitidae), which is descri1',ed in detail eages could incorrectly be placed in Ephem­ herein. Undescribed Australian larval fossils eroptera if additional characterization is not that were mentioned by Riek (i 970) include

available. Most recently, Carpenter (1987) has those recently described by Jell1 and Duncan cast doubt on the makeup and inclusion of (1986). Cretoneta (Leptophlebii

ditional specimens are not described or placed MATERIAL: 77 larvae: AMNH 43400- beyond order because of insufficient pre­ 43422,43424-43437, 43439-43452, 43454- served detail. 43471, 43478, 43482, 13485-43488, and New monospecific genera and their appli­ 43492-43498; 4 questionable larvae: AMNH cable species are described together since it 43438, 43483, 43490, an4 43491. would be impossible at this time to sort out DESCRIPTION: Dorsal, ~entral, and lateral generic vs. specific level characters. Gener­ aspects represented; generhl structural details ally, size characters may be considered spe­ represented among various fossils except cific, but not always; and whereas genitalia mouthparts not discernibl~; middle instars to are usually very valuable for specific diag­ mature specimens repres'ented. Body min­ nosis, they are very poorly represented in these nowlike, fusiform, ranging from 7.0 mm to fossils. Larval mouthparts are also valuable 13.4 mm long (excludingicaudal filaments). for either specific or generic differentiation, Head hypognathous, shprter than broad but these too are very poorly preserved. (width nearly twice lengtli), narrower than The adult and subimaginal stages are to­ thorax, tapered anteriorly; discernible com­ gether referred to as alate because it is usually pound eyes relatively smau, situated dorso­ impossible to discriminate between the two laterally; antennae thin and delicate, .atten­ different winged stages in these fossils. In for­ uated, relatively short (1.4 mm long on mal descriptions of newly named taxa, alate specimen with body 8. 9 mb and head width forms are described under the heading of 2.2 mm), inserted anteriorly on head capsule. Adult since the structural characters de­ Thorax widening posteriorly; all legs narrow scribed would apply to the adult even if the and relatively short, becoming shorter rela­ subimago was actually represented by the fos­ tive to body size as individual body size in­ sil(s). Body lengths that are given always ex­ creases; forelegs oriented anteriorly ventral clude caudal filaments. Discussions accom­ or lateral to head, middle and hindlegs ap­ panying individual descriptions include parently oriented posterolaterally;1 claws sin­ accounts of pertinent fossil records, lineage gle, relatively small, only slightly curved, with ages and relationships, classifications, and sharp apex, denticulation not discernible; on paleoecology, if possible. Concluding re­ 10.9 mm long specimen \(in millimeters): marks following the taxonomic treatment forelegs ca. 3.0 long and mi<;ldle and hindlegs concern evolution and historical biogeogra­ ca. 4.0 long, forefemur 1.0, f9retibia 0.8, fore­ phy as well as the paleoenvironment that are tarsus 0.8, claw 0.2, middle\femur 1.5, hind­ inferred from all the data presented. femur 1.6, middle and hindtibia 1.0, middle and hindtarsus 0.8, middle ~d hindclaw 0.3, ACKNOWLEDGMENTS coxae and trochanters not cl~arly discernible; mature forewing pads of schistonote type, I wish to thank Arwin Provonsha of Pur­ clearly divided for almost entire length, 3.4 due University for line illustrations and Dave mm long on specimen wi~ body 11.2 mm McShaffrey of Purdue University for assis­ long, somewhat abruptly narrowing along in­ tance in photographing fossils. I also thank ner margin anteriorly, and \subtriangular in George Edmunds, University of Utah, and posterior fourth. Abdomen\ with sharp, al­ Bill and Jan Peters, Florida A&M University, most spinelike posterolaterjtl processes on for their many helpful suggestions for the segments 1-9 (fig. 2) and with platelike gills manuscript. inserted posterolaterally on s~ents 1-7 (no fibrilliform portion discernible) (fig. l); gill I FAMILY HEXAGENITIDAE elliptical, slightly shorter thap. gills 2-6, with elongate rib (or possibly tracheal trunk) run­ Protoligoneuria limai Demoulin ning slightly anterior to longitudinal midline; Figures 1-& gills 2-6 subtriangular, all shbequal, some­ Baetidae (Siphlonurinae), Costa Lima, 1950: 419. what longer than corresponding segment Protoligoneuria limai Demoulin, 1955: 271. · length, with sclerotized rib ruhning along an­ Palaeobaetodes costalimai Brito, 1987. NEW terior margin, rounded inner posterior por­ SYNONYM. tion oflamellae folded ventra}ly on some gills

...... _.... ______~~- •

1990 McCAFFERTY: EPHEMEROPTERA 23

"·.

1111

Ill

1 2 \ Figs. 1, 2. Proto/igoneuria limai mature larva. I. Whole dorsal composite. 2. Ventral. abdomen composite. I I I 24 BULLETIN AMERICAN MUSEUM OF NATURAL HISTdlRY NO. 195 l!t :J ~- ·'! on some specimens giving unnatural appear­ and Australian realms, lwhereas extant oli­ i ance of straight inner margin, possible tra­ goneuriids are common in tropical South ~! cheation not discernible; gill 7 elliptical-elon­ America). As will be discussed below his gate, strikingly longer than gills l-6, up to conclusion was erroneous.I ' nearly twice as long in large, mature speci­ According to the International Code of mens (in millimeters) (e.g., body = 11.1, gill Zoological Nomenclat~re [Art. 73b (i)], 4 = 1.0, gill 7 = 1.9) but as little as 20 percent ..Syn types may include specimens ... not seen longer in small, young specimens (e.g., body by the author but whichl form the bases of = 8.5, gill 4 = 1.0, gill 7 = 1.2), with scler­ previously published deso/iptions or illustra­ otized rib running along anterior border and tions upon which the author founded the new indication of median longitudinal tracheal nominal species-group ta~on ...." As such, trunk at least in some specimens. Caudal fil­ all of Costa Lima's specimens are syntypes. aments relatively short and robust, becoming Unfortunately, the existence and deposition shorter relative to body as individuals be­ of this material remains 1 questionable, and come larger (e.g., body = 7.2, cerci = 4.0; even the number of specimens is not known body= 12.5, cerci = 4.2); median terminal because Costa Lima referred to the quantity filament ca. 3/4 length of cerci, with dense row of them as simply ''algun~" (= some). I can oflong setae along entire length ofboth lateral only be sure that there were three specimens borders; cerci with dense row of long setae because he included photographs of three in along inner border only. his publication. \ DISCUSSION: Costa Lima (1950) first re­ Based on additional Lo"Yer Cretaceous lar­ ported the existence of this species with a vae from Ceara, Brazil, Brito ( 1987) renamed very incomplete description of at least three this species PaiaeobaetodeS. costaiimai. I agree fossil specimens from Ceara Crato, Brazil: that at least Brito's holotype and the two Riacho do Salgado, Fazenda Santa Rosa, near paratypes that he figured represent the same Pousada Santa Fe. He did not mention the species as the Costa Lima (1950) material; age of this material, but it undoubtedly is however, the name Protoligoneuria /imai from the Santana Formation. The speci­ must be regarded as the lvalid senior syn­ mens which he reported on are presumably onym. The latter name, although perhaps not residing with the National Department of applied under preferred prcumstances by Mineral Production in Brazil. He did not Demoulin ( 195 5), does nevertheless meet no­ name this material, but ascribed them to the menclatural regulations and therefore cannot family Baetidae, subfamily Siphlotiurinae be regarded as a nomen ~anum or nomen (presently equivalent either to the family nudum. If a neotype is deemed necessary in Siphlonuridae or a complex of families in­ the future, it would best b~ chosen from the cluding Siphlonuridae). Costa Lima's illus­ better preserved AMNH niaterial. trations of the general facies of the body, in­ Protoligoneuria limai is clearly a minnow­ cluding gill morphology and the strikingly like mayfly, typical of mayfues of the super­ large and elongate gill 7 (fig. l), indicate that family Baetoidea, as pres~ntly constituted, the AMNH materials are the same species. that have streamlined bodies, narrow legs, Demoulin (l 95 5) subsequently proposed and schistonote wing pads; Its short anten­ the new scientific name Protoligoneuria limai nae, three setaceous caudal \filaments (fig. 1), for these fossils based strictly on the descrip­ and abdominal posterolatetal processes fur­ tion of Costa Lima (1950) and without des­ thermore suggest that it is a 'primitive mayfly ignating types. Demoulin, believing Costa typical of known Jurassic: mayflies (Sinit­ Lima's specimens to be members of the fam­ shenkova, 1984), and evide~tly related to the ily Oligoneuriidae rather than Siphlonuridae, Siphlonuridae or a complex of primitive fam­ devised his generic nomen to reflect this. This ilies including Siphlonuri&e. The lack of family reassignment was based entirely on characteristics associated with larvae of the the known distributions of the two families family Oligoneuriidae, especially forelegs with (siphlonurids had been known only from long filtering setae, precludes its inclusion in

southern South America, and the Holarctic that family. 1 -.-

1990 McCAFFERTY: EPHEMEROPTERA 25

Figs. 3-5. Protoligoneuria limai larvae. 3. Dorsal, middle-late instar habitus, AMNH 43455. 4. Ventral, late instar habitus, AMNH 43469. S. Dorsal, middle instar habitus, AMNH 43452. 26 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 195

Figs. 6-8. Protoligoneuria limai larvae. 6. Ventral, middle instar habitus, AMNH 43435. 7. Ventral, middle instar habitus, AMNH 43415. 8. Lateral, late instar habitus, AMNH 43418. '990 McCAFFERTY: EPHEMEROPTERA 27

The Hexagenitidae are extinct but appear tionality. Other Hexagenitidae possess a very closely related to the extant minnowlike somewhat apically truncate gill 7 that is more mayflies, and according to Tshernova and triangulate than that ofProtoligoneuria. Also, Sinitshenkova (l 974), the family is "a special sclerotization of gills of other Hexagenitidae extinct branch sharing a common origin with is apparently present along the posterior edge the Siphlonuridae." The genus Ephemeropsis of the gills. I could find no evidence of pos­ from the Upper Jurassic is well represented terior sclerotization in Protoligoneuria. in Palearctic Laurasia (Tshernova and Sin­ U ndescribed Brazilian fossil materials from itshenkova, 1974) and typifies hexagenitid the Lower Cretaceous that Sinitshenkova larvae. A minnowlike body with abdominal ( 1984) has seen and thought possibly to be posterolateral processes, narrow legs, and Hexameropsis are quite probably specimens swimming caudal filaments are found in of P. limai. In any case, the presence of the Ephemeropsis, as they are in P. limai. Mesozoic mayfly family Hexagenitidae in The wing venation of Hexagenitidae dis­ Brazil extends the known range of the group tinguishes that family and is a more complex considerably. The five previous known gen­ type than is found in Siphlonuridae, partic­ era are from central or eastern Laurasia, al­ ularly regarding the more complex cubital ve­ though H exameropsis africana is known from nation in the forewing. Protoligoneuria adults the Lower Cretaceous ofAlgeria, which would are unknown, and nothing can be deduced place it near or abutting central Laurasia. A about the wings from the available larval wing Jurassic or older origin for the family would pads. Edmunds (l 972) has aptly pointed out have accommodated a widespread Pangaean that the Siphlonuridae (in its broadest sense) distribution that, via subsequent Lower Cre­ is a "stem group" and that fossils are difficult taceous vicariance, would have resulted in to assign when they may belong to Siphlonu­ the disjunct distribution of Protoligoneuria ridae or another family that is derived with in West Gondwana. it. Nevertheless, although in this case family Certain inferences about the habit and hab­ placement would be more definitive by using itat of Protoligoneuria can be made based on adult characters, certain similarities between larval morphology. Larvae offossil siphlonu­ the larvae of P. limai and known larvae of rids and hexagenitids, including Protoligo­ Hexagenitidae strongly suggest the placement neuria, as well as several extant genera, have ofProtoligoneuria in the Hexagenitidae. Such swimming tails. These are robust caudal fil­ classification of larvae is not without prece­ aments that possess interlocking setae on the dent, since both Siberiogenites (Sinitshen­ inner borders of the cerci and lateral borders kova, 1985) and Mongologenites (Sinitshen­ of the median terminal filament (fig. 1). In kova, 1986) were placed in Hexagenitidae modern mayfiies, swimming tails are undu­ without the benefit ofassociated adult fossils. lated up and down along with the minnowlike The dramatically enlarged and outspread abdomen to provide propulsion in swim­ gill 7 is perhaps the most evident similarity ming. This behavior is exemplified by extant between Protoligoneuria and, for example, Ameletus species. Such tails can also be used Mongologenites (Sinitshenkova, 1986) and for stabilization in positively rheophilic may­ Ephemeropsis (Sinitshenkova, 1975) of the files such as current-dwelling (un­ Hexagenitidae. This trait could easily be in­ published data). terpreted as a synapomorphy. Also, the an­ I hypothesize that the enlarged terminal terior rib or thickening of the gill lamellae of abdominal gills of Protoligoneuria and other Protoligoneuria is common to at least Hex­ hexagenitids are also an adaptation for swim­ ameropsis and Ephemeropsis, and this may ming. Such outspread gills would theoreti­ represent a synapomorphy as well. cally contribute additional thrust during the Protoligoneuria does differ in some detail dorsoventral undulations of the abdomen. from other genera of Hexagenitidae. Gill 7 Anterior gill lamellae could not be undulated in mature individuals appears larger relative as such and remain relatively small. Also, to gill 6, than in other genera; however, Mon­ extra thrust would be more important in larg­ gologenites gills approach the same propor- er individuals, and th.is may explain the al- 28 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 195

quiet-water habitats (pools or slack edge­ waters of streams as well as ponds and littoral areas of lakes). Edmunds and McCafferty (1988) gave considerable evidence showing this to be the primitive type of habitat of schistonote Ephemeroptera, although the majority of extant mayflies are adapted for current dwelling. Also, short claws (fig. I) in extant species of primitive minnow mayflies such as Siphlonuridae are usually associated with quiet waters or habitats where they are used on solid rather than fine substrates (un­ published data). Meshkova (1961) concluded that the pres­ ence of leaf-shaped gills, weak legs, and strongly pubescent caudal filaments of the larvae of Ephemeropsis indicated that they had inhabited undisturbed waters; and in fact, all other hexagenitids from Laurasia have been considered lacustrine (Sinitshenkova, 1984). It is therefore probable that Protoli­ goneuria occurred in quiet waters where lar­ vae would swim, although a rheophilic ex­ istence cannot be entirely ruled out. Given the abundance of Protoligoneuria fossils, a major lentic environment, such as a lake with a considerable littoral zone, was probably present at the fossil site.

FAMILY SIPHLONURIDAE

SUBFAMILY SIPHLONURINAE (?) Siphgondwanus occidentalis, new genus, new species Figures 9, 10 TYPE: Larva, AMNH 43404 (fig. l 0). ETYMOLOGY: Siphgondwanus is a mascu­ line gender nomen comprised of an arbitrary combination ofletters alluding to a siphlonu­ rid from Gondwana. The specific epithet oc­ cidentalis is from the Latin, meaning western. Thus the names together refer to a siphlon­ urid from West Gondwana (an identifiable Fig. 9. Siphgondwanus occidemaiis larva, dor­ biogeographic entity during the Lower Cre­ sal reconstruction without legs. taceous that included the area now known as Brazil). DIAGNOSIS: The small hypognathous head lometry found in Protoligoneuria with regard with short antennae, minnowlike form of to the disproportionately greater growth of body, three short, robust tails with swimming gill 7 as larvae develop. hairs, and unique elliptical gill lamellae with Well-developed swimming by mayflies is posterior spinelike bristles (fig. 7) will distin­ basically, but not exclusively, associated with guish Siphgondwanus occidentalis. '.990 McCAFFERTY:EPHEMEROPTERA 29

LARVA: Dorsal aspect with all legs and left abdominal gills missing, and mouthparts not discernible. Body minnowlike and 9.0 mm long. Head relatively small, broader than long but narrower than thorax, apparently hypog­ nathous; antennae short, discernible portion very thin with length less than width of head, inserted anteriorly on head capsule. Gills (fig. 9) present on abdominal segments 1-6, pre­ sumably present on 7; gills ca. LO mm long, platelike, narrow-elliptical, inserted at pos­ terolateral corners of segments, with no fi­ brillifonn portion or tracheation discernible, but with row of long spinelike bristles along the posterior and apical margins; gill bristles ca. 0.2 mm long. Caudal filaments relatively short and robust; cerci 4.5 mm long, with dense row of long setae along inner margin only; median terminal filament developed (3.0 mm long), with dense row of setae on lateral margins. ADULT: Unknown. OTHER MATERIAL EXAMINED: None. DISCUSSION: Placement of Siphgondwanus in the family Siphlonuridae (in its broadest sense) is based primarily on the presence of Figs. 10, 11. 10. Siphgondwanus occidentalis larva, dorsal habitus, AMNH 43404 (holotype). generalized characteristics that typify the 11. Siphlonuridae (?) sp. 1 alate stage, dorsal habi­ family, such as the very short antennae (com­ tus, AMNH 44306. pared to the usually long antennae of Baeti­ dae), the clearly minnowlike body with swim­ ming tails, and the sharp posterolateral the subfamilies of Siphlonuridae are pro­ processes of the abdomen that appear to be gressively becoming recognized at the family present. More importantly, however, there level by many Ephemeroptera workers, and are no synapomorphies present in Siphgond­ such revision of rank status has become al­ wanus that would place it with any other ex­ most inevitable. tant or extinct family of Ephemeroptera, in­ Little can be inferred about possible rela­ cluding Hexagenitidae. Erection of a new tionships with other siphlonurid genera be­ family for this fossil does not appear war­ cause the unique gill spination of Siphgond­ ranted at this time; nevertheless, placement wanus would appear to be an autapomorphy. in Siphlonuridae, based on plesiomorphic Short bristles are present on the margins of characteristics, certainly leaves the higher some of the platelike gills of some species of classification of Siphgondwanus open to fu­ the extant Holarctic siphlonurid genus Ame­ ture review as distinguishing apomorphies in letus and are scattered over the surface of the the primitive stem groups of Ephemeroptera very highly specialized gill lamellae ofcertain become better understood. extant Coloburiscidae. The gill spination of By considering Siphlonuridae in its broad­ Siphgondwanus(fig. 9), however, appears very est sense {e.g., McCafferty and Edmunds, different. The sharp marginal gill bristles of 1979), the absence of synapomorphies in Siphgondwanus should not be confused with Siphgondwanus would additionally place it the unsclerotized marginal filaments associ­ in the subfamily Siphlonurinae. Characters ated with the gill lamellae of the Epheme­ that would allow more definitive subfamilial roidea. classification are, unfortunately, not repre­ The oldest fossils assignable to the Siph­ sented on the fossil. It may be noted also that lonuridae appear to be the Jurassic genera 30 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 195

and Amphinotic distribution. Edmunds (per­ sonal commun.) considered this former ex­ istence in tropical South America to be pre­ dictable since the family crossed the equator at least twice and the most likely place was West Gondwana. The relationships between morphology, swimming habit, and habitat are given under the discussion of Protoligoneuria above. Be­ cause the two genera possess similar swim­ ming tails, the conclusions about the possible habitat ofProtoligoneun·a in Brazil also apply to Siphgondwanus.

SIPHLONURIDAE (!) sp. l Figure 11 MATERIAL: AMNH 44306, alate. DESCRIPTION: Dorsal aspect of head, tho­ rax, and forewing. Body length unknown. Forewing 9 .0 mm, elongate-triangular; cross­ venation well developed; Rs forked just ba­ sad of midlength of wing; MA fork in distal 1 14 of vein; MP2 and CuA not arched poste­ riorly at base; other venation not clear. Hindwings missing. DISCUSSION: The elongate-triangular shape of the forewing, along with the very distal Figs. 12, 13. Siphlonuridae (?} spp. alate stage. 12. Sp. 2 lateral habitus, AMNH 44313. 13. Sp. 3 MA fork and relatively distal Rs fork suggest lateral habitus, AMNH 43477. that this fossil represents a siphlonurid. Such a placement, however, must be considered tentative at the present. This fossil does not match any of the other alate Ephemeroptera Mesobaetis (nee Baetidae) (Brauer et al., fossils from Brazil. 1889), Olgisca (Handlirsch, 1908; Demoulin, l 970a), and Stackelbergisca (Tshernova, SIPHLONURIDAE (?) sp. 2 196 7). Proameletus (Sinitshenkova, 197 6) and Figure 12 Australurus (Jell and Duncan, 1986) from the Lower Cretaceous appear to be the only other MATERIAL: AMNH 44313, alate. described Mesozoic genera assignable to DESCRIPTION: Lateral aspect of female in­ Siphlonuridae, although the closely related cluding thorax, abdomen, legs, partial caudal families Epeoromimidae and Hexagenitidae, filaments, and forewing. Body at least 9 .0 mm as well as possibly Ameletopsidae and Co­ long (head missing). All legs well developed; loburiscidae, all of which have primitive foreleg subequal in length to other legs. Fore­ minnowlike larvae, are also represented in wing 7.8 mm long, subtriangular; crossvena­ the Mesozoic. Previously, fossil siphlonurids tion extensive; Rs forked in basal third; MA 1 have been known mainly from the Northern forked in distal /4 of vein; MP2 and CuA not Hemisphere, but also from Australia. arched posteriorly at base; other venation not Extant Siphlonuridae are not known from clear. Hindwings missing. Only one caudal tropical South America, although three filament discernible. subfamilies are amphinotic. The presence of DISCUSSION: Again, the few characters Siphlonuridae in West Gondwana is not sur­ present suggest Siphlonuridae but do not al­ prising, however, given its present Holarctic low more than a preliminary classification. 990 McCAFFERTY: EPHEMEROPTERA 31

SIPHLONURIDAE (?) sp. 3 Figure 13 MATERIAL: AMNH 43477, alate. DESCRIPTION: Lateral aspect (?adult) (?sex) with head, thorax, abdomen, partial caudal filaments, forewing, and partial hindwing. Body 12.0 mm long. Forewing 9.2 mm long, distinctly triangular; costal, subcostal, and radial triad crossvenation well developed, other crossvenation not as well developed; marginal venation apparently developed, de­ tails not discernible, but at least some inter­ spaces with free, short intercalaries; Rs forked in basal third, MA forked in distal third of vein, connection of MA11 IMA, and MA2 not 14 discernible; MP 1 and IMP attached by cross­ Figs. 14, 15. Colocrus indivicum larva, dorsal vein near base, connection of MP2 not dis­ aspect. 14. Left foreleg. IS. Right half of partial cernible; CuA paralleling MP2; other vena­ abdomen, with gills 1-3 present, gills 4-7 missing. tion not clear. Hindwing at least 4.0 mm long; venation not discernible. Three caudal fila­ ments present; median terminal filament well developed. dicative. It alludes to the indication ofphyletic DISCUSSION: The apparent connecting bas­ relationships that this species provides with es of MA veins as well as MP veins is some­ respect to the Oligoneuriidae. what suggestive of Baetidae, but overall the DIAGNOSIS: The combination of a broad­ wing is more typical ofSiphlonuridae (certain ened, shortened, and rounded head capsule, Siphlonurus, e.g., have IMP and MP attach­ filtering setae on the forelegs, broadened fem­ 2 ora, and an abdomen with posterolateral pro­ ing to MP1 by crossveins only). The left cer­ cus and median terminal filament are intact cesses and lateral elliptical gill lamellae will on this fossil, making it initially appear to be distinguish the larva of this taxon. The alate two-tailed. However, close examination also form is distinguishable by the lack of most reveals a short broken base ofthe right cercus. intercalary veins in the forewings, crossvena­ Placement in the Siphlonuridae is tentative tion restricted to the costal and radial region, because the cubital region of the forewing is and gemination ofcertain major convex veins obscured. with concave veins as shown in figure 19. LARVA: Dorsal aspect of fossil. Antennae, abdominal segments 8-10, caudal filaments, FAMILY OLIGONEURIIDAE tarsi, and claws missing. Body length (from SUBFAMILY COLOCRURINAE, apex of head through abdominal segment 7) NEW SUBFAMILY 10 mm. Head 3 mm wide and 2 mm long; anterior and lateral margins rounded, form­ Colocru.s indivicum, ing hemispherical, apparently depressed head new genus, new species capsule; eyes, antennae, and mouthparts not Figures 14-17, 19 discernible. All legs with femur broad and TYPE: Larva holotype, AMNH 43484 (fig. flattened, ca. twice width of tibia; forelegs 16). Adult paratype, AMNH 43499 (fig. 17). with rows of long, dense setae on inner mar­ ETYMOLOGY: Colocrus is a neuter gender gins of femur and tibia (fig. 14); hindlegs with nomen from the Latin colatus (filter or strain­ coxae not apparent and therefore not abnor­ er) and crus (leg), an allusion to the filtering mally overdeveloped. Abdominal segments sieve formed by the filtering hairs of the fore­ (fig. 15) with well-developed, pointed pos­ legs of the larva. The specific epithet indivi­ terolateral extensions with slightly curved in­ cum is from the Latin indivicus, meaning in- ner and outer margins on at least segments 32 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 195

consisting of C, Sc, R 1, Rs (giving rise to at least R2, ?R3, and ~+ 5 ), MAI> MA2, MP1, IMP, MP2, CuA1, CuA2, CuP, A 1, A2, and A3 (fig. 19); crossveins apparent only in costal region and distal R1 and R 2 areas; Rs forked at ca. 1/3 distance from base; other major forks near base; cubital region with one and pos­ sibly more intercalaries; R4+s and MA1 run­ ning near each other; MA2 and MP 1 gemi­ nating very near each other for entire length; MP2 and CuA, forming close paralleling pair for entire length. CuP paralleling CuA basally

and CuA2 distally. Hindwings ca. 3.3 mm long; venation not discernible. OTHER MATERIAL EXAMINED: None. DISCUSSION: The Oligoneuriidae belong to a complex offamilies that include Coloburi­ scidae, Isonychiidae, Oligoneuriidae, and Heptageniidae (Edmunds et al., 1976; McCafferty and Edmunds, 1979; Landa and Soldan, 1985). These mayflies share a com­ mon siphlonurid-like ancestor (McCafferty and Edmunds, 1979) and do not appear to have given rise to any other taxa. Coloburi­ scidae and Isonychiidae were considered subfamilies of Siphlonuridae earlier (Ed­ Figs. 16, 17. Colocrns indivicum larval and alate munds et al., 1963), but, as their affinities habitus. 16. Larva, dorsal habitus, AMNH 43484 became better known, they were recognized (holotype). 17. Alate stage, ventral habitus, AMNH 43499 (paratype). as subfamilies of the Oligoneuriidae by Riek (197 3), McCafferty and Edmunds (1979), and Landa and Soldan (1985). Based on my cla­ 4-7 (other present segments not discernible distic analyses, I now recognize the Coloburi­ in this area); abdominal segment 4 extended scidae and Isonychiidae as families separate posteriorly at lateral margin ca. 1$ length of from the Oligoneuriidae. Recognition offam­ segment 5, 5 and 6 extended ca. 1h length of ily status for Isonychiidae has been followed segments 6 and 7 respectively; lateral margins by several authors (e.g., Demoulin, 1958; of segments with short robust setae, or spurs; Tshemova, 1970). The exact phyletic posi­ single, dorsal platelike gills apparent on seg­ tion of Coloburiscinae has not been known ments 1-3 (and presumably 5-7), elliptical, (McCafferty and Edmunds, 1979; Landa and as long as or slightly longer than segment Soldan, 1985), but its newly discovered ear­ length, oriented posterolaterally, and with no liest branched position within the complex fibrilliform portion discernible, however, (fig. 18) necessitates its elevation to family small darkened area at base of gill 2 possibly status: Using cladistic principles, if two or indicating presence of fibrilliform portion. more families are recognized in this complex, Development of median terminal filament Coloburiscidae must be one of them. unknown. I also herein recognize the monophyletic ADULT: Ventral aspect of fossil (sex?) with grouping of the families Coloburiscidae, legs and caudal filaments missing; hindwings Isonychiidae, Oligoneuriidae, and Heptagen­ present but venation not discernible. Body iidae (fig. 18) to constitute a separate super­ 13.2 mm long. Head short, nearly as wide as family Heptagenioidea (nee Edmunds and pronotum; compound eyes apparently situ­ Traver, 1954a). I do not agree with Tsher­ ated dorsolaterally, ca. 1$ as wide as head. nova (1970) that the Epeoromimidae (= Forewing ca. 11.0 mm long, longitudinal veins EpeoromirrJUS + Foliomimus), which is f

990 McCAFFERTY: EPHEMEROPTERA 33

known from Jurassic and Cretaceous Pale­ arctic larvae (Tshernova, 1969; Sinitshen­ kova, 1976, 1985), should be placed in the Heptagenioidea as was done by Landa and Soldin (1985) andas is suggested by its place­ ment in Hubbard's (1987) catalog. Larvae of Epeoromimidae are basically minnowlike, have an apparent hypognathous head, lack filtering setae on the forelegs, and demon­ strate no apomorphies that would associate them with the Heptagenioidea. Edmunds ( 1972) even suggested a placement of Epeo­ romimus in Siphlonuridae but took no formal action. -HEPTAGENIOIOEA My interpretation of the sequence of der­ ivation of familial lineages in this superfam­ ily is shown in figure 18. It differs significantly Fig. 18. Oadogram of the families of Hepta­ from the scheme formerly presented by genioidea. McCafferty and Edmunds (1979). My cladis­ tic evidences for this new evolutionary hy­ pothesis and higher classification, which are ancestral passive filtering capacity and as­ based on considerable morphology, including sociated foreleg characteristics and became new cephalic characters and functional mor­ highly depressed dorsoventrally and oriented phology of the larvae, as well as adult char­ to bottom feeding in streams. acters and internal anatomy, will be elabo­ The larva of Colocrus possesses character­ rated in another publication. Briefly, however, istics typically found in many of the Hepta­ they indicate that ancestrally this superfamily genioidea that clearly place the fossil in this had left the primitive quiet-water habitat of superfamily. The filtering forelegs (fig. 14) of mayflies (Sinitshenkova, 1984; Edmunds and Colocrus are similar to those found in Co­ McCafferty, 1988) and invaded fl.owing waters loburiscidae, Isonychiidae, and Oligoneuri­ where the habit of passive filter feeding of idae. The abdominal gills and posterolateral seston and the correlated well-developed fil­ processes of the abdomen (fig. 15) are of a tering setae on the forelegs and mouthparts type that could be found in any of the four had evolved. families, except Coloburiscidae, wherein the The earliest grade of evolution in the su­ gills are highly modified or absent. The flat­ perfamily is represented by the Coloburisci­ tened body and femora are typical ofthe Hep­ dae, a lineage that also demonstrates a con­ tageniidae and the more depressed forms of siderable number of autapomorphies. The Oligoneuriidae. The head is most typical of Isonychiidae, Oligoneuriidae, and Heptagen­ Heptageniidae, but certain oligoneuriids ap­ iidae share a subsequent common ancestor proach the degree of flattening, shortening, evidenced by several shared apomorphies. and broadening seen in Colocrus (fig. 16). The Isonychiidae,just as the Coloburiscidae, Therefore, from all available characters, the remained ancestrally minnowlike (the gen­ larva of Colocn-IS would be placed in the fam­ eral facies oflsonychiidae remain most sim­ ily Oligoneuriidae. ilar to the hypothetical siphlonurine ancestor The forewings ofthe alate form of Colocrus ofthis superfamily). Additional apomorphies are unique. Both intercalaries and crossvena­ are shared by Oligoneuriidae and Heptagen­ tion are reduced. Venation assignment (fig. iidae, although the Oligoneuriidae are also 19) is based on alternating convexity and intermediate in numerous characteristics be­ concavity oflongitudinal veins, which are re­ tween the more ancestral minnowlike fami­ liable landmarks for ascertaining venation in lies and the Heptageniidae. Both Oligoneu­ mayflies (Edmunds and Traver I954b) and riidae and Heptageniidae possess numerous are evident in this fossil. The fact that the autapomorphies, but Heptageniidae lost the easily located Sc vein appears as a furrow, 34 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 195

20

21

22

23 Figs. 19-23. Alate stage wings. 19. Colocrus indivicum, forewing reconstructed from right and left forewings, dorsal ( + = convex vein, - = concave vein). 20, 21. Australiphemera revelata. 20. Forewing composite. 21. Partial hindwing from paratype (CA= costal angulation). 22. Microphemera neotropica forewing. 23. Pristiplocia rupestris forewing (DCul = distal cubital intercalary vein). 990 McCAFFERTY:EPHEMEROPTERA 35

and R 1 as a ridge in this fossil clearly indicates rived with Colocrus (Colocrus appearing to that the ventral side of the wings is being represent a transitional form), the Oligoneu­ viewed (fig. 17) (the dorsal aspect is shown riinae nevertheless possess wings that are dis­ in fig. 19). This follows because in the dorsal tinctly more derived than those of Colocrus. aspect these veins are always concave and Oligoneuriinae also possess a ventral gill 1 in convex, respectively. The venation appears the larval stage, whereas Colocrus retains a to be an intermediate form between a plesio­ dorsal gill 1. I therefore recognize two sepa­ morphic type, or wings with the full comple­ rate subfamilies for these clades, the Oligo­ ment of basic ephemeropteran venation, and neuriinae and the new subfamily Colocruri­ the highly reduced apomorphic type with nae. The new subfamily is distinct from extant geminating longitudinal veins that is found oligoneuriids both by its wing venation, and in the subfamily Oligoneuriinae of the Oli­ by its shortened and broadened head in the goneuriidae. In the oligoneuriine wing the larva. highly geminating convex and concave lon­ It may be significant that the general ap­ gitudinal pairs ofveins have the effect of flat­ pearance of the outspread wings of the fossil tening the wing (Edmunds, personal com­ (fig. 17) gives an impression very similar to mun.). This is evidently an adaptation that that of the outspread wings of dried speci­ neutralizes fluting, which is typical ofephem­ mens of oligoneuriine mayflies. Also, the eropteran wings. Flattening instead allows more faint impressions of the fossil wings in sculling and hence highly atypical rapid for­ comparison with the sharp image of the body ward flight. may indicate that these wings retained sub­ Major intercalaries, except for IMP, are ab­ imaginal sheaths. Adults of modem-day oli­ sent in Colocrus as they are in Oligoneuriinae, goneuriines are known to shed the subima­ but part of the radial and cubital area of the ginal exuviae everywhere but from the wings Colocrus wing retains remnants of the basic (Edmunds and McCafferty, 1988). The pro­ ephemeropteran venation. Oligoneuriine nounced convexity and concavity ofthe main forewings have the Rs, MA, and MP forked longitudinal veins of Colocrus (much more near the base, and Ri+s-MA1> MA:rCuA1> so than in other fossils studied) are also often and CuA2-CuP represent the major geminat­ more typical of subimaginal wings in extant ed pairs of convex and concave veins. A ten­ specimens of mayflies. dency toward this gemination is clearly evi­ True oligoneuriid fossils have not been dent in Colocrus. particularly with regards to known previous to this. Protoligoneuria is a R4+s and MA1, MA2 and MP1, and MP2 and hexagenitid, not an oligoneuriid as indicated CuA1 (fig. 19). This pairing corresponds to by Demoulin (1955) (see discussion under P. those major geminated pairs in oligoneuriine limai, above). Other than the Lower Creta­ wings (see e.g., Crass, 1947; Edmunds et al., ceous Colocrurinae and six genera of Hep­ 1976). Vein MA1 (fig. 19) is very faint in the tageniidae known from the Tertiary, the only fossil and does not appear as either a ridge other fossils that have been assigned to fam­ or furrow, although its position is where a ilies within Heptagenioidea, sensu novum, convex vein would be found. Although IMP are Afogzonurella larvae and Afogzonurus is well developed in Colocrus, other inter­ adults from the Palearctic Jurassic (Sinit­ calaries are missing in the open areas between shenkova, 1985), Siph/urites adults from the MA1 and MA2, MP1 and IMP, and IMP and Nearctic Miocene (Cockerell, 1923); Isony­ MP2 • The cubitoanal region of the forewing chia larvae from the Nearctic Oligocene of Colocrus is not highly reduced as it is in (Lewis, 1977); and Cronicus adults and sub­ Oligoneuriinae. imagos from the Palearctic Eocene (Eaton, Because of the several shared apomorphies 1871). Demoulin (1970b) placed Siph/urites in of the wings of Colocrus and Oligoneuriinae, the Isonychiinae; Demoulin (1974) placed the placement of this genus in Oligoneuri­ Cronicus in the Coloburiscinae; and Sinit­ idae, as was also indicated by the larval fossil, shenkova (1985) placed Afogzonurel/a and is strongly supported. Furthermore, although Afogzonurus in the Coloburiscinae. No known the Oligoneuriinae would appear to be de- fossil wings of Heptagenioidea, or any other BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 195

~ I interpret them as the same species. When ! dealing with fossils only, this is always a ":"!,, t ~ somewhat subjective decision, but in this sit­ ~ j uation the association appears most proba­ (.) a "" J ble. Considering them the same species is '.Il i I more prudent and conservative than describ­ '' ing two tax.a that demonstrate the same phy­ letic position within the Oligoneuriidae. Because of the monophyletic relationship of Chromarcys, Colocrus, and the Oligoneu­ riinae, I would prefer to recognize a single family with three subfamilies, which reflects

- OLIGONEURllDAE an evolutionary phenocline with regard to wing venation (fig. 24). However, since Chro­ - larval specializalion marcyinae represents the earliest derived group, it is a sister lineage to the common lsonychiine·like ancestor ancestor ofColocrurinae and Oligoneuriinae, Fig. 24. Oadogram of the subfamilies of Oli­ and family status within a strict cladistic clas­ goneuriidae. sification could be argued. The similarity of the larvae of Chromarcyinae to all other Oli­ goneuriidae generally, and the fact that its Ephemeroptera for that matter, are similar dorsal gill l is no longer unique in the family to those of Colocrus, and no fossil mayfly since Colocrurinae also possesses this char­ larvae are similar to Colocrus. acter state, would support retention of Chro­ McCafferty and Edmunds (1979) consid­ marcys in the Oligoneuriidae. ered the Oriental genus Chromarcys to be in With regard to a possible relationship of the subfamily Chromarcyinae of the Oligo­ Oligoneuriidae to Hexagenitidae, Tshernova neuriidae. Demoulin (1971) considered it in and Sinitshenkova (19 74) believed Hexage­ the family Chromarcyidae, believing it in­ nitidae to be an ancestral group closely re­ termediate between the fossil family Hex­ lated to Siphlonuridae but not continuing be­ agenitidae (see discussion under Protoligo­ yond the Mesozoic (probably not beyond the neuria. above) and Oligoneuriidae. The genus Lower Cretaceous) nor giving rise to other is of considerable evolutionary interest be­ modern taxa. Since the common ancestor of cause its larvae are typical of the Oligoneu­ the Heptagenioidea was derived from a siph­ riinae (a strong resemblance to the Neotrop­ lonurid-like ancestor, Demoulin's (1971) ical Spaniophlebia), but its adult retains a full phyletic conclusions about a close relation­ complement of ephemeropteran wing veins. ship between Hexagenitidae and Oligoneu­ It is therefore obviously not as derived as the riidae are not surprising, nor can a hypothesis Colocrurinae or Oligoneuriinae (fig. 24). Gill that the Hexagenitidae are cladistically relat­ 1 of Chromarcys is dorsal, a clearly plesio­ ed to the ancestor of the Heptagenioidea be morphic trait that is also retained in Colo­ dismissed out-of-hand. However, venational crus. All Oligoneuriinae, however, possess an characteristics of Hexagenitidae are not like apomorphic ventral gill 1. those found ancestrally within the Heptageni­ The phyletic position of Colocrus (fig. 24), oidea, and the larvae of Hexagenitidae were being more derived than Chromarcys and not stream-dwelling passive suspension feed­ being a sister group to the Oligoneuriinae but ers as are the larvae ofplesiomorphic lineages more plesiomorphic than that group, is easily of Heptagenioidea. deduced based on its distinct but interme­ Ancestral genera of Oligoneuriidae along diate type of wing venation. The retention of with the family as a whole are essentially Pan­ a dorsal gill l as well as a highly specialized tropical. The origin of the highly derived head of the larva does not falsify this deduc­ north-temperate elements of Oligoneuriinae tion. It is because of this general cladistic ( and Oligoneurisca) may very agreement of the alate and larval fossils that well have been in the Neotropics (see Mc- 1990 McCAFFERTY: EPHEMEROPTERA 37

Cafferty et al, 1990). Edmunds (1975) re­ garded the minnowlike Afrotropical genus Elassoneuria to be the most ancestral oligo­ neuriine, and he hypothesized that the family evolved on the South America-Africa-Mad­ agascar-India land mass. The age and loca­ •ion of Colocrus add considerable credence to that hypothesis and indicate that the Oli­ goneuriidae originated at the latest in the Lower Cretaceous. The biogeography and common derivation ofHeptageniidae and Oligoneuriidae (fig. 18) seem to suggest that the Heptageniidae also originated in the Southern Hemisphere. The family Heptageniidae is not known from South America either as fossils or extant fau­ na; however, it is represented by an extant Afrotropical and Oriental fauna along with its profuse Holarctic representation. I deduce that the habitat of Colocrus larvae was running water because all known colobu­ riscids, isonychiids, and all other oligoneu­ riids also possess forelegs with highly devel­ oped rows of filtering setae. Modem mayflies that possess highly developed rows offiltering Figs. 25, 26. Australiphemera reve/ata alate stage. 25. Lateral habitus, AMNH 44300 (holo­ setae on the forelegs passively filter seston type). 26. Dorsal and dorsolateral habitus, AMNH from water and require a current to accom­ 44310 (paratype). plish this mode of feeding (Wallace and O'Hop, 1979; Keltner and McCatferty, l 986). Virtually all heptageniids, although not nec­ forked at midlength, a relatively distinct dis­ essarily filter feeders, are also stream dwell­ tal arch of CuP, A1 nearly straight, and no ers. cubital intercalaries; and a hindwing with a well-developed costal angulation, and Rs FAMILY EPHEMERIDAE much shorter than R2 and ~+s· LARVA: Unknown. Australiphemera revelata, ADULT: Lateral and dorsal head, lateral and new genus, new species dorsolateral thorax, forewings and partial Figures 20, 21, 25, 26 hindwings of fossils. Sex unknown but prob­ TYPES: Adult holotype, AMNH 44300 (fig. ably female based on small size of eyes. Body 25). Adult paratypes, AMNH 44310 (fig. 26). length unknown. Head ca. 2.0 mm wide, with ETYMOLOGY: Australiphemera is a femi­ ca. 0. 5 mm diameter compound eyes widely nine gender nomen comprised of an arbitrary separated and appearing lateral. Forelegs ap­ combination ofletters incorporating the Lat­ parently well developed. Forewings (fig. 20) in root australis, meaning southern, and al­ 11.0-11.5 mm long; crossvenation and mar­ luding to an ephemerid from the Southern ginal venation moderately developed; Sc vis­ Hemisphere. The specific epithet revelata is ible for entire length; longitudinal veins not from the Latin, meaning revelatory, an illu­ geminating; venation of radial triad exten­ sion to the historical information that this sive; Rs forked in basal t/4; MA fork at ca. species has revealed. m.idlength of wing; MP2 and CuA strongly DIAGNOSIS: This taxon can be distin­ decurved at base, running singularly for en­ guished by the following combination: rela­ tire length, not connected basally; series of tively small, widely separated compound eyes; forked and single veinlets attaching CuA to a typical ephemeroid forewing, and with MA anal margin, distal veinlets running subpar-

~------·-- 38 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY

.. tant to differentiating the remaining families ...... "' J;! "O :i;"' Potamanthidae and Ephemeridae, is not dis­ I i c.. j .~ E I>. E cernible on the fossils. In addition the costaI .c.. "'c: :5 0. 'i; :::I ~ Q. angulation of the hindwing (fig. 21) could ap­ CD I w w i ply to either of these families. Two characteristics of Australiphemera, however, have prompted my placement of it in the Ephemeridae. Typical ofEphemeridae and quite unlike that found in Potamanthi­ dae, Rs of the hindwing is shorter than R2 LEFTOPHLEBIOlDEA- and R4+s (a plesiomorphic trait). A Rs longer than the fork formed by R 2 and R4+s is one of the distinguishing apomorphs in Pota­ manthidae. Also cubital veinlets in the fore­ siphlonurid·like ancestor wing of this species are attached as is typical Fig. 27. Cladogram of Leptophlebioidea and of Ephemeridae. Many potamanthids have Ephemeroidea. some short, free veinlets that are unattached to CuA in this region of the forewing. Placement of Australiphemera in the fam­ allel to CuA, most distal veinlets ending in ily Ephemeridae is nonetheless somewhat by outer margin, at least basal veinlet forked; default because venational characteristics of cubital intercalaries lacking, free cubital vein­ this family are plesiomorphic within the lets not apparent; CuP slightly sigmoid, Ephemeroidea, and defining apomorphies of strongly arching toward anal margin; Ai near­ this family are found mainly in genitalic and ly straight, apparently not forked, no other larval characteristics. McCafferty ( 1979) veins or veinlets discernible in anal area. showed the phyletic relationships of the Hindwing (fig. 21) 4.4 mm long; costal an­ Ephemeroidea, and figure 27 is adapted from gulation well developed, anterior base of wing this. Because of the plesiomorphic nature of and costa forming near right angle; crossve­ the wings and lack of other data, these fossils nation moderately to well developed; basal could theoretically be represented on the

attachment ofR1 apparently at wing base; Rs cladogram anywhere along the ephemeroid shorter than R 2 and ~+s; posterior venation line to the Ephemeridae. This species, there­ not discernible. Abdomen and caudal fila­ fore, although at present classifiable in the ments missing. Ephemeridae, may actually represent a pro­ OTHER MATERIAL EXAMINED: None. to-Ephemeridae or possibly a proto-Pota­ DlsCtJSSION: The venational data available manthidae-Ephemeridae. Moreover, the ori­ on the two specimens clearly indicate place­ entation of veins in the cubital region of the ment in the superfamily Ephemeroidea. Al­ forewing of Australiphemera is somewhat though anal venation beneath Ai cannot be suggestive of a condition intermediate be­ discerned, the well-developed costal cross­ tween a plesiomorphic ephemeroid state (e.g., venation of the forewings as well as the quad­ fig. 22) and that found in certain Euthyplo­ rate costal arch and short Rs of the hindwing ciidae, such as Mesoplocia. do not indicate placement in the Neoephem­ Fossil Ephemeridae have been found pre­ eridae (). Also on the basis of ve­ viously only from the Eocene and Oligocene national characteristics, which are relatively of the Northern Hemisphere. The Jurassic plesiomorphic within the Ephemeroidea, the fossil of Weyenbergh (1874) that was classi­ ephemeroid families Behningiidae, Palinge­ fied as was incorrectly placed to niidae {Palingeniinae), , and superfamily and requires reevaluation. Euthyplociidae can be excluded from consid­ Tshemova (1977) described the fossil larva eration for placement of these fossils. Un­ Mesogenesia from the Palearctic Upper Ju­ fortunately, venation in the anal region of the rassic and placed it in the Palingeniidae. The forewing, particularly an indication of wheth­ placement of this larva was based on its pos­ er A1 is truly forked or not, which is impor- session of expanded foretibiae, possibly in- 1990 McCAFFERTY: EPHEMEROPTERA 39

dicating a fossorial habit. However, the specifics of the burrow formed varies with strongly developed tarsi, which appear as if the functional morphology of the different they may oppose the tibiae in a chelate or genera (see Keltner and McCafferty, 1986). raptorial fashion, are atypical of extant It should be emphasized, however, that the ephemeroid larvae. The larva described as occurrence of Australiphemera and other Archaeobehningia (Behningiidae) from the possible ephemerids from the Lower Creta­ Palearctic Upper Jurassic by Tshernova ceous does not absolutely indicate that this (1977) is not convincingly an ephemeroid type of burrowing habit was already evolved based on the published data. Nevertheless, at that time because the plesiomorphic ve­ from the age of Australiphemera and the po­ national characters apparent in these fossils sition of Behningiidae being more ancestrally could have been present before this type of derived, Behningiidae certainly could have fossorial habit evolved. With respect to this, been present prior to the Lower Cretaceous. it should also be kept in mind that both early Mccafferty and Edmunds (1979) hypoth­ branched and later branched lineages in the esized that the Ephemeroidea were derived Ephemeroidea, the Potamanthidae and Eu­ from the Leptophlebiidae. It now appears, on thyplociidae, respectively (see fig. 27), as well the basis of unpublished data, that the two as some Polymitarcyids, do not demonstrate groups actually shared a common ancestor true burrow-forming fossorial habit and pre­ and that both are monophyletic (fig. 27). sumably represent a more ancestral habit of Tshemova (1971) placed Cretoneta from the sprawling or living in crevices among mixed Palearctic Upper Cretaceous and Mesoneta substrate. from the Jurassic in the subfamily Mesone­ tinae of the Leptophlebiidae. Hubbard and Microphemera neotropica, Savage (19 8 1) believed the familial place­ ment of Mesoneta to be tenuous. The family new genus, new species Leptophlebiidae is, however, at least as old Figures 22, 28 as the Lower Cretaceous and probably older TYPE: Adult (sex?) 43301 (fig. 28). in light of the fossil Ephemeroidea reported ETYMOLOGY: Microphemera is a feminine herein (see also fossils tentatively described nomen consisting of an arbitrary combina­ as leptophlebiids herein). The family Lep­ tion ofletters derived in part from the Greek tophlebiidae is older if Mesogenesia is an micros, meaning little, and alluding to an ephemeroid, or Mesoneta a leptophlebiid. ephemeroid genus of small size. The specific Extant Ephemeridae are nearly cosmopol­ name neotropica is a noun in apposition and itan but are poorly represented in the Neo­ an allusion to the Neotropical distribution of tropics (McCafferty et al., 1990) by three the species. species of ; Ephemera and Afro­ DIAGNOSIS: This taxon can be distin­ mera are represented in the Afrotropics by a guished by the combination of a relatively few species, and, although the genus Ichthy­ small body and wing size, in the forewing the botus occurs in New Zealand, ephemerids are position of the MA fork basad of midlength absent from Australia. One may assume from of the wing, the composition of the cubital the presence of Australiphemera and other intercalaries, the moderate distal arch ofCuP, ephemeroids reported below from Brazil that and A 1 that is apparently unforked and pos­ some early stock was widespread (perhaps sesses only one veinlet. Pangaean) but became extinct later in South LARVA: Unknown. America. McCafferty et al. ( 1990) hypothe­ ADULT: Partial lateral head, lateral thorax, sized that Hexagenia became distributed in forewing, partial hindwing, lateral abdomen, South America in the Tertiary with a center and partial caudal filament of fossil. Sex un­ of origin in North America. known. Body 8.5 mm. Forewing 7.8 mm long; Larvae of Ephemeridae burrow into soft costal and other crossvenation well devel­ substrates of slower reaches or depositional oped; Sc visible for entire length; longitudinal areas of streams and rivers as well as shallow veins not geminating; venation ofradial triad silt bottoms of lakes and occasionally ponds. extensive; Rs forked in basal l/<1 of wing; MA The exact consistency of the substrate and forked at ca. 3/s distance from wing base; MP2 40 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 195

evidence is available, the apparent absence

ofan A 1 fork leads me to place Microphemera in the Ephemeridae rather than the Pota­ manthid.ae. Four short crossveins are visible in the anal area of the forewing (fig. 28). These could initially be interpreted as a series of anal vein­ lets, but close examination reveals that the supposed veinlets are actually costal cross­ veins of the underlying remnant of the hindwing, and what may have been inter­

preted as the A 1 is actually the costa of that hindwing. This is evidenced by the fact that this costa (supposed A 1) with its attached crossveins crosses the distal end of CuP in the forewing. The forewing venation ofAustra/iphemera and Microphemera is very similar (figs. 20, 22), except the MA fork is slightly more ba­ sad, and CuP is less arched distally, and the cubital veinlets are slightly shorter and more posteriorly oriented in Microphemera. These characteristics and the distinctly smaller size of Microphemera clearly distinguish the two taxa. Any suggestion that Australiphemera and Microphemera are merely dimorphic Figs. 28, 29. Alate habitus. 28. Microphemera sexes of the same species is not supportable. neotropica lateral habitus, AMNH 44301 (holo­ Although some sexual size dimorphism is type). 29. Pristiplocia rupestris dorsal habitus, present in extant species of Ephemerid.ae, it AMNH 44308 (holotype). is not exaggerated. The history of Ephemeridae and Ephem­ eroidea, and the biogeography and larval and CuA strongly arched posteriorly at base habitat of extant ephemerids are treated un­ (based on wing dimension and distal orien­ der the discussion ofAustraliphemera, above. tation); forked basal veinlet and four distal, As a footnote, it may be of some interest that long, single veinlets attaching CuA to anal in several respects, including the reduced anal margin, no cubital intercalaries present, no area, the forewing of Microphemera is rem­ free cubital veinlets present; CuP sigmoid and iniscent of those of the primitive extant genus only moderately arched distally toward mar­ lchthybotus, which is the only ephemerid gin; A, not discernible distally but with one known from the Australian biogeographic strong veinlet attaching it to anal margin, ap­ realm, specifically New Zealand. parently unforked. Hindwing and caudal fil. aments too incomplete to describe. FAMILY EUTHYPLOCIIDAE OTHER MATERIAL ExAMINED: None. Pristiplocia rupestris, DISCUSSION: Much of the discussion re­ new genus, new species garding superfamily and family placement Figures 23, 29 under the genusAustraliphemera, above, also applies to Microphemera. In Microphemera TYPE: Adult (sex?) AMNH 44308 (fig. 29). the A 1 vein (fig. 22) is visible to the distal ETYMOLOGY: The feminine generic nomen point where it is attached to one strong vein­ Pristiplocia is an arbitrary combination oflet­ let, and from the space distad of this it would ters based in part on the Latin pristinus, be difficult to envision a fork having been meaning early or primitive, and alluding to present. Therefore, although no hindwing a primitive euthyplociid. The trivial name

;_,. ·~ !990 McCAFFERTY: EPHEMEROPTERA 41

rupestris is Latin, meaning '"of rocks" and forewing (fig. 23) as follows: a basally de­ alluding to the fossil nature of the species. c~rv~d MPz. and CuA; nongeminating lon­ DIAGNOSIS: This taxon is distinguishable gitudmal veins; an Sc visible for its entire by its size and a forewing consistent with fam­ length; a MA fork in the basal third; a cubital ily characteristics of Euthyplociidae as well region with sigmoid veinlets attaching CuA as with a uniquely veined cubital area. This to margin; no free anal veinlets at the anal cubital area consists of an intercalary vein margin; and very extensive crossvenation that originates in CuA and terminates in the throughout the wings. outer margin and runs parallel to CuA, and The fossil genus is very similar to extant other, more basal, sigmoid intercalaries sub­ genera in the Euthyplociinae but differs in the paralleling CuA and attaching CuA to the detail of the cubital intercalary venation of anal margin. the forewing. Its venation is most similar to LARVA: Unknown. that of the Neotropical genus Campylocia (see ADULT. Dorsal aspect of fossil with head, Needham and Murphy, 1924). Both possess body, and most of forewing and part of a cubital intercalary that originates on CuA hindwing present; sexual characters not dis­ and ends near the posterior end of the outer cernible but possibly female due to small head margin of the wing. In Campylocia there is impression (small eyes). Body 13.2 mm long. sometimes a second such intercalary origi­ Head distinctly narrower than thorax but nating basad of the first on CuA but also par­ pronotum not apparent. Forelegs developed, alleling CuA. From the second intercalary, or at least 4.0 mm long; claws, segment ratios, first if it is the only one, sigmoid veinlets run and other legs not discernible. Forewing (fig. to the anal margin and additional similar 23) 13.0 mm long, with full complement of veinlets also attach CuA to the anal margin longitudinal veins and profuse crossvena­ in the more basal area. Pristiplocia differs from tion; Rs fork at 1/4 distance from base; MA this by having very long veinlets that run fork in basal 1/3 of wing, distad of Rs fotk; almost parallel to the first cubital intercalary MP2 and CuA strongly decurved posteriorly (fig. 23). These veinlets could thus be con­ at base; marginal areas with anastomosed sidered intercalaries, as I have done here. short intercalaries, at least three short inter­ Their homology, however, is apparently with calaries in areas between MP1 and IMP, IMP the veinlets of other euthyplociids. and MPz, and MP2 and CuA; cubital region Other variations in the euthyplociines in­ with one distal intercalary originating on CuA clude no cubital intercalaries but simply a and terminating in outer margin, three short series of parallel veinlets (such as in the Neo­ somewhat reticulated marginal veinlets at­ tropical genus Euthyplocia) and cubital in­ taching this cubital intercalary with margin; tercalaries present but not attached to CuA three or four additional, more basal, long, (such as in the Indonesian genus Polyplocia). slightly sigmoid intercalaries running diago­ The subparalleling, long veinlets (intercala­ nal from CuA to anal margin, paralleling each ries) of Pristiplocia represent an additional other and subparalleling distal cubital inter­ variation. calary; CuA and CuP attached basally by three The origin ofthe Euthyplociidae within the crossveins. Hindwing 4.8 mm long, with ve­ Ephemeroidea (McCafferty, 1979) (fig. 27) can nation and shape not discernible. Caudal fil­ now be interpreted as at least the Lower Cre­ aments not discernible. taceous. No other fossils of Euthyplociidae OTHER MATERIAL EXAMINED: Two adults have been known previous to this study, and (probably males) (AMNH 44304, 44307) of fossils ofits sister lineage Polymitarcyidae are questionable identity because of very incom­ known only from possible fossil burrows from plete wings. the Miocene. DISCUSSION: Pristiploda is very character­ The family Euthyplociidae has a disjunct istic of present-day Euthyplociidae and would Pantropical distribution, including the Neo­ easily be keyed to that family using keys to tropics, Afro tropics plus Madagascar, and the extant mayflies. All of the characters avail­ tropical Orient (Indonesia and North Bor­ able in the fossil agree with the family's char­ neo). The discovery of Pristiplocia indicates acteristics. These include characters of the that the family was probably well established

...... -.~·-·-·· .... , 42 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 195

ilies is problematic. The anal area ofthe fore­

wings is difficult to interpret. A 1 is evident and no veinlets can be seen. A second vein that is proximal and posterior to A 1 is nearly straight for its visible length, but is slightly curved in much the same manner as A 1• It could be interpreted as A2, in which case an argument could be made for placing the fossil in the Ephemeridae. However, because ofthe space and orientation in the anal area and the proximity ofthe visible basal aspects of these veins, it is more reasonable to envision that Fig. 30. Potamanthidae (?) sp. 1 alate stage, they were connected basally and actually rep­ dorsal habitus, AMNH 4431 l. resent the two branches ofa forked A 1• Iftheir lines are reconstructed by extending them ba­ sally they form a fork very typical ofthe Pota­ when such disjunct areas were joined in manthidae. Gondwana or southern Pangaea, and thus of­ This fossil, although also similar in size to fers, by way ofvicariance, an explanation for Microphemera (described above), differs from the present distribution of the family. it in that the A1 anal veinlet ofMicrophemera Modern euthyplociine larvae are for the bends in an opposite direction at its base and most part sprawlers and always occur in run­ forms an abrupt angle with A 1 (fig. 22). The ning water environments. Given the close re­ anal veinlet of Microphemera is thus typical lationship of Pristiplocia with present-day ofephemerid attaching veinlets in this region members of the family, one may assume the and not like Potamanthidae, wherein the in­ presence ofsome stream environment at least ner branch of A1 follows a continuous line in the vicinity of the Brazilian fossil site. (no abrupt angles) with the stem of Ai. and the outer branch is most divergent from the Ephemeroidea sp. 1 line of the stem of A 1• The small laterally Figure 30 oriented eyes of the fossil, particularly if it is MATERIAL: AMNH 44311, alate. a male, would lend some support to the ten­ DESCRIPTION: Dorsal head, thorax, partial tative potamanthid classification. abdomen, and partial forewings present. Sex If this fossil is indeed a potamanthid, it unknown. Head I. 3 mm wide including com­ would be the first fossil known of this family. pound eyes; eyes 0.3 mm wide in dorsal view, The fossil from Baltic amber described as widely separated and laterally oriented on priscus was correctly shown to head. Forewing 8.5 mm long; costal cross­ be a member of the Leptophlebiidae by De­ venation well developed; Sc visible for entire moulin (1968). Extant Potamanthidae are re­ length; longitudinal veins not geminating; ve­ stricted to the Holarctic and Oriental realms. nation of radial triad not discernible; Rs Their larvae are stream inhabitants found in forked in basal 1/4 of wing; MA forked slightly areas with mixed substrate and moderate cur­ less than midlength of wing; MP2 and CuA rent speed. arched posteriorly at base, running singularly for entire length, not connected basally; cubi­ Ephemeroidea sp. 2 tal region not clear; CuP sigmoid and strongly Figure 31 bent toward anal margin; A1 apparently forked, with no veinlets. MATERIAL: AMNH 43480, alate. DISCUSSION: This fossil is clearly an DESCRIPTION: Dorsal aspect of alate fossil ephemeroid as per Australiphemera and Mi­ (? adult female) with body and partial fore­ crophemera, discussed above, but has general wings, forelegs, and caudal filaments. Body plesiomorphic wing venation found in both 11. l mm long. Forewings at least 9.3 mm the Ephemeridae and Potamanthidae. The long; crossvenatiop well developed; longitu­ placement to one or the other of these fam- dinal veins not geminating; MA forked just McCAFFERTY:EPHEMEROPTERA 43

gills missing. Body 9 .2 mm long; cerci and median terminal filament at least 15 mm long. Head subquadrate, narrowing in cervical re­ gion, narrower than thorax, and possibly hy­ pognathous. Body elongate. Setae on caudal filaments not discernible. DISCUSSION: The shape of the head and body along with the relatively very long cau­ dal filaments give a strong indication that this fossil belongs in the extant family Lepto­ phlebiidae. A number of extant genera of Leptophlebiidae give a similar overall impression and fit this general size. Unfor­ tunately, gills, which would resolve the exact placement of this fossil, are missing. How­ ever, the narrow lanceolate-like gills com­ mon to some extant leptophlebiids, such as Paraleptophlebia. are easily and often broken off. No other fossil Ephemeroptera larvae from Brazil have caudal filaments approach­ ing the relative length of those of this spec­ imen. An unnamed species of Leptophlebiidae (Atalophlebia? sp. A) from the Eocene was Figs. 31, 3 2. A.late and larval habitus. 31. described from larval fossils from Argentina Ephemeroidea sp. 2 alate stage, dorsal habitus, by Rossi de Garcia ( 1983). This represents AMNH 43480. 32. Leptophlebiidae (?) sp. l larva, the only Ephemeroptera fossil known from dorsal habitus, AMNH 43474. South America other than those treated herein.

basad of midlength of wing; MP2 and CuA strongly arched posteriorly at base; cubital Leptophlebiidae (?) sp. 2 and anal regions not discernible. Median ter­ Figure 33 minal filament present and apparently well MATERIAL: AMNH 43476, alate. developed. DESCRIPTION: Lateral aspect offossil(? adult DISCUSSION: Not enough characters are female) including body, forewings, and par­ available on this fossil to allow even tentative tial head and caudal filaments. Body 6.1 mm identification beyond the superfamily long. Forewing 5.9 mm long; not triangular, Ephemeroidea. The ephemeroid venation of but posterior margin extensive and gradually .. ·. the posteriorly arched base ofMP2 and CuA curved; venation difficult to interpret because and the well-developed costal crossvenation one forewing overlying other; crossvenation are apparent and characteristic of epheme­ weak to moderately developed; marginal ve­ roid venation. The body and wing size of this nation not discernible; Rs forked at ca. lf4 specimen does not match any of the other length of wing; radial triad expansive; MA species of ephemeroids described herein. forked in distal half, fork apparently sym­ metrical; connection of MP2 and MP1 near base of wing; MP2 gradually and slightly Leptophlebiidae (?) sp. l curving toward anal margin; CuA closely par­ Figure 32 alleling MP2; cubital and anal areas not dis­ MATERIAL: AMNH 43474, larva. cernible. Hindwing, if present, not discern­ DESCRIPTION: Dorsal aspect of larval fossil ible. Median terminal filament apparently including head and body, left cercus, and me­ well developed. dian terminal filament; other appendages and DISCUSSION: The general shape of the wing, r!

44 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 195 l the generally weak crossvenation, the expan­ sive radial triad taking up a large portion of the wing, and the position of the MA fork all suggest the family Leptophlebiidae. The pres­ ence of three tails would also fit this family, and the small body size might exclude it from the Siphlonuridae, a group also possessing most of these characteristics. The paucity of characteristics on the fossil precludes any rea­ sonable assessment of relationships at this time. The terminal area of the abdomen of the organism appears to be damaged, and thus the base of one of the caudal filaments is not in alignment with the others. The displaced caudal filament base might be interpreted as an ovipositor since ovipositorlike structures are not uncommon in certain Leptophlebi­ idae. The base of this structure on the ab­ domen, however, is too posterior for it to be an ovipositor.

Leptophlebiidae (?) sp. 3 Figure 34 MATERIAL: AMNH 44312, alate. Figs. 33, 34. Leptophlebiidae (?) spp. alate stage, lateral habitus. 33. Sp. 2, AMNH 43476. DESCRIPTION: Lateral aspect (? adult fe­ 34. Sp. 3, AMNH 44312. male) with head, body, forewing, hindwing, and partial caudal filaments. Body 7 .5 mm long. Forewing 7.8 mm long; crossvenation FAMILY INCERTAE not clear; Rs forked at ca. 113 distance from base of wing; radial triad expansive· MA Incertae Sedis sp. I forked in distal half; MP forked near' base· MATERIAL: AMNH 43453. larva MP 1 intercalary present; cubital and anal ar~ not clear. Hindwing highly reduced, 0.6 mm DESCRIPTION: Dorsal aspect ofla~al fossil with head and body, partial forelegs. and par­ long; venation not discernible. Only two cau­ tial median terminal filament. Body.13.5 mm dal filaments discernible. long; median terminal filament at least 1.0 D1s~ss10N: A~ain, shape of the wing, wing mm long. Head flattened, slightly shorter than venat10_n, and size of the body are highly broad, rounded to slightly pointed anteriorly sugge~t1ve of Leptophlebiidae. Importantly, nearly straight and subparallel laterally; the highly reduced hindwing is reminiscent com~ pound eyes dorsal, width of one ca. 1/4 head of a number of extant Neotropical genera of width, antenna! bases dorsal, anteromedial Leptophlebiidae (e.g., Thrauiodes, Hagenu­ lus, Borinquena. Careospina, and Neohagen­ to compound eyes. Structural details of tho­ rax and abdomen missing; body apparently ulus). The basally oriented fork of MP how­ narrow-elongate. Median terminal filament e"'.er, is i:;ot as common among leptophlebiids very thin; cerci missing. with which I am familiar. It is relatively basal but not as basal in certain Neohagenulus and Incertae Sedis. sp. 2 Traverina. If this is indeed a leptophlebiid, ~hen a ~audal filament is either missing or it MATERIAL: AMNH 43423, larva. is atypically two-tailed as, for example, the DESCRIPTION: Dorsal aspect oflarval fossil leptophlebiid genus Careospina. including head, thorax, and abdomen, with 1990 McCAFFERTY:EPHEMEROPTERA 45

left and hind femora, and left cercus and pos­ 1975) because of differential rates of evolu­ sibly partial median terminal filament intact; tion in larval and adult stages. Whereas the other appendages and structural details miss­ family is easily defined on the basis of nu­ ing. Body 12.5 mm long; cerci at least 12.0 merous larval synapomorphies, only the gen­ mm long. Head capsule rounded anteriorly, era other than Chromarcys demonstrate the laterally subparallel, appearing longer than highly specialized rapid-flight wings. Thus, a broad. Femora not broadened. Thorax and lingering question has been whether to in­ abdomen giving impression of stout, non­ clude or exclude Chromarcys from the fam­ minnowlike body form; gills missing. Cerci ily. The newly discovered extinct genus Colo­ with sparse, long setae along both inner and crus appears to be an intermediate form with outer margins: median terminal filament ap­ regard to specialized wing venation (figs. 19, parently developed, marginal setae not dis­ 24). The intermediate, transitional forewing cernible. of Colocrus further supports the proposed re­ lationships of Chromarcys and Oligoneuri­ inae (Edmunds, 197 5). It would also CONCLUDING REMARKS strengthen the case for basing family classi­ New information gained from the study of fication in this instance on the larval syn­ the Lower Cretaceous mayflies from Ceara apomorphies, and including the Oligoneuri­ Crato, Brazil, directly impacts our under­ inae, Colocrurinae, and Chromarcyinae in the standing of mayfly evolution in a number of Oligoneuriidae. Although Colocrurinae is ways. It contributes to mayfly phylogeny and clearly a sister lineage to the Oligoneuriinae, the augmentation of theories based primarily it does not share the apomorphic ventral gill on extant taxa. It supplies times of origins or l in the larval stage with Oligoneuriinae but at least minimum ages oflineages and allows retains the plesiomorphic dorsal condition comparisons of Cretaceous fauna with mod­ also found in Chromarcyinae. ern and more ancient faunas. It also contrib­ New minimum times of existence for sev­ utes West Gondwanian distributions that are eral mayfly lineages are now demonstrable. of historical consequence and that allow ex­ The origin of the Oligoneuriidae was at least planations of some modern distributional Lower Cretaceous, and the Heptageniidae patterns. Besides this, the diversity and abun­ could be as old as the Lower Cretaceous (fig. dance of both terrestrial and aquatic life stages 18). If minimum ages are considered in re­ of mayflies strongly suggest the habitats and lationship to the branching sequences of lin­ paleoecology of the fossils. eages (figs. 18, 24) then one can further infer The mayfly superfamily Heptagenioidea is that the origins of the Heptagenioidea and here recognized as a monophyletic group of even Oligoneuriidae were actually earlier. The basically stream-dwelling larvae consisting of Heptagenioidea is derived from a siphlonu­ the families Coloburiscidae, Isonychiidae, rid like ancestor, and siphlonurid fossils are Oligoneuriidae, and Heptageniidae. Cladistic known from as early as the Lower or Middle relationships (fig. 18), which are being treated Jurassic. Moreover, recently discovered in detail elsewhere, support recognition of the Mogzonurella and /vfogzonurus fossils from families Coloburiscidae and Isonychiidae, the Jurassic were placed in the Coloburiscin­ separate from the Oligoneuriidae or Siphlo­ ae by Sinitshenkova ( 1985). Ifthis placement nuridae. The family Oligoneuriidae, which is is correct, the common ancestor to the Hep­ represented among the fossils studied, is tagenioidea dates to at least the Jurassic. shown to be phyletically intermediate be­ Both the Ephemeridae and Euthyplociidae tween the suspension feeding minnowlike can now be dated to at least the Lower Cre­ mayflies of the more plesiomorphic families taceous (fig. 27), based on the Brazilian fos­ Colo buriscidae and Isonychiidae and the bot­ sils. Again, extrapolating from phyletic tom-feeding and fiat-headed larvae of the branching sequences (fig. 27), the origin of more apomorphic family Heptageniidae. the Leptophlebioidea and the Ephemeroidea The systematics of the family Oligoneuri­ is expected to be earlier than this. This is idae has always been problematic (Edmunds, further supported by the discovery of ten ta- 46 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 195

tive leptophlebiids from Brazil. Finds by resented in the Neotropics today by only a Tshernova {1977) of possible Behningiidae few species with probable Nearctic affinities. and Palingeniidae fossils from the Upper Ju­ The extinct genera, however, appear to be rassic, along with the new discoveries of very primitive and may actually represent a Ephemeridae (or proto-Ephemeridae), Eu­ proto-Ephemeridae-Potamanthidae group thyplociidae, and possibly Potamanthidae based on their plesiomorphic wing venation. from the Lower Cretaceous of Brazil indicate It is not known whether these new genera are that the Ephemeroidea was well established close relatives to Ephemera. which is an an­ and already radiated into several of its fa­ cestral extant genus of Ephemeridae that is milial lineages in the Mesozoic. found in the Holarctic, Oriental, and Afro­ Biogeographically significant discoveries tropical realms but not the Neotropics. include finds oftaxa that may have been pre­ The extinct family Hexagenitidae, ~·­ dicted to have been in West Gondwana be­ though relatively well known from the North­ cause of present-day distributions and taxa ern Hemisphere Jurassic is shown to have that are quite unexpected. Examples of the also existed in the Southern Hemisphere former are Pristiplocia (Euthyplociidae) and Lower Cretaceous, and to have coincided Colocrus (Oligoneuriidae). These families during this era with many families that have have not actually been represented in the fos­ continued into the present, including the sil record previous to this study. The family closely related Siphlonuridae. Euthyplociidae is Pantropical in distribution Although the family Siphlonuridae (in its as is the family Oligoneuriidae, except for broadest sense) is not represented by extant some highly specialized genera. Widespread taxa in tropical South America or Africa, Gondwanian distribution followed by vicar­ Amphinotic genera are known from Trans­ iance associated with continental drift is antarctic areas, including the Chilean-Pata­ therefore suggested as the underlying basis for gonian area, and the group is widespread in the present-day disjunctions. the Holarctic. The discovery of siphlonurids Given the relationships between the Oli­ from West Gondwana provides the predicted goneuriidae and Heptageniidae (fig. 18), there geographic link between these disjunct extant exists a possibility of a Gondwanian origin distributions of this family. Most likely, the of Heptageniidae despite the fact that the family was widespread in Pangaea during the family is primarily a Holarctic group today. Jurassic. The fact that the family does not A place of origin still cannot be demonstrated now occur in tropical areas is somewhat per­ for Heptageniidae. However, the discovery plexing. of Colocrurinae in West Gondwana in ad­ The fossil mayflies found in Brazil are dis­ dition to a consideration of Chromarcyinae tinctly advanced from Paleozoic and Triassic in tropical Asia, in light of these groups' rel­ forms, and all, including Hexagenitidae, fall atively basal phyletic positions, adds some within the Schistonote grouping of modern plausibility to an origin of Heptageniidae in mayflies (McCafferty and Edmunds, 1979). the Southern Hemisphere. However, although they are characteristically The possible presence ofPotamanthidae in typical of modern schistonotes, several of the West Gondwana was unexpected. Potaman­ taxa have unique combinations of morpho­ thidae had not actually been known from the logical traits not found in extant forms. This fossil record previously, and today the family is especially profound in the Hexagenitidae is strictly Holarctic and Oriental in distri­ and Colocrurinae (Oligoneuriidae). bution. Unfortunately, the one alate fossil that Of the higher taxa found to have existed might be in Potamanthidae is too incomplete or possibly existed in West Gondwana during to allow a positive identification. Neverthe­ the Lower Cretaceous, Hexagenitidae and less, a legitimate question remains as to why Colocrurinae are now extinct, and Siphlonu­ Potamanthidae is not presently represented ridae and Potamanthidae are not now rep­ in the Afrotropical or Neotropical realms. resented by extant tax.a in tropical South The presence of two new genera ofEphem­ America. Also, the Ephemeridae of Lower eridae in West Gondwana was somewhat un­ Cretaceous Brazil do not appear to be directly expected because the family is poorly rep- related to the few extant Neotropical species f no McCAFFERTY:EPHEMEROPTERA .+7

of Ephemeridae. since the latter are thought was derived more recently than the Lower to be of Tertiary or possibly even Quarter­ Cretaceous, and much of its widespread dis­ nary North American origin (McCafferty et tribution may be due to dispersal rather than al., 1990). This leaves only the possible Lep· continental vicariance. Alternatively, how­ tophlebiidae and the Euthyplociidae repre­ ever, it may be that Mesozoic baetids simply senting a continuous existence of connected have yet to be found. Potentially giving cre­ lineages in the area since the Lower Creta­ dence to this alternative is some indication ceous, although the Colocrurinae also ap­ that certain undescribed baetids in South pears to represent an ancestral branch related America and Africa represent sister lineages, in origin to the modem oligoneuriines pres­ which theoretically could have resulted from ently found in tropical South America. continental drift in the Southern Hemi­ From the above, and as suggested earlier, sphere. it appears that many of these mayfly groups The presence of siphlonurid and hexag­ were more widespread during the Mesozoic. enitid larvae with swimming-adapted tails In addition, there apparently was significant strongly suggests that a quiet-freshwater hab­ extinction and emigration of mayfly lineages itat was present at the fossil site in the past. since the Lower Cretaceous in that area of This could have been a shallow lake envi­ the world that includes Brazil and the present ronment, littoral ponds, or possibly even pool Neotropics, with only a relatively few of those areas within streams, although the sheer lineages having possibly survived there to the numbers of Protoligoneuria larvae found present. would perhaps favor the suggestion of the Pannote mayflies are an apomorphic lacustrine environment. The presence of monophyletic grouping of the extant families Ephemeridae could also indicate these types , , Neoephem­ of depositional habitats. The presence of a eridae, , , and Prosopis­ stream habitat, in addition to a lentic habitat, tomatidae (McCafferty and Edmunds, 1979). is supported by the presence of the filter-feed­ No representatives of pannote mayflies were ing larva of Colocrus that presumably would found in Brazil as far as I could discern. It have required fl.owing water. Also, the pres­ could very well be that the pannote lineage ence of the alate forms ofOligoneuriidae, Eu­ did not originate until the Upper Cretaceous thyplociidae, and possibly Potamanthidae, or Tertiary. The placement ofcertain Jurassic the larvae of which are all obligate stream­ fossils in the Ephemerellidae by Demoulin dwelling forms, indicates that a stream en­ (1954) is highly doubtful based on the rem­ vironment was at least in the vicinity. nants available. Pannote fossils are not def­ The new data here modify the general initely known until the Eocene and Oligo­ paleoecological conclusions ofSinitshenkova cene. ( 1984) regarding the Mesozoic mayflies. Based The family Baetidae is a large schistonote on known Palearctic fossils, that author con­ group that is prevalent in the Neotropics as cluded that Jurassic and Lower Cretaceous well as almost every other area of the world mayflies were basically lacustrine. This may today. Together with the Leptophlebiidae, it be generally valid for Lower and Middle Ju­ demonstrates the most adaptive radiation rassic mayflies, but by the Lower Cretaceous, among modem mayflies. It is of interest to fl.owing-water mayflies, including filter feed­ me, however, that baetids were not repre­ ers, had definitely evolved, as evidenced by sented in the Brazilian find. Whereas lepto­ the Gondwanian fossils. The new Brazilian phlebiids are possibly as old as the Jurassic data show the evolution of mayflies, from and definitely as old as the Lower Cretaceous, still-water forms to predominantly fl.owing­ baetids are not known previous to the Eocene. water forms, as had been proposed by Ed­ I agree with Sinitshenkova (1985) that Me­ munds and McCafferty (1988), to have ac­ sobaeris, which is known from the Jurassic, tually begun much earlier than the Cenozoic. should be placed in the Siphlonuridae, not A collection of alate forms of mayflies, as the Baetidae as per Hubbard (1987). in the Brazilian find, would usually be indic­ The lack of an early fossil record could lead ative of a riparian environment adjacent to one to conjecture that the family Baetidae a freshwater habitat. However, most of the 48 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 195 alate mayfly fossils from Brazil, although di­ being washed up on bank areas or washed verse, appear to be females (at least male gen­ downstream and deposited in a lake. Such italia are not discernible), and it may there­ expired females of modern mayflies can be fore be primarily ovipositing females that had common along the shores of streams, ponds, expired on the water and were fossilized after and shallow embayments of lakes.

REFERENCES Brauer, F., J. Redtenbacher, and L. Ganglbauer 1971. Contribution a l'etude morphologique, 1889. Fossile insekten aus der Juraformation systematique, et phylogenique des Ost-Sibiriens. Mem. Acad. Imp. Sci. St. Ephemeropteres Jurassiques VI. L'aile Petersbourg 36(15): 1-20. posterieure des H exagenites Scudder et Brito, I. M. Jes rappons Hexagenitidae-Chromar­ J 987. Note preliminar sobre uma nova Efem­ cyidae-Oligoneuriidae. Bull. R. Inst. ero do Cretaceo do Ceara (Insecta Nat. Sci. Belgique 4 7(29): 2-10. Ephemeroptera). Ann. 10th Congr. Bras. 1974. Quatrieme contribution a la connais­ Paleontol. 2: 593-597. sance des Ephemeropteres de l'ambre Carpenter, F. M. Oligocene de la Baltique. Bull. R. Inst. 1987. Review of the extinct family Syntonop­ Nat. Sci. Belgique 50(3): 1-4. teridae (Order Uncenain). Psyche 94: Eaton, A. E. 373-388. 18 71. Monograph on the Ephemeridae. Trans. Cockerell, T. D. A. Entomol. Soc. London 19: 1-164. 1923. A new genus of mayflies from the Mio­ Edmunds, G. F., Jr. cene of Florissant, Colorado. Psyche 30: 1972. Biogeography and evolution of Ephem­ 170-172. eroptera. Annu. Rev. Entomol. 17: 21- Costa Lima, A. da 42. 19 50. Ninfa de Efemerideo fossil do Ceara. 197 5. Phylogenetic biogeography of mayflies. Ann. Acad. Bras. Ciencias 22: 419-420. Ann. Missouri Bot. Gard. 62: :?51-263. Crass, R. S. Edmunds, G. F., Jr., R. K. Allen, and W. L. Peters I 94 7. The may-flies (Ephemeroptera) of Natal 1963. An annotated key to the nymphs of the and Eastern Cape. Ann. Natal Mus. I I: families and subfamilies of mayflies 37-110. (Ephemeroptera). Univ. Utah Biol. Ser. Demoulin, G. 8(1): 1-49. I 954. Les Ephemeropteres Jurassiques du Edmunds. G. F., Jr., S. L. Jensen, and L. Berner Sinkiang. Bull. Ann. Entomol. Soc. Bel­ 1976. The mayflies of North and Central gique 90: 322-326. America. Minneapolis: Univ. Minn. I 9 55. Sur une larve siphlonuridienne d'E­ Press. phemere fossile du Bresil. Bull. Ann. R. Edmunds, G. F., Jr., and W. P. McCafferty Entomol. Soc. Belgique 91: 27 l. 1988. The mayfly subimago. Annu. Rev. 1958. Nouveau schema de classification des Entomol. 33: 509-529. Archodonates et des Ephemeropteres. Edmunds, G. F., Jr., and J. R. Traver Bull. R. Inst. Nat. Sci. Belgique 34(27): 19 54a. An outline of reclassification of the 1-19. Ephemeroptera. Proc. Entomol. Soc. 1968. Deuxieme contribution a la connais­ Washington 56: 236-240. sance des Ephemeropteres de l'ambre 1954b. The flight mechanics and evolution of Oligocene de la Baltique. Deutsche the wings ofEphemeroptera, with notes Entomol. Z. 15: 233-276. on the archetype insect wing. J. Wash­ l 970a. Contribution a l'etude morphologique, ington Acad. Sci. 44: 390-400. systematique, et phylogenique des Handlirsch, A. Ephemeropteres Jurassiques d'Europe 1908. Die fossilen Insekten und die phyloge­ centrale V. Hexagenitidae = Paede­ nie der rezenten formen. Leipzig. phemeridae (syn. nov.). Bull. R. Inst. Hubbard, M. D. Nat. Sci. Belgique 46(4): 1-8. 1987. Ephemeroptera. Fossilium catalogus, I: l 970b. Contribution a la connaissance des Anirnalia, pars 129. Amstelveen, The Ephemeropteres du Miocene 1. Siph­ Netherlands: Kugler Puhl. lurites explanatus Cockerell. Bull. R. Hubbard, M. D., and J. Kukalova-Peck Inst. Nat. Sci. Belgique 46(5): 1-4. 1980. Permian mayfly nymphs: new taxa and McCAFFERTY: EPHEMEROPTERA 49

systematic characters. In J. F. Aanna­ mayflies. Proc. Symp. Biogeog. Me­ f gan and K. E. Marshall (eds.), Advances soamerica. 1984. Tulane Univ. Biol. ! in Ephemeroptera biology, pp. 19-33. Ser., in press. I New York: Plenum Press. McCatferty, W. P .• and A. V. Provonsha Hubbard. M. D., and H. M. Savage 1988. Revisionary notes on predaceous Hep­ 198 l. The fossil Leptophlebiidae (Ephemer­ tageniidae based on larval and adult as­ I optera): a systematic and phylogenetic sociations (Ephemeroptera). Great Lakes review. J. Paleontol. 55: 810-813. Entomol. 21: 15-17. '. P. A., and P. M. Duncan McCatferty, W. P., and N. D. Sinitshenkova i 986. Invenebrates. mainly insects from the 1983. from the Oligocene in east­ freshwater, Lower Cretaceous, Koon­ ern Asia (Ephemeroptera: Ephemeri­ warra fossil bed (Korumburra Group), dae). Ann. Entomol. Soc. Am. 76: 205- South Gippsland, Victoria. Mem. As­ 208. soc. Australasian Paleontol. 3: l l l-205. McShatfrey, D., and W. P. McCatferty Keltner, J., and W. P. McCatfeny 1986. Feeding behavior of Stenacron inter­ 1986. Functional morphology of burrowing in punctatum (Ephemeroptera: Heptage­ the burrowing mayflies H exagenia lim­ niidae). J. North Am. Benthol. Soc. 5: bata and vittigera. Zool. J. 200-210. Linn. Soc. 87: 139-162. 1988. Feeding behavior of Rhithrogena pel­ :ukalovii-Peck, J. lucida (Ephemeroptera: Heptageni­ 1985. Ephemeroid wing venation based upon idae). J. NonhAm. Benthol. Soc. 7: 87- new gigantic Carboniferous mayflies and 99. basic morphology, phylogeny, and Meshkova, N. P. metamorphosis ofpterygote insects (In­ 1961. New data on Ephemeropsis trisetalis secta, Ephemerida). Can. J. Zoo!. 63: Eichwald larvae (Insekten, Ephemer­ 933-955. optera). Paleontol. Zh. 1961(4): 164- Landa. V. 168. [in Russian] 1973. A contribution to the order Ephemer­ Needham, J. G .. and H. E. Murphy optera based on comparative anatomy. 1924. Neotropical mayflies. Bull. Lloyd Libr. Proc. lst Internat. Conf. Ephem., Tal­ 24, Entomol. Ser. 4: 1-79. lahassee, Aa.. 1970: 155-159. Riek, E. F. Landa. V., and T. Soldan 1970. Fossil history. In CSIRO, The insects 19 8 5. Phylogeny and higher classification of of Australia, 168-186. Melbourne: Mel­ the order Ephemeroptera: a discussion bourne Univ. Press. from the comparative anatomical point 1973. Classification of the Ephemeroptera. of view. Prague: Ceskoslav. Akad. Ved. Proc. !st Intemat. Conf. Ephem., Tal­ Le\vis. S. E. lahassee, Aa., 1970: 160-178. 1977. Two new species of fossil mayflies Rossi de Garcia, E. (Ephemeroptera: and 1983. Insectos fosiles en la formacion Ventana Siphlonuridae) from the Ruby River (Eoceno). Provincia de Neuquen. Riv. Basin (Oligocene) of southwestern Assoc. Geo!. Argentina 38: 17-23. Montana. Proc. Entomol. Soc. Wash­ Sinitshenkova, N. D. ington 79: 583-587. 1975. Mayfly larvae of the family Hexageni­ McCatferty, W. P. tidae. Paleontol. Zh. 1975(1): 82-87. [in 1979. Evolutionary trends among the families Russian] of Ephemeroidea. Proc. 2nd Internat. 1976. New early mayflies (Insecta: Ephemer­ Conf. Ephem., Krakow, Poland, 1975: optera) from eastern Transbaykalia. Pa­ 45-50. leontol. Zh. 1976(4): 85-93. [in Rus­ 1987. New fossil mayfly in amber and its re­ sian] lationship among extant Ephemeridae 1984. The Mesozoic mayflies (Ephemerop­ (Ephemeroptera). Ann. Entomol. Soc. tera) with special reference to their ecol­ Am. 80: 472-474. ogy. Proc. 4th Internat. Conf. Ephem., McCatferty, W. P., and G. F. Edmunds, Jr. Prague, 1983: 61-66. 1979. The higher classification of the Ephem­ 1985. Jurassic mayflies (Ephemerida-Ephem­ eroptera and its evolutionary basis. Ann. eroptera) of southern Siberia and west­ Entomol. Soc. Am. 72: 5-12. ern Mongolia. Tr. Paleontol. Inst. (Mos­ McCafferty, W. P., R. W. Aowers, and R. D. Waltz cow) 211: 11-23. [in Russian] 1990. The biogeography of Mesoamerican 1986. Mayflies, Ephemerida (= Ephemerop- 50 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 195

tera). In Insects in the early Cretaceous 1977. Unusual new larval mayflies (Ephem­ ecosystems of the West Mongolia. 45- eroptera: Palingeniidae, Behningiidae) 46. Trans. Joint Soviet-Mongolian Pa­ from the Jura Mountain area of the leontol. Exped. 28 (Moscow) [in Rus­ Transbayk.al. Paleontol. Zh. 1977(2): 91- sian] 96. [in Russian] Tshemova, 0. A. 1980. Order Ephemerida Latreille, 1810. Tr. 196 7. Mayflies of a recent family in Jurassic Paleontol. Inst. 175: 31-36. [in Russian] deposits of Transbayk.alia (Ephemer­ Tshemova, 0. A., and N. D. Sinitshenkova optera, Siphlonuridae). Entomol. Obozr. 1974. New fossil genus and species of the fam­ 46: 322-326. [in Russian] ily Hexagenitidae (Ephemeroptera) from 1969. New early Jurassic mayflies (Ephemer­ the southern European part ofthe USSR optera, Epeoromimidae and Mesoneti­ and its connection with recent Ephem­ dae). Entomol. Obozr. 48: 153-161. [in eroptera. Entomol. Obozr. 54: 130-136. Russian] [in Russian] 1970. On the classification of the fossil and Wallace, J.B., and J. O'Hop recent Ephemeroptera. Entomol. Obozr. 1979. Fine particle suspension feeding capa­ 49: 124-145. [in Russian] bilities of lsonychia spp. (Ephemerop­ l 971. A mayfly (Ephemeroptera, Leptophle­ tera: Siphlonuridae). Ann. Entomol. Soc. biidae) from fossil resin of Cretaceous Am. 72: 353-357. deposits in the polar regions of Siberia. Weyenbergh, H. Entomol. Obozr. 50: 612-618. [in Rus­ 1874. Varia zoologica et paleontologica. Pe­ sian] riod. Zoo!. l: 77-109.